study of charm production from the measurement …...目次 1．主論文 study of charm production...

学位論文

Study of Charm Production from the Measurement of

Single Electrons in Au + Au Collisions at

NNs = 200 GeV

(核子対あたりの重心系衝突エネルギー200 GeVの金＋金原子核衝突における単電子測定によるチャーム生成の研究)

２０１０年１月

広島大学大学院理学研究科物理科学専攻

蜂谷崇

目次１．主論文

Study of Charm Production from the Measurement of Single Electrons in Au+Au Collisions at = 200 GeV (核子対あたりの重心系衝突エネルギー200GeV の金＋金原子核衝突における単電子測定によるチャーム生成の研究)

蜂谷崇２．公表論文

(1) Centrality Dependence of Charm Production from a Measurement of Single Electrons in Au+Au Collisions at =200 GeV

S. S. Adler et al., (別紙、共著者リスト①参照) Physical Review Letters 94, 082301 (2005).

３．参考論文

(1) Measurement of Single Electrons and Implications for Charm Production in Au+Au Collisions at = 130 GeV

K. Adcox et al., (別紙、共著者リスト②参照) Physical Review Letters 88, 192303 (2002).

(2) Measurement of Bottom Versus Charm as a Function of Transverse Momentum

with Electron-Hadron Correlations in p + p Collisions at = 200 GeV A. Adare et al., (別紙、共著者リスト③参照) Physical Review Letters 103, 082002 (2009).

(3) PHENIX inner detectors

M. Allen et al., (別紙, 共著者リスト④参照) Nuclear Instruments and Methods in Physics Research A 499, 549 (2003).

NNs

NNs

NNs

s

共著者リスト① S. S. Adler, S. Afanasiev, C. Aidala, N. N. Ajitanand, Y. Akiba, J. Alexander, R. Amirikas, L. Aphecetche, S. H. Aronson, R. Averbeck,

T. C. Awes, R. Azmoun, V. Babintsev, A. Baldisseri, K. N. Barish, P. D. Barnes, B. Bassalleck, S. Bathe, S. Batsouli, V. Baublis,

A. Bazilevsky, S. Belikov, Y. Berdnikov, S. Bhagavatula, J. G. Boissevain, H. Borel, S. Borenstein, M. L. Brooks, D. S. Brown,

N. Bruner, D. Bucher, H. Buesching, V. Bumazhnov, G. Bunce, J. M. Burward-Hoy,4 S. Butsyk, X. Camard, J.-S. Chai, P. Chand,

W. C. Chang, S. Chernichenko, C. Y. Chi, J. Chiba, M. Chiu, I. J. Choi, J. Choi, R. K. Choudhury, T. Chujo, V. Cianciolo, Y. Cobigo,

B. A. Cole, P. Constantin, D. G. d'Enterria, G. David, H. Delagrange, A. Denisov, A. Deshpande, E. J. Desmond, A. Devismes,

O. Dietzsch, O. Drapier, A. Drees, R. du Rietz, A. Durum, D. Dutta, Y. V. Efremenko, K. El Chenawi, A. Enokizono, H. En'yo, S. Esumi,

L. Ewell, D. E. Fields, F. Fleuret, S. L. Fokin, B. D. Fox, Z. Fraenkel, J. E. Frantz, A. Franz, A. D. Frawley, S.-Y. Fung, S. Garpman,

T. K. Ghosh, A. Glenn, G. Gogiberidze, M. Gonin, J. Gosset, Y. Goto, R. Granier de Cassagnac, N. Grau, S. V. Greene,

M. Grosse Perdekamp, W. Guryn, H.-A° . Gustafsson, T. Hachiya, J. S. Haggerty, H. Hamagaki, A. G. Hansen, E. P. Hartouni,

M. Harvey, R. Hayano, N. Hayashi, X. He, M. Heffner, T. K. Hemmick, J. M. Heuser, M. Hibino, J. C. Hill, W. Holzmann, K. Homma,

B. Hong, A. Hoover, T. Ichihara, V.V. Ikonnikov, K. Imai, D. Isenhower, M. Ishihara, M. Issah, A. Isupov, B.V. Jacak, W.Y. Jang,

Y. Jeong, J. Jia, O. Jinnouchi, B. M. Johnson, S. C. Johnson, K. S. Joo, D. Jouan, S. Kametani, N. Kamihara, J. H. Kang,

S. S. Kapoor, K. Katou, S. Kelly, B. Khachaturov, A. Khanzadeev, J. Kikuchi, D. H. Kim, D. J. Kim, D.W. Kim, E. Kim, G.-B. Kim,

H. J. Kim, E. Kistenev, A. Kiyomichi, K. Kiyoyama, C. Klein-Boesing, H. Kobayashi, L. Kochenda, V. Kochetkov, D. Koehler,

T. Kohama, M. Kopytine, D. Kotchetkov, A. Kozlov, P. J. Kroon, C. H. Kuberg, K. Kurita, Y. Kuroki, M. J. Kweon, Y. Kwon, G. S. Kyle,

R. Lacey, V. Ladygin, J. G. Lajoie, A. Lebedev, S. Leckey, D. M. Lee, S. Lee, M. J. Leitch, X. H. Li, H. Lim, A. Litvinenko, M. X. Liu,

Y. Liu, C. F. Maguire, Y. I. Makdisi, A. Malakhov, V. I. Manko, Y. Mao, G. Martinez, M. D. Marx, H. Masui, F. Matathias,

T. Matsumoto, P. L. McGaughey, E. Melnikov, F. Messer, Y. Miake, J. Milan, T. E. Miller, A. Milov, S. Mioduszewski, R. E. Mischke,

G. C. Mishra, J. T. Mitchell, A. K. Mohanty, D. P. Morrison, J. M. Moss, F. Mu¨hlbacher, D. Mukhopadhyay, M. Muniruzzaman,

J. Murata, S. Nagamiya, J. L. Nagle, T. Nakamura, B. K. Nandi, M. Nara, J. Newby, P. Nilsson, A. S. Nyanin, J. Nystrand, E. O’Brien,

C. A. Ogilvie, H. Ohnishi, I. D. Ojha, K. Okada, M. Ono, V. Onuchin, A. Oskarsson, I. Otterlund, K. Oyama, K. Ozawa, D. Pal,

A. P. T. Palounek, V. S. Pantuev, V. Papavassiliou, J. Park, A. Parmar, S. F. Pate, T. Peitzmann, J.-C. Peng, V. Peresedov,

C. Pinkenburg, R. P. Pisani, F. Plasil, M. L. Purschke, A. K. Purwar, J. Rak, I. Ravinovich, K. F. Read, M. Reuter, K. Reygers,

V. Riabov, Y. Riabov, G. Roche, A. Romana, M. Rosati, P. Rosnet, S. S. Ryu, M. E. Sadler, N. Saito, T. Sakaguchi, M. Sakai, S. Sakai,

V. Samsonov, L. Sanfratello, R. Santo, H. D. Sato, S. Sato, S. Sawada, Y. Schutz, V. Semenov, R. Seto, M. R. Shaw, T. K. Shea,

T.-A. Shibata, K. Shigaki, T. Shiina, C. L. Silva, D. Silvermyr, K. S. Sim, C. P. Singh, V. Singh, M. Sivertz, A. Soldatov, R. A. Soltz,

W. E. Sondheim, S. P. Sorensen, I.V. Sourikova, F. Staley, P.W. Stankus, E. Stenlund, M. Stepanov, A. Ster, S. P. Stoll, T. Sugitate,

J. P. Sullivan, E. M. Takagui, A. Taketani, M. Tamai, K. H. Tanaka, Y. Tanaka, K. Tanida, M. J. Tannenbaum, P. Tarja´n, J. D. Tepe,

T. L. Thomas, J. Tojo, H. Torii, R. S. Towell, I. Tserruya, H. Tsuruoka, S. K. Tuli, H. Tydesjo¨, N. Tyurin, H.W. van Hecke, J. Velkovska,

M. Velkovsky, V. Veszpre´mi, L. Villatte, A. A. Vinogradov, M. A. Volkov, E. Vznuzdaev, X. R. Wang, Y. Watanabe, S. N. White,

F. K. Wohn, C. L. Woody, W. Xie, Y. Yang, A. Yanovich, S. Yokkaichi, G. R. Young, I. E. Yushmanov, W. A. Zajc, C. Zhang, S. Zhou,

S. J. Zhou, and L. Zolin

共著者リスト②

K. Adcox, S. S. Adler, N. N. Ajitanand, Y. Akiba, J. Alexander, L. Aphecetche, Y. Arai, S. H. Aronson, R. Averbeck, T. C. Awes,

K. N. Barish, P. D. Barnes, J. Barrette, B. Bassalleck, S. Bathe, V. Baublis, A. Bazilevsky, S. Belikov, F. G. Bellaiche, S. T. Belyaev,

M. J. Bennett, Y. Berdnikov, S. Botelho, M. L. Brooks, D. S. Brown, N. Bruner, D. Bucher, H. Buesching, V. Bumazhnov, G. Bunce,

J. Burward-Hoy, S. Butsyk, T. A. Carey, P. Chand, J. Chang, W. C. Chang, L. L. Chavez, S. Chernichenko, C.Y. Chi, J. Chiba, M. Chiu,

R. K. Choudhury, T. Christ, T. Chujo, M. S. Chung, P. Chung, V. Cianciolo, B. A. Cole, D. G. D’Enterria, G. David, H. Delagrange,

A. Denisov, A. Deshpande, E. J. Desmond, O. Dietzsch, B.V. Dinesh, A. Drees, A. Durum, D. Dutta, K. Ebisu, Y.V. Efremenko,

K. El Chenawi, H. En'yo, S. Esumi, L. Ewell, T. Ferdousi, D. E. Fields, S. L. Fokin, Z. Fraenkel, A. Franz, A. D. Frawley, S.-Y. Fung,

S. Garpman, T.K. Ghosh, A. Glenn, A. L. Godoi, Y. Goto, S.V. Greene, M. Grosse Perdekamp, S. K. Gupta, W. Guryn, H.-A. Gustafsson,

T. Hachiya, J. S. Haggerty, H. Hamagaki, A. G. Hansen, H. Hara, E. P. Hartouni, R. Hayano, N. Hayashi, X. He, T. K. Hemmick,

J.M. Heuser, M. Hibino, J. C. Hill, D. S. Ho, K. Homma, B. Hong, A. Hoover, T. Ichihara, K. Imai, M. S. Ippolitov, M. Ishihara,

B.V. Jacak, W.Y. Jang, J. Jia, B. M. Johnson, S. C. Johnson, K. S. Joo, S. Kametani, J. H. Kang, M. Kann, S. S. Kapoor, S. Kelly,

B. Khachaturov, A. Khanzadeev, J. Kikuchi, D. J. Kim, H. J. Kim, S.Y. Kim, Y. G. Kim, W.W. Kinnison, E. Kistenev, A. Kiyomichi,

C. Klein-Boesing, S. Klinksiek, L. Kochenda, V. Kochetkov, D. Koehler, T. Kohama, D. Kotchetkov, A. Kozlov, P. J. Kroon, K. Kurita,

M. J. Kweon, Y. Kwon, G. S. Kyle, R. Lacey, J. G. Lajoie, J. Lauret, A. Lebedev, D. M. Lee, M. J. Leitch, X. H. Li, Z. Li, D. J. Lim,

M. X. Liu, X. Liu, Z. Liu, C. F. Maguire, J. Mahon, Y. I. Makdisi, V. I. Manko, Y. Mao, S. K. Mark, S. Markacs, G. Martinez,

M. D. Marx, A. Masaike, F. Matathias, T. Matsumoto, P. L. McGaughey, E. Melnikov, M. Merschmeyer, F. Messer, M. Messer,

Y. Miake, T. E. Miller, A. Milov, S. Mioduszewski, R. E. Mischke, G. C. Mishra, J.T. Mitchell, A. K. Mohanty, D. P. Morrison, J. M.Moss,

F. Muhlbacher, M. Muniruzzaman, J. Murata, S. Nagamiya, Y. Nagasaka, J. L. Nagle, Y. Nakada, B. K. Nandi, J. Newby, L. Nikkinen,

P. Nilsson, S. Nishimura, A. S. Nyanin, J. Nystrand, E. O'Brien, C. A. Ogilvie, H. Ohnishi, I. D. Ojha, M. Ono, V. Onuchin,

A. Oskarsson, L. Osterman, I. Otterlund, K. Oyama, L. Paffrath, A. P. T. Palounek, V. S. Pantuev, V. Papavassiliou, S. F. Pate,

T. Peitzmann, A. N. Petridis, C. Pinkenburg, R. P. Pisani, P. Pitukhin, F. Plasil, M. Pollack, K. Pope, M. L. Purschke, I. Ravinovich,

K. F. Read, K. Reygers, V. Riabov, Y. Riabov, M. Rosati, A. A. Rose, S. S. Ryu, N. Saito, A. Sakaguchi, T. Sakaguchi, H. Sako,

T. Sakuma, V. Samsonov, T. C. Sangster, R. Santo, H. D. Sato, S. Sato, S. Sawada, B. R. Schlei, Y. Schutz, V. Semenov, R. Seto,

T. K. Shea, I. Shein, T.-A. Shibata, K. Shigaki, T. Shiina, Y. H. Shin, I. G. Sibiriak, D. Silvermyr, K. S. Sim, J. Simon-Gillo,

C. P. Singh, V. Singh, M. Sivertz, A. Soldatov, R. A. Soltz, S. Sorensen, P.W. Stankus, N. Starinsky, P. Steinberg, E. Stenlund, A. Ster,

S. P. Stoll,M. Sugioka, T. Sugitate, J. P. Sullivan, Y. Sumi, Z. Sun, M. Suzuki, E. M. Takagui, A. Taketani, M. Tamai, K. H. Tanaka,

Y. Tanaka, E. Taniguchi, M. J. Tannenbaum, J. Thomas, J. H. Thomas, T. L. Thomas, W. Tian, J. Tojo, H. Torii, R. S. Towell,

I. Tserruya, H. Tsuruoka, A. A. Tsvetkov, S. K. Tuli, H. Tydesjo, N. Tyurin, T. Ushiroda, H.W. van Hecke, C. Velissaris, J. Velkovska,

M. Velkovsky, A. A. Vinogradov, M. A. Volkov, A. Vorobyov, E. Vznuzdaev, H. Wang, Y. Watanabe, S. N. White,C. Witzig, F. K. Wohn,

C. L. Woody, W. Xie, K. Yagi, S. Yokkaichi, G. R. Young, I. E. Yushmanov, W. A. Zajc, Z. Zhang, and S. Zhou

共著者リスト③

A. Adare, S. Afanasiev, C. Aidala, N. N. Ajitanand, Y. Akiba, H. Al-Bataineh, J. Alexander, K. Aoki, L. Aphecetche, R. Armendariz,

S. H. Aronson,J. Asai, E. T. Atomssa, R. Averbeck, T. C. Awes, B. Azmoun, V. Babintsev, M. Bai, G. Baksay, L. Baksay, A. Baldisseri,

K. N. Barish, P. D. Barnes, B. Bassalleck, A. T. Basye, S. Bathe, S. Batsouli, V. Baublis, C. Baumann, A. Bazilevsky, S. Belikov,

R. Bennett, A. Berdnikov, Y. Berdnikov, A. A. Bickley, J. G. Boissevain, H. Borel, K. Boyle, M. L. Brooks, H. Buesching, V. Bumazhnov,

G. Bunce, S. Butsyk, C. M. Camacho, S. Campbell, B. S. Chang, W. C. Chang, J.-L. Charvet, S. Chernichenko, J. Chiba, C.Y. Chi,

M. Chiu, I. J. Choi, R. K. Choudhury, T. Chujo, P. Chung, A. Churyn, V. Cianciolo, Z. Citron, C. R. Cleven, B. A. Cole, M. P. Comets,

P. Constantin, M. Csana´d, T. Cso¨rgo? , T. Dahms, S. Dairaku, K. Das, G. David, M. B. Deaton, K. Dehmelt, H. Delagrange,

A. Denisov, D. d'Enterria, A. Deshpande, E. J. Desmond, O. Dietzsch, A. Dion, M. Donadelli, O. Drapier, A. Drees, K. A. Drees,

A. K. Dubey, A. Durum, D. Dutta, V. Dzhordzhadze, Y.V. Efremenko, J. Egdemir, F. Ellinghaus, W. S. Emam, T. Engelmore,

A. Enokizono, H. En'yo, S. Esumi, K. O. Eyser, B. Fadem, D. E. Fields, M. Finger, Jr., M. Finger, F. Fleuret, S. L. Fokin, Z. Fraenkel,

J. E. Frantz, A. Franz, A. D. Frawley, K. Fujiwara, Y. Fukao, T. Fusayasu, S. Gadrat, I. Garishvili, A. Glenn, H. Gong, M. Gonin,

J. Gosset, Y. Goto, R. Granier de Cassagnac, N. Grau, S.V. Greene, M. Grosse Perdekamp, T. Gunji, H.-A° . Gustafsson, T. Hachiya,

A. Hadj Henni, C. Haegemann, J. S. Haggerty, H. Hamagaki, R. Han, H. Harada, E. P. Hartouni, K. Haruna, E. Haslum, R. Hayano,

M. Heffner, T. K. Hemmick, T. Hester, X. He, H. Hiejima, J. C. Hill, R. Hobbs, M. Hohlmann, W. Holzmann, K. Homma, B. Hong,

T. Horaguchi, D. Hornback, S. Huang, T. Ichihara, R. Ichimiya, Y. Ikeda, K. Imai, J. Imrek, M. Inaba, Y. Inoue, D. Isenhower,

L. Isenhower, M. Ishihara, T. Isobe, M. Issah, A. Isupov, D. Ivanischev, B.V. Jacak, J. Jia, J. Jin, O. Jinnouchi, B. M. Johnson, K. S.Joo,

D. Jouan, F. Kajihara, S. Kametani, N. Kamihara, J. Kamin, M. Kaneta, J. H. Kang, H. Kanou, J. Kapustinsky, D. Kawall,

A.V. Kazantsev, T. Kempel, A. Khanzadeev, K. M. Kijima, J. Kikuchi, B. I. Kim, D. H. Kim, D. J. Kim, E. Kim, S. H. Kim, E. Kinney,

K. Kiriluk, A. Kiss, E. Kistenev, A. Kiyomichi, J. Klay, C. Klein-Boesing, L. Kochenda, V. Kochetkov, B. Komkov, M. Konno, J. Koster,

D. Kotchetkov, A. Kozlov, A. Kra´l, A. Kravitz, J. Kubart, G. J. Kunde, N. Kurihara, K. Kurita, M. Kurosawa, M. J. Kweon, Y. Kwon,

G. S. Kyle, R. Lacey, Y.-S. Lai, Y. S. Lai, J. G. Lajoie, D. Layton, A. Lebedev, D. M. Lee, K. B. Lee, M. K. Lee, T. Lee, M. J. Leitch,

M. A. L. Leite, B. Lenzi, P. Liebing, T. Lis?ka, A. Litvinenko, H. Liu, M. X. Liu, X. Li, B. Love, D. Lynch, C. F. Maguire, Y. I. Makdisi,

A. Malakhov, M. D. Malik, V. I. Manko, E. Mannel, Y. Mao, L. Mas?ek, H. Masui, F. Matathias, M. McCumber, P. L. McGaughey,

N. Means, B. Meredith, Y. Miake, P. Mikes?, K. Miki, T. E. Miller, A. Milov, S. Mioduszewski, M. Mishra, J. T. Mitchell, M. Mitrovski,

A. K. Mohanty, Y. Morino, A. Morreale, D. P. Morrison, T.V. Moukhanova, D. Mukhopadhyay, J. Murata, S. Nagamiya, Y. Nagata,

J. L. Nagle, M. Naglis, M. I. Nagy, I. Nakagawa, Y. Nakamiya, T. Nakamura, K. Nakano, J. Newby, M. Nguyen, T. Niita, B. E.Norman,

R. Nouicer, A. S. Nyanin, E. O'Brien, S. X. Oda, C. A. Ogilvie, H. Ohnishi, H. Okada, K. Okada, M. Oka, O. O. Omiwade, Y. Onuki,

A. Oskarsson, M. Ouchida, K. Ozawa, R. Pak, D. Pal, A. P. T. Palounek, V. Pantuev, V. Papavassiliou2 J. Park, W. J. Park, S. F. Pate,

H. Pei, J.-C. Peng, H. Pereira, V. Peresedov, D.Yu. Peressounko, C. Pinkenburg, M. L. Purschke, A. K. Purwar, H. Qu, J. Rak,

A. Rakotozafindrabe, I. Ravinovich, K. F. Read, S. Rembeczki, M. Reuter, K. Reygers, V. Riabov, Y. Riabov, D. Roach, G. Roche,

S. D. Rolnick, A. Romana, M. Rosati, S. S. E. Rosendahl, P. Rosnet, P. Rukoyatkin, P. Ruz?ic?ka, V. L. Rykov, B. Sahlmueller,

N. Saito, T. Sakaguchi, S. Sakai, K. Sakashita, H. Sakata, V. Samsonov, S. Sato, T. Sato, S. Sawada, K. Sedgwick, J. Seele,

R. Seidl, A.Yu. Semenov, V. Semenov, R. Seto, D. Sharma, I. Shein, A. Shevel,3 T.-A. Shibata, K. Shigaki, M. Shimomura, K. Shoji,

P. Shukla, A. Sickles, C. L. Silva, D. Silvermyr, C. Silvestre, K. S. Sim, B. K. Singh, C. P. Singh, V. Singh, S. Skutnik, M. Slunec?ka,

A. Soldatov, R. A. Soltz, W. E. Sondheim, S. P. Sorensen, I.V. Sourikova, F. Staley, P.W. Stankus, E. Stenlund, M. Stepanov, A. Ster,

S. P. Stoll, T. Sugitate, C. Suire, A. Sukhanov, J. Sziklai, T. Tabaru, S. Takagi, E. M. Takagui, A. Taketani, R. Tanabe, Y. Tanaka,

K. Tanida, M. J. Tannenbaum, A. Taranenko, P. Tarja´n, H. Themann, T. L. Thomas, M. Togawa, A. Toia, J. Tojo, L. Toma´s?ek,

Y. Tomita, H. Torii, R. S. Towell, V-N. Tram, I. Tserruya, Y. Tsuchimoto, C. Vale, H. Valle, H.W. van Hecke, A. Veicht, J. Velkovska,

R. Vertesi, A. A. Vinogradov, M. Virius, V. Vrba, E. Vznuzdaev, M. Wagner, D. Walker, X. R. Wang, Y. Watanabe, F. Wei, J. Wessels,

S. N. White, D. Winter, C. L. Woody, M. Wysocki, W. Xie, Y. L. Yamaguchi, K. Yamaura, R. Yang, A. Yanovich, Z. Yasin, J. Ying,

S. Yokkaichi, G. R. Young, I. Younus, I. E. Yushmanov, W. A. Zajc, O. Zaudtke, C. Zhang, S. Zhou, J. Zima´nyi, and L. Zolin

共著者リスト④ M. Allen, M.J. Bennett, M. Bobrek, J.B. Boissevain, S. Boose, E. Bosze, C. Britton, J. Chang, C.Y. Chi, M. Chiu, R. Conway,

R. Cunningham, A. Denisov, A. Deshpande, M.S. Emery, A. Enokizono, N. Ericson, B. Fox, S.-Y. Fung, P. Giannotti, T. Hachiya,

A.G. Hansen, K. Homma, B.V. Jacak, D. Jaffe, J.H. Kang, J. Kapustinsky, S.Y. Kim, Y.G. Kim, T. Kohama, P.J. Kroon, W. Lenz,

N. Longbotham, M. Musrock, T. Nakamura, H. Ohnishi, S.S. Ryu, A. Sakaguchi, R. Seto, T. Shiina, M. Simpson, J. Simon-Gillo,

W.E. Sondheim, T. Sugitate, J.P. Sullivan, H.W. van Hecke, J.W. Walker, S.N. White, P. Willis, N. Xu

主論文

Study of Charm Production from the Measurement ofSingle Electrons in Au + Au Collisions at√

sNN = 200 GeV

Takashi Hachiya

Department of Physical Science, Graduate School of ScienceHiroshima University

Kagamiyama 1-3-1, Higashi-Hiroshima, 739-8526, Japan

January 2010

Abstract

Quark Gluon Plasma (QGP) is a new state of matter in which constituents of nucleon(quarks and gluons) are deconfined. High energy heavy ion collision is a unique tool tocreate and study QGP. Two important findings, jet suppression and strong elliptic flow,have been reported from the experiments at RHIC. They are considered to be evidence ofQGP formation in high energy heavy ion collisions.

Heavy quarks (charms and bottoms) are clean probes to study characteristics of QGPcreated in heavy ion collisions. Because of their large masses, heavy quarks are onlyproduced in the initial stage of heavy ion collisions. Therefore, heavy quark production issensitive to the initial gluon density. Moreover, the measurement of heavy quarks providean important baseline of quarkonium suppression which is a proposed signature of QGPformation in heavy ion collisions.

The PHENIX experiment has measured transverse momentum spectra (0.4 < pT <4.0 GeV/c) of single electrons from the semi-leptonic decays of heavy flavors as a functionof centrality in Au + Au collisions at

√sNN = 200 GeV. Signal electrons from heavy flavor

decays are statistically extracted from large amount of photonic electron background whichcomes from Dalitz decays of light neutral mesons and photon conversions. A thin (1.7% ofradiation length) photon converter has been installed to the PHENIX detector acceptanceduring a part of the data taking period. The converter increases only photonic backgroundelectrons by a fixed factor. Thus, by comparing the data sets with and without theconverter, the photonic component and the signal component are accurately separated.After subtracting a small background of Ke3 and vector mesons decays, the remainingcontribution of electrons is the semi-leptonic decays of heavy flavors, mainly charms atlow pT.

The centrality dependence of the total charm production has been studied. For allcentralities, we found that the total charm production in Au + Au collisions is scaled bythe number of nucleon-nucleon collisions. This indicates that most of charm quarks areproduced in the initial stage of space time evolution in high energy heavy ion collisions.For minimum bias Au + Au collisions, the total charm cross section per binary collision isσcc = 622 ± 57(stat) ± 160(sys) µb. The measured total charm cross section is consistentwith a next-to-leading-order perturbative quantum chromodynamics calculation withinthe theoretical and the experimental uncertainties.

Contents

1 Introduction 11.1 Quark Gluon Plasma . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.2 Thesis Motivation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7

2 Theoretical and Experimental Overview 92.1 Heavy Quark Measurement . . . . . . . . . . . . . . . . . . . . . . . . . . 92.2 Heavy Quark Production . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10

2.2.1 Heavy Quark in pQCD . . . . . . . . . . . . . . . . . . . . . . . . . 102.2.2 Fragmentation Functions . . . . . . . . . . . . . . . . . . . . . . . . 132.2.3 Semi-Leptonic Decays . . . . . . . . . . . . . . . . . . . . . . . . . 16

2.3 High Energy Heavy Ion Collisions . . . . . . . . . . . . . . . . . . . . . . . 182.3.1 Space Time Evolution . . . . . . . . . . . . . . . . . . . . . . . . . 182.3.2 Energy Density . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 202.3.3 Collision Geometry . . . . . . . . . . . . . . . . . . . . . . . . . . . 21

2.4 Heavy Quarks in Nucleus-Nucleus Collisions . . . . . . . . . . . . . . . . . 232.4.1 Initial Production . . . . . . . . . . . . . . . . . . . . . . . . . . . . 232.4.2 Pre-equilibrium Production . . . . . . . . . . . . . . . . . . . . . . 232.4.3 Cold Nuclear Effects . . . . . . . . . . . . . . . . . . . . . . . . . . 232.4.4 Final State Effects in QGP . . . . . . . . . . . . . . . . . . . . . . . 262.4.5 Aim of This Study . . . . . . . . . . . . . . . . . . . . . . . . . . . 27

3 Experimental Setup 303.1 Relativistic Heavy Ion Collider . . . . . . . . . . . . . . . . . . . . . . . . . 303.2 PHENIX Detector Overview . . . . . . . . . . . . . . . . . . . . . . . . . . 303.3 Global Detector . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33

3.3.1 Beam Beam Counter . . . . . . . . . . . . . . . . . . . . . . . . . . 333.3.2 Zero Degree Calorimeter . . . . . . . . . . . . . . . . . . . . . . . . 353.3.3 Multiplicity Vertex Detector . . . . . . . . . . . . . . . . . . . . . . 37

3.4 Central Arm Spectrometer . . . . . . . . . . . . . . . . . . . . . . . . . . . 393.4.1 Central Magnet . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 393.4.2 Drift Chambers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 393.4.3 Pad Chambers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 413.4.4 Ring Imaging Cerenkov Detector . . . . . . . . . . . . . . . . . . . 423.4.5 Time Of Flight Counters . . . . . . . . . . . . . . . . . . . . . . . . 433.4.6 Electromagnetic Calorimeter . . . . . . . . . . . . . . . . . . . . . . 44

I

II CONTENTS

3.5 Data Acquisition System . . . . . . . . . . . . . . . . . . . . . . . . . . . . 463.6 Photon Converter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50

4 Analysis 514.1 Outline . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 514.2 Event Selection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51

4.2.1 Minimum Bias Trigger . . . . . . . . . . . . . . . . . . . . . . . . . 524.2.2 Centrality Determination . . . . . . . . . . . . . . . . . . . . . . . . 524.2.3 Glauber Calculation . . . . . . . . . . . . . . . . . . . . . . . . . . 54

4.3 Charged Particle Tracking . . . . . . . . . . . . . . . . . . . . . . . . . . . 554.3.1 Matching to the Outer Detector . . . . . . . . . . . . . . . . . . . . 59

4.4 Electron Identification . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 594.4.1 Electron ID with RICH . . . . . . . . . . . . . . . . . . . . . . . . . 594.4.2 Electron ID with EMCal . . . . . . . . . . . . . . . . . . . . . . . . 604.4.3 Summary of Electron ID Conditions . . . . . . . . . . . . . . . . . 614.4.4 Fiducial Cut . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63

4.5 Hadron Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 664.6 Run Selection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 674.7 Raw Yield of Inclusive Electrons . . . . . . . . . . . . . . . . . . . . . . . . 704.8 Detector Response in Simulation . . . . . . . . . . . . . . . . . . . . . . . 70

4.8.1 Comparison between Real Data and Simulation . . . . . . . . . . . 704.8.2 Material Budget in Simulation . . . . . . . . . . . . . . . . . . . . . 72

4.9 Correction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 774.9.1 Acceptance Correction . . . . . . . . . . . . . . . . . . . . . . . . . 774.9.2 Multiplicity Dependent Efficiency Loss . . . . . . . . . . . . . . . . 81

4.10 Heavy Flavor Electrons . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 824.10.1 Converter Method . . . . . . . . . . . . . . . . . . . . . . . . . . . 824.10.2 Simulation Study . . . . . . . . . . . . . . . . . . . . . . . . . . . . 844.10.3 Determination of Rγ . . . . . . . . . . . . . . . . . . . . . . . . . . 864.10.4 Blocking Effect . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 884.10.5 Extraction of Non-photonic Electrons . . . . . . . . . . . . . . . . . 914.10.6 Background from Ke3 decays and vector meson decays . . . . . . . 93

4.11 Bin Width Effect . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 974.12 Systematic Uncertainties . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97

4.12.1 Systematic Uncertainty of Inclusive Electrons . . . . . . . . . . . . 984.12.2 Systematic Uncertainty of Electrons from Ke3 and Vector Meson

Decays . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 984.12.3 Uncertainty of The Converter Method . . . . . . . . . . . . . . . . 101

5 Results and Discussions 1055.1 Invariant Yield of Heavy Flavor Electrons . . . . . . . . . . . . . . . . . . 1055.2 Comparison with p+ p result . . . . . . . . . . . . . . . . . . . . . . . . . 1055.3 Centrality Dependence of Charm Electron Yield . . . . . . . . . . . . . . . 1095.4 Comparison with d+ Au collisions . . . . . . . . . . . . . . . . . . . . . . 1135.5 Medium Modification of Heavy Quarks . . . . . . . . . . . . . . . . . . . . 114

CONTENTS III

5.6 Total Charm Cross Section . . . . . . . . . . . . . . . . . . . . . . . . . . . 1155.6.1 Calculation of Cross Section . . . . . . . . . . . . . . . . . . . . . . 1155.6.2 Comparison with Di-electron Measurement . . . . . . . . . . . . . . 1225.6.3 Collision Energy Dependence . . . . . . . . . . . . . . . . . . . . . 122

6 Summary and Conclusions 125

A Simulation Condition 129A.1 Comparison between Read Data and Simulation . . . . . . . . . . . . . . . 129A.2 Photon Converter in Simulation . . . . . . . . . . . . . . . . . . . . . . . . 135

B Cocktail Calculation for Centralities 136

C Data Table 139C.1 Invariant Differential Yield of Heavy Flavor Electrons . . . . . . . . . . . . 139

D Centrality Dependence of Charm Production 142D.1 Systematic Uncertainty of α for Run 2 data . . . . . . . . . . . . . . . . . 142D.2 Systematic Uncertainty of α for Run 4 and Run 5 data . . . . . . . . . . . 144D.3 Uncorrelated Systematic Uncertainty of Ncoll . . . . . . . . . . . . . . . . . 150

E Charm Cross Section in Run 4 154

List of Figures

1.1 Running coupling constant αs of strong interaction as a function of themomentum transfer, Q. These values are measured in several experimentsand are compared to the QCD prediction. . . . . . . . . . . . . . . . . . . 2

1.2 The energy density / T 4 from the recent lattice QCD calculation as a func-tion of T/Tc [4]. The significant increase appears around T ≈ Tc. . . . . . . 3

1.3 Theoretical phase diagram of nuclear matter [5]. The curves represent thephase boundaries between these phases. . . . . . . . . . . . . . . . . . . . . 4

1.4 π0 RAA(pT) for central and peripheral Au + Au collisions [9] and π0RdAu(pT)for minimum bias d+Au collisions [10]. The boxes on the left and the rightside of the figure show the systematic errors for the Au + Au and the d+Aupoints, respectively. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5

1.5 Azimuthal anisotropy (v2) of π, K and protons as a function of pT in mini-mum bias Au + Au collisions at

√sNN = 200 GeV [13]. These are compared

to the hydrodynamical calculation [15]. . . . . . . . . . . . . . . . . . . . 6

1.6 Scaled azimuthal anisotropy (v2/n) as a function of pT/n [14]. The nindicates the number of constituent quarks in a hadron. . . . . . . . . . . 7

2.1 Schematic view of heavy flavor measurement. . . . . . . . . . . . . . . . . . 10

2.2 The LO and some NLO diagrams of heavy quark production. (a) gluonfusion. (b) qq annihilation. (c) pair creation with gluon emission. (d) flavorexcitation. (e) gluon splitting (f) gluon splitting but of flavor excitationcharacter [26]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12

2.3 Parton Distribution Function as a function of x at Q = 5 GeV calculatedby the CTEQ group [29]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13

2.4 pT spectra of B hadrons measured at CDF compared with the FONLLpredictions [35, 36]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14

2.5 (a) Differential invariant cross section of electrons from heavy-flavor decayscompared with the FONLL calculation. (b) Ratio of the data and theFONLL calculation. [39]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14

2.6 (a) Inclusive cross section measurement for the production of D0 and D∗+

in e+ +e− collisions in the CLEO and BELLE as a function of xp = p/pmax

which is approximately z [49, 50]. (b) Fragmentation function of b quarksinto B hadrons as a function of xB = z measured in ALEPH, OPAL andSLD [51, 52, 53]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16

2.7 Semi-leptonic decay diagram of the D and B meson in “spectator model”. 17

IV

LIST OF FIGURES V

2.8 The normalized decay width spectra for charms (left panel) and bottoms(right panel) as a function of the relative momentum of electrons (2Ee/me).Both CKM favored and disfavored spectra are plotted. . . . . . . . . . . . 18

2.9 A light-cone diagram of space-time evolution in high energy heavy ion col-lisions. The values of time and temperature for various phases are takenfrom [57]. The mixed phase exists if the phase transition is first order. . . 19

2.10 Participant-spectator model of high energy heavy ion collision. Two collid-ing nuclei approach with impact parameter b before the collision. After thecollision, the spectator nucleons pass away into the longitudinal direction. 22

2.11 A theoretical calculation of charm quark production as a function of pT

in central Au + Au collisions at√sNN = 200 GeV. The initial (solid),

pre-equilibrium (dot-dashed) and thermal production (dashed) are plotted,respectively [65]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24

2.12 Modification of nuclear structure function per nucleon. (top) a phenomeno-logical curve. (bottom) a set of experimental results [71]. . . . . . . . . . 25

2.13 The ratio of the cross section for the cc (left) and the bb (right) in p + Aand A + A collisions. The nuclear effect of the PDF’s are calculated usingthe EKS98 nuclear weight functions [72]. . . . . . . . . . . . . . . . . . . . 26

2.14 Both the left and right panel show π0RAA for top 5% central Au + Aucollisions at

√sNN = 200 GeV [83]. π0RAA are compared with the BDMPS

(left) and the GLV (right) model with various parameters. The red curvesindicate the best fit curve of the models. . . . . . . . . . . . . . . . . . . . 28

2.15 RAA of electrons from heavy flavor decays for top 10% central Au + Aucollisions at

√sNN = 200 GeV [84]. The electron RAA is compared with

the models. The green dot curve is the BDMPS, and the yellow band isthe GLV models. The green band shows the combined collisional and theradiative energy loss model. . . . . . . . . . . . . . . . . . . . . . . . . . . 28

2.16 The fractional energy loss of the collisional and radiative process as a func-tion of momentum for charm (left) and bottom (right) quarks [82]. The solidand dot-dashed curves correspond to the collisional and radiative process,respectively. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29

3.1 The RHIC accelerator complex. . . . . . . . . . . . . . . . . . . . . . . . . 31

3.2 The cutaway drawing of the PHENIX detector. . . . . . . . . . . . . . . . 32

3.3 The composition of PHENIX detector in Run 2 operation: (top) the beamview of the central arm. (bottom) the side view of the PHENIX detector. . 34

3.4 (a) A BBC array comprising 64 BBC elements. (b) A BBC element. . . . 35

3.5 The z vertex distribution. The shaded area corresponds to the events trig-gered by LVL1. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35

3.6 (top) The location of the ZDC. (bottom) A side view of a ZDC module. . . 36

3.7 The schematic view of the MVD . . . . . . . . . . . . . . . . . . . . . . . . 37

3.8 A cutaway view of the PHENIX magnets. The line shows the contour ofthe magnetic field. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39

3.9 (top) A schematic view of the DC with 20 DC sectors. (bottom) A pictureof the DC covered by the PC. . . . . . . . . . . . . . . . . . . . . . . . . . 40

VI LIST OF FIGURES

3.10 (left)The layout ofX and U ,V stereo wires in a sector. (right) The schematicdiagram of X, U and V orientation. . . . . . . . . . . . . . . . . . . . . . . 41

3.11 (left) The pad and pixel geometry. (right) A cell defined by three pixels. . 423.12 A cutaway view of the RICH gas vessel with the spherical mirrors. . . . . . 433.13 The top view of the RICH. An electron emits Cerenkov photons. . . . . . . 443.14 Schematic view of the single TOF panel which is composed of the 96 scin-

tillator slats. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 453.15 The PbSc module consisting 4 towers. . . . . . . . . . . . . . . . . . . . . . 463.16 An unit array of the PbGl modules. . . . . . . . . . . . . . . . . . . . . . . 473.17 Schematic diagram of the PHENIX DAQ. . . . . . . . . . . . . . . . . . . 473.18 Block diagram of the PHENIX timing system. . . . . . . . . . . . . . . . . 483.19 Block diagram of the EVB architecture. . . . . . . . . . . . . . . . . . . . . 493.20 (left) Photon converter installed to wrap around the MVD outer shell in

Run 2. (right) Converter placed around the beam pipe in Run 4 becausethe MVD was removed. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50

4.1 The centrality determined by the correlation between the ZDC total energyand the BBC multiplicity [105]. . . . . . . . . . . . . . . . . . . . . . . . . 53

4.2 Schematic view of the colliding nuclei. . . . . . . . . . . . . . . . . . . . . 544.3 Systematic uncertainty of the Npart. . . . . . . . . . . . . . . . . . . . . . . 554.4 Systematic uncertainty of the Ncoll. . . . . . . . . . . . . . . . . . . . . . . 564.5 (left) The schematic view of a reconstructed track by the DC in the x − y

plane. (right) The schematic view of a reconstructed track by the DC inthe r − z plane. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57

4.6 (left) The DC hits in the x− y plane. (right) The corresponding hit distri-bution in the feature space [107]. . . . . . . . . . . . . . . . . . . . . . . . 57

4.7 The schematic view of the reflected line by the RICH mirrors and hit PMTsin the PMT array with the tight association radius. . . . . . . . . . . . . . 60

4.8 (left) Energy to momentum (E/p) ratio (right) n0 distribution in RICH. Inboth panels, electron signals are clearly seen. . . . . . . . . . . . . . . . . . 62

4.9 Fiducial volume for the East-South (left) and the East-North (right). Thered and the blue lines show the boundaries of the fiducial cut E and F ,respectively. The shaded areas are removed out in the analysis. . . . . . . . 63

4.10 φ distributions of e+ in the East-South. The red and the blue histogramscorrespond to the data with and without the converter. . . . . . . . . . . . 64

4.11 φ distributions of e− in the East-South. The red and the blue histogramscorrespond to the data with and without the converter. . . . . . . . . . . . 65

4.12 φ distributions of e+ in the East-North. The red and the blue histogramscorrespond to the data with and without the converter. . . . . . . . . . . . 65

4.13 φ distributions of e− in the East-North. The red and the blue histogramscorrespond to the data with and without the converter. . . . . . . . . . . . 66

4.14 Schematic view of the flip and slide method. . . . . . . . . . . . . . . . . . 674.15 E/p distributions for four pT ranges. The net signal in each panel is ex-

tracted by subtracting the backgrounds (BG) from the total electron can-didates (Total). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68

LIST OF FIGURES VII

4.16 The run dependence of the raw electron yield for East-South. The red andthe blue points correspond to the positron and the electron yield, respectively. 69

4.17 The raw yield of inclusive electrons as a function of pT in minimum biasAu + Au collisions. The blue and the red points represent the electronyields in the converter run and the non-converter run, respectively. . . . . . 71

4.18 The raw yield of inclusive electrons as a function of pT. The blue and thered points represent the electron yields in the converter run and the non-converter run. These six panels correspond to the six centrality classes,0-10%, 10-20%, 20-40%, 40-60% and 60-92%, respectively. . . . . . . . . . . 71

4.19 Comparisons of the detector response between the real data and the simula-tions. The detector response of emcsdphi e (top-left), emcsdz e (top-right),n0 (bottom-left) and dep (bottom-right) are shown, respectively. The blackand red lines correspond to the real data and the simulation. . . . . . . . . 73

4.20 Invariant mass distribution of e+e− pairs. The red and black points corre-spond to the real and the combinatorial background pairs estimated by anevent mixing method. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74

4.21 Schematic view of track reconstruction. The electrons produced at R > 0are reconstructed with incorrect momenta. This results in the fake invariantmass of the reconstructed conversion pairs. . . . . . . . . . . . . . . . . . . 74

4.22 Invariant mass distribution of e+e− pairs in the real data (left) and thesimulation (right). The red and blue histogram correspond to the non-converter and the converter run, respectively. . . . . . . . . . . . . . . . . . 75

4.23 Raw pT distribution of e+e− pairs from Me+e− < 40 MeV (left) and 60 <Me+e− < 100 MeV (right). Red histogram shows real data, and the othersare a set of simulation data . . . . . . . . . . . . . . . . . . . . . . . . . . 76

4.24 pT distribution of electrons. The black and the red histograms correspondto the input and the reconstructed electrons, respectively. . . . . . . . . . 78

4.25 The acceptances for e+/e− at the East-South/East-North, respectively. Ineach panel, the black and the red histograms (from top 2 histograms) showthe pure geometrical acceptance with and without the fiducial cut. Thegreen and the thick-black histograms show the acceptance determined byonly 5 σ matching cut and all the electron ID cuts. . . . . . . . . . . . . . 79

4.26 The acceptance functions for e+/e− at the East-South/East-North, respec-tively. In each plot, the green histogram is calculated from the simulation,and the black is calculated from the real data. The red histogram is theaverage of the green and the black one. . . . . . . . . . . . . . . . . . . . 80

4.27 Relation between the inclusive electron yield and the thickness of the de-tector material. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83

4.28 The pT spectra of pions with the fit functions. The red and the blue pointscorrespond to charged and neutral pion, respectively . . . . . . . . . . . . 85

4.29 The ratio of the pion spectra to the fit functions. . . . . . . . . . . . . . . 854.30 Rπ0

γ as a function of pT. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 874.31 Rη

γ as a function of pT. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 874.32 The ratio of the electron yields from Dalitz decays of the other hadrons

relative to π0. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89

VIII LIST OF FIGURES

4.33 the ratio of the electron yield from Dalitz decays of all other hadrons (h =η+η′+ω+φ) to π0. The red, the green and the blue histograms correspondto η/π0 = 0.45, 0.35(0.45 − 0.1) and 0.55(0.45 + 0.1), respectively. . . . . 89

4.34 The total Rγ from all hadrons decaying to electrons. . . . . . . . . . . . . . 90

4.35 The ratio of electrons from Dalitz decays with and without the converter.The ratio is determined by the PISA simulation of π0 Dalitz decays. . . . . 90

4.36 RCN in real data and Rγ in simulation for minimum bias and five centralityclasses. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91

4.37 The 1 + RNP distributions as a function of pT for minimum bias and fivecentrality classes. The points above 1.0 show a fraction of non-photonicelectrons relative to photonic electrons. . . . . . . . . . . . . . . . . . . . . 92

4.38 Raw pT spectrum of inclusive electrons (green) in minimum bias Au + Aucollisions are decomposed into non-photonic (magenta) and photonic elec-trons (light blue). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92

4.39 The averaged pT spectrum of K+ and K− with the fitting function. . . . . 94

4.40 Raw pT distributions of electrons from K+, K−, K0L and K0

S decays, respec-tively . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 94

4.41 The ratio of kaon spectra in minimum bias to that in five centralities . . . 95

4.42 The ratio of electrons from Ke3 decays to photonic electrons . . . . . . . . 95

4.43 pT spectra of ω, ρ and φ→ e+e− for each centrality . . . . . . . . . . . . . 96

4.44 pT spectra of kaon (left) and proton (right) compared with 2 mT scalingfunctions. The red and blue correspond to mT scaling with m = 1.0 andm = 1.7, respectively. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97

4.45 Comparison of the e+ yield as a function of the DC phi. . . . . . . . . . . . 99

4.46 Comparison of the e− yield as a function of the DC phi. . . . . . . . . . . . 99

4.47 Ratio between the statistical error and the value of the geometrical accep-tance in the simulation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100

4.48 Systematic uncertainty of the acceptance and the electron ID variables . . 101

4.49 Systematic uncertainty of electrons from Ke3 decays (left) and two electrondecays of light vector mesons (right). . . . . . . . . . . . . . . . . . . . . . 102

4.50 Relative error of Rγ calculated by changing the η/π0 ratio. . . . . . . . . . 103

5.1 Invariant differential yield of heavy flavor electrons in minimum bias Au + Aucollisions at

√sNN = 200 GeV. . . . . . . . . . . . . . . . . . . . . . . . . 106

5.2 Invariant differential yield of heavy flavor electrons in 0-10% central Au + Aucollisions at

√sNN = 200 GeV. . . . . . . . . . . . . . . . . . . . . . . . . 106


√sNN = 200 GeV. . . . . . . . . . . . . . . . . . . . . . . . . . 106


√sNN = 200 GeV. . . . . . . . . . . . . . . . . . . . . . . . . . 106


√sNN = 200 GeV. . . . . . . . . . . . . . . . . . . . . . . . . . 107


√sNN = 200 GeV. . . . . . . . . . . . . . . . . . . . . . . . . . 107

LIST OF FIGURES IX

5.7 Invariant differential yields of heavy flavor electrons for different Au + Aucentralities scaled by several powers of ten for clarity. Error bars and brack-ets correspond to the statistical and the systematic uncertainties, respec-tively. The curves are the best fit curves of electrons in p+ p collisionsscaled by TAA for the corresponding Au + Au centrality. . . . . . . . . . . . 108

5.8 The ratio of bottom contribution to heavy flavor electrons as a function ofpT in p+ p collisions at

√s = 200 GeV [123]. . . . . . . . . . . . . . . . . 109

5.9 (top)Invariant differential cross sections of electrons from heavy flavor de-cays in p+ p collisions at

√s = 200 GeV. The red and the black points

correspond to the Run 2 data [120] and Run 5 data [39] with statisticalerrors, respectively. The brackets and the yellow bands are systematic un-certainty for Run 2 and Run 5, respectively. The curves are the FONLLcalculation. (bottom) Ratio of the data and the FONLL calculations. Theupper (lower) curves shows the theoretical upper (lower) limit of the FONLLcalculation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 110

5.10 dNe/dy/Ncoll as a function ofNcoll in Au + Au and p+ p collisions at√sNN =

200 GeV. The gray box and the black circles correspond to the minimumbias data and five centrality data. The cross symbol shows the p+ p data.The error bars and the brackets represent the statistical and systematicuncertainties, respectively. . . . . . . . . . . . . . . . . . . . . . . . . . . . 111

5.11 dNe/dy/TAA as a function of Ncoll in Au + Au and p+ p collisions measuredin Run 4 [121] and Run 5 [39]. The gray box and the black circles corre-spond to the minimum bias data and the five centrality data. The crosssymbol shows the p+ p data. The error bars and the brackets represent thestatistical and systematic uncertainties, respectively. . . . . . . . . . . . . 112

5.12 dNe/dy/Ncoll(0.8 < pT < 4.0 GeV/c) in d+Au collisions at√sNN = 200 GeV.113

5.13 The parameter α of open charm and J/ψ as a function of rapidity (left)and pT (right). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 114

5.14 D+/D0 ratio (left) and Ds/D0 ratio (right) measured at CDF [128]. . . . . 116

5.15 Invariant yield of electrons from charm and bottom decays calculated byPYTHIA. The green and the blue points represent the charm and the bot-tom contributions, and the red point is the combined contribution of charmsand bottoms, respectively. . . . . . . . . . . . . . . . . . . . . . . . . . . . 118

5.16 Scaled differential invariant yields 1TAA

12πpT

dσe

dydpTfor minimum bias and the

five centrality classes. The spectrum is fit by the PYTHIA electron functionwith 0.8 < pT < 4.0 GeV/c (blue) . . . . . . . . . . . . . . . . . . . . . . . 118

5.17 Rapidity distribution of charmed hadrons. . . . . . . . . . . . . . . . . . . 1205.18 (top) The invariant mass spectrum of e+e− pairs in p+ p collisions [132].

The points show the measured data with the statistical(bar) and the sys-tematic(shades) errors. The curves represent the cocktail calculation of thelight neutral mesons and the vector mesons. The contributions from c andb quarks and Drell-Yan are also shown. (bottom) The ratio of the data tothe cocktail. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123

5.19 The comparison of the total charm cross sections from the single electronsmeasurements and the e+e− pair measurement. . . . . . . . . . . . . . . . 123

X LIST OF FIGURES

5.20 The total charm cross sections as a function of the collision energy. Thetheoretical curves by PYTHIA and NLO pQCD are also shown. . . . . . . 124

A.1 Comparisons of emcsdphi e measured at the PbSc EMCal for several mo-mentum classes. The black and the red histograms correspond to the realdata and the simulation, respectively. . . . . . . . . . . . . . . . . . . . . . 130

A.2 Comparisons of emcsdz e measured at the PbSc EMCal for several momen-tum classes. The black and the red histograms correspond to the real dataand the simulation, respectively. . . . . . . . . . . . . . . . . . . . . . . . . 130

A.3 Comparisons of emcsdphi e measured at the PbGl EMCal for several mo-mentum classes. The black and the red histograms correspond to the realdata and the simulation, respectively. . . . . . . . . . . . . . . . . . . . . . 131

A.4 Comparisons of emcsdz e measured at the PbGl EMCal for several momen-tum classes. The black and the red histograms correspond to the real dataand the simulation, respectively. . . . . . . . . . . . . . . . . . . . . . . . . 131

A.5 Comparisons of n0 for several momentum classes. The black and the redhistograms correspond to the real data and the simulation, respectively. . . 132

A.6 Comparisons of chi2/npe0 for several momentum classes. The black and thered histograms correspond to the real data and the simulation, respectively. 132

A.7 Comparisons of disp for several momentum classes. The black and the redhistograms correspond to the real data and the simulation, respectively. . . 133

A.8 Comparisons of dep measured at the PbSc EMCal for several momentumclasses. The black and the red histograms correspond to the real data andthe simulation, respectively. . . . . . . . . . . . . . . . . . . . . . . . . . . 133

A.9 Comparisons of dep measured at the PbGl EMCal for several momentumclasses. The black and the red histograms correspond to the real data andthe simulation, respectively. . . . . . . . . . . . . . . . . . . . . . . . . . . 134

A.10 Comparisons of emcdt for several momentum classes. The black and thered histograms correspond to the real data and the simulation, respectively. 134

A.11 The photon converter wrapped the MVD in simulation. The panels showsthe beam view (left) and the side view (right), respectively. The blue partat the bottom of the MVD in the beam view (left) is the overlap width ofthe converter. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 135

B.1 The pion spectra for minimum bias and five centrality classes. The blue andthe green points are the charged and the neutral pion, respectively. Thespectral shape is obtained by fitting. . . . . . . . . . . . . . . . . . . . . . 137

B.2 The pT spectra of photonic background electrons for minimum bias andfive centrality classes. These contributions are determined by the cocktailcalculation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 138

D.1 The dNe/dy(0.8 < pT < 4.0)/Ncoll for the source (1) (left) and (2) (right).The gray, the blue and the red points correspond to the center value, thehigher and the lower uncertainties, respectively. The curves represent thefit functions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 143

LIST OF FIGURES XI

D.2 dNe/dy(pT > 0.4)/TAA as a function of pT. The data points are moved-up(down) by the systematic uncertainty of εmult. The magenta and the light-blue points correspond to the higher and the lower systematic uncertainties.The curves are the fit functions for the corresponding data. The fit areperformed for the Au + Au (left) and for the Au + Au and p+ p data (right)separately. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 145

D.3 Ratio k of inclusive to non-photonic electrons. Electrons are measured inRun 4 Au + Au [121] and Run 5 p+ p collisions [39]. . . . . . . . . . . . . 147

D.4 dNe/dy(pT > 0.4)/TAA as a function of pT. The data points are moved-up(down) by the systematic uncertainty ofRγ. The magenta and the light-bluepoints correspond to the higher and the lower systematic uncertainties. Thecurves are the fit functions for corresponding data. The fit are performedfor Au + Au (left) and for Au + Au + p+ p data (right) separately. . . . . 148

D.5 dNe/dy(pT > 0.4)/TAA as a function of pT. The data points are moved-up(down) by the systematic uncertainty of Ncoll. The magenta and the light-blue points correspond to the higher and the lower systematic uncertainties.The curves represent the fit functions for the corresponding data. The fitare performed for Au + Au (left) and for Au + Au + p+ p data (right). . . 149

D.6 The correlated terms of the uncertainty as a function of centrality. Thesepanels from the top-left to the bottom-right correspond to the errors fromcomponents (i) to (ix), respectively. The line in each panel shows theweighted average. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 151

D.7 The uncorrelated terms of the uncertainty as a function of centrality. Ineach panel, the closed circles are the errors from these components and theopen circles are the averages of these error pairs. . . . . . . . . . . . . . . 152

D.8 Systematic uncertainty of Ncoll as a function of centrality. The points rep-resent the difference of the calculated Ncoll compared to the default value.The yellow and gray bands correspond to the total and the uncorrelatederror of Ncoll. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 153

List of Tables

2.1 The summary of Bjorken energy density εBj for several collision systemsand energies. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21

3.1 Summary of RHIC operation. The integrated luminosity is recorded inPHENIX. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31

3.2 Summary of the MVD composition and their radiation lengths . . . . . . . 383.3 The basic parameters of the photon converter . . . . . . . . . . . . . . . . 50

4.1 Summary of centrality, b, Npart, Ncoll and TAA . . . . . . . . . . . . . . . . 564.2 Bit definition of the track quality. . . . . . . . . . . . . . . . . . . . . . . . 584.3 Summary of the required cut values for the electron ID. . . . . . . . . . . . 624.4 Summary of the fiducial cuts E and F for the East-South and the East-North. 644.5 Summary of the run lists for the converter run and the non-converter run. 684.6 Summary of event samples for the converter run and the non-converter run. 694.7 Summary of the required cuts for the comparison between the real data and

the simulation. The symbol “©” and “×” mean whether the cut is requiredor not required, respectively. The variables, sdphi and sdz, are shortenedforms of emcsdphi e and emcsdz e, respectively. . . . . . . . . . . . . . . . 72

4.8 Summary of NBP and NMVD in the non-converter run and the converter run. 774.9 Summary of the material amounts in real data and simulation. . . . . . . . 774.10 The parameters for the electron spectrum . . . . . . . . . . . . . . . . . . 784.11 Electron ID efficiencies determined by the conversion pair analysis . . . . . 814.12 Summary of the efficiencies determined in the PISA simulation. . . . . . . 814.13 Summary of the multiplicity dependent efficiency loss . . . . . . . . . . . . 814.14 Parameters for the fit functions of π0. . . . . . . . . . . . . . . . . . . . . . 844.15 Mesons to pion ratios . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 864.16 Branching ratios of the mesons decaying to electrons and photons . . . . . 864.17 The parameters of the fit function in Eq. 4.37 . . . . . . . . . . . . . . . . 884.18 The parameter obtained by fitting the kaon spectrum. . . . . . . . . . . . . 934.19 dN/dy of K+, K− and 〈K〉 in minimum bias Au + Au collisions [105]. . . . 934.20 The Summary of the systematic uncertainties for the electron ID cut vari-

ables and the others. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100

5.1 Summary of α values obtained by fitting. . . . . . . . . . . . . . . . . . . . 1125.2 D+/D0, Ds/D

0, Λc/D0 ratios in the experimental data and PYTHIA. . . . 117

5.3 Electron branching ratio of charmed hadrons . . . . . . . . . . . . . . . . . 117

XII

LIST OF TABLES XIII

5.4 The parameters obtained by fitting the PYTHIA electron spectrum. . . . . 117

5.5 dσe/dy (0.8 < pT < 4.0 GeV/c) for minimum bias and the five centralityclasses. The systematic uncertainty represents the combined uncertainty ofthe spectrum and TAA. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119

5.6 The systematic uncertainty of the source (4). . . . . . . . . . . . . . . . . . 120

5.7 All the systematic uncertainties are summarized and the total uncertaintyis obtained by adding these uncertainties in quadrature. . . . . . . . . . . . 121

5.8 Centrality class, Ncoll, nuclear overlap function (TAA), charm cross sec-tion (dσcc/dy) and total charm cross section (σcc) in Au + Au collisionsat

√sNN = 200 GeV. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121

B.1 The parameters obtained by fitting the pion spectra for each centrality. . . 136

C.1 The invariant differential yield of heavy flavor electrons in minimum biasAu + Au collisions in

√sNN = 200 GeV. . . . . . . . . . . . . . . . . . . . . 139

C.2 The invariant differential yield of heavy flavor electrons in 0-10% centralAu + Au collisions in

√sNN = 200 GeV. . . . . . . . . . . . . . . . . . . . . 140


√sNN = 200 GeV. . . . . . . . . . . . . . . . . . . . 140


√sNN = 200 GeV. . . . . . . . . . . . . . . . . . . . 140


√sNN = 200 GeV. . . . . . . . . . . . . . . . . . . . 141


√sNN = 200 GeV. . . . . . . . . . . . . . . . . . . . 141

C.7 The Ncoll scaled integrated yield of heavy flavor electrons dNe/dyNcoll

(0.8 < pT <

4.0 GeV/c). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 141

D.1 The systematic uncertainty of α for the source (1) and (2). The totalsystematic error is estimated adding these errors in quadrature. . . . . . . 143

D.2 Systematic uncertainty of the multiplicity dependent efficiency loss [122]. . 144

D.3 The systematic uncertainty of α for the source (1). The parameter α areobtained for the Au + Au data and for the Au + Au and p+ p data sepa-rately. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 146

D.4 Summary of the ratio k and the relative error of non-photonic electrons inRun 4 analysis. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 146

D.5 The systematic uncertainty of α for the source (2). The parameter α areobtained for Au + Au and for Au + Au and p+ p separately. . . . . . . . 147

D.6 Uncorrelated systematic uncertainty of Ncoll described in Appendix D.3. . 148

D.7 The systematic uncertainty of α for the source (3). The parameter α areobtained for Au + Au and for Au + Au and p+ p separately. . . . . . . . 149

D.8 Summary of the α systematic uncertainty from all three sources . . . . . . 149

D.9 Summary of the uncorrelated Ncoll uncertainty for the five centrality classes. 152

D.10 Summary of the total and the uncorrelated systematic uncertainties for thefive centrality classes. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 153

XIV LIST OF TABLES

E.1 The summary of dσe/dy(pT > 0.4) calculated in the step 1. . . . . . . . . . 155E.2 Summary of dσe/dy and then b contribution is subtracted. . . . . . . . . . 156E.3 Summary of dσcc/dy and σcc in minimum bias Au+Au collisions and five

centrality classes. The charm cross section in p+ p collisions is also shown[39]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 156

Chapter 1

Introduction

One of the most important issues in natural science is to search for and study the fun-damental constituents of matter. In the last half of the 20th century, it is realized thatquarks and gluons are the constituents of nucleon (the proton and the neutron). Thistheory is experimentally confirmed by deep inelastic scattering of electron and proton [1].

Quantum Chromodynamics (QCD) describes the strong interaction of quarks and glu-ons with color charge exchange and is a component of the standard model of particlephysics [27]. A feature of QCD is the momentum transfer (Q) dependence of the strongcoupling constant αs(Q) which is expressed as Eq. 1.1 [27]:

αs(Q) ≈ 12π

(33 − 2nf ) ln (Q2/Λ2QCD)

. (1.1)

Here, nf is the number of active quark flavors with quark mass less than Q and ΛQCD isthe scale parameter of the strong interaction (ΛQCD ≈ 200 MeV). Figure 1.1 shows therunning αs(Q) as a function of the momentum transfer measured in several experiments.The curve in Fig. 1.1 indicates the QCD prediction.

At a large momentum transfer (Q > ΛQCD) and equivalently at a short distance, αs(Q)decreases logarithmically, and quarks and gluons behave almost freely. This is called as“asymptotic freedom”. The feature of asymptotic freedom allows us to describe the stronginteraction at large Q2 in term of perturbation theory. In contrast, αs(Q) becomes large atsmall momentum transfer (Q ≈ ΛQCD), at long distance. Thus, the perturbative approachis not applicable. In this region, quarks strongly attract each other and form color neutralstate such as mesons (quark-antiquark pairs) and baryons (bound state of three quarks).This phenomenon is called as “color confinement”. Because of this “confinement” featureof QCD, a single colored-quark has never been observed.

1.1 Quark Gluon Plasma

According to asymptotic freedom of QCD, under extreme conditions such as high tempera-ture and/or high density, the matter does not consist of normal hadrons, but of many-bodysystem of deconfined quarks and gluons. This state of matter is called as “Quark GluonPlasma” (QGP) [2, 3]. Based on a simple dimensional calculation, a phase transition from

1

2 CHAPTER 1. INTRODUCTION

Figure 1.1: Running coupling constant αs of strong interaction as a function of the mo-mentum transfer, Q. These values are measured in several experiments and are comparedto the QCD prediction.

hadronic matter to QGP would occur the critical temperature Tc ∼ 180 MeV, where thestrong coupling constant is order of unity. Thus, the perturbative description is suspicious.Lattice QCD is a powerful tool to study the phase transition in such a strong couplingregime. Lattice QCD predicts a phase transition to QGP [4]. Figure 1.2 shows the lat-tice QCD prediction of the energy density/T 4 as a function of temperature. The energydensity rapidly increases around the critical temperature (Tc) and approach the arrowsshown at right side of the figure. The arrows represent the energy density εSB of “StefanBoltzmann” limit in the QGP phase. The predicted critical temperature is Tc ≈ 180 MeVand the corresponding critical energy density is εc ≈ 1 GeV/fm3 which is roughly 10 timeslarger than that of the normal nuclear matter (ε = 0.14 GeV/fm3).

If QGP is assumed as an ideal gas of massless quarks and gluons (“Stefan-Boltzmann”limit), the energy density εSB is expressed as follows:

εSB =

{7

8dq + dg

}π2

30T 4 (1.2)

=

{7

8· 2f · 2s · 2q · 3c + 2s · 8c

}π2

30T 4 (1.3)

= 37 · π2

30T 4 (1.4)

where dq and dg stand for the degree of freedom of quarks and gluons in QGP and 7/8 is afactor from which quark is fermion. The dq consists of flavor (2), spin (2), quark/antiquark(2) and color (3) and the dg is spin(2) and color(8) degree of freedom, respectively. Onthe other hand, for lower temperature, if hadronic matter were a ideal gas of massless

1.1. QUARK GLUON PLASMA 3

0.0

2.0

4.0

6.0

8.0

10.0

12.0

14.0

16.0

1.0 1.5 2.0 2.5 3.0 3.5 4.0

T/Tc

ε/T4 εSB/T4

3 flavour2+1 flavour

2 flavour

Figure 1.2: The energy density / T 4 from the recent lattice QCD calculation as a functionof T/Tc [4]. The significant increase appears around T ≈ Tc.

pions, the degree of freedom of pion gas (dπ) is 3 and its energy density in the hadron gasis 3 π2

30T 4. The energy density in QGP is roughly ten times larger than that in hadron

matter. This is due to the explicit appearance of the color degree of freedom in QGP.

Strikingly speaking, the phase transition from hadronic matter to QGP at high tem-perature and zero net baryon density is now considered to be a rapid cross over. Figure1.3 shows a theoretical phase diagram of nuclear matter as a function of baryo-chemicalpotential µ and temperature T [5]. In this figure, there are two kinds of phase transitions.One is a transition to the QGP phase at high temperature and low density. The other isa transition to a color superconductivity phase (2SC) of matter at low temperature andhigh density. Such a condition of high temperature and low baryon density is consideredto have existed in the early Universe in the first 10−5 second after the Big Bang [2]. Sinceit is difficult to study directly about the early Universe, high energy heavy ion collisionis an only possible tool to reproduce QGP in the laboratory. Therefore, the high energyheavy ion collision provides a great opportunity to realize the QGP state.

Relativistic Heavy Ion Collider (RHIC) at Brookhaven National Laboratory (BNL)began to collide the Au nucleus in

√sNN = 130 GeV in year 2000, then archived its full

energy of√sNN = 200 GeV in year 2001. A large set of experimental observations strongly

implies the formation of QGP [6]. Two most important observations are “suppression ofparticle production at high transverse momentum (pT)” and “strong elliptic flow”. Webriefly review these observations.


Figure 1.3: Theoretical phase diagram of nuclear matter [5]. The curves represent thephase boundaries between these phases.

High pT Suppression

High pT particles are produced in point-like hard-scatterings of high energy partons, andthen fragmented into hadron jet. In nucleus-nucleus collisions (A+A), the scattered highpT partons pass through the dense matter created in the collision. Therefore, these partonsmay suffer an energy loss in the matter due to the parton-matter interactions, e.g. gluonbremsstrahlung radiations [7, 8]. By contrast, if partons propagate through the matterwithout any energy losses, the production of high pT particles should scale linearly withthe number of point-like parton scatterings.

In order to study the high pT particle production, “nuclear modification factor (RAA)”is defined as follows:

RAA =dNAA/dpT

〈TAA〉 dσpp/dpT

, (1.5)

where dNAA/dpT and dσpp/dpT are the invariant cross section as a function of pT for A+Aand p+ p collisions, respectively. 〈TAA〉 is the nuclear overlap function which is equivalentto the N + N integrated luminosity in a A + A collision event. 〈TAA〉 depends on theimpact parameter of the collision event. 〈TAA〉 as a function of impact parameter (b) canbe calculated from geometrical overlap of the colliding nuclei. If RAA = 1, this indicatesthat the particle production in A+A collisions is described by the simple superimpositionof the particle production in p+ p collisions (binary scaling).

PHENIX experiment reported the RAA of neutral pions in central and peripheralAu + Au [9] and minimum bias d + Au collisions [10] at

√sNN = 200 GeV shown in

Fig. 1.4. The central Au + Au collisions demonstrates the strong suppression of the factor0.3 for π0 production compared to the peripheral Au + Au collisions which is consistent to

1.1. QUARK GLUON PLASMA 5

0

0.2

0.4

0.6

0.8

1

1.2

1.4

1.6

0 2 4 6 8 10

Au-Au, Cent

Au-Au, Periph

d-Au, min-bias

pT (GeV/c)

RA

A

Figure 1.4: π0 RAA(pT) for central and peripheral Au + Au collisions [9] and π0RdAu(pT)for minimum bias d + Au collisions [10]. The boxes on the left and the right side of thefigure show the systematic errors for the Au + Au and the d+ Au points, respectively.

unity (RAA = 1). In contrast to Au + Au collisions, RdAu (RAA in d + Au) presents onlythe contribution of the initial-state normal nuclear effect such as Cronin effect [66] andgluon shadowing [70], since the small size of deuteron is not enough to produce hot-densematter. In Fig. 1.4, RdAu is larger than unity for pT > 2 GeV/c and it increases slightlywith increasing pT. This behavior of RdAu is consistent with the expectation based onCronin effect. By comparison with RAA and RdAu, it is obvious that the central Au + Aucollisions produce extremely dense matter.

Elliptic Flow

“Elliptic flow” is an azimuthal anisotropy of particle production. The strength of ellipticflow is sensitive to the early stage of space-time evolution of the colliding system. Innon-central collisions, the overlap region of two colliding nuclei has an almond shapeat t = 0. This initial spatial anisotropy is transferred to the momentum anisotropy ofproduced particles. In the case of that the mean free path (L) of partons in the systemis much smaller than the system size (L � R), the scatterings of the partons generatethe pressure gradient. The pressure gradient in short axis of the almond is steeper thanthat in long axis. Then, the elliptic flow is developed. The collective expansion of thesystem reduces the initial spatial anisotropy and the pressure gradient becomes smallwith time. This indicates that the earlier thermalization of the system is required for thestronger elliptic flow since the initial almond shape disappears gradually by the collective


(GeV/c)T

p0 1 2 3 4

2v

0

0.1

0.2

0.3-π++π-

+K+K

pp+

πHydro.

Hydro. K

Hydro. p

Minimum bias

Figure 1.5: Azimuthal anisotropy (v2) of π, K and protons as a function of pT in minimumbias Au + Au collisions at

√sNN = 200 GeV [13]. These are compared to the hydrody-

namical calculation [15].

expansion. Therefore, the strength of the elliptic flow provides an information about thethermalization time of which the system reaches the thermal equilibration [11].

Experimentally, the azimuthal distribution of particle emissions is studied in term ofFourier expansion expressed in Eq. 1.6:

Ed3N

d3p=

d2N

2πpTdpTdy

(1 + 2

∞∑n=1

vn(pT) cos [n(φ− Ψr)]

), (1.6)

where φ is the azimuthal angle of the produced particle, Ψr is the azimuthal angle of thereaction plane and vn(pT ) is the magnitude of the n th order harmonics. The reaction planeis defined as the plane span by the direction of the impact parameter and the direction ofthe beam axis. The second order harmonic v2 of Fourier expansion represents the strengthof the elliptic flow.

STAR experiment first reported the strong elliptic flow in relatively peripheral Au + Aucollisions at

√sNN = 130 GeV [12]. The detailed measurements of v2 were continued by

the RHIC experiments in Au + Au collisions at√sNN = 200 GeV. Figure 1.5 shows the

elliptic flow strength (v2) of π, K and proton as a function of pT measured by PHENIX[13]. The curves in Fig. 1.5 represent a hydrodynamical calculation including a first orderphase transition with a freeze-out temperature of 120 MeV [15]. The measured v2 has aclear particle-mass dependence. This mass effect is a consequence of radial expansion inwhich the heavier particles flow out to higher pT. The v2 of π, K and p are well reproducedby the model calculation at lower pT (pT < 1.5 GeV/c). The strong v2 implies that thesystem reaches the thermal equilibration quickly at τ ' 0.6 fm/c [2].

By contrast, the model calculation fails to reproduce the data at high pT, since hy-drodynamical picture is not applicable in higher pT region. The quark coalescence model

1.2. THESIS MOTIVATION 7

/n 2vD

ata

/ Fit

/n (GeV/c)tp

0

0.02

0.04

0.06

0.08

systematic error

Polynomial Fit

0SK

-+K+K

pp+

Λ+Λ

0 0.5 1 1.5 2 2.50.4

0.6

0.8

1

1.2

1.4

1.6

1.8 -π++π

Figure 1.6: Scaled azimuthal anisotropy (v2/n) as a function of pT/n [14]. The n indicatesthe number of constituent quarks in a hadron.

[16] is a possible scenario that can explain the scaling property of constituent-quark num-ber. Figure 1.6 shows the quark number scaling [14]. Therefore, the result suggests QGPformation since the strong elliptic flow is developed in the quark-level reaction.

1.2 Thesis Motivation

The property of the dense matter and its space-time evolution needs to be studied in moredetail, although it has been observed by previous measurements that the dense matter iscreated in high energy heavy ion collisions. Heavy quarks (charms and bottoms) are cleanprobes to observe initial stage of the dense matter. Heavy quarks are only produced inpoint-like hard-scatterings in nucleon-nucleon and nucleus-nucleus collisions, since theirmasses (MQ) are larger than the typical energy scale of QCD (ΛQCD). At the energyscale of Q2 ∼ M2

Q, the strong coupling constant is αs ∼ 0.3 (see Fig. 1.1) which is smallenough to apply perturbative QCD calculation to the production of heavy quarks. Inhigh energy nucleon-nucleon collisions, the heavy quarks are primarily produced via gluon


fusions (e.g. gg → cc). Therefore, the total yield of heavy quarks is sensitive to the initialgluon density [17, 18]. Once heavy quarks are produced, they traverse the dense matter andsuffer an possible energy loss via parton-matter interactions, e.g. bremsstrahlung radiationof gluons. But it is expected that the energy loss of heavy quarks is smaller than that oflighter quarks since their heavy mass reduces the available phase space for gluon radiation(“dead cone effect”)[19]. In addition, the cold nuclear effects such as “Cronin effect” [66]and “nuclear shadowing”[70] could also modify the yield and pT distribution of heavyquarks. Therefore, these effects need to be studied through the systematic measurementsof heavy quarks in p+ p, p+ A and A+ A.

In this thisis, we measured the centrality dependence of the total charm productionvia a measurement of single electrons from their semi-leptonic decays. The PHENIXexperiment has an unique ability to measure electrons with wide pT range. During RHICyear 2002 to 2004 (Run 2 to Run 4), the PHENIX detector collected the data in p+ p,d + Au and Au + Au collisions at

√sNN = 200 GeV. Using these data, the initial gluon

density via the total charm production is studied in this thesis.This thesis is organized as follows. Chapter 2 introduces the theoretical and the ex-

perimental background of high energy heavy ion collisions and heavy flavor productions.Chapter 3 explains the RHIC accelerator complex and the PHENIX detector. Chapter 4describes the analysis method. Especially, the separation between the signal and the back-ground is explained in detail. In chapter 5, the result of electron measurement is shownand the interpretation is discussed. Chapter 6 is finally the conclusion of this thesis.

Chapter 2

Theoretical and ExperimentalOverview

Heavy quarks (charms and bottoms) provide clean informations of property of dense mattercreated in high energy heavy ion collisions. Since heavy quarks are mainly produced ininitial hard scattering, the total produced yield of heavy quarks is sensitive to initial gluondensity. As same as light quark, heavy quarks suffer energy losses when they propagatethrough the dense matter, although their larger masses are expected to induce smallerenergy losses. At the beginning of this chapter, the theoretical and the experimentalapproaches of heavy quark production are overviewed.

In latter part, the standard picture of high energy heavy ion collisions is introduced.This describes space-time evolution of the dense matter created in the collisions. Themodification of heavy quark production in heavy ion collisions is reviewed.

2.1 Heavy Quark Measurement

There are two types of methods for heavy quark measurement, “direct” and “indirect”method. In direct methods, heavy flavor mesons (D and B mesons) are reconstructed bycatching all decay products from their hadronic decays (e.g. D → Kπ). Because of thelimited acceptance of the PHENIX detector and the large amount of background hadronsin heavy ion collisions, it is difficult to measure heavy quark production in PHENIX bythe direct methods. On the other hand, the measurement of single electrons from semi-leptonic decays of heavy flavor hadrons is one of the most powerful methods in indirectmethods. These electrons are called as “heavy flavor electrons” in this thesis. Figure2.1 shows a schematic view of heavy flavor measurement by the direct and the indirectmethod.

The overall history from the heavy quark production to the electron measurement areexpressed as Eq.2.1:

p+ p (A+ A)pQCD−→ c(b)

fragmentation−→ D(B)weak decays−→ e. (2.1)

Here, “pQCD” denotes the heavy quark production which is calculable by perturbativeQCD, “fragmentation” is the fragmentation (hadronization) process of heavy quarks,and “weakdecays” is the semi-leptonic decays into electrons (e).

9

10 CHAPTER 2. THEORETICAL AND EXPERIMENTAL OVERVIEW

cc

+K -l

0D0D

-K

+π

DirectMethod

IndirectMethod

Figure 2.1: Schematic view of heavy flavor measurement.

In 1970’s, the indirect method was carried out in p+ p collisions at CERN IntersectingStorage Ring (ISR) (

√s = 30 − 63 GeV). Heavy flavor electrons were first measured in

e/π ratio (e/π ∼ 10−4 for pT > 1.3 GeV/c) [21, 22, 23, 24]. The measured large yield wasinterpreted as evidence of charm production [25].

The difficulty of indirect method are large amount of background from internal (Dalitz)and external conversions of photons from the hadron decays (e.g. π0 → γ + γ) which iscalled as “photonic electrons”. In order to reduce background of photonic electrons fromphoton conversions, the PHENIX detector is designed to minimize an amount of materialaround the beam pipe. In addition, the external converter is used to determine photonicelectrons experimentally. These are explained in section 4

Each steps from the production to the measurement expressed in Eq. 2.1 are describedin following sections.

2.2 Heavy Quark Production

2.2.1 Heavy Quark in pQCD

The hard-scatterings of partons containing in the colliding protons can produce the heavyquark pairs. The general perturbative expression of the partonic cross section for theheavy quark production can be written by the following equation (Eq. 2.2) [20]:

σij(s,m2Q, µ

2R) =

α2s(µ

2R)

m2Q

∞∑k=0

(4παs(µ2R))k

k∑l=0

f(k,l)ij (η) lnl

(µ2

R

m2Q

). (2.2)

Here, s = x1x2s is the partonic energy squared in the center of mass. The index i andj are the parton types (q,q,g) which interact in the particular Feynman diagrams. The

dimensionless scaling function f(k,l)ij represents the amplitude of a given partonic scattering

diagrams. The dimensionless parameter η = s/4m2Q −1 expresses the production threshold

2.2. HEAVY QUARK PRODUCTION 11

of the heavy quark. It means that total partonic energy s must be at least larger than2mQ to create a heavy quark pair. µR is the renormalization scale and µR = mQ is usuallyused for the heavy quark production. The index k indicates the order of subprocessdiagrams. The k = 0 and k = 1 correspond to Leading Order (LO) and Next to LeadingOrder diagram (NLO) which the cross section of LO and NLO is a function of α2

s and α3s,

respectively. When k = 1 and l = 1, the logarithmic term ln (µ2R/m

2Q) appears. The LO

processes are “gluon fusion (gg → QQ)” and “q − q annihilation (qq → QQ)” shown inFig. 2.2 (a) and (b). The gluon fusion process dominates the qq annihilation in high energycollisions, since there are much large number of gluons in proton at small-x. The someNLO processes are shown in Fig. 2.2 (c), (d) and (e). In general, the NLO processes aresmaller contribution than the LO processes. However, the logarithmic term are developedin the NLO process for l = 1, so that this logarithmic term (ln

(µ2

R/m2Q

)) can make the

NLO process to be even larger contribution than the LO process.Using the partonic cross section described above, the total cross section of heavy quark

production in p+ p collisions is formulated as following function (Eq. 2.3) [20]:

σpp(s,m2Q) =

∑i,j=q,q,g

∫ 1

4m2Q

s

dτ

∫ 1

τ

dx1

x1

fpi (x1, µ

2F )fp

j

(τ

x1

, µ2F

)σij(τs,m

2Q, µ

2R). (2.3)

Here, fpi (x1, µ

2F ) is Parton Distribution Function (PDF) in a proton. PDF is a probability

density of partons in proton described in term of Bjorken variable (x) and momentumtransfer scale (µ2

F ), where x is the momentum fraction of proton carried by a parton andµ2

F is called as factorization scale. The PDFs are experimentally determined by deepinelastic lepton-nucleon scatterings (DIS) and hard scatterings in p+ p collisions with awide range of x and µ2

F . The DGLAP equation [27] can describe the PDF evolution withrespect to the momentum scale. The shape of PDF are derived from the comparisonof the theoretical calculation to the experimental measurement. Several theorist groupspublished their own calculations of PDFs [29, 30, 31]. Fig. 2.3 shows the PDF publishedby the CTEQ group [29].

There are 3 parameters for the perturbative calculation of the total cross section inEq. 2.3, mR, µR and µF . The variation of these parameters can make the uncertaintyof the calculation. Thus, the uncertainty is usually determined using these conditions :1/2mQ < µR, µF < 2mQ, 1.2 < mc < 1.8 GeV and 4.5 < mb < 5.0 GeV.

In a recent work, Fixed-Order plus Next-to-Leading-Log calculation (FONLL) is per-formed to evaluate the heavy quark production [32, 33]. The FONLL is based on purter-bative QCD including the NLO fixed-order calculation (FO) and the resummation of thelogarithm of pT/mQ (αn

s (logk(pT/mQ))) with next-to-leading logarithmic accuracy (NLL).The logarithmic term due to multiple gluon radiations can contribute the heavy quarkproduction at higher pT (pT > mQ). In order to compare with the experimental datadirectly, the FONLL framework introduces additionally the non-purterbative informationin self-consistent way. The non-purtertative information related to the hadronization ofheavy quark into hadrons is determined by the experimental data in e+ + e− collisions atLEP [34].

Figure 2.4 shows the pT spectrum of B hadrons measured in CDF compared with theFONLL calculation in p + p collisions at

√s = 1960 GeV [37, 38, 34, 35]. Figure 2.5


gg

Q�Q

(a)

gg

Q�Q

( )g

�qq

�QQ(b)

gg

g�Q

(e)Q

gg g

�QQ(d)

gg g

�QQ(f)

g

Figure 2.2: The LO and some NLO diagrams of heavy quark production. (a) gluon fusion.(b) qq annihilation. (c) pair creation with gluon emission. (d) flavor excitation. (e) gluonsplitting (f) gluon splitting but of flavor excitation character [26].


x

0

0.2

0.4

0.6

0.8

1

1.2

x f(

x,Q

)

10-4 10-3 10-2 10-1 .2 .3 .4 .5 .6 .7

Q = 5 GeV

.8

Gluon / 15dbarubarscuvdv(dbar-ubar) * 5

Figure 2.3: Parton Distribution Function as a function of x at Q = 5 GeV calculated bythe CTEQ group [29].

(a) shows the differential invariant cross section of electrons from semi-leptonic decays ofheavy flavors compared with the FONLL calculations in p+ p collisions at

√s = 200 GeV,

and Fig. 2.5 (b) is the ratio of the data and the calculation. The FONLL calculations forboth collision energies are in good agreement with the experimental data. The FONLLpredicts the total cross section in p+ p collisions at

√s = 200 GeV for charm (σFONLL

cc =256+400

−146 µb) and bottom (σFONLLbb

= 1.87+0.99−0.67 µb) quarks, respectively.

2.2.2 Fragmentation Functions

Quarks produced in the hard scattering are not directly measured. The nature of asymp-totic freedom in QCD prohibits the isolation of colored-quark. Therefore, quarks (q)combine each other so that the colorless state of a hadron is formed. The bound stateof q (colored-q) and q (anti-colored) is called meson, and the state of three quarks withthree different colors is baryon, respectively. The process that quarks are bound to formcolorless hadron is generally called as “Fragmentation” or “Hadronization”.

After quark and anti-quark pair is initially produced, the quark pair separates in op-posite direction. When the distance between the pair exceeds about 1 fm, the potentialenergy due to the color confinement (string) become so large that one or more q− q pairsare created. Eventually all the initial quark energy is converted into two jets of hadrons.

The fragmentation of a quark into hadrons are described by the fragmentation functionDh

q (z), which is the probability that a parton (q) is fragmented into a particular hadron(h) with z. Here, z ≡ Eh/Eq is a fraction of the initial quark (q) energy carried by thehadron. The Dh

q (z) satisfies the momentum and probability conservation (Eq 2.4).


100

101

102

103

0 5 10 15 20 25 30 35 40

d!

/dp

T(p

pb

ar"

B+ +

X,

|y| <

1)

(n

b/G

eV

)

pT (GeV)

fB+ = 0.389All data rescaled to B

+ and |y| < 1

FONLL, CTEQ6M, Kart. # = 29.1NLO, same as above

NLO, no fragm.CDF B

+" J/$ K

+

CDF Hb " µ- D

0

CDF Hb " J/$ X

Figure 2.4: pT spectra of B hadrons measured at CDF compared with the FONLL pre-dictions [35, 36].

)3

c-2

(m

b G

eV

3/d

pσ

3E

d

-1010

-910

-810

-710

-610

-510

-410

-310

-210

-110

(a)=200 GeVs)/2 + X at - + e+

(e→ p+p

PHENIX data

FONLL(total)

e)→FONLL(c

e)→FONLL(b

e)→ c →FONLL(b

(GeV/c)Tp0 1 2 3 4 5 6 7 8 9 10

0.5

1

1.5

2

2.5

3

DA

TA

/FO

NL

L (b)

Figure 2.5: (a) Differential invariant cross section of electrons from heavy-flavor decayscompared with the FONLL calculation. (b) Ratio of the data and the FONLL calculation.[39].


∑ ∫zDh

q (z, µ2) dz = 1 (2.4)

Based on “factorization theorem”, the differential cross section of the heavy flavor hadronscan be written as follows:

dσH

dpT

=

∫dpT dz

dσQ

dpT

DHQ (z) δ(pT − zpT), (2.5)

where pT and pT are the transverse momenta of heavy flavor hadrons and heavy quarks.The dσQ / dpT is the differential cross section of heavy quarks, and the DH

Q is the frag-mentation function of heavy flavors.

A heavy flavor meson retains a large fraction of the momentum of the primordial heavyquark [40]. The heavy quark has large mass so that the energy of the heavy flavor mesonis less affected by the fragmentation with the light quark. Therefore, the fragmentationfunction of heavy quarks is much harder than that of light quarks.

The heavy quark fragmentation function DhQ(z, µ2) needs to be described by phe-

nomenological (non-perturbative) models including the perturbative calculation of theDGLAP evolution equation. The most popular parameterizations of Dh

Q(z, µ2) are listedas follows:

Peterson et al . [41] : D(z) ∝ 1

z

(1 − 1

z− ε

1 − z

)−2

, (2.6)

Kartvelishvili et al . [42] : D(z) ∝ zα (1 − z) , (2.7)

Collins&Spiller [43] : D(z) ∝(

1 − z

z+

(2 − z)εC1 − z

)×

(1 + z2

)(1 − 1

z− εC

1 − z

)−2

, (2.8)

Colangelo&Nason [44] : D(z) ∝ (1 − z)α zβ, (2.9)

Bowler [45] : D(z) ∝ (1 − z)a

z−(1+bm2T )

exp

(−bm

2T

z

), (2.10)

Braaten et al . [46] : (see Eq. (31), (32) in [46]). (2.11)

where ε, εC , a, bm2T , α and β are non-perturbative parameters depending on the heavy

flavor hadron. In general, the non-perturbative parameters in these formulae do not havean absolute meaning. These parameters are determined by the experimental results in thee+ +e− collisions, since the fragmentation function is independent of the type of the initialhard scattering processes.

In PYTHIA [28] (a widely used pQCD event generator) with the default setting of theparameters, the Bowler fragmentation function (2.10) is used. The default parameters inPYTHIA are a = 0.3 and b = 0.58 GeV−2. The peterson fragmentation function (Eq. 2.6)is most popular for heavy flavor hadrons. The ε parameter in the Eq. 2.6 was obtained forcharms and bottoms in [47], for example, the εc ≈ 0.05 and the εb ≈ 0.006 in the caseof using the leading-logarithmic approximation in the perturbative calculation. The most


0

2

4

6

8

10

12

14

16

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1xp=p/pmax

sB d

σ/dx

p (G

eV2 n

b) D0 D*

CLEOBELLE

(a)

0

0.5

1

1.5

2

2.5

3

3.5

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1xB

1/σ

dσ/

dxB

ALEPH 91 GeVOPAL 91 GeVSLD 91 GeV

(b)

Figure 2.6: (a) Inclusive cross section measurement for the production of D0 and D∗+

in e+ + e− collisions in the CLEO and BELLE as a function of xp = p/pmax which isapproximately z [49, 50]. (b) Fragmentation function of b quarks into B hadrons as afunction of xB = z measured in ALEPH, OPAL and SLD [51, 52, 53].

accurate approach to derive the fragmentation function is to use the Mellin transform ofthe DH

Q and determine the moment of the Mellin transform from the experimental result[48].

Experimental studies of the heavy quark fragmentation function have been performed.Figure 2.6 (a) shows the inclusive cross-section of D0 and D∗ times branching ratios as afunction of xp measured in e+ + e− collisions at the CLEO and the BELLE experiment[49, 50]. The variable xp approximates the z. Figure 2.6 (b) shows the fragmentationfunction of the b quarks into B hadrons measured in the ALEPH and the CLEO at LEPand the SLD at SLAC [51, 52, 53].

2.2.3 Semi-Leptonic Decays

We measure electrons from the semi-leptonic decays of hadrons containing charm (c) andbottom (b) quarks. The semi-leptonic decays shown in Figure 2.7 are simply explained inthe standard technique, “spectator model”. In the spectator model, the light antiquark(q) which accompany the heavy quark Q in the hadron is assumed to play no role in thedecay [54, 55]. Therefore, the decay of Q can be treated in same way as the leptonic decayof a free muon.

The semi-leptonic decay width in the model can be written as follows:

Γ(Q)sl =

mQ

28π3

∫dx dy θ(x+ y − xm)θ(xm − x− y + xy) ×

∑|M (Q)|2 (2.12)

where x and y are the rescaled energies of the charged and neutral leptons, x = 2Ee/mQ,y = 2Eν/mQ, in the heavy quark rest frame. xm = 1 − (mq/mQ)2 is the kinematic limitof energy transfer. The mq/mQ is equal to ms/mc for c decay and mc/mb for b decay.


cq

ν

+e

sq

+Wbq

-e

ν

cq

-W

Figure 2.7: Semi-leptonic decay diagram of the D and B meson in “spectator model”.

The matrix elements for c and b decay are :∑|M c|2 = 64G2

F |Vcs|2 c · e+ s · ν, (2.13)∑|M b|2 = 64G2

F |Vcb|2 b · ν c · e−, (2.14)

where b, c, s, ν, ν, ν, e+ and e− are the four-momenta of the decay particles. Vcs and Vcb areelements of Cabibbo-Kobayashi-Maskawa (CKM) matrix. GF = 1.16632 × 10−5(GeV−2)is Fermi constant, the coupling constant of weak interaction.

The differential decay width for charms and bottoms can be calculated from Eq. 2.12as follow:

dΓ(c)sl

dx= |Vcs|2 Γ0(mc)

[12x2(xm − x)2

(1 − x)

], (2.15)

dΓ(b)sl

dx= |Vcb|2 Γ0(mb)

[2x2(xm − x)2

(1 − x)3

](6 − 6x+ xxm + 2x2 − 3xm), (2.16)

where Γ0 is the rescaled muon decay width

Γ0(mQ) =G2

Fm5Q

192π3.

The resulting electron spectra for charm and bottom decays are shown in Fig. 2.8. Inthis figure, both the electron spectra from CKM favored (b → c and c → s) and CKMdisfavored (b→ u and c→ d) decay mode are included. The CKM disfavored spectra areharder than the CKM favored spectra. But the contribution from CKM disfavored modeis small and negligible (e.g. |Vub|/|Vcb| = 0.08 ± 0.02).

If Mc = 1.4 GeV, the theoretical estimate of the decay width is Γsl(D) = 1.1 ×10−10 MeV. On the other hand, using the measured branching ratios (e.g. 17.2% for D+)and the lifetimes (e.g. 1051 × 10−15(s)) of D mesons [27], the experimental decay widthis calculated as Γsl(D

+) = (1.07 ± 0.13) × 10−10 MeV. These two values are consistenteach other. Therefore, the theoretical model can describe nicely the semi-leptonic decaysof heavy flavors.


Figure 2.8: The normalized decay width spectra for charms (left panel) and bottoms (rightpanel) as a function of the relative momentum of electrons (2Ee/me). Both CKM favoredand disfavored spectra are plotted.

2.3 High Energy Heavy Ion Collisions

High-energy heavy-ion collision is a possible way to realize a high temperature and a highenergy density condition in the laboratory, where quark-gluon plasma (QGP) is expectedto form. In heavy ion collisions, two Lorentz-contracted nuclei pass through each otherand a lot of nucleon-nucleon collisions take place in the overlap region of the collidingnuclei. Their deposited energies generate the high energy density object spread in thespace, like a fire ball which produces many secondary hadrons at a later time.

The experimental investigations had started at Bevalac (1975 - 1985) of the LawrenceBerkeley National Laboratory (LBL), and continued at Alternating Gradient Synchrotron(AGS) (1987 - 1995) of the Brookhaven National Laboratory (BNL), Super Proton Syn-chrotron (SPS) (1987 - present) of the European Organization for Nuclear Research(CERN) and Relativistic Heavy Ion Collider (RHIC) of the BNL. However the fixed tar-get experiments at Bevalac, AGS and SPS could not observe the clear evidence of QGPformation. The first collider experiment at RHIC finally confirmed the existence of QGP.

2.3.1 Space Time Evolution

The phenomenon of the high-energy heavy-ion collision is very complicated since the mat-ter produced in the collision undergoes several phase from the initial hard scattering tothe final hadron emission. J.D.Bjorken illustrated space-time evolution of high energyheavy ion collision based on hydrodynamic [56]. In the Bjorken’s picture assuming “cen-tral plateau” (high energy limit), the space-time evolution can be separated into fourindividual phases characterized by a proper time τ =

√t2 + z2. Figure 2.9 shows a light-

2.3. HIGH ENERGY HEAVY ION COLLISIONS 19

Beam Beam

Coordinate space

Deconfined quarks and gluons

Equilibrated QGP

Mixed phase

Hadron gas

Freeze–out

0.6 fm/s, T 350 MeV

Time

4 fm/s, T 180 MeV

8 fm/s, T 160 MeV

16 fm/s, T 100 MeV

Figure 2.9: A light-cone diagram of space-time evolution in high energy heavy ion colli-sions. The values of time and temperature for various phases are taken from [57]. Themixed phase exists if the phase transition is first order.

cone diagram of the space-time evolution in heavy ion collision. The proper times for eachphases are defined as a hyper-surface in Fig. 2.9. The value of the proper times τ inFig. 2.9 are described in [57].

We consider that the two Lorentz contracted “disks” along the longitudinal (z) axiscollides head-on at z = 0 and time t = 0 in the center of mass frame. The thickness ofthe disk represents R/γ ≈ 1 fm where γ is the Lorentz factor. After the collision, thedisk carrying a large amount of baryon number recedes from the overlap region. In theoverlap region, the hard scatterings between partons, quarks and gluons, occur which canbe described by perturbative QCD following parton cascade, and a huge amount of energyis deposited in the matter. This initial phase is named as pre-equilibrium phase.

The multiple scatterings of partons continue and these partons spread their momentafor shorter time, so that the matter would reach the local thermal equilibrium, and theQGP phase is formed. The time to form the QGP phase is called as “thermalizationtime” τ0 which is expected to be 0.6 fm/s [57]. Once the local equilibration is archived,the QGP matter would expand hydrodynamically until the matter cools down to thecritical temperature Tc of the QGP phase transition (Tc ≈ 180MeV).

At Tc, the QGP matter begins to hadronize so that quarks and gluons are confinedinto color singlet hadrons. If the phase transition between the QGP to the hadron phaseis the first order phase transition, a mixture of QGP and hadronic matter would appearduring the transition (mixed phase). In the mixed phase, the volume fraction of hadronicmatter in the QGP phase increases gradually with expansion. The temperature in thesystem stays at Tc because hadronic matter releases latent heat of the transition. Even


though the phase transition from hadronic matter to QGP is now considered to be a rapidcross over at zero baryon density.

After the system completes to hadronize at τ ≈ 8 fm, the produced hadrons interactwith each other in the hadronic matter. This is called as “hadron gas phase”. The hadronicmatter keeps the collective expansion until the system temperature drops to freeze-outtemperature (TF ≈ 100MeV) where the system size is still larger than the mean free pathof the interacting hadrons. Finally the hadrons freely move away from the hadronic matterunder TF and those are detected in our measurement at approximately infinite distance.

2.3.2 Energy Density

The formation of QGP requires a sufficiently large energy density. The density is expectedto be on the order of 1 GeV/fm3 which is about 10 times larger than that in normalnucleus (≈ 0.14 MeV/fm3). It is interesting how much energy density can be reached atthe formation time (τ0) of QGP in high energy heavy ion collisions.

In Bjorken’s picture, the expansion of the system is one dimensional along with z andis cylindrically symmetry until the time reaches τ0. The volume of the system at τ0 iswritten as ∆V = πR2dz where R is a radius of the colliding nucleus. Energy (E) in thevolume is written as follow:

E = 〈mT〉dN

dyδy =

dET

dyδy, (2.17)

where 〈mT〉 =√p2

T +m2 is a transverse energy of the produced hadrons (transverse mass),dN/dy is the rapidity density of particle multiplicity and y is rapidity defined as follows:

y =1

2ln

(t+ z

t− z

). (2.18)

The transverse energy of the produced hadrons can contribute to the energy densityof the initial system.

Using both the volume ∆V and the energy (E), the energy density (εBj) of the systemat τ0 can be expressed as:

εBj =E

∆V, (2.19)

=〈mT〉πR2

dN(τ0)

dz=

〈mT〉πR2 τ0

dN(τ0)

dy, (2.20)

=1

πR2 τ0

dET(τ0)

dy, (2.21)

where ET is the total energy and dz = τdy at central rapidity (y = 0). We equated〈mT〉dN

dy= dET

dy. This energy density εBj is generally referred as “Bjorken energy density”

[56].Although τ0 is not well known, the energy density εBj was evaluated for the AGS,

SPS and RHIC experiments using the measured value of dET/dy and the normally used

2.3. HIGH ENERGY HEAVY ION COLLISIONS 21

value of τ0 = 1 fm/c. These εBj are summarized in Tab. 2.1. The energy density εBj inRHIC low energy (

√sNN = 19.6 GeV) is compared to the recalculated εBj in SPS energy

(√sNN = 17.2 GeV), and found to be consistent [62]. There is a factor of 2.6 increase of

the εBj between RHIC low energy and highest energy (√sNN = 200 GeV).

Table 2.1: The summary of Bjorken energy density εBj for several collision systems andenergies.

Accelerator Colliding Nucleus√sNN (GeV) εBj (GeV/fm3)

AGS Au + Au 5 1.5 [58]SPS Pb + Pb 17 2.9 [59, 60]

RHIC Au + Au 19.6 2.2 [62]Au + Au 130 4.7 [61, 62]Au + Au 200 5.4 [62]

2.3.3 Collision Geometry

The geometrical aspect of high energy heavy ion collisions has an important role in collisiondynamics. Two colliding nuclei only interact in the region of the geometrical overlapas shown in Fig 2.10. The region is characterized by impact parameter b which is thedistance between the centers of the colliding nuclei. The nucleons in the colliding nucleiare separated into 2 groups, “participant” and “spectator” illustrated in Fig. 2.10. Theparticipant is the nucleons in the overlap region where nucleon-nucleon collisions takeplace. On the other hand, the spectator passes away into longitudinal direction with littleinteraction. This geometrical treatment of the collision is known as “Participant-Spectatormodel”.

In experiment, high energy heavy ion collision is characterized by “centrality” to studythe matter produced in the collision systematically, since impact parameter can not directlymeasured. Centrality represents a percentile of total cross section of inelastic nucleusnucleus collision.

The Glauber model [63] is used to associate b with centrality and to evaluate thegeometric parameters such as b, the number of participants (Npart), the number of nucleon-nucleon collisions (Ncoll) and the nuclear overlap function (TAB(b)). In the Glauber model,the nucleus-nucleus collisions are treated as multiple nucleon interactions. Nucleon ofcolliding nucleus is assumed to travel in a straight line and not to deflect after the collision.

The nuclear thickness function of nucleus A is defined as follow [2, 64]:

TA(s) =

∫dzρA(z, s), (2.22)

where ρA is the mass number density normalized by its mass number and TA(s) is thenumber of nucleons per unit area along with z axis at the point from the center of thenucleus represented by a 2D vector s. The vector s is in the transverse plane with respect


b

Spectator

Participant

Before Collision After Collision

Spectator

Figure 2.10: Participant-spectator model of high energy heavy ion collision. Two collidingnuclei approach with impact parameter b before the collision. After the collision, thespectator nucleons pass away into the longitudinal direction.

to z. Using TA(s) and b, the nuclear overlap function of nucleus A and B is defined asfollow:

TAB(b) =

∫d2s TA(s) TB(s − b), (2.23)

where d2s = 2πsds is the 2 dimensional area element. Eq. 2.23 means that the nucleonin nucleus A has a chance of a collision with the nucleon in nucleus B at the same 2dimensional position. Therefore, Ncoll is defined as follows:

Ncoll(b) =

∫d2s σin

NN TA(b)TB(s − b) = σinNN · TAB(b), (2.24)

where σinNN is the inelastic cross section of the nucleon-nucleon interaction.

In order to evaluate these geometrical parameters described above, Monte Carlo calcu-lations of the Glauber model are performed. The Woods-Saxon parameterization is usedfor a realistic nuclear mass density ρA(r) as follows:

ρA(r) =ρ0

1 + exp ( r−RA

a), (2.25)

where RA is the radius of the nucleus and a is the diffusion parameter. When the distancebetween the nucleons becomes less than

√σin

NN/π in the calculation, we consider a collisionbetween these two nucleons take places. The result of the Glauber model calculation issummarized in section 4.2.3.

2.4. HEAVY QUARKS IN NUCLEUS-NUCLEUS COLLISIONS 23

2.4 Heavy Quarks in Nucleus-Nucleus Collisions

2.4.1 Initial Production

The production of heavy quarks in p+ p collisions can be calculated in pQCD formal-ism, since their large masses requires hard-scatterings with large momentum transfer, asdiscussed in the previous section. Heavy quarks are produced via initial hard-scatteringsin Nucleus-Nucleus (A + A) collisions. Therefore, it is expected that the yield of heavyquarks is proportional to the number of nucleon-nucleon hard-scatterings. This can beexpressed in terms of the nuclear overlap function, TAA:

NHQAA = TAA · σHQ

pp , (2.26)

whereNHQAA is the invariant yield of heavy quarks in A+A collisions and σHQ

pp is the invariantcross section in p+ p collisions. Here, we neglect the initial and final state effects in A+Acollisions. These initial and final state effects can modify the heavy quark production andare discussed in the next sections.

2.4.2 Pre-equilibrium Production

In addition to the initial production, it is predicted [18] that there are the pre-equilibriumproduction and the thermal production.

In the pre-equilibrium stage of the space-time evolution, secondary parton scatteringsin the dense matter could lead to heavy quark production. The yield of heavy quarksproduced in the pre-equilibrium stage could be approximately proportional to the ther-malization time of the dense matter. If the pre-equilibrium production is separated fromthe initial production, we can measure the thermalization time.

In the thermal stage, the heavy quark production is treated in the same way as thethermal production of light quarks (u, d, s). The heavy quark in the thermal stage isproduced through gluon fusion gg → QQ and quark-antiquark annihilation qq → QQ.However their large mass suppress the thermal production of heavy quarks relative to thatof light quarks. If a reasonable temperature is assumed (T ∼ 200−400 MeV), the thermalproduction is far below the initial production.

Figure 2.11 shows calculated pT distributions of charm quarks produced in the initial,pre-equilibrium and thermal stage in central Au + Au collisions at

√sNN = 200 GeV [65].

The pre-equilibrium production is about 10% compared with the initial production. Onthe other hand, the thermal production is negligibly small.

2.4.3 Cold Nuclear Effects

There are some known effects in normal nuclear matter which can affect the yield and pT

spectra of produced particles. They are generally called as “cold nuclear matter effect”or “initial state effect”. In order to study the property of the matter created in the highenergy heavy ion collisions, it is necessary to take these nuclear effects into account. Thecold nuclear effects, “Cronin effect” and “nuclear shadowing effect”, are described in thissection.


initial fusion

pre-thermalthermalideal thermal

pT (GeV/c)

Au+Au(b=0)s1/2=200 GeV

dN/d

yd2 p T

(y=

0) (

GeV

-2c2 )

10-9

10-8

10-7

10-6

10-5

10-4

10-3

10-2

10-1

1

0 1 2 3 4 5 6 7 8

Figure 2.11: A theoretical calculation of charm quark production as a function of pT

in central Au + Au collisions at√sNN = 200 GeV. The initial (solid), pre-equilibrium

(dot-dashed) and thermal production (dashed) are plotted, respectively [65].

Cronin Effect

Cronin effect [66] is an enhancement of the particle production in proton + nucleus (p +A) collisions compared to the number of nucleon-nucleon collisions scaling (Ncoll). Theenhancement is explained that the incident parton undergoes the multiple scatteringsin the target nucleus before hard-collision, so that its momentum become broad [67].Therefore, the Cronin enhancement depends on the nuclear thickness or the number ofcollision. The cross section in p+ A collisions is expressed as follows:

Ed3σ

d3p(pT, A) = E

d3σ

d3p(pT, p = 1) × Aα(pT), (2.27)

where α(pT) represents a parameter of the modification compared to Ncoll scaling. If theα(pT) = 1, the cross section in (p+A) is consistent with Ncoll scaling. From the identifiedcharged hadron measurement in PHENIX experiment, the α(pT) > 1 is observed forpT > 1.5 GeV/c in d+ Au collisions at

√sNN = 200 GeV [68].

Nuclear Shadowing Effect

The fact that the structure function F2 per nucleon in nucleus is significantly differentfrom that of a free nucleon is a well known nuclear effect. This was discovered in themeasurement of a deep inelastic muon scattering by the European Muon Collaboration(EMC) in 1982 [70]. The modification of the nuclear structure function is illustrated bythe ratio of FA

2 and FD2 as a function of Bjorken x shown in Fig 2.12 [71]. The ratio can be

subdivided by four regions with respect to x. For x < 0.05, “shadowing” where the ratio


Figure 2.12: Modification of nuclear structure function per nucleon. (top) a phenomeno-logical curve. (bottom) a set of experimental results [71].

is smaller than unity. For 0.05 < x < 0.15, “anti-shadowing” where the ratio is slightlylarger than unity. For 0.15 < x < 0.6, the EMC effect where the ratio is a fall increasingx. For 0.6 < x, the ratio rises by Fermi motion.

This modification of the structure function affects the particle production in the colli-sion. The relevant x region in the RHIC energy can be estimated using Eq. 2.28 and pT

of the produced parton in the mid-rapidity.

x =2 pT√s

(2.28)

For the parton with 1 < pT < 10, the x region corresponds to 0.01 < x < 0.1 where theshadowing effect can appear.

The EKS98 model [73] parameterizes the nuclear modification of parton distribution innucleus. A cross section of heavy quarks in p+A and A+A collisions is calculated usingthe EKS98 model [72]. Figure 2.13 shows the ratio of the cross-sections for charm (left)and bottom (right) at mid-rapidity in p + A and A + A collisions. At the RHIC energy,as shown in Figure 2.13, the charm production is not much modified since the nuclear


(GeV)s10

210

310

4

cro

ss-s

ecti

on

rati

oc

c

0.6

0.7

0.8

0.9

1

1.1

1.2 Pb-Pb/pp

p-Pb/pp

EKS 98

SPS

FNAL RHIC

LHC

(GeV)s10

210

310

4

cro

ss-s

ecti

on

rati

ob

b

0.6

0.7

0.8

0.9

1

1.1

1.2 Pb-Pb/pp

p-Pb/pp

EKS 98SPS

FNAL

HERA-B

RHIC

LHC

Figure 2.13: The ratio of the cross section for the cc (left) and the bb (right) in p+A andA+A collisions. The nuclear effect of the PDF’s are calculated using the EKS98 nuclearweight functions [72].

shadowing at the x region relevant for charm production in the EKS98 model is small. Bycontrast, the bottom production is slightly enhanced by the anti-shadowing effect.

2.4.4 Final State Effects in QGP

Medium Modification

When an energetic parton passes through the dense matter created in heavy ion collisions,the parton suffers the energy loss due to parton-matter interaction, e.g. gluon radiation.We can see a similar process in electromagnetics. A high-energy charged particle traversingcharged matter lose its energy by “bremsstrahlung” which is photon radiation due tointeraction of the particle with the matter. The bremsstrahlung is described using theelectromagnetic potential from the moving charged particle, so called Lienard−Wiechertpotential.

It is predicted that the radiative energy loss of heavy quarks is smaller than that of lightquarks, since its larger mass reduces the available phase space for gluon radiation (deadcone effect [19]). A similar mass dependence of energy loss can be seen in electromagnetics.The energy loss of muons by bremsstrahlung is much smaller than that of electrons.

There are two theoretical calculations to treat the energy loss process quantitativelyfor light and heavy quarks. One is the BDMPS model [75] in which the radiated gluonssuffer the final state interaction (non-abelian effects) in the colored-dense matter. Thetotal energy loss in the BDMPS model is described in terms of transport coefficient q andis proportional to L2 as follows:

δE ∝ qL2

(q ≡ 〈q2

T〉λ

), (2.29)


where q means the average squared transverse momentum (qT) transfered from the matterto the parton per mean free path (λ). The L is the path length of the parton in the matter.

The other is the GLV (DGLV) model [76, 77, 78] in which the radiative energy loss ischaracterized by the gluon density (dNg/dy) in the dense matter. The energy loss in theGLV model can be analytically calculated for the limit of the thin matter. The fractionalenergy loss of a fast parton with energy E is expressed as follow [79]:

dE

E∝ 1

A⊥

dNg

dyL log

E

µ, (2.30)

where A⊥ is the transverse size of the dense matter and µ is the Debye screening scale.The nuclear modification of PDF and the longitudinal Bjorken expansion of the matter isalso taken into account in the GLV model for the realistic calculation.

Figure 2.14 shows the nuclear modification factor RAA of π0 in top 5% central Au + Aucollisions at

√sNN = 200 GeV [83]. The measured π0RAA is compared to the BDMPS

(left) and the GLV calculation (right) with various values of the model parameters, re-spectively. These models can describe the measured π0RAA with q = 13.2+2.1

−3.2 GeV2/fmfor the BDMPS and dNg/dy = 1400+270

−150 for the GLV models.On the other hand, these models can not describe the measured RAA of electrons from

heavy flavor decays as shown in Fig. 2.15 [84]. The measured electron RAA is smaller thanthese model predictions for high pT where q = 14 GeV in the BDMPS and dNg/dy = 1000in the GLV are used. The data shows the opposite trend compared to the prediction ofthe less energy loss for heavy quarks. This indicates that the radiative energy loss due togluon bremsstrahlung is not adequate to describe fully the measured suppression of heavyflavors, and the additional energy loss mechanism is required.

In recent theoretical studies [80, 81, 82], it is proposed that the contribution from thecollisional energy loss due to the elastic scattering becomes important for heavy quark,although the collisional process is negligibly small for light quark. Figure 2.16 showsthe theoretical calculation of the collisional and radiative energy loss as a function ofmomentum for charm (left) and bottom (right) quarks. The collisional energy loss forboth charm and bottom quarks makes a larger contribution than the radiative loss for lowpT and becomes gradually smaller for high pT. The combined approach of the collisionaland radiative energy loss can describe the substantial part of the heavy quark energy lossshown as the green band in Fig. 2.15. But there is a still small difference between the dataand the model.

2.4.5 Aim of This Study

The motivation of this thesis is to study the initial gluon density of the dense mattercreated in heavy ion collision. Because their large mass, heavy quarks are only producedin point-like hard-scatterings in A+ A collisions. Such hard scatterings take place in theearly time of space-time evolution before the possible QGP formation. Thus, the heavyquark production is sensitive to the initial condition of the collision, especially the initialgluon density.

The first aim of this study is to investigate whether the total production of heavyquarks in Au + Au collisions is scaled by Ncoll, especially charms at low pT. If the Ncoll


)c(GeV/T

p

0 2 4 6 8 10 12 14 16 18 20

AA

R

0

0.1

0.2

0.3

0.4

0.5

0.6 (Au+Au 0-5% Central)0πPHENIX

12%±Global Systematic Uncertainty

)c(GeV/T

p

0 2 4 6 8 10 12 14 16 18 20

AA

R

0

0.1

0.2

0.3

0.4

0.5

0.6 (Au+Au 0-5% Central)0πPHENIX

12%±Global Systematic Uncertainty

Figure 2.14: Both the left and right panel show π0RAA for top 5% central Au + Au colli-sions at

√sNN = 200 GeV [83]. π0RAA are compared with the BDMPS (left) and the GLV

(right) model with various parameters. The red curves indicate the best fit curve of themodels.

Figure 2.15: RAA of electrons from heavy flavor decays for top 10% central Au + Aucollisions at

√sNN = 200 GeV [84]. The electron RAA is compared with the models. The

green dot curve is the BDMPS, and the yellow band is the GLV models. The green bandshows the combined collisional and the radiative energy loss model.


30 0 5 10 15 20 25 30

p GeV

0

0.05

0.1

0.15

0.2

E

E

CHARM

0 5 10 15 20 25 30

p GeV

0

0.02

0.04

0.06

0.08

E

E

BOTTOM

Figure 2.16: The fractional energy loss of the collisional and radiative process as a functionof momentum for charm (left) and bottom (right) quarks [82]. The solid and dot-dashedcurves correspond to the collisional and radiative process, respectively.

scaling of the total charm production can be observed, it will strongly suggest that thehigh pT suppression of charm quarks is caused by the final state effects in QGP, such asenergy losses of partons. The cold nuclear effects can modify both the total yield andthe spectral shape of produced charms. Therefore, the measurement of the total charmproduction provides an important baseline for the high pT suppression.

In addition, this study can put quantitative constraints on the gluon shadowing effect,even though the modification of the charm production due to the shadowing effect isexpected to be small. According to the EKS98 model for the shadowing of nuclear PDF,the gluon density of Pb with Q2 = 5.39 GeV2 is roughly 5% smaller than that of proton atx ∼ 0.02. This means that, if the EKS parameterization is correct, charm production dueto gluon fusions in heavy ion collisions will be 10% smaller than that with Ncoll scaling inp+ p. If the other processes are included, the modification of the charm production dueto the shadowing will be reduced.

Once the cold nuclear effects can be understood, we can study additional productionsof charms in the pre-equilibrium and thermal stage of space-time evolution in heavy ioncollisions. The additional productions are expected to provide some informations relatedto the thermalization time and the temperature of the dense matter. Therefore, thesystematic measurement of the charm productions in p+ p, d+Au and Au + Au collisionsis necessary to study these effects due to the initial production, the additional production,the cold nuclear effects and the final state effects.

As a first step of this important study, we measured the centrality dependence ofthe total charm productions from single electron measurement in Au + Au collisions at√sNN = 200 GeV. These result are compared with p+ p and d+ Au data. The measure-

ments and the results are described in the next sections.

Chapter 3

Experimental Setup

3.1 Relativistic Heavy Ion Collider

The Relativistic Heavy-Ion Collider (RHIC) at Brookhaven National Laboratory (BNL)was constructed to study the characteristics of a new phase of sub-atomic matter createdin heavy ion collisions. A collision energy at RHIC is one order of magnitude higher thanthe AGS at BNL and the SPS at CERN [85]. RHIC can accelerate a variety of nuclearbeams from protons (p) to light ions to gold ions (Au) with its maximum energy up to250 GeV for p and 100 GeV per nucleon for Au. Figure 3.1 shows a schematic view of theaccelerator complex. Protons and Au ions are initially produced and accelerated by theLinac and the Tandem Van de Graaff pre-accelerator, respectively. The produced beam issent to the Booster Synchrotron followed by the Alternating Gradient Synchrotron (AGS).AGS then accelerates the beam upto 28 GeV for p and 10.8 GeV for Au per nucleon, andinjected into the RHIC rings. The beams are finally accelerated up to its maximum energy.

RHIC has two individual accelerator rings with 3.83 km circumference. The two ringsintersect at six intersection points. There were initially four experiments, PHENIX, STAR,BRAHMS and PHOBOS placed at the intersection points of RHIC, and they are designedto study the various aspects of dense matter created in heavy ion collisions and the spinstructure of nucleon using polarized p beams. RHIC had started the operation from year2000 and it is continuing its operation. All four experiments has started taking data at thestart of RHIC. BRAHMS and PHOBOS had stopped data taking in year 2006. PHENIXand STAR is continuing the operation . During that period, RHIC had collided variousnuclear species with various energies. These are summarized in Tab. 3.1.

3.2 PHENIX Detector Overview

The Pioneering High Energy Nuclear Interaction eXperiment (PHENIX) is designed tomeasure several fundamental probes of high-energy nuclear interactions [86]. In orderto measure photons, leptons and hadrons simultaneously, and to separate leptons frommuch larger number of hadrons, the PHENIX detector has a good particle identificationcapability in broad pT range with good momentum and energy resolution.

The PHENIX detector is organized by three parts of detector systems: the global

30

3.2. PHENIX DETECTOR OVERVIEW 31

Figure 3.1: The RHIC accelerator complex.

Table 3.1: Summary of RHIC operation. The integrated luminosity is recorded inPHENIX.

Run Species√sNN

∫Ldt year

1 Au + Au 130 1 µb−1 2000

2 Au + Au 200 24 µb−1 2001/02p+ p 200 0.15 pb−1

3 d+ Au 200 2.74 nb−1 2002/03p+ p 200 0.35 pb−1

4 Au + Au 200 241 µb−1 2003/04Au + Au 62.4 9 µb−1

5 Cu + Cu 200 3 nb−1 2005Cu + Cu 62.4 0.19 nb−1

Cu + Cu 22.4 2.7 µb−1

p+ p 200 3.8 pb−1

6 p+ p 200 10.7 pb−1 2006p+ p 62.4 0.1 pb−1

7 Au + Au 200 813 µb−1 2007

8 d+ Au 200 80 nb−1 2007/08p+ p 200 5.2 pb−1

9 p+ p 500 14 nb−1 2008/09p+ p 200 16 pb−1

32 CHAPTER 3. EXPERIMENTAL SETUP

Figure 3.2: The cutaway drawing of the PHENIX detector.

detectors, the central arm spectrometers and the muon arm spectrometers. Figure 3.2shows a bird’s eye view of the PHENIX detector.

The global detector consists of a pair of the beam-beam counters (BBC) and a pairof the zero-degree calorimeters (ZDC). They produce a minimum bias trigger and give anevent characterization of the heavy ion collisions.

The central arm spectrometer measures electrons, photons and charged hadrons atcentral rapidity region. The magnetic field for the central arm spectrometer is supplied bythe central magnet which provides an axial field parallel to the beam axis. The central armspectrometer is constructed by two almost symmetrical spectrometer arms, named as theWest arm and the East arm. The West arm and the East arm covers the central rapidityregion (pseude-rapidity, |η| ≤ 0.35) and a quarter of full azimuthal angle (∆φ = π/2). Thecentral arms consist of several detector subsystems for charged particle tracking, particleidentification and calorimetry. Drift chambers (DC) and a layer of pad chambers (PC) areplaced at the most-inner of the central arm and measure charged particle trajectories withgood momentum resolution. The time projection chamber (TEC) in the East arm providesadditional particle tracking and identification. The time-of-flight counters (TOF) in theEast arm identifies charged hadrons up to 2.4 GeV/c for π/K and 4 GeV/c for p. The ringimaging Cerenkov detectors (RICH) have a good capability of electron separation fromlarge number of produced hadrons. The electro-magnetic calorimeter (EMCal) is placedat the most outside of the central arm and measures total energy of photons and electrons.

3.3. GLOBAL DETECTOR 33

Figure 3.3 (top) shows the schematic view of the central arm spectrometer from the beamaxis.

The muon arm spectrometers are instrumented to measure muons and consist of apair of the spectrometer arm, named as the North arm and the South arm. They coverthe forward and the backward pseudo rapidity (1.15 ≤ |η| ≤ 2.25 for the South arm and1.15 ≤ |η| ≤ 2.44 for the North arm) and the full azimuthal angle (∆φ = 2π). Eachof the muon spectrometer arms consists of a muon tracker (MuTr) followed by a muonidentifier (MuID). Figure 3.3 (bottom) shows a side view of the PHENIX detector. Whenthe PHENIX detector is seen from the East arm side, the South and North muon armspectrometers are placed in the left and right side of the central magnet, respectively.

In order to measure both soft and hard/rare probes such as J/ψ → e+e− process andhigh pT particle production, the PHENIX detector has the first (LVL1) and the second(LVL2) level trigger and a very fast data acquisition system (DAQ).

We give the detailed description of these detector subsystems and DAQ systems in thenext section.

3.3 Global Detector

3.3.1 Beam Beam Counter

The beam beam counter (BBC) is a pair of identical Cerenkov detector arrays [87]. TheBBC is placed ±144 cm far from the center of the PHENIX detector along with thebeam axis. The BBC covers 3.1 < |η| < 3.9 and full azimuthal angle (∆φ = 2π). ABBC array (Figure 3.4 (a)) is composed of 64 Cerenkov detector elements. One element(Figure 3.4 (b)) consists of a hexagonal-cylinder shaped quartz Cerenkov radiator and a 15stage mesh-dynode type photomultiplier tube (Hamamatsu R6178). The BBC is requiredto work in a strong magnetic field (3 kG).

The BBC plays several important roles in the measurement of the collisions. The BBCprovides the minimum bias trigger for level-1 trigger. The BBC is used to determine thecollision vertex along with the beam axis (z vertex) and the centrality for event char-acterization. In addition, the BBC measures the collision timing for the time-of-flightmeasurement.

The BBC measures the arrival time of charged particles emitted from the collisions.The arrival times (TS and TN for the south and the north side of the BBC) are determinedto calculate a truncated average of hit timings in individual BBC elements with σ =40 ± 0.5 ps of timing resolution. The vertex position (z) and the start timing (t0) of acollision are determined using TS and TN :

z =c

2× (TS − TN) (3.1)

t0 =1

2× (TS + TN − L

c) (3.2)

Here, L(= 144 cm) is a known distance from the center of PHENIX to the BBC locationalong with beam axis and c is the speed of light. The z information is used for bothoffline analysis, and level-1 trigger (LVL1). The combination of z position and more than


West

South Side View

Beam View

PHENIX Detector

North

East

BB

MuTr

MuID MuID

ZDC NorthZDC South

MVD

BB

MVD

PbSc PbSc

PbSc PbSc

PbSc PbGl

PbSc PbGl

TOF

PC1 PC1

PC3

PC2

Central Magnet

CentralMagnet

NorthMuon MagnetSouth M

uon Magnet

TEC

PC3

RICH RICH

DC DC

Figure 3.3: The composition of PHENIX detector in Run 2 operation: (top) the beamview of the central arm. (bottom) the side view of the PHENIX detector.


(a) (b)

Figure 3.4: (a) A BBC array comprising 64 BBC elements. (b) A BBC element.

Figure 3.5: The z vertex distribution. The shaded area corresponds to the events triggeredby LVL1.

2 hit PMT’s is required to produce LVL1. Figure 3.5 shows z vertex distribution. Theshaded area in Fig. 3.5 corresponds to the event selected by LVL1 condition. The positionresolution of z vertex is σz = 0.7 mm, and this corresponds to 23 ps of t0 resolution [88].In Fig. 3.5, The peak around z = ±144 cm corresponds collisions outside of the BBCs,e.g. beam-gas collisions.

Each of BBC element is calibrated using a MIP peak. The pulse height informationmeasured in all BBC elements is summed up to measure total charged particle multiplicityin BBC acceptance. This multiplicity information combined with the ZDC is used fordetermination of collision centrality as described in section 4.2.2.

3.3.2 Zero Degree Calorimeter

The zero degree calorimeter (ZDC) [89, 90] is a pair of hadron calorimeters which areplaced at 18 m up and downstream of the center of the PHENIX along with the beam


Figure 3.6: (top) The location of the ZDC. (bottom) A side view of a ZDC module.

axis and covers |θ| < 2 mrad with respect to the beam axis shown in top of Fig. 3.6.The ZDC consists of three identical modules. One module is a sandwich type samplingcalorimeter with a photomultiplier tube read-out (Hamamatsu R329-2) consisting of 5 mmthick tungsten absorber layers interleaved with optical fiber (0.5 mmφ) read-out layers.The total thickness of a ZDC module is 2 interaction length (ΛI). A ZDC module samplesCerenkov lights emitted from secondary charged particles in the fiber layers. Figure 3.6(bottom) shows a side view of one ZDC module.

The ZDC measures spectator neutrons emitted from heavy ion collisions at RHIC. Sincethe ZDC is located behind the beam bending magnets (DX) shown in Figure 3.6 (top),all charged particles are swept out by the DX magnets. The energy resolution at singleneutron peak is approximately 21% [91], where the single neutron peak is measured in lowmultiplicity events. A coincidence hit of both the ZDC gives a minimum bias selectionof the collisions. The neutron multiplicity calculated from the total energy in the ZDC iscorrelated with the event geometry. Therefore, the ZDC multiplicity combined to the BBCare used to determine collision centralities. The ZDC also measures neutrons from mutualCoulomb dissociations in heavy-ion collisions. This is used for luminosity monitoring. Thecross section of the Coulomb dissociation in Au + Au collisions at

√sNN = 130 GeV is

measured to be σMCD = 3.67 ± 0.26 barns [91].


Figure 3.7: The schematic view of the MVD

3.3.3 Multiplicity Vertex Detector

Multiplicity Vertex Detector (MVD) [87] is a array of silicon (Si) detectors comprising abarrel strip detector and a pair of end-cap pad detectors. The barrel detector consists oftwo layers of Si strip elements (300 µm thick and 200 µm pitch). The inner and outerlayer cover 2π in azimuth and ±32 cm in z-position. The bottom part of the outer layer isonly installed to reduce the amount of material in the central arm acceptance, since mostof the background electrons is from photon conversions.

MVD is placed at the center of the PHENIX detector with |z| < ±35 cm along withthe beam pipe. The radius of the MVD outer shell is about 30 cm. Figure 3.7 shows aschematic view of the MVD.

The purpose of the MVD is to measure charged particle multiplicity and to determinea position of collision vertex. The MVD information is not used in this analysis. But thecomposition of the MVD needs to be described in detail for electron analysis. The amountof material in the MVD is approximately 1.1% radiation length (X0), and it is a majorsource of background electrons from photon conversions. During the “converter run”, aphoton converter described in section 3.6 was installed wrapping around the MVD outershell. The detailed list of the MVD material is summarized in Tab. 3.2. The MVD wasinstalled for the first three years of the PHENIX operation. After these periods, the MVDwas removed. Therefore, background electrons from photon conversions are significantlyreduced in Run 4 analysis compared to Run 2 analysis.


Table 3.2: Summary of the MVD composition and their radiation lengths

Description MaterialThickness

(mm)Rad. Length

(mm)Thickness X0

(g/cm2)

Beam Pipe Be 1.016 353 0.29 %MVD

Inner Shell Al 0.01 89 0.01 %adhesive 0.0508 254 0.02 %Roha cell 3 5000 0.06 %adhesive 0.0508 0.02 %

Al 0.01 0.01 %Inner Cell Si 0.3 94 0.32 %

Rohacell Cage Rohacell 24.4 0.11 %Outer Si Si 0.3 0.32 %

Kapton Cable Kapton 0.05 250 0.02 %Cu 0.005 14 0.02 %Cu 0.005 0.02 %

Outer Shell Al 0.01 0.01 %adhesive 0.0508 0.02 %Rohacell 6 0.11 %adhesive 0.0508 0.02 %

Al 0.01 0.01 %

Air before DCH Air 1980 304200 0.65 %Ar Ar 252.5 109708.2 0.23 %

3.4. CENTRAL ARM SPECTROMETER 39

Figure 3.8: A cutaway view of the PHENIX magnets. The line shows the contour of themagnetic field.

3.4 Central Arm Spectrometer

3.4.1 Central Magnet

The central magnet [92] is energized by two pairs of concentric coils and provides a axiallysymmetric magnetic field parallel to the beam pipe. The field strength at R = 0 cm is 0.5T in Run 2 (0.9 T in Run 4) and gradually decrease with approximately Gaussian profile.Then, the field reaches almost zero for region R > 250 cm. Therefore, electrons can passthrough the RICH with a straight line, because there is no magnetic field in the RICH.Figure 3.8 shows a cutaway view of the PHENIX magnet systems with the field lines ofthe central and the muon magnet.

3.4.2 Drift Chambers

The Drift Chambers (DC) are placed at the front of both the East and the West armswith a radius of 2.0 to 2.4 m far from the beam axis [93]. Each DC has a cylindrical shapewith |z| < 90 cm and π/2 azimuthal angle (φ) coverage shown in Fig. 3.9 (top).

The DC is supported by a titanium frame and filled with 20 equal sectors covering 4.5degrees in azimuth. There are 6 types of wire modules stacked radially in each sector.They are called X1, U1, V 1, X2, U2 and V 2. Each module contains 4 sense planes and4 cathode planes forming wire cells with a 2 − 2.5 cm drift space in the φ direction. TheX1 and X2 wire cells run in parallel to the beam axis to measure particle trajectories in


R = 2.02 m

90o

R = 2.46 m2.

5 m

Ti frame

mylar window

Figure 3.9: (top) A schematic view of the DC with 20 DC sectors. (bottom) A picture ofthe DC covered by the PC.


cathode plane

X1 − planes

U1 − planes

V1 − planes

X2 − planes

U2 − planes

anode plane

wire

sense (anode)

Potential wire

Gate wires

Back wire

Termination wires

V2 − planes

degr.4.5

~ 18

00 m

m

VXU

Sector, side view

Wire orientations, top view

Figure 3.10: (left)The layout of X and U ,V stereo wires in a sector. (right) The schematicdiagram of X, U and V orientation.

r − φ. These wire cells are followed by two set of small angle U , V wire planes. The U1,V 1, U2 and V 2 wires have stereo angles of approximately 6◦ relative to the X wires andmeasure the z coordinate of the tracks. The X, U and V cells contain 12, 4 and 4 sensewires, respectively. As a result, there are 40 sense wires in each cell. Figure 3.10 showsthe layout of sense planes and sense wire cells in a sector. The U and V stereo wires startin a sector on one side and end in a neighboring sector on the other side of the DC.

The DC measures the charged particle trajectories in the r−φ plane to determine theirmomentum with good resolution. It is required that the DC recognizes the track for thehighest multiplicity. To satisfy the requirement, the DC is designed to have single wireresolution better than 150 µm in r − φ, single wire track separation better than 1.5 mm,single wire efficiency better than 99%, and spatial resolution in z axis better than 2 mm.Moreover, to reduce the single wire occupancy, the wires are electrically isolated into twohalves at the middle of the DC (z = 0) using kapton of 100 µm thickness. Therefore, theDC system has 6400 sense wire (40/2 wires/cell * 80 cells/side * 2 sides/arm * 2 arms =6400 wires). This corresponds to 12800 readout channels The DC operates with a mixtureof 50% Ar, 50% ethane gas and small fraction of alcohol. Figure 3.9 (b) shows a pictureof the DC covered by PC.

3.4.3 Pad Chambers

The Pad Chambers (PC) consist of three individual layers of multi-wire proportionalchambers [93]. The first inner layer of the PC (PC1) are placed between the DC and theRICH in both the East and the West arms. The second layer (PC2) is located behind theRICH in the West arm only. The third layer (PC3) is mounted in front of the EMCal.


1.5mm2.7mm

8.4mm

Anode wire

Field wire

8.2mm

Center pixel

Side pixel

Figure 3.11: (left) The pad and pixel geometry. (right) A cell defined by three pixels.

The installed radii of these three PC layers correspond to 248, 419 and 490 cm from thebeam pipe, respectively. A layout of three PC layers in both the East and the West armscan be seen in Fig. 3.3. Each PC contains a single wire plane inside a gas volume boundedby two cathode planes. One cathode is finely segmented into an array of pixels to readout the induced charge through specially designed electronics. The PC operates with agas mixture of 50% argon and 50% ethane.

The PCs are the only non-projective detectors in the central arm tracking system andprovide three-dimensional space-point of charged particles with high efficiency. The PC1is also used to determine z-coordinate of the track measured by the DC. The PC2 andPC3 are needed to recognize particle trajectories, because many background tracks areproduced by particle decays and particle-detector interactions outside the DC.

A special pad design was invented to archive a high pixel granularity and small readoutchannels. Figure 3.11 (left) shows the pixel configuration of a pad. The nine by nineinterleaved pixels are grouped together to form single readout channel. Each cell recognizesa valid hit and contains three pixels which are connected to different but neighboring padsshown in Fig. 3.11 (right). So, the hit information can be broken down to the cell level byidentifying the triplets of the pads. The cell size is 8.4 mm for the PC1 and PC2 , and istwice for the PC3, since the PC3 is twice far from the beam pipe compared to the PC1.The spacial resolutions of the PC1 and PC 3 in z direction is ±1.7 mm and ±3.6 mm,respectively.

3.4.4 Ring Imaging Cerenkov Detector

The Ring Imaging Cerenkov detector (RICH) [94, 95] is a gas Cerenkov detector and isa primary device for identifying electrons among the very large number of charged pions.The RICH is placed behind the PC1 in both the East and the West arms. Figure 3.3shows the RICH location.

The RICH consists of a 40m3 gas vessel with an entrance and exit windows of thealuminized Kapton, two intersecting spherical mirrors and two arrays of 16 × 80 photomultiplier tubes (PMT). A PMT array is placed behind the central magnet so that particlesfrom the collision do not directly hit the PMT array. Each PMT (Hamamatsu H3171S)is housed in a magnetic shield. A Winston cone is attached at the entrance of the shield


Figure 3.12: A cutaway view of the RICH gas vessel with the spherical mirrors.

with 50 mm diameter to collect Cerenkov photon efficiently. Figure 3.12 shows a cutawayview of the RICH.

An electron passing through the RICH emits Cerenkov lights. The spherical mirrorreflects the Cerenkov photons and focuses on the PMT array forming a ring-shaped pat-tern. The diameter of the Cerenkov ring which corresponds the emission angle of theCerenkov photon is defined by a type of a gas radiator. The vessel is filled with CO2 gasat 1 atm as Cerenkov radiator. This corresponds to Cerenkov threshold of 4.9 GeV/c forpion. Figure 3.13 shows the top view of the RICH. The Cerenkov photons emitted by anelectron are reflected to the PMT arrays, although a charged hadron passes away withoutCerenkov light emissions.

3.4.5 Time Of Flight Counters

The Time-Of-Flight counters (TOF) [94] play an important role for particle identificationof charged hadrons. The TOF is designed to provide π/K separation at momenta up to2.4 GeV/c and K/p up to 4.0 GeV/c within 4 σ.

The TOF is placed at 5.1 m far from the beam pipe in the East arm and covers therange of 70 − 110◦ in poler angle (θ) and 30◦ in azimuth (φ). The location of the TOFis shown in Fig. 3.3. The TOF consists of 10 panels which is composed of 96 plasticscintillator slats with read-out PMT (Hamamatsu R3478S) attached at both ends of theslat. Thus, 960 slats of the scintillators and 1920 channels of PMTs are totally installed.A slat is 1.5 × 1.5 cm2 in width and depth and is aligned to r − φ direction. Figure 3.14shows the schematic view of one panel in which 96 slats are arranged.

The TOF measures an arrival time (stop time) by taking an average time at both endof PMTs. This time-of-flight of a charged particles is determined by using the arrival timeand the start time provided by the BBC. The particle identification is performed using


Cerenkov photons from e+ or e- are detected by array of PMTs

mirrorMost hadrons do not emit Cerenkov light

Electrons emit Cerenkov photonsin RICH.

Central Magnet

ＲＩＣＨ

PMT arrayPMT array

Figure 3.13: The top view of the RICH. An electron emits Cerenkov photons.

the squared mass of the measured particle expressed as:

m2 =p2

c2

[(tTOF

L/c

)2

− 1

], (3.3)

where p is the momentum, tTOF is the time of flight, L is the flight path length and c isthe speed of light. The time resolution for the time-of-flight measurement is archived tobe σ ' 120 ps [105].

3.4.6 Electromagnetic Calorimeter

The electromagnetic calorimeter (EMCal) [96] is used to measure the spatial position andenergy of electrons and photons produced in heavy ion collisions. The EMCal is locatedat the most outside of the central arm. The EMCal consists of two different detectorsystems. One is a Pb-Scintillator calorimeter (PbSc) which provides 4 and 2 sectors forthe West and East arms, and the other is Pb-Glass calorimeter (PbGl) which occupies 2lower sectors of the East arm. The location of the PbSc and the PbGl calorimeters areshown in Fig. 3.3.

The PbSc is a shashlik type sampling calorimeter made of alternating tiles of Pb andscintillator and consists 15552 individual towers. Each PbSc tower contains 66 samplingcells (55 × 55 × 375 mm3) and has a thickness of 18 radiation length (X0) and about30 mm of Moliere radius. These cells are optically connected 36 longitudinally penetrating


Scintillator

PMT

µ-metal

Light guide

Ho

ney

com

b s

tru

ctu

red

pap

erC

arb

on

fib

er s

hee

t

200

cm

50 cm

Bleeder

Alu

min

ium

fra

me

Figure 3.14: Schematic view of the single TOF panel which is composed of the 96 scintil-lator slats.

wavelength shifting fibers for light collection. Light is readout by 30 mm PMT (FEU115M)at the back of the towers. A single basic structure is called as a module which consistsfrom 4 towers. Figure 3.15 shows a PbSc module.

On the other hand, the PbGl is Cerenkov calorimeter which consists 9216 lead glassmodules with PMT readout (FEU84). Each PbGl module is 40 × 40 × 400 mm3 and has14.4 X0 and 36 mm of Moliere radius. The PbGl has been used previously in CERN WA98experiment [97]. Figure 3.16 shows an unit array of the PbGl modules.

The EMCal has a important role in particle identification using the informations ofmeasured energy, time and shower shape of electromagnetic cascade. The energy resolu-tions of the PbSc and the PbGl are obtained using electron and positron beam at a testexperiment for the PbSc and the PbGl respectively:

σE

E

∣∣∣∣PbSc

=8.1%√E(GeV)

⊕ 2.1%,

σE

E

∣∣∣∣PbGl

=5.9%√E(GeV)

⊕ 0.8%.

The position resolutions are also obtained:

σPbScx (E, θ) =

(1.55(mm) ⊕ 5.7(mm)√

E(GeV)

)⊕ ∆ × sin(θ),

σPbGlx (E) = 0.2(mm) ⊕ 8.4(mm)√

E(GeV),

where σPbScx depends on both energy and impact angle (θ) of incident particle and ∆ is

given by radiation length.


Wavelengthshifting fibersModule

Towers

Layers of lead andscintillator tiles(sampling cells)

Front

Back

Phototubesattached here

Figure 3.15: The PbSc module consisting 4 towers.

Timing information is used for both particle identification and pattern recognition tofind overlapped clusters. In particular, the timing information is a good tool to separateneutral baryons from photons. The time resolutions of the PbSc are obtained as 120 and400 ps for electrons and charged pions, respectively. On the other hand, time measurementis difficult for the PbGl, since the arrival time of hadrons are faster than the Cerenkovlight of electro-magnetic shower produced near the front of the module.

3.5 Data Acquisition System

The PHENIX detector is designed to measure a wide range of observables such as softhadron production and rare probes such as photons, single leptons and lepton pairs on avariety of colliding systems from p+ p to Au + Au. The number of tracks passed throughthe detector varies from only a few tracks in p+ p to several hundred tracks in centralAu + Au collisions. The expected interaction rate also varies from a few kHz from centralAu + Au to approximately 500 kHz for p+ p collisions. To handle a variety of event sizeand rate seamlessly, the PHENIX Data Acquisition system (DAQ) [98] consists of thedetector front ends equipped with the pipelined and dead-timeless features and higherlevel trigger systems. Figure 3.17 shows a general schematic diagram of the PHENIXDAQ.

The RHIC delivers the 9.4 MHz fundamental clock which corresponds to 106 ns oftime interval between beam crossings. All of PHENIX Front End Electronics modules aresynchronized to the RHIC clock. Signals produced in the various subsystems are processedby Front End Modules (FEMs) that converts the detector analog signals into digital data.The FEM for each subsystem is placed in the PHENIX Intersection Region (IR). A FEM

3.5. DATA ACQUISITION SYSTEM 47

photodiode withpreamplifier

reflective cover

LED board

lead glass matrix withcarbon fibre/epoxy

steel plates

mirror foil

photomultiplierwith housing

Figure 3.16: An unit array of the PbGl modules.

RHIC clocks

20MB/s

250TB/yr

~350000 ch

ATM Switch

MTM

SEB

DCMFEM

Linux Farm

~600 CPU's

Online

Monitoring

ZD

C S

outh

ZD

C N

orth

A S

ubsyst

em

(e.g

. Eas

t DC

)

Data packets

over fiber

Clock, Trigger,

Mode bits

over fiber

Sh

ield W

all

a few

miles la

ter

DCB

PartitionerData Packets

Dat

a Pac

kets

over

NEV

IS T

oken

passing

Bus

over Parallel

FIFO Interface

GTM

GL1

L1

Bu

sy

Bu

sy

Trigger

tape

storage

1.2PB

Clock

Interaction regionCountingroom

RHIC

Computing

Facility

GL1 Global Level 1 Trigger

GTM Granual Timing Module

MTM Master Timing Module

DCB Data Collection Board

DCM Data Collection Module

FEM Front End Module

SEB Sub Event Buffer

ATP Assembly Trigger Processor

Missing:

• ARCNET Serial Interface to FEMs

• High and Low Voltage Control and

associated control systems

• Alarm system

'trigger' is a placeholder

for a much more complex

'Local Level 1' trigger system

SEBSEBSEB

ATP

End

User

Figure 3.17: Schematic diagram of the PHENIX DAQ.


PHENIX Timing System

RHIC Clock MTM

GTM

GTM Fanout

FEM

Rev Counter

RHIC Clock

GL1

Level 1 Accept

Clk, Rev, L1

Legend:

MTM: Master Timing Module

GTM: Granule Timing Module

FEM: Front End Module

GL1: Global Level 1 Trigger

Copper

Fiber

Figure 3.18: Block diagram of the PHENIX timing system.

module can buffer the data of up to 40 beam crossings to wait for the LVL1 decisionsand to wait for digitization and readout of triggered event. We have two types of datacollection methods in the FEM. The first type is that the signal is digitized in real time inevery clock period. The BBC, ZDC, DC and PC use the method. The second type is thatthe signal is stored in analog form in Analog Memory Units (AMU) and is only digitizedafter the event is accepted by the LVL1 trigger. The RICH, EMCal and MVD adoptedthe second method.

All the FEM’s are synchronized by the master clock produced in the PHENIX timingsystem. Figure 3.18 shows a block diagram of the timing system. The RHIC clock isreceived by the Master Timing Module (module). The MTM distributes the master clockby fanning out the RHIC clock, where the internal phase locked loop is used to minimize theclock jitter. The master clock is transmitted to the FEM via the Granual Timing Module(GTM). Thus, all the FEM’s are processed in a single clock. The GTM is prepared for aunit of detector elements. In addtion, the GTM send the control commands (mode bits)and trigger decisions.

The purpose of the Level-1 trigger (LVL1) is to select interesting events and provideevent rejection for the limited DAQ rate. The Global Level-1 (GL1) system generates theLVL1 decision based on logical combinations of the Local Level-1 (LL1) decisions. The LL1is generated by individual trigger detector such as the BBC and the ZDC independently.The GL1 can treat totally 128 bits of the LL1 inputs to make the LVL1 decision.

Once the LVL1 decision is generated, the decision signal is sent to the FEM via theGTM, then the data buffered in the FEM are transfered to the Data Collection Module(DCM) located at the PHENIX Counting House. The FEM, GTM and DCM are connectedvia a fiber optic cable to eliminate a large number of noise, cross-talk and grounding

3.5. DATA ACQUISITION SYSTEM 49

ONCS(Archiving)

Data & Control

Fast Control

Slow Control

Level-1

DCM's

Detector

B DCM's

Detector

A DCM's

Detector

C DCM's

Level-1 Sub-

eventBuffer(SEB)

ATM Switch

Detector

D DCM's

DetectorSEB's

Event

BuilderController

ONCS (Control)Assembly/TriggerProcessors

(ATP's)

Figure 3.19: Block diagram of the EVB architecture.

problems. At the maximum LVL1 trigger rate, the whole FEM’s send over 100 Gbytes ofdata per second. The DCM is designed to receive this large amount of data. After receivingthe data, the DCM performs the zero suppression, error checking and data formatting togenerate data packets. The zero suppression is to compress the data by comparing thepreset threshold. In addition, the DCM provides a busy signal which are returned to theGL1 to hold off further triggers.

Many parallel streams of the data packets from the DCMs are sent to the event builder(EVB). The EVB performs the final stage of event assembly. Figure 3.19 shows a blockdiagram of the EVB architecture. In the EVB, the streams of the data packets from theDCMs are first received and buffered in a set of the Sub Event Buffers (SEB). These dataare transfered on request to a set of Assembly/Trigger Processors (ATP) via AsynchronousTransfer Mode switch (ATM). The final events assembled in the ATPs are transmitted tothe PHENIX Online Control System (ONCS) for logging and monitoring processes. Theformat of the final event is called “PHENIX Raw Data Format (PRDF)”. The raw dataare sent to High Performance Storage System (HPSS) at RHIC Computing Facility (RCF)and converted to physical quantities for analysis.

Typical data logging rates for Au + Au collisions increased from 100 Hz in Run 2 and1 kHz Run 4, then finally reached 5 kHz in Run 7.


Figure 3.20: (left) Photon converter installed to wrap around the MVD outer shell inRun 2. (right) Converter placed around the beam pipe in Run 4 because the MVD wasremoved.

3.6 Photon Converter

A photon converter is a key component to separate signal and background electrons in theconverter method described in section 4.10.1. The converter is a thin brass sheet of 1.7%radiation length (X0). In a part of Run 2, the converter was installed wrapping aroundthe MVD outer shell. On the other hand, the converter was placed directly around thebeam pipe in Run 4 since the MVD was removed. Figure 3.20 shows the picture of theconverter in Run 2 (left) and Run 4 (right), respectively. The basic parameters of theconverter is summarized in Tab. 3.3. The radiation length of the converter was preciselydetermined since the size and weight of the converter was measured.

Table 3.3: The basic parameters of the photon converter

Parameters

Length 190.5 (cm)Width 60.96 (cm)Thickness 0.0254 (cm)Weight 2500.405 (g)Area density 0.215313 (g/cm2)Mass density 8.476889 (g/cm3)X0 1.68%

Component (X0 (g/cm2)) Composition

Cu (12.86) 70%(68.5%-71.5%)Zn (12.43) 30%Fe (13.84) < 0.05%Pb (6.37) 0.07%

Chapter 4

Analysis

4.1 Outline

In this chapter, the analysis of heavy flavor electron measurement in Au + Au collisionsat

√sNN = 200 GeV is described. The data used in this analysis is recorded by the

PHENIX detector with a minimum bias trigger (MB) during the Run 2 period of theRHIC operation. The definition of the MB trigger and the event characterization aredescribed in section 4.2. The reconstruction of charged particle tracks and the electronidentification are described in section 4.3 and 4.4.3

Inclusive electrons consists of three components: (1) signal electrons from semi-leptonicdecays of open heavy flavors (namely heavy flavor electrons), (2) “photonic” backgroundelectrons from Dalitz decays of light neutral mesons and photon conversions in the detectormaterials, (3) background electrons from Ke3 decays and two electron decays of vectormesons. In addition, electrons from quarkonium decays and Drell-Yan are also backgroundincluded in the component (3). Here, both the component (1) and (3) are called as“non-photonic” electrons. To separate and extract non-photonic and photonic electrons,“Converter method” is used. The converter method is described in section 4.10.1.

The invariant yield is obtained using the extracted signal of heavy flavor electrons.The invariant yield is a Lorentz invariant form of the yield expressed as:

Ed3N

dp3=

d2N

2πpTdydpT

, (4.1)

where N is the raw yield of heavy flavor electrons per event. To obtain the invariant yield,the correction factors are applied which are the geometrical acceptance, the track recon-struction and the electron identification efficiency. These correction factors are estimatedusing a GEANT [99] based Monte Carlo simulation of the PHENIX detector described insection 4.9. The systematic uncertainties for signal extraction and correction factors aredescribed in section 4.12.

4.2 Event Selection

In this section, the definition of the minimum bias trigger, the determination of the eventcharacterization are described.

51

52 CHAPTER 4. ANALYSIS

4.2.1 Minimum Bias Trigger

In this analysis, the minimum bias (MB) trigger is defined by a coincidence of both theBBC and the ZDC signals. Because the BBC can measure charged particles at pseudo-rapidity 3.1 to 3.9, the BBC is sensitive to both inelastic Au + Au collisions and back-grounds from beam-gas and/or upstream interactions. On the other hand, the ZDC issensitive to the inclusive Au + Au interactions which include not only inelastic collisions,but also mutual Coulomb dissociations. In the case of the mutual Coulomb dissocia-tions, there is almost no signal in the BBC. Therefore, in order to select only the inelasticAu + Au collisions and reduce the backgrounds from bean-gas interactions, the coincidenceof both the BBC and the ZDC is required as the MB trigger. The detailed definition ofthe MB trigger is a coincidence of more than 2 hits PMT in both side of the BBC and acoincidence of hits in both side of the ZDC. In addition, the position of collision vertex isselected for the MB trigger to optimize the acceptance of the central arm. The require-ments of the vertex positions are |z| < 75 cm in online and |z| < 20 cm in offline analysis,respectively.

Efficiency of the MB trigger is studied. The MB trigger efficiency is expressed as

εMB = εBBC × εZDC, (4.2)

where εBBC and εZDC correspond to the trigger efficiencies by the BBC and the ZDC,respectively. The BBC efficiency (εBBC) is estimated using a Monte Carlo simulationbased on the GEANT [99] with HIJING 1.35 event generator [100] as input. The BBCresponse in the simulation is tuned so that the charge and time distributions in all 128PMT and the LVL1 logic of the BBC are reproduced. The systematic uncertainty of theefficiency is estimated by changing the model parameters in HIJING and the threshold ofthe hit PMT. The BBC efficiency is obtained as [101]:

εBBC = 93.1 ± 0.4(stat) ± 1.6(sys)%. (4.3)

The ZDC efficiency εZDC is estimated using the relative fraction of the ZDC triggeredevents in the BBC triggered events. The ratio is obtained as 97.5%. The BBC trig-gered events include 60% of the background due to high luminosity and 40% of the ZDCinefficiency. Therefore, the ZDC efficiency εZDC is obtained as [102]:

εZDC = 99.0+1.0−1.5%. (4.4)

Finally, the MB trigger efficiency is obtained as:

εMB = εBBC × εZDC = 92.2+2.5−3.0% (4.5)

4.2.2 Centrality Determination

The nucleus collisions are classified by “centrality”. The centrality is an experimentalvariable to relate to the geometrical information of the nucleus collisions, e.g. impactparameter (b) and number of participants (Npart).

4.2. EVENT SELECTION 53

Figure 4.1: The centrality determined by the correlation between the ZDC total energyand the BBC multiplicity [105].

For the centrality determination, a correlation between the BBC multiplicity (QBBC)and the ZDC total energy (EZDC) is used. The BBC measures the charged particle multi-plicity in the rapidity range 3.1 < η < 3.9 which increases proportionally with respect toNpart. On the other hand, the ZDC measures the total energy which corresponds to thenumber of spectator neutrons. The spectator is a part of nucleus not to participate in thecollision, and thus decreases with respect to Npart. Therefore, anti-correlation is seen forthe signals between the BBC and the ZDC as shown in Fig. 4.1. Similar anti-correlationwas measured in the NA49 experiment [103].

The relation between the centrality and the impact parameter is determined using aMonte Carlo simulation of the Glauber model including the detector response [104]. Theboundary of the centrality is defined in the “clock method ” and is expressed in terms ofan angle φcent as:

φcent = tan−1

((QBBC −Q0)/Q

maxBBC

EZDC/EmaxZDC

), (4.6)

where QmaxBBC and Emax

ZDC correspond to the maximum values of the BBC multiplicity andthe ZDC energy, respectively. The origin of the clock is (0, 0.2 ·Qmax

BBC) which is determinedby the simulation. The lines shown in Fig. 4.1 indicate the centrality boundaries fromcentral to peripheral collisions. The range of the centrality is from 0 to 92% due to theMB trigger efficiency.


b

x

y

Overlap Region

nucleus nucleus

Figure 4.2: Schematic view of the colliding nuclei.

4.2.3 Glauber Calculation

The number of participants (Npart) and the number of nucleon-nucleon collisions (Ncoll)described in section 2.3.3 are useful to compare the observable (e.g. the production crosssection of charged particles) in p+ p and nucleus+nucleus (A+A) collisions. The Npart isthe number of nucleons included in the overlap region of colliding nuclei in A+A collisionsshown in Fig. 4.2. Thus, the Npart can be interpreted as the size of the the overlap regionand is related to the total volume of the dense matter formed in the collision. Ncoll is thenumber of inelastic nucleon collisions which is used to study the medium modification ofhigh pT particles.

To estimate an average number of the impact parameter (b), Npart and Ncoll for the cor-responding centrality classes, a Monte-Carlo simulation of the Glauber model is performed[63]. The Glauber model is a simple geometrical picture of a nucleus-nucleus collision. Inthe Glauber model, nucleon is assumed to travel in a straight line.

When the distance (d) between the nucleons becomes less than d <√σin

NN/π in thecalculation, a collision is considered to take place. Here, σin

NN represents to the inelasticcross section of the nucleon-nucleon collisions. The nucleon collided with other nucleonsis called as “wounded nucleon”. In the Glauber model, the wounded nucleon is consideredto be identical to the normal nucleon, although one might think that the wounded nucleonis different with the normal nucleon. For the realistic calculation of Au + Au collisions at√sNN = 200 GeV, the Woods-Saxon parameterization ρA(r) in Eq. 2.25 are used with

RA = 6.38 fm, a = 0.54 fm and the inelastic cross section is σinNN = 42 mb. The systematic

uncertainties of the Glauber calculations are estimated by changing the parameters andthe simulation conditions. The definition of the centrality class is also slightly changed tostudy the systematic uncertainty. Figure 4.3 and 4.4 show the systematic uncertainties

4.3. CHARGED PARTICLE TRACKING 55

partN0 50 100 150 200 250 300 350

def

ault

par

t )

/ N

def

ault

par

t -

NX p

art

( N

0.8

0.9

1

1.1

1.2

partN0 50 100 150 200 250 300 350

def

ault

par

t )

/ N

def

ault

par

t -

NX p

art

( N

0.8

0.9

1

1.1

1.2

= 39 mbnnσ = 45 mbnnσ

Woods-Saxon: R = 6.65 fm, d = 0.55 fmWoods-Saxon: R = 6.25 fm, d = 0.53 fm

= 0.4 fm)hardNucleons with hard core (rZDC: different neutron loss functiondifferent BBC fluctuations

gray disk nn-overlap functiongaussian nn-overlap functiondifferent origin for angle calculationBBC trigger eff. 91.4 + 2.5 %: more centralBBC trigger eff. 91.4 - 3.0 %: less centralexp. centrality selection: more centralexp. centrality selection: less central

Figure 4.3: Systematic uncertainty of the Npart.

as a function of Npart and Ncoll, respectively. The result of the Glauber calculation issummarized in Tab. 4.1 [106].

4.3 Charged Particle Tracking

Charged particles emitted from nucleus-nucleus collisions pass through the magnetic fieldand bend along with a plane perpendicular to beam pipe until these particles arrive atthe DC. Once these particles reach the DC, these particles go away in straight lines, sincethere is almost no magnetic field at the outside of the DC.

The DC reconstructs charged particle trajectories using the Combinatorial HoughTransform method (CHT) which is a general algorithm for finding straight line tracks.In the Hough transform, any pairs of the DC hits are mapped into a feature space definedby φ and α. The φ is the polar angle at the intersection of the track with a reference circlenear the mid-point of the DC, and α is the inclination angle of the track at that pointrelative to the straight line. Figure 4.5 (left) shows a schematic view of these variables.In this coordinate, the DC hit pairs from an actual track make a clear peak in the featureplane, although the random pairs from different tracks result in a flat background. If atrack has n hits in the DC, the peak height is n(n− 1)/2. Figure 4.6 shows an example ofhits in a part of the DC in the feature plane (left) and the corresponding hit distributionin the Hough transform feature space (right) [107].

Once a track is found in φ − α plane, the z-coordinate of the track is determinedusing the Hough transform with the associated PC1 cluster and the stereo U and V wireinformations of the DC. The variable zed is the z-coordinate of the intersection point withthe reference radius and the θ is the polar angle between the track and the beam axis(z-axis) in the r − z plane shown in Fig. 4.5 (right). Finally, the reconstructed track istraced back to the event vertex measured by the BBC.

The momentum (p) of a charged particle is determined using its polar angle (θ) and


collN0 200 400 600 800 1000

def

ault

coll

) /

Nd

efau

ltco

ll -

NX co

ll(

N

0.6

0.7

0.8

0.9

1

1.1

1.2

1.3

1.4

collN0 200 400 600 800 1000

def

ault

coll

) /

Nd

efau

ltco

ll -

NX co

ll(

N

0.6

0.7

0.8

0.9

1

1.1

1.2

1.3

1.4 = 39 mbnnσ = 45 mbnnσ

Woods-Saxon: R = 6.65 fm, d = 0.55 fmWoods-Saxon: R = 6.25 fm, d = 0.53 fm

= 0.4 fm)hardNucleons with hard core (rZDC: different neutron loss functiondifferent BBC fluctuations

gray disk nn-overlap functiongaussian nn-overlap functiondifferent origin for angle calculationBBC trigger eff. 91.4 + 2.5 %: more centralBBC trigger eff. 91.4 - 3.0 %: less centralexp. centrality selection: more centralexp. centrality selection: less central

Figure 4.4: Systematic uncertainty of the Ncoll.

Table 4.1: Summary of centrality, b, Npart, Ncoll and TAA

Centrality (%) b (fm) < Npart > < Ncoll > < TAA > (mb−1)0 - 10 3.2 ± 0.2 325.2 ± 3.3 955.4 ± 93.6 22.75 ± 1.56

10 - 20 5.7 ± 0.3 234.6 ± 4.7 602.6 ± 59.3 14.35 ± 1.0020 - 30 7.4 ± 0.3 166.6 ± 5.4 373.8 ± 39.6 8.90 ± 0.7230 - 40 8.7 ± 0.4 114.2 ± 4.4 219.8 ± 22.6 5.23 ± 0.4440 - 50 9.9 ± 0.4 74.4 ± 3.8 120.3 ± 13.7 2.86 ± 0.2850 - 60 11.0 ± 0.4 45.5 ± 3.3 61.0 ± 9.9 1.45 ± 0.2360 - 70 11.9 ± 0.5 25.7 ± 3.8 28.5 ± 7.6 0.68 ± 0.1870 - 80 13.5 ± 0.5 13.4 ± 3.0 12.4 ± 4.2 0.30 ± 0.1080 - 92 14.1 ± 0.6 6.3 ± 1.2 4.9 ± 1.2 0.12 ± 0.030 - 20 4.4 ± 0.2 279.9 ± 4.0 779.0 ± 75.2 18.55 ± 1.2720 - 40 8.1 ± 0.4 140.4 ± 4.9 296.8 ± 31.1 7.07 ± 0.5840 - 60 10.5 ± 0.4 60.0 ± 3.6 90.7 ± 11.8 2.16 ± 0.2660 - 92 13.0 ± 0.5 14.5 ± 2.5 14.5 ± 4.0 0.35 ± 0.10

minimum bias 9.5 ± 0.4 109.1 ± 4.1 257.8 ± 25.4 6.14 ± 0.45

4.3. CHARGED PARTICLE TRACKING 57

DC reference radius

(2.2m)

PC1 radius

(2.45m)

Z Vertex

zed

z

r

Figure 4.5: (left) The schematic view of a reconstructed track by the DC in the x − yplane. (right) The schematic view of a reconstructed track by the DC in the r − z plane.

Figure 4.6: (left) The DC hits in the x−y plane. (right) The corresponding hit distributionin the feature space [107].


the inclination angle (α). Especially, the transverse momentum (pT) is approximatelyexpressed as:

α ≈ K

pT

, (4.7)

where K = 87 (mrad GeV/c) is the effective field integral between the collision vertex andthe DC. Because of the non-uniform shape of the focusing magnetic field, an analyticalcalculation is not possible for momentum determination. Therefore, the four-dimensionalfield-integral grid (f(p, r, θ0, z)) is constructed for momentum reconstruction and for theentire radial extent of the central arms in order to trace the entire trajectory of the trackfrom the event vertex [107]. The parameters in the field grid are: the total momentum ofthe particle (p), the radius (r), the polar angle of the particle at the vertex (θ0) and thez coordinate of the vertex (z).

The momentum scale is verified by comparing the known proton mass with the mea-sured value. The mass value of proton is measured using the time of flight informationby the TOF. The accuracy of the momentum scale is better than 0.7%. The momentumresolution is directly related to the α resolution:

δp

p=δα

α=

1

K

√(σms

β

)2

+ (σαp)2, (4.8)

where δα is the measured angular spread and β is velocity. The momentum resolutionis composed from the contribution of the multiple scattering (σms) and the contributionof the intrinsic pointing resolution of the DC (σα). These values are estimated to beσms = 0.61 (mrad GeV/c) and σα = 0.86 (mrad) using the high energy hadron trackin the zero field data [108, 109]. The momentum resolution is finally determined to beδp/p ' 0.7% ⊕ 1.0% × p (GeV/c) [110].

The quality of the reconstructed tracks is defined using the hit informations of the Xand the stereo U and V wires and the associated PC1 cluster. The “quality” is a 6 bitvariable to describe the track quality. The definition of quality is described in Tab. 4.2.

Table 4.2: Bit definition of the track quality.

bit decimal description0 (LSB) 1 X1 hits used1 2 X2 hits used2 4 UV hits used3 8 UV unique (no hit sharing)4 16 PC1 cluster used5 (MSB) 32 PC1 unique (no cluster sharing)

4.4. ELECTRON IDENTIFICATION 59

4.3.1 Matching to the Outer Detector

The reconstructed tracks are extended as straight lines and projected into the outer de-tectors such as the RICH, TOF and EMCal. The inter-detector association of the trackis performed by searching for the closest hit in these detector. The closest hit from thetrack projection point is identified as “associated hit”.

The distance in both the φ and z direction between the projection point and theassociated hit is approximately Gaussian with a width (σmatch) expressed as:

σmatch =

√(σmatch

det

)2+

(σmatch

ms

pβ

)2

, (4.9)

where σmatchdet is the finite detector resolution and σmatch

ms is the contribution of the multiplescattering. The mean value of the distribution is tuned to be zero for the detector align-ment. The residual distribution between the projection point and the associated hit isreferred as the matching distribution. The matching distributions of φ and z are indepen-dent each other. In this analysis, the widths of the matching distributions of φ and z arenormalized by their standard deviations, and then these are used for the track selection.This selection of the tracks is called as “matching cut”.

4.4 Electron Identification

For electron identification, some variables are measured and calculated from the RICHand the EMCal. In this section, the definition of these variables are described.

4.4.1 Electron ID with RICH

The electron identification (ID) is performed by mainly the RICH described in section3.4.4. Cerenkov lights emitted by an electron is reflected by the spherical mirror in theRICH and is focused as a ring shape on the PMT arrays of the RICH shown in Fig. 4.7.The hit PMT is defined so that a PMT detects more than 0.1 photo-electrons (Npe).

To determine association of the RICH signal with the reconstructed track, the track isreflected by the spherical mirror and the reflected line is projected onto the PMT arrays ofthe RICH. If the track is an electron, the reflected line should be projected to the centerof a ring formed by hit PMTs. Then, we count the number of hit PMTs (n0) within aassociation radius (rasso) from the projection point. The association radius (rasso) is ideally5.9±2.5 cm and depends on both flight path length of the Cerenkov light, refraction indexof the radiator (CO2) and velocity of the electron. Therefore, the range of the associationradius is set from 3.4 to 8.4 cm. Using the association with the RICH, four variables forthe electron ID are defined as follows:

• n0 − Number of hit PMT within rasso.

• npe0 − Number of photo-electrons in all hit PMTs within rasso. Using number ofphoto-electrons in single hit PMT i (N i

pe), npe0 is expressed as:

npe0 ≡∑

N ipe (4.10)


Figure 4.7: The schematic view of the reflected line by the RICH mirrors and hit PMTsin the PMT array with the tight association radius.

• disp − Displacement between a measured ring center (rm) and a position of trackprojection. The measured ring center is the weighted average of the hit PMT position(ri) expressed as:

rm ≡∑ri ×N i

pe

npe0, (4.11)

• chi2/npe0 − χ2-like variable of the ring shape between the expected and the measuredring. Using the projection point (r0) and ri, chi2/npe0 is expressed as:

chi2/npe0 ≡∑

rasso(ri − r0)

2 ×N ipe

npe0, (4.12)

The variable n0 and npe0 represents the quality of association between the track andthe RICH hit. The variable disp and chi2/npe0 indicate the quality of the ring shapereconstructed in the RICH.

4.4.2 Electron ID with EMCal

The EMCal is another main detector for the electron ID which provides the complementaryinformation to the RICH. Four variables measured in the EMCal are used for the electronID. These four variables are defined as follows:


• emcsdphi e − Distance in φ direction between the track projection position (pemcphi)and the hit position (emcphi) in the EMCal. The hit position is the center of the en-ergy cluster at the surface of the EMCal. The distance is normalized by its standarddeviation (σemcdphi(p)). The emcsdphi e is expressed as :

emcsdphi e ≡ emcphi − pemcphi

σemcdphi(p). (4.13)

• emcsdz e − Distance in z direction between the track projection position (pemcz)and the hit position (emcz) in the EMCal. The distance is normalized by its standarddeviation (σemcdz(p)). The emcsdz e is expressed as :

emcsdz e ≡ emcz − pemcz

σemcdz(p)(4.14)

• dep − Energy(E) - momentum (p) matching, where E is the energy measured bythe EMCal and p is the momentum of the track. Since an electron deposits fullenergy in the EMCal and the mass of electron is small compared to its momentum,the energy to momentum ratio (E/p) must be about unity. On the other hand, E/pis less than unity for hadron since hadron deposits only a fraction of its full energy.Using its standard deviation (σE/p(p)), the dep is expressed as:

dep ≡ E/p− 1

σE/p(p), (4.15)

• emcdt − Time information (tEMC) measured in the EMCal. Using its standarddeviation (σtEMC

(p)), the emcdt is expressed as:

emcdt ≡ tEMC

σtEMC(p)

. (4.16)

4.4.3 Summary of Electron ID Conditions

The cut values required for the event selection, the track selection and the electron ID aresummarized in Tab. 4.3. These required cuts are determined to increase both the statisticsand purity of electrons as much as possible.

There are some comments on these cuts. (1) A large fraction of the DC in the Westarm is unstable during Run 2. Therefore, the East arm is only used in this analysis. (2)The loose quality value in the DC is required to increase statistics. In electron analysis,the coincidence of hits in the three detectors (the DC, RICH and EMCal) are requiredto identify electrons. Therefore, possible contaminations from the ghost tracks might bereduced.

Applying these cuts for the electron ID, clean electron samples can be extracted. Figure4.8 (left) shows the E/p ratio of electron sample for 0.5−0.6 GeV/c in pT. Here, all of thesecuts except the dep cut are required for the electron ID. The red and the blue histogramscorrespond to the E/p ratio associated with the RICH hit (Real) and background (BG)


E/p0 0.5 1 1.5 2 2.5

0

200

400

600

800

1000

1200

1400

1600

1800

RealBGNet signal

Ep PbScE 0.5<pT<0.6

n00 2 4 6 8 10

0

2000

4000

6000

8000

10000

12000RealBGNet

crk n0 east 0.5<p<0.6

Figure 4.8: (left) Energy to momentum (E/p) ratio (right) n0 distribution in RICH. Inboth panels, electron signals are clearly seen.

estimated by the “flip and slide” method described in section 4.5. The net signal (green)shows the peak around one. As the other example, the n0 distribution is shown in Fig.4.8 (right). The red, blue and green histograms show the associated, background andnet signal, respectively. These examples indicate that these cuts can work well for theelectron extraction from huge hadron backgrounds. After the electron ID cuts are applied,the hadron backgrounds due to the random association still remain as roughly 10% whichis described in section 4.5.

Table 4.3: Summary of the required cut values for the electron ID.

Variables Required cutsTrigger selection MB triggerCollision vertex |ZBBC| < 20 cmSpectrometer The East armTrack quality Not required

Matching of track√

emcsdphi e2 + emcsdz e2 < 2.0RICH hit n0 ≥ 3

Ring shape chi2/npe0 < 10.0Matching of ring center disp < 5.0

Energy momentum matching −2.0 < dep < 3.0EMCal time emcdt < 2.0 or ttof < 0.3


Figure 4.9: Fiducial volume for the East-South (left) and the East-North (right). The redand the blue lines show the boundaries of the fiducial cut E and F , respectively. Theshaded areas are removed out in the analysis.

4.4.4 Fiducial Cut

The selection of the fiducial volume is introduced to control the stability of the detectoracceptance. There are two definitions of the fiducial volumes which is denoted as “E” and“F”. The E fiducial cut is for removing the difference of the detector acceptance betweenthe data with and without the photon converter. These two data sets are comparedto extract non-photonic electrons described in section 4.10.1. The F fiducial cut is forremoving the discrepancy of the acceptance between the real data and the simulation.

These fiducial volumes are defined in the plane of “charge/pT vs φ” at the DC. Theφ is an azimuthal angle of the reconstructed track at the DC. Since the south and northside of the DC are independent detectors, the fiducial volumes are defined for each sideof the East-South and the East-North, respectively. The definitions of the fiducial cutsE and F for the East-South and the East-North are summarized in Tab. 4.4. Figure 4.9shows the fiducial volume in the plane of “charge/pT vs φ” for the East-South (left) andthe East-North (right), respectively. The shaded areas shown in Fig. 4.9 are removedout by these fiducial cuts. An additional fiducial cut for a sector one of the PbGl EMCalare applied so that the tracks with pT less than 0.5 GeV/c are removed. Because the φdistribution in the real data is different with that in the simulation.

In order to verify whether the fiducial cut works, the φ distribution of electrons betweenthe data with that without the converter is compared. Figure 4.10, 4.11, 4.12 and 4.13 showthe phi distributions of e+ and e− for the East-South and the East-North, respectively.In these figure, the blue and the red histograms correspond to the data with and withoutthe converter. From the comparisons, the acceptance between the data with and withoutthe converter are in good agreement.


Table 4.4: Summary of the fiducial cuts E and F for the East-South and the East-North.

Fiducial cut for the East SouthE charge/pT > -46.804 + 12.338 ×φE 2.22 < φ < 3.085E 3.20 < φF charge/pT > 36.175 -11.81 ×φF charge/pT < 33.26 - 11.86 ×φ

Fiducial cut for the East NorthE charge/pT > −46.804 + 12.338 × φE charge/pT < −551.5 + 170.0 × φE charge/pT > −546.5 + 170.0 × φE φ > 2.23F charge/pT > 36.9 − 11.85 × φF charge/pT < 33.357 − 11.809 × φ

φDch 2 2.2 2.4 2.6 2.8 3 3.2 3.4 3.6 3.8

φDch 2 2.2 2.4 2.6 2.8 3 3.2 3.4 3.6 3.8

0

200

400

600

800

1000

1200

1400

1600

1800

Phi in Pt 0.2-0.3

φDch 2 2.2 2.4 2.6 2.8 3 3.2 3.4 3.6 3.8

φDch 2 2.2 2.4 2.6 2.8 3 3.2 3.4 3.6 3.8

0

200

400

600

800

1000

1200

Phi in Pt 0.3-0.5

φDch 2 2.2 2.4 2.6 2.8 3 3.2 3.4 3.6 3.8

φDch 2 2.2 2.4 2.6 2.8 3 3.2 3.4 3.6 3.8

0

100

200

300

400

500

Phi in Pt 0.5-0.8

φDch 2 2.2 2.4 2.6 2.8 3 3.2 3.4 3.6 3.8

φDch 2 2.2 2.4 2.6 2.8 3 3.2 3.4 3.6 3.8

0

20

40

60

80

100

Phi in Pt 0.8-1.0

φDch 2 2.2 2.4 2.6 2.8 3 3.2 3.4 3.6 3.8

φDch 2 2.2 2.4 2.6 2.8 3 3.2 3.4 3.6 3.8

0

20

40

60

80

100Phi in Pt 1.0-2.2

φDch 2 2.2 2.4 2.6 2.8 3 3.2 3.4 3.6 3.8

φDch 2 2.2 2.4 2.6 2.8 3 3.2 3.4 3.6 3.8

0

1

2

3

4

5

6

Phi in Pt 2.2-10.0

Figure 4.10: φ distributions of e+ in the East-South. The red and the blue histogramscorrespond to the data with and without the converter.


φDch 2 2.2 2.4 2.6 2.8 3 3.2 3.4 3.6 3.8

φDch 2 2.2 2.4 2.6 2.8 3 3.2 3.4 3.6 3.8

0

200

400

600

800

1000

1200

Phi in Pt 0.2-0.3

φDch 2 2.2 2.4 2.6 2.8 3 3.2 3.4 3.6 3.8

φDch 2 2.2 2.4 2.6 2.8 3 3.2 3.4 3.6 3.8

0

200

400

600

800

1000

Phi in Pt 0.3-0.5

φDch 2 2.2 2.4 2.6 2.8 3 3.2 3.4 3.6 3.8

φDch 2 2.2 2.4 2.6 2.8 3 3.2 3.4 3.6 3.8

0

100

200

300

400

500

Phi in Pt 0.5-0.8

φDch 2 2.2 2.4 2.6 2.8 3 3.2 3.4 3.6 3.8

φDch 2 2.2 2.4 2.6 2.8 3 3.2 3.4 3.6 3.8

0

20

40

60

80

100

120

Phi in Pt 0.8-1.0

φDch 2 2.2 2.4 2.6 2.8 3 3.2 3.4 3.6 3.8

φDch 2 2.2 2.4 2.6 2.8 3 3.2 3.4 3.6 3.8

0

20

40

60

80

100

Phi in Pt 1.0-2.2

φDch 2 2.2 2.4 2.6 2.8 3 3.2 3.4 3.6 3.8

φDch 2 2.2 2.4 2.6 2.8 3 3.2 3.4 3.6 3.8

0

1

2

3

4

5

6

7Phi in Pt 2.2-10.0

Figure 4.11: φ distributions of e− in the East-South. The red and the blue histogramscorrespond to the data with and without the converter.

φDch 2 2.2 2.4 2.6 2.8 3 3.2 3.4 3.6 3.8

φDch 2 2.2 2.4 2.6 2.8 3 3.2 3.4 3.6 3.8

0

200

400

600

800

1000

1200

1400

1600

Phi in Pt 0.2-0.3

φDch 2 2.2 2.4 2.6 2.8 3 3.2 3.4 3.6 3.8

φDch 2 2.2 2.4 2.6 2.8 3 3.2 3.4 3.6 3.8

0

200

400

600

800

1000

1200

Phi in Pt 0.3-0.5

φDch 2 2.2 2.4 2.6 2.8 3 3.2 3.4 3.6 3.8

φDch 2 2.2 2.4 2.6 2.8 3 3.2 3.4 3.6 3.8

0

100

200

300

400

500

600

Phi in Pt 0.5-0.8

φDch 2 2.2 2.4 2.6 2.8 3 3.2 3.4 3.6 3.8

φDch 2 2.2 2.4 2.6 2.8 3 3.2 3.4 3.6 3.8

0

20

40

60

80

100

Phi in Pt 0.8-1.0

φDch 2 2.2 2.4 2.6 2.8 3 3.2 3.4 3.6 3.8

φDch 2 2.2 2.4 2.6 2.8 3 3.2 3.4 3.6 3.8

0

10

20

30

40

50

60

70

80

90Phi in Pt 1.0-2.2

φDch 2 2.2 2.4 2.6 2.8 3 3.2 3.4 3.6 3.8

φDch 2 2.2 2.4 2.6 2.8 3 3.2 3.4 3.6 3.8

0

1

2

3

4

5

6Phi in Pt 2.2-10.0

Figure 4.12: φ distributions of e+ in the East-North. The red and the blue histogramscorrespond to the data with and without the converter.


φDch 2 2.2 2.4 2.6 2.8 3 3.2 3.4 3.6 3.8

φDch 2 2.2 2.4 2.6 2.8 3 3.2 3.4 3.6 3.8

0

200

400

600

800

1000

1200

1400Phi in Pt 0.2-0.3

φDch 2 2.2 2.4 2.6 2.8 3 3.2 3.4 3.6 3.8

φDch 2 2.2 2.4 2.6 2.8 3 3.2 3.4 3.6 3.8

0

200

400

600

800

1000

1200Phi in Pt 0.3-0.5

φDch 2 2.2 2.4 2.6 2.8 3 3.2 3.4 3.6 3.8

φDch 2 2.2 2.4 2.6 2.8 3 3.2 3.4 3.6 3.8

0

100

200

300

400

500

Phi in Pt 0.5-0.8

φDch 2 2.2 2.4 2.6 2.8 3 3.2 3.4 3.6 3.8

φDch 2 2.2 2.4 2.6 2.8 3 3.2 3.4 3.6 3.8

0

20

40

60

80

100

120

Phi in Pt 0.8-1.0

φDch 2 2.2 2.4 2.6 2.8 3 3.2 3.4 3.6 3.8

φDch 2 2.2 2.4 2.6 2.8 3 3.2 3.4 3.6 3.8

0

10

20

30

40

50

60

70

80

90

Phi in Pt 1.0-2.2

φDch 2 2.2 2.4 2.6 2.8 3 3.2 3.4 3.6 3.8

φDch 2 2.2 2.4 2.6 2.8 3 3.2 3.4 3.6 3.8

0

1

2

3

4

5

6

Phi in Pt 2.2-10.0

Figure 4.13: φ distributions of e− in the East-North. The red and the blue histogramscorrespond to the data with and without the converter.

4.5 Hadron Background

A small fraction of the backgrounds from the hadron contamination remains in the electronsamples after the electron ID cuts is applied. The origin of the mis-identified electronsamples is mainly due to accidental association of the reconstructed track with the hitin the RICH. These backgrounds are estimated by an event mixing method. The eventmixing is performed in software by flipping the z coordinate of the track in a same event,and then searching for the association of the flipped track with the RICH hit. For example,if a track has 45 cm of z-coordinate, the z-value of the flipped track is −45 cm. However,a flipped track near z = 0 is almost same with its original track. To treat such a track,the z component of a track within |z| < 35 cm is slided toward the opposite side of thez coordinate. If a track has 20 cm in z, the z value of the slided track is −15cm (=20 − 35 cm). Since these flipped and slided tracks provide only random association withthe RICH hit, the random hadron backgrounds are statistically estimated. This eventmixing method is called as “flip and slide method”. Figure 4.14 shows the schematicdiagram of the flip and slide method. The backgrounds estimated by the method arestatistically subtracted from the reconstructed electrons to obtain the net signals.

Figure 4.15 shows E/p ratios for four pT ranges. The red and the blue points showthe total and background (BG) electron candidates. The green histograms represent thenet signals obtained by the subtraction of the background from the total electron candi-dates. From the figure, by comparing the total and background electron candidates, thebackgrounds from random hadron association are estimated to be about 10%.

After the subtraction of the random backgrounds, the other background componentsstill remain which are shown as a lower tail of the E/p distributions in Fig. 4.15. Thesebackgrounds are electrons from Ke3 decays and photon conversions that occur far fromthe collision vertex. These background electrons reconstructs with higher momentum

4.6. RUN SELECTION 67

TrackFlipped

Track

Flip

Z Vertex

Slide

Slided

TrackRICHRICH

Random

association

Real

association

RICH mirror

Figure 4.14: Schematic view of the flip and slide method.

compared with its actual momentum due to the tracking algorithm. Therefore, the E/pratio of the background electrons has lower value. These background components arestudied using a GEANT simulation [99] of the PHENIX detector described in section4.10.6.

4.6 Run Selection

The dead and/or the hot area of the detector is possibly changed during the Run 2 opera-tion period. The run-by-run detector condition is studied to verify whether the data usedin this analysis is taken with the stable condition. Figure 4.16 shows the run dependence ofthe inclusive electron yield for six pT classes, which correspond to 0.2-0.3, 0.3-0.5, 0.5-0.8,0.8-1.0, 1.0-2.2 and pT < 2.2 GeV/c, respectively. In these figures, the red and the bluepoints correspond to the positron and the electron yield, respectively.

The photon converter has been installed during a part of the run period. The runperiods with and without the converter are called as the converter run and the non-converter run respectively. In Fig. 4.16, the electron yield in the converter run is constantlyhigher than that in the non-converter run. On the other hand, the electron yield in thenon-converter run is less stable than that in the converter run. Thus, a good run periodin the non-converter run is used in this analysis.

The actual run lists for the converter run and the non-converter run are summarizedin Tab. 4.5. The data statistics for the converter run and the non-converter run are shownin Tab. 4.6. The minimum bias events are subdivided for five centrality classes, 0-10%,10-20%, 20-40%, 40-60% and 60-92%, respectively.


E/p0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 2.2 2.40

200

400

600

800

1000

1200

1400

1600 Total

BG (Flip and Slide)

Net Signal

<0.8T

E/p 0.6<p

E/p0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 2.2 2.40

100

200

300

400

500

600

700 Total

BG (Flip and Slide)

Net Signal

<1.0T

E/p 0.8<p

E/p0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 2.2 2.4

0

100

200

300

400

500

600 Total

BG (Flip and Slide)

Net Signal

<1.5T

E/p 1.0<p

E/p0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 2.2 2.4

0

20

40

60

80

100

120

140

160

180

<2.0T

E/p 1.5<p

Total

BG (Flip and Slide)

Net Signal

<2.0T

E/p 1.5<p

Figure 4.15: E/p distributions for four pT ranges. The net signal in each panel is extractedby subtracting the backgrounds (BG) from the total electron candidates (Total).

Table 4.5: Summary of the run lists for the converter run and the non-converter run.

the converter run28623 28625 28627 28632 28718 28749 28750 28751 2876128765 28768 28775 28777 28781 28791 28794 28795 2879828902 28903 28956 28958 28961 28962 28966 28968 2897128972 28973 29014 29015 29016 29017 29036

the non-converter run29116 29146 29171 29178 29179 29183 29184 29185 2918629212 29213 29255 29267 29268 29368 29372 29380 2938629392 29393 29401 29404 29445 29446 29454 29459 2946129510 29512 29514 29515 29529 29534 29537 29561 2956229563

4.6. RUN SELECTION 69

RunNumber28000 29000 30000 31000 32000 33000

RunNumber28000 29000 30000 31000 32000 33000

Ne/

nev

t

0

0.002

0.004

0.006

0.008

0.01

0.012

Pt : 0.2-0.3

RunNumber28000 29000 30000 31000 32000 33000

RunNumber28000 29000 30000 31000 32000 33000

Ne/

nev

t

0

0.002

0.004

0.006

0.008

0.01

0.012

Pt : 0.3-0.5

RunNumber28000 29000 30000 31000 32000 33000

RunNumber28000 29000 30000 31000 32000 33000

Ne/

nev

t

0

0.001

0.002

0.003

0.004

0.005

0.006

0.007

0.008

Pt : 0.5-0.8

RunNumber28000 29000 30000 31000 32000 33000

RunNumber28000 29000 30000 31000 32000 33000

Ne/

nev

t

0

0.0005

0.001

0.0015

0.002

0.0025

Pt : 0.8-1.0

RunNumber28000 29000 30000 31000 32000 33000

RunNumber28000 29000 30000 31000 32000 33000

Ne/

nev

t

0

0.0002

0.0004

0.0006

0.0008

0.001

0.0012

0.0014

0.0016

0.0018

Pt : 1.0-2.2

RunNumber28000 29000 30000 31000 32000 33000

RunNumber28000 29000 30000 31000 32000 33000

Ne/

nev

t0.0005

0.001

0.0015

0.002

0.0025-1x10

Pt : 2.2-10.0

Figure 4.16: The run dependence of the raw electron yield for East-South. The red andthe blue points correspond to the positron and the electron yield, respectively.

Table 4.6: Summary of event samples for the converter run and the non-converter run.

Centrality non-converter run converter runMinimum bias 2,544,577 2,253,413

0-10 % 247,060 220,41010-20 % 280,233 248,60420-40 % 553,849 490,24540-60 % 562,982 497,59560-92 % 900,453 796,559


4.7 Raw Yield of Inclusive Electrons

Applying the electron ID cut described in section 4.4.3, the raw yields of inclusive electronsare obtained for both the converter run and the non-converter run. The backgroundsfrom the hadron contaminations estimated by the flip and slide method are subtractedstatistically. Figure 4.17 shows the raw yield of inclusive electrons as a function of pT inminimum bias Au + Au collisions. The blue and the red histograms correspond to theyield in the converter run and the non-converter run, respectively. They are normalizedby their number of analyzed events independently. Since additional photon conversionstake place in the converter run, the blue histogram is always higher than the red one.The spectra become closer at high pT. This indicates that a fraction of photonic electronsbecomes smaller at higher pT. Figure 4.18 shows the raw yield of inclusive electrons for0-10%, 10-20%, 20-40%, 40-60% and 60-92% centrality classes, respectively. From thecomparison of the electron spectra with and without the converter, we can separate thephotonic electron and the non-photonic electron components. The details of the analysisis described in section 4.10.5.

4.8 Detector Response in Simulation

The geometrical acceptance and the electron ID efficiencies are estimated using the MonteCarlo simulations of the PHENIX detector based on the GEANT simulator [99]. Thisdetector simulator is called as “PISA”. In order to calculate the acceptance and the elec-tron ID efficiency, the detector responses in the simulation should be the same with thosein the data. For the electron analysis, the background is mainly electrons from photonconversions. Thus, the material budget in the simulation is crucial to estimate the back-ground. In this section, the detector responses and the material budget in the simulationare described.

4.8.1 Comparison between Real Data and Simulation

To study the detector responses in the PISA simulation, all the electron ID variables arecompared with that in the real data. If the variable in the simulation is different fromthat in the real data, the simulation is tuned to reproduce the real data accurately. Thesecomparisons are performed using the electron samples. To select the electron samples, allthe electron ID cuts except the compared variable are required (to be the same as thereal data). Table 4.7 shows the summary of the required cuts for the comparison of eachvariable. In Tab. 4.7, the symbol “©” and “×” mean whether the cut is required or notrequired, respectively. For example, when we compare the RICH variable n0 between thesimulation and the real data, the electron sample is selected by emcsdphi e, emcsdz e, depand emcdt. It should be noted that a tighter cut in emcsdz e is required for the comparisonof emcsdphi e, since the two dimensional matching cut variables are independent eachother.

Figure 4.19 shows the comparisons of the variables between the real data and thesimulation. The plots from the top-left to the bottom-right in Fig. 4.19 correspond tothe emcsdphi e, emcsdz e, n0 and dep, respectively. These distributions are normalized by

4.8. DETECTOR RESPONSE IN SIMULATION 71

(GeV/c)T

p0 0.5 1 1.5 2 2.5 3 3.5 4

evt

/N eN

-710

-610

-510

-410

-310

-210

Raw Yield of Inclusive Electrons in min. bias

Figure 4.17: The raw yield of inclusive electrons as a function of pT in minimum biasAu + Au collisions. The blue and the red points represent the electron yields in theconverter run and the non-converter run, respectively.

(GeV/c)T

p0 0.5 1 1.5 2 2.5 3 3.5 4

evt

/Ne

N

-710

-610

-510

-410

-310

-210

Raw Yield in centrality 0-10%

(GeV/c)T

p0 0.5 1 1.5 2 2.5 3 3.5 4

evt

/Ne

N

-710

-610

-510

-410

-310

-210


(GeV/c)T

p0 0.5 1 1.5 2 2.5 3 3.5 4

evt

/Ne

N

-710

-610

-510

-410

-310

-210


(GeV/c)T

p0 0.5 1 1.5 2 2.5 3 3.5 4

evt

/Ne

N

-710

-610

-510

-410

-310

-210


(GeV/c)T

p0 0.5 1 1.5 2 2.5 3 3.5 4

evt

/Ne

N

-710

-610

-510

-410

-310

-210


Figure 4.18: The raw yield of inclusive electrons as a function of pT. The blue and thered points represent the electron yields in the converter run and the non-converter run.These six panels correspond to the six centrality classes, 0-10%, 10-20%, 20-40%, 40-60%and 60-92%, respectively.


the number of the entries. These comparisons indicate that the detector responses in thesimulation are in good agreement with the read data. The detailed comparisons of theindividual electron ID variables are described in Appendix A.1.

Table 4.7: Summary of the required cuts for the comparison between the real data and thesimulation. The symbol “©” and “×” mean whether the cut is required or not required,respectively. The variables, sdphi and sdz, are shortened forms of emcsdphi e and emcsdz e,respectively.

Applied Cut Compared Variablessdphi(z) n0 chi2/npe0 disp dep emcdt√

sdphi2 + sdz2 < 2 × (sdz < 1) © © © © ©

n0 ≥ 3 © × © © © ©

chi2/npe0 < 10 © × × × © ©

disp < 5 © × × × © ©

−2 < dep © © © © × ×emcdt < 2 © © © © © ×

4.8.2 Material Budget in Simulation

Photon conversions are the main source of photonic electron backgrounds. Most of pho-ton conversions occur in the beam pipe and the MVD. If photon conversions in the beampipe and the MVD are accurately modeled in the PISA simulation, photonic electron back-grounds can be correctly subtracted using the converter method described in section 4.10.1.Therefore, it is important to verify whether the material budget of the PHENIX detectorin the simulation is the same as the real one.

The material budget of the PHENIX detector are studied using the yield of electron-positron (e+e−) pairs from photon conversions. Figure 4.20 shows the invariant massdistribution of e+e− pairs. The red and the black points correspond to the real andthe combinatorial background pairs, respectively. The combinatorial backgrounds areestimated by an event mixing method. There are 3 peaks at 5, 15 and 75 MeV/c2 inFig. 4.20 which are artificially produced by the tracking algorithm. Since the trackreconstruction algorithm assumes that all tracks originate from the collision vertex, theelectrons produced by photon conversions at the point far from the vertex are reconstructedwith incorrect momenta. Figure 4.21 shows the schematic view of the mis-reconstruction ofthe conversion pair. As a result, the reconstructed mass of pair is shifted from its originalvalue, even though its original value is very close to zero. This shift is approximatelyproportional to the magnetic field integral. This indicates that these artificial peaks canbe used to identify the positions where the photon conversions occur. The first peak around5 MeV/c2 is mainly from π0 Dalitz decays, the second peak around 15 MeV/c2 is fromphoton conversions at the beam pipe and the MVD except the outer shell and the thirdpeak around 75 MeV/c2 are from the conversions at the MVD outer shell, respectively.


σ-5 -4 -3 -2 -1 0 1 2 3 4 50

50

100

150

200

250

300

350

400

emcsdphi_e in 0.5<p<0.6emcsdphi_e in 0.5<p<0.6

σ-5 -4 -3 -2 -1 0 1 2 3 4 50

50

100

150

200

250

300

350

400

450

emcsdz_e in 0.5<p<0.6

n00 1 2 3 4 5 6 7 8 9 100

2000

4000

6000

8000

n0 in 0.5<p<0.6n0 in 0.5<p<0.6

-4 -3 -2 -1 0 1 2 3 40

50

100

150

200

250

Figure 4.19: Comparisons of the detector response between the real data and the simu-lations. The detector response of emcsdphi e (top-left), emcsdz e (top-right), n0 (bottom-left) and dep (bottom-right) are shown, respectively. The black and red lines correspondto the real data and the simulation.


)2 (GeV/ceeM0 0.02 0.04 0.06 0.08 0.1 0.12 0.14

Co

un

t

0

20

40

60

80

100

120

140

160

180

200

220 Dalitz

B.P. Conv.

MVD Conv.

Real

Mixed BG

invariant masseeM

Figure 4.20: Invariant mass distributionof e+e− pairs. The red and black pointscorrespond to the real and the combina-torial background pairs estimated by anevent mixing method.

Material

Collision vertex

e-e+

Real Track

Reconstructed Track

R

Figure 4.21: Schematic view of track re-construction. The electrons produced atR > 0 are reconstructed with incorrectmomenta. This results in the fake invari-ant mass of the reconstructed conversionpairs.

Figure 4.22 show the invariant mass distribution of e+e− pairs in the real data (left)and the simulation (right). The combinatorial backgrounds have been already subtracted.The red and blue histograms correspond to the pair yield in the non-converter run andthe converter run, respectively. In both the real data and the simulation, the peak at75 MeV/c2 in the converter run is much higher than that in the non-converter run, sincethe converter increases photon conversions around the MVD outer shell. Here, π0 simu-lations are used for the pair analysis. The conditions of π0 simulations are described insection 4.10.2.

In order to compare the material budget between the real data and the simulationquantitatively, we define the following variables:

• NBP : the pair yield forMe+e− < 40 MeV/c2 where the Dalitz decays (0.8%X0 equiv-alent) and the conversions at the beam pipe (0.29% X0) and the MVD (0.9% X0)except the outer shell are included,

• NMVD : the pair yield for 60 < Me+e− < 100 MeV/c2 where the conversions at theMVD outer shell (0.17% X0) and the converter (1.7% X0) are included,

• Rmaterial : the ratio of NMVD to NBP.

Rmaterial =NMVD

NBP

. (4.17)

where X0 is the radiation length.The converter gives a good reference to study the amount of the material. Since

the converter in the simulation is implemented using the measured thickness and the


(GeV)eeM0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 0.18 0.2

pai

rN

0

20

40

60

80

100

<0.2 (GeV))ee

Invariant mass of ee pairs (M

Real Data

No conv Run : 908.20BPN : 139.35MVDN

Conv Run : 705.90BPN : 939.00MVDN

<0.2 (GeV))ee


(GeV)eeM0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 0.18 0.2

pai

rN

0

20

40

60

80

100

120

<0.2 (GeV))ee


Simulation

No conv Run : 1074.65BPN : 177.80MVDN

Conv Run : 1045.75BPN : 1284.05MVDN

<0.2 (GeV))ee


Figure 4.22: Invariant mass distribution of e+e− pairs in the real data (left) and thesimulation (right). The red and blue histogram correspond to the non-converter and theconverter run, respectively.

measured weight, NMVD in the simulation should be the same with that in the real data inthe converter run. Therefore, Rmaterial shows the accurate information of NBP comparedwith NMVD. This means that the material amount of both the beam pipe and the MVDexcept the outer shell can be studied by comparing Rmaterial between the real data and thesimulation.

Before the Rmaterial calculation, the pT distribution of the pairs between real dataand simulation are compared since the invariant mass distribution is sensitive to the pT

spectrum of the pairs, Figure 4.23 shows the pT distributions of the pairs in 0 − 40 (left)and 60 − 100 MeV/c2 (right) in the converter run. The red histogram corresponds to thereal data, and the black, the green, the blue, the magenta and the light blue histogramscorrespond to π0 simulation with pT threshold of 0.0, 0.6, 1.0, 1.5 and 2.5 GeV/c. Inboth figures, the shape of the pT distribution in the real data is well reproduced by thesimulation above each pT threshold. From the comparison of pT distribution, we decideto use two sets of the simulation data with the pT threshold of 0.0 and 0.6 GeV/c. ForRmaterial calculation, we select the pairs with pT > 0.7 GeV/c.

In addition, in order to increase the statistics of NBP in the simulation, NBP in thenon-converter run of the simulation is combined with that in the converter run of thesimulation. But, it is not so simple to combine these NBP. In the converter run, electronsemitted inside the converter suffers the radiative energy loss in the converter. This blockingeffect modifies the invariant mass distribution within Me+e− < 40 MeV/c2 in the converterrun. On the other hand, this blocking effect does not appear in the non-converter run. Forthis reason, the mass within Me+e− < 40 MeV/c2 in the converter run is slightly smallerthan that in the non-converter run. This effect is shown in both Fig. 4.22 (left) and(right). Therefore, the blocking effect for the pair yield needs to be taken into accountwhen the NBP for the converter and non converter run are combined.


(GeV/c)Tp0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5

pai

rN

10-1

1

10

102

103

raw pT distribution (0<Mee<40)

RealSim:0.6-10.0Sim:1.0-10.0Sim:1.5-10.0Sim:2.5-10.0Sim:0.0-10.0


(GeV/c)Tp0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5

pai

rN

10-1

1

10

102

103


RealSim:0.6-10.0Sim:1.0-10.0Sim:1.5-10.0Sim:2.5-10.0Sim:0.0-10.0


Figure 4.23: Raw pT distribution of e+e− pairs from Me+e− < 40 MeV (left) and 60 <Me+e− < 100 MeV (right). Red histogram shows real data, and the others are a set ofsimulation data

The efficiency of the blocking effect εpairblock for the pair yield is estimate by the simulation

as follows:

εpairblock =

NnoconvBP

N convBP

, (4.18)

where NnoconvBP and N conv

BP are correspond to NBP in the non-converter run and the converterrun, respectively. The blocking effect of the pair is obtained to be εpair

block = 1.068. By usingthis εpair

block, NBP is combined as.

NBP =1

2· (N conv

BP + 1/εpairblock ·N

noconvBP ). (4.19)

Table 4.8 shows NBP and NMVD in the real data and the simulation. After the NBP andNMVD are normalized by number of events, Rmaterial is calculated and compared betweenthe real data and the simulation. The results are summarized in Tab. 4.9. As a result,the material budget in the simulation is found to be consistent with that in the real datawithin statistical uncertainty (99.8%±4.2%). This value is obtained from the double ratioof Rreal

material to Rsimmaterial.

The another double ratio is calculated using the pairs with pT > 0.9 GeV/c. Thatdouble ratio with pT > 0.9 GeV/c is found to be 101.0% ± 5.33%. The difference ofthese two values is 1.2% (101.0%/99.8%). We assigned 4.4% (=

√(4.2%)2 + (1.2%)2) for

systematic uncertainty of the material budget.

4.9. CORRECTION 77

Table 4.8: Summary of NBP and NMVD in the non-converter run and the converter run.

Real datadataset Nevt NMVD NBP

noconv. 2544590 226 ± 18.87 1320.75 ± 36.89conv. 2253420 1532.4 ± 42.50 1110.3 ± 34.66

π0 simulationNevt NMVD NBP

noconv. (pT ≥ 0.0) 667129 151.7 ± 13.51 801 ± 28.47noconv. (pT ≥ 0.6) 666825 308.6 ± 19.77 1808.8 ± 42.80conv. (pT ≥ 0.0) 667328 1057.3 ± 34.27 740.25 ± 27.83conv. (pT ≥ 0.6) 666714 2349.65 ± 51.22 1703.4 ± 42.25

Table 4.9: Summary of the material amounts in real data and simulation.

dataset NMVD NBP Rmaterial

sum of real data 1532.4 ± 42.45 1101.8 ± 22.94 1.391 ± 0.0482sum of sim. with pT > 0.0 1057.3 ± 34.27 745.46 ± 19.25 1.418 ± 0.0588sum of sim. with pT > 0.6 2349.65 ± 51.22 1698.14 ± 29.07 1.384 ± 0.0384combined the 2 sim. set 1.394 ± 0.0321

Double ratio (Rrealmaterial/R

simmaterial) 0.998 ± 0.042

4.9 Correction

4.9.1 Acceptance Correction

The measured electron yield is smeared by the detection efficiency of the PHENIX detector.In order to obtain the true invariant yield, the acceptance and the detector efficiency needsto be corrected. The acceptance correction factor (εacc) is composed from the geometricalacceptance(εgeom), the track reconstruction efficiency (εtrack) and the electron ID efficiency(εeID) expressed as:

εacc = εgeom × εtrack × εeID. (4.20)

These efficiencies are determined simultaneously using the PISA simulation with singleelectron and position inputs. In the PISA simulation, the 1 M single electron and positronevents are generated with flat pT slope up to 10 GeV/c, |y| < 0.6, 2π in azimuth and|z| < 30 cm. The correction factor (εacc) is determined as follows:

εacc =Measured electrons with eID cut

Input electrons× 1.2, (4.21)

where a factor 1.2 is a normalization factor of the input electrons to unit rapidity (0.6/0.5).


(GeV/c)Tp0 1 2 3 4 5 6 7 8

(ar

b. u

nit

)e

N

10-7

10-6

10-5

10-4

10-3

10-2

10-1

1

10

102

103

104

105

distributionTRaw p distributionTRaw p

Figure 4.24: pT distribution of electrons. The black and the red histograms correspond tothe input and the reconstructed electrons, respectively.

Table 4.10: The parameters for the electron spectrum

p0 (GeV/c) 2.329p1 -14.30

In order to determine the pT dependence of εacc, the pT slope of the input and mea-sured electrons is modified using a pT weight of electron spectrum which is previouslymeasured in Au + Au collisions at

√sNN = 200 GeV [111]. The pT spectrum of electrons

is parameterized as follows:

EdN3

e

dp3∝ f(pT) =

(1 +

pT

p0

)p1

, (4.22)

where p0 and p1 are parameters summarized in Tab. 4.10. Figure 4.24 shows the raw pT

distributions of input and reconstructed electrons in the simulation. These pT shapes arealready modified by the pT weight.

The efficiencies εacc, εgeom, εtrack and εeID are determined by the PISA simulation.Figure 4.25 shows the pT dependence of the efficiencies for e+ and e− at the East-Southand the East-North, respectively. The black and the red histograms (from top 2 histogramsin each panel) show the pure geometrical acceptance before and after applying the fiducialcut. Those are calculated assuming what the detectors efficiency is 100%. The green andblack (thick) histograms show the efficiencies of εgeom × εtrack, εgeom × εtrack × εeID. Theseefficiencies are determined using only 5 σ matching cut at the EMCal and all the electronID cuts, respectively.

The obtained electron ID efficiency (εeID) is checked and confirmed by the real data.The electron ID efficiency in the real data is determined by using e+e− pairs from photonconversions [112]. These conversion pairs are selected by requiring the invariant mass

4.9. CORRECTION 79

(GeV/c)T

p0 1 2 3 4 5 6 7

0

0.02

0.04

0.06

0.08

0.1

0.12

+East-South e

+Acceptance correction function, East-South e

(GeV/c)T

p0 1 2 3 4 5 6 7

0

0.02

0.04

0.06

0.08

0.1

0.12

-East-South e

-Acceptance correction function, East-South e

(GeV/c)T

p0 1 2 3 4 5 6 7

0

0.02

0.04

0.06

0.08

0.1

0.12

+East-North e

+Acceptance correction function, East-North e

(GeV/c)T

p0 1 2 3 4 5 6 7

0

0.02

0.04

0.06

0.08

0.1

0.12

-East-North e

-Acceptance correction function, East-North e

Figure 4.25: The acceptances for e+/e− at the East-South/East-North, respectively. Ineach panel, the black and the red histograms (from top 2 histograms) show the puregeometrical acceptance with and without the fiducial cut. The green and the thick-blackhistograms show the acceptance determined by only 5 σ matching cut and all the electronID cuts.

within 60 < Me+e− < 90 MeV/c2 in the converter run. The combinatorial background isestimated by an event mixing method and subtracted to get the net yield of the conversionpairs. In order to determine the electron ID efficiency, the yields of the conversion pairsare compared with and without the standard electron ID cut. In the first case, the pairyield is calculated by selecting a track with the standard electron ID cut and the secondtrack in the pair without any eID cuts. In the second case, the pair yield is calculated byselecting both two track with the standard electron ID cut. Then, the ratio of these twoyields provide the electron ID efficiency. The efficiencies for several electron ID cuts aresummarized in Tab. 4.11.

In order to confirm the electron ID efficiency, we calculate an alternative correctionfunction by combining the pure geometrical acceptance in the simulation and the electronID efficiency by the real data. Then, the alternative function is compared with that byonly the simulation. Figure 4.26 shows the pT dependence of the acceptance correctionfunctions. The black and the green histograms correspond to the acceptance functions bythe simulation and the combined method, respectively. These acceptance functions are ingood agreement each other up to 5 GeV/c. We decided to use an simple average of thesetwo acceptance functions as the acceptance correction factor. The red histogram in Fig.4.26 shows the average of the acceptance.

Finally, we determine the geometrical acceptance (εgeom), the tracking efficiency (εtrack)and the eID efficiency (εeID) separately. These are calculated as follows.


(GeV/c)Tp0 1 2 3 4 5 6 7

0

0.01

0.02

0.03

0.04

0.05

0.06

0.07

0.08

0.09

0.1

+Acceptance correction function, East-South e

(GeV/c)Tp0 1 2 3 4 5 6 7

0

0.01

0.02

0.03

0.04

0.05

0.06

0.07

0.08

0.09

0.1

-Acceptance correction function, East-South e

(GeV/c)Tp0 1 2 3 4 5 6 7

0

0.01

0.02

0.03

0.04

0.05

0.06

0.07

0.08

0.09

0.1

+Acceptance correction function, East-North e

(GeV/c)Tp0 1 2 3 4 5 6 7

0

0.01

0.02

0.03

0.04

0.05

0.06

0.07

0.08

0.09

0.1



Figure 4.26: The acceptance functions for e+/e− at the East-South/East-North, respec-tively. In each plot, the green histogram is calculated from the simulation, and the blackis calculated from the real data. The red histogram is the average of the green and theblack one.

For the geometrical acceptance (εgeom),

εgeom =Ne

input, (4.23)

where Ne is the number of reconstructed tracks with the fiducial cut and “input” is thenumber of input electrons in the simulation. Here, the εgeom is normalized to dN/dy. Inthis calculation, no track and electron ID cut is required.

For the tracking efficiency (εtrack),

εtrack =Newith |dr| < 5

Ne

, (4.24)

where dr represents the matching cut (√

emcsdphi e2 + emcsdz e2). A numerator is selectedby 5 σ of the matching cut.

For the electron ID efficiency (εeID),

εeID =Newith eID cut

Newith |dr| < 5, (4.25)

where all of the standard electron ID cuts are required in the numerator. These obtainedvalues are summarized in Tab. 4.12. The systematic uncertainty of these values are de-scribed in section 4.12.

4.9. CORRECTION 81

Table 4.11: Electron ID efficiencies determined by the conversion pair analysis

Variables Efficiency√emcsdphi e2 + emcsdz e2 < 2.0 0.956

n0 >= 3 0.860chi2/npe0 < 10 0.933

disp < 5 0.966−2 < dep 0.958emcdt < 2 0.961

Table 4.12: Summary of the efficiencies determined in the PISA simulation.

Name Efficiency (%)εgeom 14.1εtrack 75.3εeID 64.7

4.9.2 Multiplicity Dependent Efficiency Loss

The detector performance depends on the detector occupancy since multi-particle signalsis merged into the single hit in the high multiplicity environment such as central Au + Aucollisions. In addition to the track reconstruction and the electron ID efficiencies, themultiplicity dependent efficiency loss (εmult) due to the detector occupancy need to betaken into account. The εmult is determined by embedding a simulated single electrontrack into a real event and measuring the probability that the embedded electron track iscorrectly reconstructed [113].

The efficiency loss is independent of pT. The efficiency losses are determined for min-imum bias and five centrality classes as shown in Tab. 4.13. The systematic uncertaintyon the εmult is assigned as 7%.

Table 4.13: Summary of the multiplicity dependent efficiency loss

Centrality εmult

minimum bias 0.7810-10% 0.738

10-20% 0.76020-40% 0.81740-60% 0.88760-92% 0.953


4.10 Heavy Flavor Electrons

Inclusive electrons contain three components as follows:

1. “Photonic electrons” : Background electrons from Dalitz decays of light neutralmesons (π0, η, η′, ρ, ω and φ) and photon conversions in the detector material.

2. “Non-Photonic electrons”

(a) “Heavy Flavor” : Signal electrons from semi-leptonic decays of hadrons con-taining charm and bottom quarks,

(b) “The Other Background”: Background electrons from Ke3 decays (K → πeν)and di-electron decays of ρ, ω and φ. The small contributions from J/ψ andDrell-Yan process are also included.

Heavy flavor electrons is the signal in this analysis and needs to be extracted frominclusive electrons. In order to separate the non-photonic and the photonic electron com-ponents, the converter method is used. After the separation, the third component (theother background) still remains. The third component is subtracted step by step. Theconverter method is described in section 4.10.1. The subtraction of the other backgroundis described in section 4.10.6 and 4.10.6.

4.10.1 Converter Method

The converter method is used to separate non-photonic and photonic electrons with highprecision. To extract non-photonic electrons, the inclusive electron yield in the converterrun is directly compared with that in the non-converter run. The photon converter is athin brass sheet of 1.7% radiation length which is described in section 3.6. The photonconverter increases only the yield of photonic electrons by a fixed factor Rγ determinedby its radiation length. Therefore, the comparison of these two datasets provides a clearseparation of these components. The yield of inclusive electrons with (NConv−in

e ) andwithout the converter (NConv−out

e ) is expressed as:

NConv−oute = Nγ

e +Nnon−γe , (4.26)

NConv−ine = RγN

γe + (1 − ε)Nnon−γ

e , (4.27)

where Nγe (Nnon−γ

e ) is the yield of photonic and non-photonic electrons, and ε representssmall loss of electrons due to the converter. This small loss is denoted as “blocking effect”.Next, the ratio of the electron yield (RCN) with and without the converter is defined asfollows:

RCN =NConv−in

e

NConv−oute

=Rγ + (1 − ε)RNP

1 +RNP

, (4.28)

where RNP is the ratio of non-photonic to photonic electrons (Nnon−γe /Nγ

e ). If there were nocontribution from non-photonic electrons, then RCN = Rγ. Eq. 4.28 can be transformedfor RNP:

RNP =Rγ −RCN

RCN − 1 + ε=Nnon−γ

e

Nγe

. (4.29)

4.10. HEAVY FLAVOR ELECTRONS 83

Ne

0

1.1% 1.7%

Dalitz : 0.8% X0 equivalent

0

With converter

W/O converter

0.8%

Conversions in the detector

Conversion in converter

Photonic

Non-photonic

Figure 4.27: Relation between the inclusive electron yield and the thickness of the detectormaterial.

RNP expresses a significance of non-photonic electrons relative to photonic electrons.

Finally, photonic and non-photonic electrons are determined as:

Nγe =

NConv−ine − (1 − ε)NConv−out

e

Rγ − 1 + ε, (4.30)

Nnon−γe =

RγNConv−oute −NConv−in

e

Rγ − 1 + ε. (4.31)

These equations indicate that Rγ is the key parameter to extract non-photonic elec-trons. Figure 4.27 shows the relation between the inclusive electron yield and the thicknessof the detector material in radiation length. In this figure, the non-photonic electron yieldis same in both with and without the converter. However, the photonic electron yield isincreased. The photonic electron yield per photon is approximately given by Y ∝ δ + 7

9t,

where δ is the branching ratio of Delitz decay per photon relative to 2 γ (for π0, η andη′) and 1 γ (for ρ, ω and φ) decay, and t is the thickness of the conversion material inradiation length. The factor 7

9is the approximate probability of photon conversions in one

unit of radiation length. For π0, the parameter δ is 0.6% which is half of the branchingratio of π0 Dalitz decays (1.198/2). By applying δ ' 0.8 and t ≈ 1.1% (t ≈ 2.8%) without(with) the converter, Rγ ≈ 1.9 is roughly estimated. Since there are some pT dependencefor δ, the complete formula of Rγ is determined using the PISA simulation.


4.10.2 Simulation Study

The simulation study is one of the most important parts in the converter method sinceRγ is determined by the PISA simulation.

The photonic source of electrons is a mixture of light neutral mesons (π0, η, η′, ρ, ωand φ) with different pT slopes. For these photonic sources, π0 is the main source andη is the second main source. The contributions of π0 and η relative to the all photonicsources are roughly 80% and 20%, respectively. The other sources contribute only smallfractions. Therefore, Rγ for π0 and η are first determined separately by the simulations,then they are combined with a weight of their relative yield. Here, the pT dependences ofthese particle yields are modeled by the cocktail calculation of the simple event generator[114] which is described in section 4.10.2. In the following, we denote that Rπ0

γ and Rηγ is

Rγ for π0 and η. In next sections, we describe the simulation study step by step.

Input for Simulation

The most important source of photonic electrons is π0. For the input of the PISA simu-lation, the pT slope of π0 is obtained by fitting simultaneously the pT spectra of π+, π−

[105] and π0 [115] in minimum bias Au + Au collisions measured by PHENIX. The twofitting functions corresponding the lower and higher pT ranges are expressed as:

f(pT) =1

2πpT

d2N

dydpT

, (4.32)

= p0 · (1 + pT/p1)p2 (0.5 < pT < 5.0 GeV/c), (4.33)

= p0 · (1 + pT/p1)p2 + p3 · p−8.0T (1.5 < pT < 12.0 GeV/c), (4.34)

where p0 − p3 are parameters summarized in Tab. 4.14.

Table 4.14: Parameters for the fit functions of π0.

pT range p0 p1 p2 p30.5 − 5.0 GeV/c 416.54 2.32603 -13.8865 -1.5 − 12.0 GeV/c 0.41898 356909 -763859 23.2689

Figure 4.28 shows the pT spectra of charged and neutral pions with the fit functions.The charged pion means an average of π+ and π−. Figure 4.29 shows the ratio of the pionspectra to the fit functions.

Cocktail Calculation

The other light mesons contributing to photonic electrons are η, η′, ρ, ω and φ. The shapesof the pT distributions for these mesons are determined by a mT scaling of π0 spectrum.In the mT scaling, a spectral shape of a hadron is assumed to be reproduced by replacing

the input π0 pT (pπ0

T ) with a hadron pT of phT =

√(pπ0

T )2 +m2h −m2

π0 . Here, mh, mπ0 are


(GeV/c)T

p0 2 4 6 8 10 12

inva

rian

t yi

eld

-710

-610

-510

-410

-310

-210

-110

1

10

210

pion spectra

Figure 4.28: The pT spectra of pionswith the fit functions. The red and theblue points correspond to charged andneutral pion, respectively

(GeV/c)T

p0 2 4 6 8 10 12

Rat

io

0

0.2

0.4

0.6

0.8

1

1.2

1.4

1.6

1.8

2Ratio of spectra to function

Figure 4.29: The ratio of the pion spec-tra to the fit functions.

mass of its hadron and π0, respectively. In addition, the absolute normalizations of the mT

scaled spectra are given by the ratio of the meson to pion yield at high pT (pT ∼ 5 GeV/c).The meson to pion ratios used in the analysis are summarized in Tab. 4.15. Here, the η/π0

ratio is a world average of measurement [116], and the φ/π0 ratio is consistent with thePHENIX measurement [119]. Using the pT spectra of these mesons, the pT distributionsof electrons from Dalitz decays and others are determined using the simple hadron decaygenerator [114]. The branching ratios of Dalitz decays for these mesons are summarizedin Tab. 4.16.

Simulation Dataset

Using the pT spectra determined by the mT scaling, extensive PISA simulations are per-formed. The several sets of the simulation data for π0 and η are generated to determineRγ. All simulations are performed at the RIKEN-CCJ.

The simulation datasets are summarized in the following list. All the particles inthe simulations are generated with a flat distribution in |y| < 0.6, 2π in azimuth and|z| < 30 cm.

• Five sets of π0 with different pT thresholds. The five pT thresholds correspond to0, 0.6, 1.0, 1.5 and 2.5 (GeV/c), respectively. The pT slope is determined by fittingthe measured pion spectra. All the datasets are produced with and without theconverter.

• Five sets of η with different pT thresholds. The five pT thresholds correspond to 0,0.6, 1.0, 1.5 and 2.5 (GeV/c), respectively. The pT slope is determined by the mT

scaling. All the datasets are produced with and without the converter.


Table 4.15: Mesons to pion ratios

η/π0 0.45 ± 0.1 [116, 117]η′/π0 0.25 ± 0.13 [118]ρ/π0 1.0 ± 0.5 [118]ω/π0 1.0 ± 0.5 [118]φ/π0 0.4 ± 0.2 [118]

Table 4.16: Branching ratios of the mesons decaying to electrons and photons

π0 → 2γ 98.798 ± 0.032 (%)π0 → γe+e− 1.198 ± 0.032 (%)η → 2γ 39.43 ± 0.26 (%)η → γe+e− 0.6 ± 0.08 (%)η′ → γe+e− < 9 × 10−4

ρ → e+e− (4.54 ± 0.1) × 10−5

ω → π0e+e− (5.9 ± 1.9) × 10−4

ω → e+e− (6.95 ± 0.15) × 10−5

φ → ηe+e− (1.15 ± 0.1) × 10−4

φ → e+e− (2.96 ± 0.04) × 10−4

The pT thresholds are introduced to increase statistics for high pT effectively. All of thesimulation data are generated with and without the converter for the Rγ determination.The π0 simulation is used not only for the electron analysis, but also for the electron pairanalysis.

4.10.3 Determination of Rγ

Determination of Rπ0

γ

We determine Rπ0

γ from the π0 simulation data. Rπ0

γ is the ratio of the photonic electronyields from π0 with and without the converter. Since π0 is the most dominant sourceof photons and photonic electrons, Rπ0

γ is very close to Rγ. A small correction due to ηcontribution need to be taken into account. The η contribution is described in the nextsections.

Figure 4.30 shows Rπ0

γ as a function of pT. Here, all five datasets of π0 simulations are

combined to increase the statistics. As a result, Rπ0

γ is approximately constant at 1.83 athigh pT, but there is a small pT dependence at low pT.

Determination of Rηγ

The η is the second dominant source of photonic electrons. Since the mass of η is largerthan π0, the available phase space of Dalitz decays for η is slightly larger than that of π0.


(GeV/c)Tp0 1 2 3 4 5 6

0 π γ

R

1

1.2

1.4

1.6

1.8

2

2.2

2.4

2.6

2.8

3

0πγcombined R

0πγcombined R

Figure 4.30: Rπ0

γ as a function of pT.

(GeV/c)Tp0 1 2 3 4 5 6

η γR

1

1.2

1.4

1.6

1.8

2

2.2

2.4

ηγcombined Rηγcombined R

Figure 4.31: Rηγ as a function of pT.

Therefore, the Dalitz branching ratio relative to two photon decay of η is slightly largerthan that of π0. The relative branching ratio (Dalitz decay / two photons) is 1.2% forπ0 and 1.5% for η. This difference makes Rη

γ smaller than Rπ0

γ . Figure 4.31 shows Rηγ

as a function of pT. All five datasets of η simulations are combined as same as the Rpi0

γ

calculation. The Rηγ is approximately constant at 1.6.

Total Rγ: Combine All Contributions

To obtain the total Rγ, Rπ0

γ and Rηγ are combined based on their relative yield in Au + Au

collisions at RHIC. The combined Rγ (Rπ0+ηγ ) is expressed as follows:

Rπ0+ηγ =

Rπ0

γ ·Nπ0

e +Rηγ ·Nη

e

Nπ0

e +Nηe

=Rπ0

γ +Rηγ · εη/π0

1 + εη/π0 , (4.35)

where Nπ0

e and Nηe are the electron yields from π0 and η decays, respectively. εη/π0

is theratio of the electron yields from Dalitz decays between π0 and η.

In addition to π0 and η, the contributions from the other hadrons (η’, ω and φ) needto be taken into account for the total Rγ determination. We treat that these contributionsto Rγ is small and and have similar shapes to Rη

γ, because the branching ratios of Dalitzdecays relative to photon decays for η’, ω and φ are similar to η. Therefore, we replaceεη/π0

with εh/π0in order to include these hadron contributions in the Rγ calculation. Thus,

the equation 4.35 is modified as follows:

Rπ0+ηγ → Rγ =

Rπ0

γ +Rηγ · εh/π0

1 + εh/π0 , (4.36)

where εh/π0is the ratio of the electron yields from Dalitz decays of all hadrons (h) except

π0 to π0, Here, the numerator (h) includes the contributions of η, η’, ω and φ.


The ratio εh/π0is obtained using the cocktail calculation of photonic electrons described

in section 4.10.2. Figure 4.32 shows the ratio of the electron yields from Dalitz decays of η,η’, ω, φ relative to π0, respectively. From these ratios, the ratio εh/π0

is obtained as shownin Fig. 4.33. The red, the green and the blue histograms correspond to the center value(e.g. η/π0 = 0.45), the lower (e.g. η/π0 = 0.35) and the upper limit (e.c. η/π0 = 0.55) ofthe systematic uncertainty, respectively.

The εh/π0is parameterized as follows:

εh/π0

= f(pT) = p0 + p1/√pT, (4.37)

where the p0 and p1 are the parameters as summarized in Tab. 4.17.

Table 4.17: The parameters of the fit function in Eq. 4.37

p0 0.313909p1 -0.098408

Finally, we obtained the total Rγ. Figure 4.34 shows the Rγ as a function of pT. Thered and the blue points are Rπ0

γ and Rηγ, respectively. The green represents the total Rγ.

4.10.4 Blocking Effect

In the converter run, electrons emitted inside the converter suffer a small energy loss,mainly due to bremsstrahlung, when electrons passes through the converter. This energyloss causes the reduction of the electron yields. We call this effect as “blocking effect”.Here, we consider the blocking effect for the single electron yield, even though the blockingeffect affects both the single electron and e+e− pair yields. The blocking effect for the pairsis already described in section 4.8.2.

The blocking effect should affect equally both photonic and non-photonic electrons.Since Rγ is simply calculated as the ratio of the electron yields with and without theconverter, Rγ automatically includes the blocking effect for photonic electrons. Thus, theblocking effect only appears for non-photonic electrons in Eq. 4.31. The factor ε in Eq. 4.31represents a partial loss of the non-photonic electron yield due to the blocking effect.

The blocking effect is studied by the PISA simulation of π0 Dalitz decays. Sinceelectrons from Dalitz decays originate at a collision vertex, these electrons undergo thesame blocking effect with non-photonic electrons in both the simulation and the real data.Therefore, we can study the blocking effect by the simulation. In the simulation study,the electron yields from π0 Dalitz decays are compared with and without the converter.Figure 4.35 shows the ratio of electrons from Dalitz decays with and without the converter.The ratio represents (1 − ε). As the result, the loss ε due to the blocking effect is ε =(1−0.979) = 0.021(2.1%). We assigned 25% of the systematic uncertainty for the blockingeffect.


(GeV/c)Tp0 1 2 3 4 5

Rat

io

0

0.5

1

1.5

2

2.5

3

3.5

4-1x10

Dalitzπ Dalitz / η

(GeV/c)Tp0 1 2 3 4 5

Rat

io

0

0.05

0.1

0.15

0.2

0.25

0.3-1x10

Dalitzπ’ Dalitz / η

(GeV/c)Tp0 1 2 3 4 5

Rat

io

0

0.2

0.4

0.6

0.8

1

1.2

-1x10 Dalitzπ Dalitz / ω

(GeV/c)Tp0 1 2 3 4 5

Rat

io

0

0.02

0.04

0.06

0.08

0.1

0.12

0.14

0.16

0.18

0.2-1x10

Dalitzπ Dalitz / φ

Figure 4.32: The ratio of the electron yields from Dalitz decays of the other hadronsrelative to π0.

(GeV/c)Tp0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5

0

0.05

0.1

0.15

0.2

0.25

0.3

0.35

0.4

0.45

0.5

Dalitz ratio0πhadron Dalitz /

= 0.450π/η : 0.35 (0.45 - 0.1)0π/η : 0.55 (0.45 + 0.1)

0π/η

Dalitz ratio0πhadron Dalitz /

Figure 4.33: the ratio of the electron yield from Dalitz decays of all other hadrons (h =η+η′+ω+φ) to π0. The red, the green and the blue histograms correspond to η/π0 = 0.45,0.35(0.45 − 0.1) and 0.55(0.45 + 0.1), respectively.


(GeV/c)Tp0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5

γR

1

1.2

1.4

1.6

1.8

2

2.2

2.4

0πγRηγR

γCombined R

γcombined R

Figure 4.34: The total Rγ from all hadrons decaying to electrons.

p_T(GeV/c)0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5

Rat

io o

f D

alit

z el

ectr

on

in C

on

v/N

oco

nv

0.75

0.8

0.85

0.9

0.95

1

1.05

1.1

1.15

1.2

1.25

Figure 4.35: The ratio of electrons from Dalitz decays with and without the converter.The ratio is determined by the PISA simulation of π0 Dalitz decays.


p_T (GeV/c)0 0.5 1 1.5 2 2.5 3 3.5 4

0.6

0.8

1

1.2

1.4

1.6

1.8

2

2.2

2.4

min. biasCNR

p_T (GeV/c)0 0.5 1 1.5 2 2.5 3 3.5 4

0.6

0.8

1

1.2

1.4

1.6

1.8

2

2.2

2.4

0-10%CNR

p_T (GeV/c)0 0.5 1 1.5 2 2.5 3 3.5 4

0.6

0.8

1

1.2

1.4

1.6

1.8

2

2.2

2.4

10-20%CNR

p_T (GeV/c)0 0.5 1 1.5 2 2.5 3 3.5 4

0.6

0.8

1

1.2

1.4

1.6

1.8

2

2.2

2.4

20-40%CNR

p_T (GeV/c)0 0.5 1 1.5 2 2.5 3 3.5 4

0.6

0.8

1

1.2

1.4

1.6

1.8

2

2.2

2.4

40-60%CNR

p_T (GeV/c)0 0.5 1 1.5 2 2.5 3 3.5 4

0.6

0.8

1

1.2

1.4

1.6

1.8

2

2.2

2.4

60-92%CNR 60-92%CNR

Figure 4.36: RCN in real data and Rγ in simulation for minimum bias and five centralityclasses.

4.10.5 Extraction of Non-photonic Electrons

The ratio of inclusive electrons (RCN) with and without the converter in real data isexpressed in Eq. 4.28. RCN is compared with Rγ. If there were no contribution from non-photonic electrons, then RCN = Rγ. Figure 4.36 shows the RCN and the Rγ as a functionof pT. The blue points and the red curves are the RCN and Rγ. From the top-left to thebottom-right, the panels correspond to minimum bias and five centrality classes (0-10%,10-20%, 20-40%, 40-60% and 60-92%), respectively. The difference between RCN and Rγ isseen clearly, and the RCN gradually decreases for high pT compared to Rγ. This differenceindicates that the existence of non-photonic electrons.

In order to show directly the significance of non-photonic electrons, the ratio RNP iscalculated from RCN and Rγ as expressed in Eq. 4.29. Figure 4.37 shows the “1 + RNP”distribution for minimum bias and five centrality classes. It denotes that “1 + RNP”is the ratio of the inclusive electron yield relative to the photonic electron yield. Thesignal of non-photonic electrons can appear above unity in Fig. 4.37. From the figure, thecontribution of non-photonic electrons increases with pT. The fraction of non-photonicelectrons are 40% relative to photonic electrons for pT > 1.0 GeV/c.

The raw yield of non-photonic electrons can be statistically extracted using Eq. 4.31.Figure 4.38 shows the raw pT distributions in minimum bias Au + Au collisions. The red,the light blue and the magenta spectra correspond to the pT distribution of inclusive,non-photonic and photonic electrons, respectively.


(GeV/c)T

p0 0.5 1 1.5 2 2.5 3 3.5 4

NP

1 +

R

0.5

1

1.5

2

2.5

for min.biasNP1+ R

(GeV/c)T

p0 0.5 1 1.5 2 2.5 3 3.5 4

NP

1 +

R0.5

1

1.5

2

2.5

for 0-10%NP1+ R

(GeV/c)T

p0 0.5 1 1.5 2 2.5 3 3.5 4

NP

1 +

R

0.5

1

1.5

2

2.5

for 10-20%NP1+ R

(GeV/c)T

p0 0.5 1 1.5 2 2.5 3 3.5 4

NP

1 +

R

0.5

1

1.5

2

2.5

for 20-40%NP1+ R

(GeV/c)T

p0 0.5 1 1.5 2 2.5 3 3.5 4

NP

1 +

R

0.5

1

1.5

2

2.5

for 40-60%NP1+ R

(GeV/c)T

p0 0.5 1 1.5 2 2.5 3 3.5 4

NP

1 +

R0.5

1

1.5

2

2.5

for 60-92%NP1+ R for 60-92%NP1+ R

Figure 4.37: The 1 + RNP distributions as a function of pT for minimum bias and fivecentrality classes. The points above 1.0 show a fraction of non-photonic electrons relativeto photonic electrons.

(GeV/c)Tp0 0.5 1 1.5 2 2.5 3 3.5 410

-8

10-7

10-6

10-5

10-4

10-3

10-2

10-1

raw yield in min. bias

inclusive electrons

photonic electrons

non-photonic electrons

raw yield in min. bias

Figure 4.38: Raw pT spectrum of inclusive electrons (green) in minimum bias Au + Aucollisions are decomposed into non-photonic (magenta) and photonic electrons (light blue).


4.10.6 Background from Ke3 decays and vector meson decays

After non-photonic electrons are extracted by the converter method, small backgroundsstill remain. These backgrounds are electrons from Ke3 decays and two electron decays oflight vector mesons. Electrons from J/ψ decays and Drell-Yan process are also background,but they are negligibly small especially for low pT. In this section, the backgrounds fromKe3 decays and light vector meson decays are described.

Ke3 Decay Background

The background electrons from Ke3 decays (K → πeν) are determined by the PISAsimulations. The input pT distribution for the simulation is obtained by fitting the pT

spectrum of kaon in minimum bias Au + Au collisions measured by PHENIX [105]. Here,the K+ and K− spectra are averaged for the fitting. The fit function is expressed asfollows:

1

2πpT

dN

dydpT

= f(pT) = p0 · e−pT/p1 , (4.38)

where p0 and p1 are the parameters. Figure 4.39 show the kaon spectrum with the fitfunction. The fit parameters are summarized in Tab. 4.18.

Table 4.18: The parameter obtained by fitting the kaon spectrum.

Parameter Valuep0 23.1648p1 (GeV/c) 0.317162

10M events are generated for each of K+, K− K0S and K0

L, respectively. Figure 4.40shows the raw pT distributions of electrons from Ke3 decays. From the left to the right,the panels correspond to electrons from K+, K−, K0

L and K0S decays, respectively.

The absolute normalization of electrons fromKe3 decays in the simulation is determinedusing dN/dy of kaons in minimum bias collisions measured in PHENIX. These dN/dy aresummarized in Tab. 4.19. The average dN/dy of K+ and K− are used for K0

L and K0S.

Table 4.19: dN/dy of K+, K− and 〈K〉 in minimum bias Au + Au collisions [105].

K+ K− 〈K〉dN/dy 13.6 ± 1.5 12.7 ± 1.2 13.1 ± 1.0

In order to obtain the electron spectra from Ke3 decays for the five centrality classes,the electron spectrum in simulation is modified using the pT weight which is the ratio ofthe kaon yield measured in minimum bias to that in each centrality. Figure 4.41 showsthe ratio of the kaon yield for minimum bias and the five centrality classes. For the


(GeV/c)Tp0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 210

-2

10-1

1

10

102

kaon spectra with a fit function

Figure 4.39: The averaged pT spectrum of K+ and K− with the fitting function.

(GeV/c)T

p0 0.5 1 1.5 2 2.5 3 3.5 4

-810

-710

-610

-510

-410

+e from K

(GeV/c)T

p0 0.5 1 1.5 2 2.5 3 3.5 4

-810

-710

-610

-510

-410

-e from K

(GeV/c)T

p0 0.5 1 1.5 2 2.5 3 3.5 4

-810

-710

-610

-510

-410

L0e from K

(GeV/c)T

p0 0.5 1 1.5 2 2.5 3 3.5 4

-810

-710

-610

-510

-410

S0e from KS0e from K

Figure 4.40: Raw pT distributions of electrons from K+, K−, K0L and K0

S decays, respec-tively

centrality dependence of electrons from Ke3 decays, the multiplicity dependent efficiencyloss described in section 4.9.2 is also taken into account.

Since the simulation accurately reproduces the detector response, the electron spectrafrom Ke3 decays in the simulation can be directly compared with the real data. Figure4.42 shows the ratio between electrons from Ke3 decays and photonic electrons. Fromthe top-left to the bottom-right, the panels correspond to the minimum bias and the fivecentrality classes, respectively. The contribution of electrons from Ke3 decays relative tophotonic electrons is 5% at low pT and smaller at high pT.


(GeV/c)T

p0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2

0

0.5

1

1.5

2

2.5

3

3.5

0-10% / min. bias

(GeV/c)T

p0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2

0

0.5

1

1.5

2

2.510-20% / min. bias

(GeV/c)T

p0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2

0

0.2

0.4

0.6

0.8

1

1.2

20-40% / min. bias

(GeV/c)T

p0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2

0

0.1

0.2

0.3

0.4

0.5

40-60% / min. bias

(GeV/c)T

p0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2

0

0.02

0.04

0.06

0.08

0.1

60%- / min. bias

Figure 4.41: The ratio of kaon spectra in minimum bias to that in five centralities

(GeV/c)T

p0 0.5 1 1.5 2 2.5 3 3.5 4

0

0.01

0.02

0.03

0.04

0.05

0.06

0.07

0.08

0.09

0.1 min.biasγ to eKratio of e

(GeV/c)T

p0 0.5 1 1.5 2 2.5 3 3.5 4

0

0.01

0.02

0.03

0.04

0.05

0.06

0.07

0.08

0.09

0.1 0-10%γ to eKratio of e

(GeV/c)T

p0 0.5 1 1.5 2 2.5 3 3.5 4

0

0.01

0.02

0.03

0.04

0.05

0.06

0.07

0.08

0.09


(GeV/c)T

p0 0.5 1 1.5 2 2.5 3 3.5 4

0

0.01

0.02

0.03

0.04

0.05

0.06

0.07

0.08

0.09


(GeV/c)T

p0 0.5 1 1.5 2 2.5 3 3.5 4

0

0.01

0.02

0.03

0.04

0.05

0.06

0.07

0.08

0.09


(GeV/c)T

p0 0.5 1 1.5 2 2.5 3 3.5 4

0

0.01

0.02

0.03

0.04

0.05

0.06

0.07

0.08

0.09

0.1 60-92%γ to eKratio of e 60-92%γ to eKratio of e

Figure 4.42: The ratio of electrons from Ke3 decays to photonic electrons


(GeV/c)T

p0 1 2 3 4 5 6

-1110

-1010

-910

-810

-710

-610

-510

-410

-310

-210Total

-e+ e→ ω-e+ e→ ρ-e+ e→ φ

in min. bias-e+ e→ φ, ρ, ωe from

(GeV/c)T

p0 1 2 3 4 5 6

-1110

-1010

-910

-810

-710

-610

-510

-410

-310

-210Total

-e+ e→ ω-e+ e→ ρ-e+ e→ φ

in 0-10%-e+ e→ φ, ρ, ωe from

(GeV/c)T

p0 1 2 3 4 5 6

-1110

-1010

-910

-810

-710

-610

-510

-410

-310

-210Total

-e+ e→ ω-e+ e→ ρ-e+ e→ φ


(GeV/c)T

p0 1 2 3 4 5 6

-1110

-1010

-910

-810

-710

-610

-510

-410

-310

-210Total

-e+ e→ ω-e+ e→ ρ-e+ e→ φ


(GeV/c)T

p0 1 2 3 4 5 6

-1110

-1010

-910

-810

-710

-610

-510

-410

-310

-210Total

-e+ e→ ω-e+ e→ ρ-e+ e→ φ


(GeV/c)T

p0 1 2 3 4 5 6

-1110

-1010

-910

-810

-710

-610

-510

-410

-310

-210


Total-e+ e→ ω

-e+ e→ ρ-e+ e→ φ


Figure 4.43: pT spectra of ω, ρ and φ→ e+e− for each centrality

Background from two electron decays of light vector mesons

The background electrons from two electron decays of light vector mesons (ρ, ω andφ → e+e−) are determined by the cocktail calculation described in section 4.10.2 andappendix B. Figure 4.43 shows the pT spectra of electrons from the decays of these lightvector mesons. From the top-left to the bottom-right, the panels correspond to minimumbias and the five centrality classes.

It should be noted that the mT scaling is slightly modified, since the mT scaling of piondoes not well reproduce the spectral shape of kaon and proton. The modified formula ofmT scaling is expressed as follows:

phT =

√(pπ0

T )2 +m · (M2h) −M2

π0 , (4.39)

where m is a factor to change the weight of the mass. In case of that m > 1 is set, thepT slope becomes more harder than the original mT scaling. Figure 4.44 shows the pT

spectra of kaon (left) and proton (right) with two mT scaled functions. The red and theblue curves correspond to the functions with m = 1.0 and m = 1.7, respectively. The bluecurve reproduces better than the red curve. Therefore, we used m = 1.7 for the modifiedmT scaling.

4.11. BIN WIDTH EFFECT 97

(GeV/c)Tp0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2

10-2

10-1

1

10

102

Kaon spectrum and a pion fit with mt scaling

(GeV/c)Tp0 0.5 1 1.5 2 2.5 3 3.5 4 4.5

10-4

10-3

10-2

10-1

1

10

102

Proton spectra and a pion fit with mt scaling

Figure 4.44: pT spectra of kaon (left) and proton (right) compared with 2 mT scalingfunctions. The red and blue correspond to mT scaling with m = 1.0 and m = 1.7,respectively.

4.11 Bin Width Effect

After the background subtractions and the acceptance corrections are performed, theinvariant differential yield of heavy flavor electrons is obtained. There still remains a binwidth effect due to the finite bin width of the obtained pT spectrum. In this analysis, wehave so far assumed that the mean pT (〈pT〉) in a bin is equal to the center of its bin in theelectron histogram. But this is not really correct if the electron spectrum has a steeplyfalling shape. In this case, the 〈pT〉 is shifted to lower pT than its bin center. This binwidth effect is larger for the wider bin width. Thus, we correct the bin width effect.

This correction is performed base on the PYTHIA heavy flavor spectrum described innext section. We first fit the PYTHIA spectrum by an empirical function as follows:

f(pT) =

(a

1 + b · pT + ec+d·pT · p3T

)e

, (4.40)

where a, b, c, d and e are the fit parameters. Using this function, we calculate the 〈pT〉 foreach pT bin. The data point in each pT bin is moved to its 〈pT〉 value.

4.12 Systematic Uncertainties

The systematic uncertainties are studied. The sources of the systematic uncertainties inthis analysis are mainly classified as follows:

1. Uncertainty of inclusive electrons : geometrical acceptance and electron identifica-tion.


2. Uncertainty of background subtraction : background from decays of Ke3 and vectormesons.

3. Uncertainty of the converter method.

These three uncertainties, 1, 2 and 3 are independent each other. Thus, these uncer-tainties are estimated separately.

4.12.1 Systematic Uncertainty of Inclusive Electrons

The systematic uncertainty of inclusive electrons are composed of the errors of the geo-metrical acceptance and the electron ID efficiency due to the applied cuts.

First, the uncertainty of the geometrical acceptance is considered. In this analysis,the dead and the hot area of the detectors are removed by applying the fiducial cut andthe acceptance in the simulation is tuned as described in section 4.8.1. To estimate theuncertainty of the geometrical acceptance, the position dependence of the electron yieldare compared in the real data and simulation. Figure 4.45 and 4.46 show the e+ and e−

yield as a function of the DC phi value. The red and the green correspond to the realdata and the simulation, respectively. From the comparison, we assign 10 % uncertaintyfor the geometrical acceptance.

Second, the statistical fluctuation of the acceptance correction function is assignedas a systematic uncertainty. The acceptance correction function is described in section4.9.1. This uncertainty is calculated as the ratio of the statistical error to the contentvalue in each pT bin of the correction function. Figure 4.47 shows the ratio of the errorand the content as a function of pT. From the figure, we assigned conservatively that thesystematic uncertainty is 3.0% at pT = 0.3 GeV/c and 2.5% for higher pT.

Third, the systematic uncertainties of the electron ID efficiency for each cut variable areestimated using inclusive electrons. We measured the fluctuation of the inclusive electronyields by changing the electron ID cuts tighter and looser. Then, the fluctuation of theelectron yield is assigned as the systematic uncertainty for each electron ID cut. Thetighter and looser cuts for the electron ID used in this study are summarized in Tab. 4.20.The obtained systematic uncertainties for each cut variables are also summarized in Tab.4.20. The systematic uncertainty of εmult is described in section 4.9.2. Figure 4.48 showsthe pT dependence of the systematic uncertainties for the electron ID variables. Finally,these uncertainties are added in quadrature to obtain the total systematic uncertainty ofthe electron ID. The total systematic uncertainty of the electron ID is 11.8%.

4.12.2 Systematic Uncertainty of Electrons from Ke3 and VectorMeson Decays

The systematic uncertainty of electrons from decays of Ke3 and light vector mesons areconsidered separately. The uncertainty from Ke3 decays includes two components. Oneis the systematic uncertainty of the kaon measurement (11.2%), and the other is theuncertainty of the electron ID (11.8%). These two components are added in quadratureto obtain the total uncertainty of Ke3 decays. To evaluate the uncertainty relative tonon-photonic electrons, these values are multiplied by the ratio of the electron yields from

4.12. SYSTEMATIC UNCERTAINTIES 99

phi2 2.2 2.4 2.6 2.8 3 3.2 3.4 3.6 3.8 40

50

100

150

200

250

300

350

400

450

Phi in Pt 0.3-0.5

phi2 2.2 2.4 2.6 2.8 3 3.2 3.4 3.6 3.8 40

20

40

60

80

100

120

140

160

180

200

220

Phi in Pt 0.5-0.8

phi2 2.2 2.4 2.6 2.8 3 3.2 3.4 3.6 3.8 4

0

10

20

30

40

50

Phi in Pt 0.8-1.0

phi2 2.2 2.4 2.6 2.8 3 3.2 3.4 3.6 3.8 4

0

5

10

15

20

25

30

35

40

45

Phi in Pt 1.0-2.2

Figure 4.45: Comparison of the e+ yield as a function of the DC phi.

phi2 2.2 2.4 2.6 2.8 3 3.2 3.4 3.6 3.8 40

100

200

300

400

500

Phi in Pt 0.3-0.5

phi2 2.2 2.4 2.6 2.8 3 3.2 3.4 3.6 3.8 40

50

100

150

200

250

Phi in Pt 0.5-0.8

phi2 2.2 2.4 2.6 2.8 3 3.2 3.4 3.6 3.8 4

0

10

20

30

40

50

Phi in Pt 0.8-1.0

phi2 2.2 2.4 2.6 2.8 3 3.2 3.4 3.6 3.8 4

0

10

20

30

40

50

Phi in Pt 1.0-2.2

Figure 4.46: Comparison of the e− yield as a function of the DC phi.


(GeV/c)T

p0 1 2 3 4 5 6

Rat

io (

%)

0

0.01

0.02

0.03

0.04

0.05

0.06

0.07

0.08

0.09

0.1

Ratio of Error to Data

Figure 4.47: Ratio between the statistical error and the value of the geometrical acceptancein the simulation.

Table 4.20: The Summary of the systematic uncertainties for the electron ID cut variablesand the others.

Variables Cut Systematic Commentstighter std. looser uncertainty√

emcsdphi e2 + emcsdz e2 1.5 2.0 2.5 3.0%n0 4.0 3.0 2.0 3.0%chi2/npe0 7.0 10.0 15.0 5.0%disp 3.0 5.0 7.0 3.0%dep −1.5 −2.0 −2.5 2.0%emcdt 1.0 2.0 3.0 2.0%MC statistics 2.5%Acceptance 5.0% 10% at low pT

εmult 7.0%Total 11.8 %


(GeV/c)T

p0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5

Rel

ativ

e E

rro

r (%

)

0

0.05

0.1

0.15

0.2

0.25

Systematic Uncertainty

quadratic sum of eID sys. errSystematic error on MC statistics CutSystematic error on Acc. matching in real and MCSystematic error on Matching CutSystematic error on N0Systematic error on Chi2Systematic error on DispSystematic error on Temc and TTofSystematic error on Dep(E-p/p/sigma)Systematic error on Embedding

Systematic Uncertainty

Figure 4.48: Systematic uncertainty of the acceptance and the electron ID variables

Ke3 decays to non-photonic electrons. Figure 4.49 (left) shows the uncertainty of electronsrelative to non-photonic electrons as a function of pT. From the figure, the uncertainty ofthis component contributes only at lower pT.

Next, the uncertainty from two electron decays of light vector mesons (ρ, ω and φ) isalso estimated. The uncertainty also includes two components. One is the meson to thepion ratios which are assigned 50% uncertainty for all mesons at high pT. The second isthe systematic uncertainty of the charged and the neutral pion measurement. These twocomponents are added in quadrature. Figure 4.49 (right) shows the pT dependence of theuncertainty of electrons from light vector meson decays. The uncertainty increases slightlyat high pT.

4.12.3 Uncertainty of The Converter Method

The uncertainty of the converter method consists of four components as listed below:

A Material budgets in the real data and the simulation (4.4%).

B η/π0 ratio (0.45 ± 0.1).

C Acceptance with and without the converter (2.5%).

D Blocking effect (25% of 0.021).

In the component A, the uncertainty of the material budget is estimated as 4.4% ine+e− pair analysis. Since the material budget affects directly Rγ, this 4.4% uncertainty can


(GeV/c)Tp0 0.5 1 1.5 2 2.5 3 3.5 4

Rel

ativ

e er

ror

0

0.01

0.02

0.03

0.04

0.05

0.06

0.07

0.08

0.09

0.1

Systematic uncertainty on e from kaon decays

(GeV/c)Tp0 0.5 1 1.5 2 2.5 3 3.5 4

Rel

ativ

e er

ror

0

0.01

0.02

0.03

0.04

0.05

0.06

0.07

0.08

0.09

0.1

Systematic uncertainty on e from VM->ee

Figure 4.49: Systematic uncertainty of electrons from Ke3 decays (left) and two electrondecays of light vector mesons (right).

propagate to the uncertainty of Rγ. In the section 4.8.2, the uncertainty of the materialbudget is estimated using Rmaterial = NMVD/NBP. Here, NBP and NMVD are the yieldof the conversion pairs at the beam pipe and the converter. The NBP also contains thecontribution from Dalitz decays. Those are expressed as:

NBP = Dalitz + beampipe + MVD(excluding outershell) (4.41)

∼ Dalitz + beampipe + MVD, (4.42)

NMVD = MVD outershell + converter (4.43)

∼ converter, (4.44)

where the contribution from the MVD outershell is small compared with the other contri-butions, then it is approximately neglected.

In the single electron analysis, Rγ is expressed as:

Rγ =Dalitz + beampipe + MVD + converter

Dalitz + beampipe + MVD, (4.45)

= 1 +converter

Dalitz + beampipe + MVD, (4.46)

∼ 1 +NBP

NMVD

= 1 +Rmaterial. (4.47)

Using this relation, the uncertainty of Rγ is determined as:

4.4% × Rγ − 1

Rγ

= 2.0%. (4.48)

We assign 2.0% uncertainty for Rγ.For the component B, the modification of Rγ due to the η/π0 ratio is studied by

changing the ratio within its error, η/π0 = 0.45 ± 0.1. The other mesons (η′, ω and φ) are


(GeV/c)T

p0 1 2 3 4 5 6

Rel

ativ

e E

rro

r (%

)

-0.05

-0.04

-0.03

-0.02

-0.01

0

0.01

0.02

0.03

0.04

0.05

ratio0π/η error due to γR ratio0π/η error due to γR

Figure 4.50: Relative error of Rγ calculated by changing the η/π0 ratio.

also taken into account for the modification of Rγ. Figure 4.50 shows the relative errorof Rγ calculated by changing the η/π0 ratio. The error is roughly 1% at 0.5 GeV/c andincrease slightly at high pT. To calculate the total uncertainty of Rγ, the uncertaintiesfrom the material budget and the η/π0 ratio are added in quadrature. The obtained totaluncertainty of Rγ is

√(2.0%)2 + (1.0%)2 = 2.2%.

In the component C, the systematic uncertainty of RCN is considered. If the acceptancebetween the converter run and the non-converter run is different, the electron yield in thesetwo runs can not be directly compared. This difference should include as a systematicuncertainty. Figure 4.10, 4.11, 4.12 and 4.13 show the phi angle dependence of the electronyield for e+ and e− at the sorth and nouth side of the DC, respectively. These phidistributions represent the acceptances of electrons. By comparing the acceptances withand without the converter, we assign that the systematic uncertainty of RCN is 2.5%.

Finally, we consider the error propagation of these four components to obtain the sys-tematic uncertainty of the non-photonic electron yield. These four components differentlycontributes to the uncertainty of Nnon−γ

e as shown in Eq. 4.31. Therefore, these contribu-tions are estimated separately and added in quadrature. The component A and B affectRγ. Thus, this can be expressed by changing Rγ to Rγ(1 ± δRγ ) in Eq. 4.31 as follows:

Nnon−γe =

Rγ(1 + δRγ )NConv−oute −NConv−in

e

Rγ(1 + δRγ ) − 1 + ε, (4.49)

where δRγ denotes the error value. The combined uncertainty of A and B is 2.2% deter-mined above.

The component C affects RCN. Since the uncertainty of RCN is due to the difference ofthe acceptance with and without the converter, NConv−in

e is changed to NConv−ine /(1+δRCN

)to represent the changing of the electron yield in the converter run as follows:

Nnon−γe =

Rγ(1 + δRCN)NConv−out

e −NConv−ine

(1 + δRCN)(Rγ − 1 + ε)

, (4.50)

where δRCN= ±0.025 is used as described above.


The component D affects ε. As same as the other components, In order to evaluatethe error of the blocking effect, ε is changed to ε+ δε as follows:

Nnon−γe =

RγNConv−oute −NConv−in

e

Rγ − 1 + ε+ δε, (4.51)

where δε = ±0.005 is used which is 25% of 0.021. Then, the uncertainties calculated bythese four components are added in quadrature.

Chapter 5

Results and Discussions

5.1 Invariant Yield of Heavy Flavor Electrons

The invariant differential yield of single electrons from semi-leptonic decays of heavy flavorshas been measured in Au + Au collisions at

√sNN = 200 GeV.

As described in previous sections, non-photonic electrons are extracted by the convertermethod from a large amount of photonic background electrons. The background electronsfromKe3 and vector meson decays are subtracted. After these backgrounds are subtracted,the only remaining significant source of non-photonic electrons is electrons from semi-leptonic decays of heavy flavors. Therefore, we denote the remaining electrons as heavyflavor electrons. Here, we neglect the contributions from J/ψ and Drell-Yan processes dueto their small contributions.

Figure 5.1 shows the invariant differential yield of heavy flavor electrons in minimumbias Au + Au collisions at

√sNN = 200 GeV. The error bars and brackets represent the

statistical and the systematic uncertainty, respectively. Figure 5.2, 5.3, 5.4, 5.5 and 5.6show the heavy flavor electron spectra for 0-10%, 10-20%, 20-40%, 40-60% and 60-92%centralities, respectively. The data points of these spectra are summarized in Appendix C.

5.2 Comparison with p + p result

The invariant differential yield of heavy flavor electrons in Au + Au collisions is comparedwith that in p+ p collisions at

√s = 200 GeV. The heavy flavor electron spectrum in p+ p

collisions was measured in PHENIX Run 2 [120]. Figure 5.7 shows the invariant yield ofheavy flavor electrons for minimum bias collisions and for the five centrality classes. Thesespectra are scaled by several powers of ten for clarity. The curves in Fig. 5.7 show the bestfit curve of the heavy flavor electron spectrum in p+ p collisions. The curves are scaled byTAA for the corresponding Au + Au centrality. Here, TAA is the nuclear overlap functioncalculated by the Glauber model described in section 4.2.3. The curves are also scaled bythe same factor as the Au + Au data point for clarity. The Au + Au data points are inreasonable agreement with the TAA scaled p+ p fit curve in all centrality classes withinrelatively large error at high pT.

Recently, the heavy flavor electron spectrum in p+ p collisions is extended to high pT

105

106 CHAPTER 5. RESULTS AND DISCUSSIONS

(GeV/c)T

p0 0.5 1 1.5 2 2.5 3 3.5 4

)2

dy

((c/

GeV

)T

)dN

/dp

T pπ

(1/2

-710

-610

-510

-410

-310

-210

-110

1

)/2 in min. bias Au+Au collisions-+e+Non-photonic (e )/2 in min. bias Au+Au collisions-+e+Non-photonic (e

Figure 5.1: Invariant differential yield ofheavy flavor electrons in minimum biasAu + Au collisions at

√sNN = 200 GeV.

(GeV/c)T

p0 0.5 1 1.5 2 2.5 3 3.5 4

)2

dy

((c/

GeV

)T

)dN

/dp

T pπ

(1/2

-710

-610

-510

-410

-310

-210

-110

1

)/2 in 0 - 10 % Au+Au collisions-+e+Non-photonic (e )/2 in 0 - 10 % Au+Au collisions-+e+Non-photonic (e

Figure 5.2: Invariant differential yield ofheavy flavor electrons in 0-10% centralAu + Au collisions at

√sNN = 200 GeV.

(GeV/c)T

p0 0.5 1 1.5 2 2.5 3 3.5 4

)2

dy

((c/

GeV

)T

)dN

/dp

T pπ

(1/2

-710

-610

-510

-410

-310

-210

-110

1



√sNN = 200 GeV.

(GeV/c)T

p0 0.5 1 1.5 2 2.5 3 3.5 4

)2

dy

((c/

GeV

)T

)dN

/dp

T pπ

(1/2

-710

-610

-510

-410

-310

-210

-110

1



√sNN = 200 GeV.

5.2. COMPARISON WITH P + P RESULT 107

(GeV/c)T

p0 0.5 1 1.5 2 2.5 3 3.5 4

)2

dy

((c/

GeV

)T

)dN

/dp

T pπ

(1/2

-710

-610

-510

-410

-310

-210

-110

1



√sNN = 200 GeV.

(GeV/c)T

p0 0.5 1 1.5 2 2.5 3 3.5 4

)2

dy

((c/

GeV

)T

)dN

/dp

T pπ

(1/2

-710

-610

-510

-410

-310

-210

-110

1



√sNN = 200 GeV.

with smaller statistical uncertainty in Run 5 [39]. The pT spectrum in p+ p collisionsin Run 2 is consistent with that in Run 5 within statistical and systematic uncertainty.The top panel in Fig. 5.9 shows the invariant differential cross sections of heavy flavorelectrons in p+ p collisions at

√s = 200 GeV. The red and black points are the electrons

measured in Run 2 and Run 5 with statistical error bars. The brackets and the yellowbands represents the systematic uncertainties. These electron spectra in p+ p collisions arealso compared with a Fixed-Order-Plus-Next-to-Leading-Log (FONLL) pQCD calculation[33] as described in section 2.2.1. The top curve in Fig. 5.9 shows the central value ofthe FONLL calculation. The contributions of charm and bottom are also shown. Thecontribution of bottom decays becomes larger than that from charm decays above pT >4 GeV/c. The bottom panel in Fig. 5.9 shows the ratio of the data to the FONLLcalculation. The ratio is nearly independent of pT over the entire pT range and is 1.71 ±0.02(stat)± 0.18(sys) obtained by fitting. This indicates that the spectral shape of heavyflavor electrons in p+ p collisions is explained by the FONLL. From these comparisons ofthe spectral shape in Au + Au, p+ p and the FONLL, the spectral shape of heavy flavorelectrons in Au + Au collisions is also explained by the FONLL within large error

In p+ p collisions, the ratio of electrons from bottom decays to heavy flavor electronshas been measured by the electron-hadron correlations [123]. The ratio is compared withthe FONLL calculation. Figure 5.8 shows the ratio of (b → e)/[(b → e) + (c → e)] asa function of pT. The solid and dashed curves represent the FONLL calculation and itssystematic uncertainties. The FONLL calculation agrees with the measured ratio. Thisresult confirms that the main contribution is charms up to pT < 4 GeV/c for the spectra


(GeV/c)T

p0 0.5 1 1.5 2 2.5 3 3.5 4

)2 (

(c/G

eV)

T)d

N/d

ydp

T pπ

(1/2

-810

-710

-610

-510

-410

-310

-210

-110

1

10

210

310 =200GeVNNs)/2 + X @ -+e+ (e→Au+Au

min. bias3 10× 0 - 10 % 2 10× 10 - 20 % 1 10× 20 - 40 % -1 10× 40 - 60 % -2 10× 60 - 92 %

Best fit curve of pp

Figure 5.7: Invariant differential yields of heavy flavor electrons for different Au + Aucentralities scaled by several powers of ten for clarity. Error bars and brackets correspondto the statistical and the systematic uncertainties, respectively. The curves are the bestfit curves of electrons in p+ p collisions scaled by TAA for the corresponding Au + Aucentrality.

5.3. CENTRALITY DEPENDENCE OF CHARM ELECTRON YIELD 109

(GeV/c)T

Electron p2 3 4 5 6 7

e)

→ e

+b→

e/(

c→b

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1PHENIX |y| <0.35

FONLL y=0

FONLL error band y=0

=200 GeVs p+p at

90% C.L.

90% C.L.

Figure 5.8: The ratio of bottom contribution to heavy flavor electrons as a function of pT

in p+ p collisions at√s = 200 GeV [123].

in Au + Au collisions.

5.3 Centrality Dependence of Charm Electron Yield

In order to quantify the centrality dependence of the charm production, the integratedyield (dNe/dy) of heavy flavor electrons is calculated with 0.8 < pT < 4.0 GeV/c foreach centrality class. In this pT range, the charm contribution is dominant as shown inFig. 5.9. Therefore, we denote that heavy flavor electron within this pT range is charmelectron. Although the integration pT range should be as wide as possible to study the totalproduction of charm electrons, the high pT is limited up to 4.0 GeV/c due to the increaseof bottom contribution for higher pT. By contrast, the low limit of the pT range is chosenfor pT > 0.8 GeV/c where the systematic uncertainty is comparable with the statisticalerror. Then, the calculated integrated yield is fit by an empirical function, ANα

coll, whereA is a normalization parameter and α is the parameter which describes the centralitydependence. If there is no initial and final state effects on the total charm production, weexpect α = 1. Figure 5.10 shows dNe/dy(0.8 < pT < 4.0)/Ncoll as a function of Ncoll forminimum bias and the five centrality classes in Au + Au and p+ p collisions. The graybox and the black circles correspond to the minimum bias data and the five centralityclasses. The cross symbol shows the p+ p data. The error bars and the brackets representthe statistical and systematic uncertainties. We find α = 0.938± 0.075(stat)± 0.018(sys).If p+ p data is included for the fitting, α = 0.958 ± 0.035(stat). In this fitting, most ofthe systematic uncertainties will cancel. The systematic uncertainty of α is described inAppendix D.1. This shows that the total yield of charm electrons for all centralities isconsistent with Ncoll scaling.

This result is confirmed by the updated data with much higher statistics in measured


)2/G

eV3

(m

b c

3/d

p3 σ

E d

-910

-810

-710

-610

-510

-410

-310

-210

-110

PHENIX data Run 2PHENIX data Run 5FONLL (total)

e)→FONLL (c e)→FONLL (b

e)→ c →FONLL (b

=200GeVs)/2 + X at -+e+ (e→p+p

(GeV/c)T

p0 1 2 3 4 5 6 7 8 9 10

DA

TA

/FO

NL

L

0.5

1

1.5

2

2.5

3

3.5

4

Figure 5.9: (top)Invariant differential cross sections of electrons from heavy flavor decaysin p+ p collisions at

√s = 200 GeV. The red and the black points correspond to the Run

2 data [120] and Run 5 data [39] with statistical errors, respectively. The brackets and theyellow bands are systematic uncertainty for Run 2 and Run 5, respectively. The curvesare the FONLL calculation. (bottom) Ratio of the data and the FONLL calculations. Theupper (lower) curves shows the theoretical upper (lower) limit of the FONLL calculation.

5.3. CENTRALITY DEPENDENCE OF CHARM ELECTRON YIELD 111

collN0 200 400 600 800 1000 1200

Co

ll/d

y

/ N

ed

N

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

-4x10

(m

b)

AA

/dy

/ Te

dN

0

0.05

0.1

0.15

0.2

0.25

0.3

0.35-2x10

centrality binned

min-bias

= 200 GeVsp + p at

<4.0

)T

(0.8

<p

<4.0

)T

(0.8

<p

Figure 5.10: dNe/dy/Ncoll as a function of Ncoll in Au + Au and p+ p collisions at√sNN =

200 GeV. The gray box and the black circles correspond to the minimum bias data andfive centrality data. The cross symbol shows the p+ p data. The error bars and thebrackets represent the statistical and systematic uncertainties, respectively.

in PHENIX Run 4 [121]. Figure 5.11 shows the updated dNe/dy/TAA as a function ofNcoll. The Au + Au and the p+ p data are measured in Run 4 and Run 5[39]. Theintegration range is slightly broadened for the updated data due to higher statistics. Theupdated α is α = 1.009 ± 0.012(stat) ± 0.051(sys). If the p + p data is included, α =1.010 ± 0.009(stat) ± 0.040(sys). All α values are consistent with unity within statisticaland systematic uncertainty.

This is an important finding that the total charm production in Au + Au collisions isconsistent with Ncoll scaling of the production in p+ p collisions, as expected for point likepQCD processes. Therefore, the result rules out a large enhancement of total charm yielddue to the secondary and/or the athermal charm production predicted in [18]. In addition,this result also gives the important baseline for charm suppression. It is found that theheavy flavor electrons are suppressed and its RAA approaches the π0 value at pT > 4 GeV/c[121]. Therefore, it can be understood that most of charm quarks are produced in initialhard scatterings and they suffer the energy loss when they pass through the dense matter.

Even though the total charm production is consistent with the Ncoll scaling, theremight be small room for an possible secondary/thermal charm production.


collN0 200 400 600 800 1000 1200

(m

b)

AA

>0.4

) / T

T/d

y (p

ed

N

0.003

0.004

0.005

0.006

0.007

0.008

0.009 centrality binned

min. biasp + p

Figure 5.11: dNe/dy/TAA as a function of Ncoll in Au + Au and p+ p collisions measuredin Run 4 [121] and Run 5 [39]. The gray box and the black circles correspond to theminimum bias data and the five centrality data. The cross symbol shows the p+ p data.The error bars and the brackets represent the statistical and systematic uncertainties,respectively.

Table 5.1: Summary of α values obtained by fitting.

Data for fitting αAu + Au 0.938 ± 0.075(stat) ± 0.018(sys)Au + Au and p+ p 0.958 ± 0.035(stat)Updated Au + Au 1.009 ± 0.012(stat) ± 0.0506(sys)Updated Au + Au and p+ p 1.010 ± 0.009(stat) ± 0.0403(sys)

5.4. COMPARISON WITH D + AU COLLISIONS 113

collN0 2 4 6 8 10 12 14 16 18

in d

Au

coll

<4.0

) / N

T/d

y(0.

8<p

ed

N

-0.02

0

0.02

0.04

0.06

0.08

0.1

0.12

0.14

0.16

0.18-310×

CentralityBinmin.biasp+p Run 5

=200GeVs<4.0) in d + Au @ T

(0.8<pcoll

/dy / NedN

Figure 5.12: dNe/dy/Ncoll(0.8 < pT < 4.0 GeV/c) in d+Au collisions at√sNN = 200 GeV.

5.4 Comparison with d + Au collisions

Heavy flavor electrons in d + Au collisions at√sNN = 200 GeV are also measured by

PHENIX [74]. The Ncoll scaled yields of heavy flavor electrons (dNe/dy/Ncoll(0.8 < pT <4.0 GeV/c)) for minimum bias d+ Au collisions and four centrality classes are calculatedwith 0.8 < pT < 4.0 GeV/c in the same way as the Au + Au data. The four centralityclasses correspond to 0-20%, 20-40%, 40-60% and 60-88%, respectively. Figure 5.12 showsthe dNe/dy/Ncoll(0.8 < pT < 4.0) as a function of Ncoll in d+Au collisions. The error barsand the brackets represent the statistical and the systematic uncertainties, respectively.The p+ p data point is also shown in this figure. From the figure, the dNe/dy/Ncoll(0.8 <pT < 4.0) for all the centrality classes in d+Au collisions and for p+ p collisions are foundto be consistent within their statistical and systematic uncertainties. Thus, this resultindicates that the total production of charms in d + Au collisions is consistent with Ncoll

scaling. This implies that the modification of heavy flavor production due to the shadowingeffect of the cold nuclear effect is small. However, the statistical and systematic error toolarge to make any strong constraints.

The FNAL-E866 experiment was reported the heavy flavor production in p+A collisionsat

√s = 38.8 GeV [124]. In order to compare the heavy flavor production in d + Au

collisions with that in p+ p collisions, they define the following relation:

σccpA = Aα · σcc

pp, (5.1)

where A is the mass number of the target nucleus and α is the parameter to study thecold nuclear effect. If there is no modification due to the cold nuclear effect, then, α = 1.


Figure 5.13: The parameter α of open charm and J/ψ as a function of rapidity (left) andpT (right).

Figure 5.13 show α of open charm and J/ψ as a function of rapidity (left) and pT (right).In the left panel of Fig. 5.13, α of open charm is unity near y = 0 and decreases graduallywith increasing y. This result of α = 1 at y = 0 could indicate that the shadowing effectis small. This is the same tendency with that in d + Au collisions at

√sNN = 200 GeV,

even though the small effect is expected since the related x region in the target nucleusis x ∼ 0.1 where the shadowing effect is expected to be small compared with that ofthe RHIC energy (x ∼ 0.02). For the forward rapidity region, α is smaller than unity.This might indicate that the modification of the charm production due to the shadowing.In a model [125], the intrinsic charm in a projectile is expected to contribute the charmproduction. Since the intrinsic charm in a nucleus is depend on A2/3 in the model, thecharm production in p+A collisions is expected to be reduced compared with that in p+pcollisions. This is also a possible model to describe the experimental result at the forwardrapidity. Therefore, it is difficult to discuss the modification of the charm production atthe forward rapidity region.

5.5 Medium Modification of Heavy Quarks

PHENIX reported that the RAA of heavy flavor electrons shows a strong suppression athigh pT as described in section 2.4.4. The RAA value of heavy flavor electrons is aboutRAA = 0.35 for pT > 4 GeV/c, which is similar with that of π0. This clear suppressionat high pT is seen for 0-10%, 10-20%, 20-40% centrality and minimum bias events. Thisresult indicates that not only light quarks, heavy quarks also lose their energy in the densematter created in Au + Au collisions.

The BDMPS and GLV models including gluon radiations failed to describe this strongsuppression of heavy flavor electrons. These predictions show a larger RAA compared withthe data. The collisional energy loss is another possible model to describe the measuredsuppression of heavy flavor. The updated GLV model including both the radiative andthe collisional energy loss can describe the substantial part of the energy loss. But thereis still small difference between the model and the data.

Recently, the ratio of the bottom contribution to heavy flavor electrons was measured

5.6. TOTAL CHARM CROSS SECTION 115

in p+ p collisions at√s = 200 GeV as shown in Fig. 5.8. The bottom contribution is

about 40% relative to heavy flavor electrons and is increasing with pT. Using this result,we consider the energy loss of bottoms, although bottoms are expected to suffer smallerenergy loss than that of charms due to their large mass. If the initial composition ofcharms and bottoms in Au + Au collisions were the same with that in p+ p collisions andbottoms suffer no energy loss in the dense matter, the expected RAA is same or even alittle larger than the measure RAA within the statistical and the systematic uncertainty,even though in the case of all of charms are suppressed. This may interpret that bottomsalso suffer an possible energy loss in the dense matter. But the current measurement cannot set significant limits of the bottom energy loss.

5.6 Total Charm Cross Section

5.6.1 Calculation of Cross Section

In this section, we evaluate the total charm cross section per nucleon-nucleon collision.The procedure to obtain the total charm cross section is described step by step:

(1) We calculate the integrated invariant yield of charm electrons per nucleon-nucleoncollision (dNe

dy(0.8 < pT < 4.0)/TAA) with the limited pT range (0.8 < pT < 4.0 GeV/c).

The dNe

dy(0.8 < pT < 4.0)/TAA is equivalent to the integrated cross section of charm

electrons per N +N collisions (dσe

dy(0.8 < pT < 4.0)).

(2) Assuming the PYTHIA calculation can reproduce the spectral shapes of charm andbottom quarks, the dσe

dy(0.8 < pT < 4.0) is extrapolated to the full pT range. The

conversion factor (RpT= dσe

dy(0.8 < pT < 4.0)/dσe

dy) is 12.6%. The RpT

is determinedby the PYTHIA electron spectrum from charm and bottom decays.

dσe

dy=

1

RpT

dσe

dy(0.8 < pT < 4.0) (5.2)

(3) The electron cross section dσe

dyis translated to the charm cross section dσcc

dyusing a total

branching ratio of charm decaying to electron. The total branching ratio of BR(c→ e)is 9.5%. The bottom is neglected due to a small contribution for the integrated dσe

dy.

dσcc

dy=

1

BR(c→ e)

dσe

dy(5.3)

(4) The total charm cross section (σcc) is finally obtained to extrapolate to the full rapidityrange. It is assumed that the rapidity distribution of charm can be modeled byPYTHIA. The extrapolation factor (Ry) is 23.0%.

σcc =1

Ry

dσcc

dy(5.4)


(GeV/c)Tp6 8 10 12 14 16 18 20

0

0.1

0.2

0.3

0.4

0.5

0.6

D+/D0 ratio at CDFD+/D0 ratio at CDF

p_T[GeV/c]6 8 10 12 14 16 18 20

0

0.05

0.1

0.15

0.2

0.25

0.3

0.35

Ds/D0 ratio at CDFDs/D0 ratio at CDF

Figure 5.14: D+/D0 ratio (left) and Ds/D0 ratio (right) measured at CDF [128].

Since the conversion and the extrapolation factors used in this analysis are determinedby PYTHIA, we first describe a PYTHIA calculation with a modified set of parameters.We used PYTHIA 6.205 with Mc = 1.25 GeV/c2, Mb = 4.1 GeV/c2, 〈kT〉 = 1.5 GeV/c,K factor = 3.5 and PDF = CTEQ5L for the input. Here, Mc and Mb are charm andbottom quark mass, 〈kT〉 is the width of intrinsic parton kT smearing in a proton, andthe K factor is a constant to take into account the higher order correction (e.g. NLO).These parameters are determined to reproduce the charm production data in lower energyexperiments at SPS and FNAL [127, 126] and the charm electron production data at ISRexperiments [22, 23, 24]. This set of the PYTHIA parameters were used for previousPHENIX electron analysis [114].

The total branching ratio of charm decaying to electron is obtained to combine thecharmed hadron ratio and their electron branching ratios. The branching ratio is sensitiveto the charmed hadron ratio, since the ratio of the charged D+ meson is quit differentfrom that of the neutral D0 meson. The charmed hadron ratio in default PYTHIA doesnot agree with the experimental data. The CDF experiment reported the D0 and D+

cross section in√s = 1.96 TeV p + p collisions [128]. Figure 5.14 show the D+/D0 ratio

(left) and Ds/D0 ratio (right) measured at CDF. The STAR experiment also reported the

D+/D0 ratio at√sNN = 200 GeVd+ Au collisions [129]. In addition, the ratio in e+ + e−

experiments is summarized by the Particle Data Group [27].

From these experimental data, we assume in this analysis the these ratios to be 0.45,0.25 and 0.1 for D+/D0, Ds/D

0 and Λc/D0, respectively, as shown in Tab. 5.2. The total

electron branching ratio (BR(c → e)) is obtained by weighted averaging these electronbranches with their charmed hadron ratios. This modification of the electron branch isapplied to the PYTHIA calculation. This total branching ratio (BR(c → e)) is also usedwhen dσe

dyis translated to dσcc

dyin step 3.

Figure 5.15 shows the differential cross section of electrons from charm and bottomdecays calculated by PYTHIA. The green and blue points represent the charm and bot-tom contributions, and the red point is the combined contribution of charm and bottom,


Table 5.2: D+/D0, Ds/D0, Λc/D

0 ratios in the experimental data and PYTHIA.

CDF STAR PDG PYTHIA This analysis(p + p) (d + Au) (e+ + e− at

√s = 91GeV)

D+/D0 0.45 0.4 0.3 0.45 ± 0.1Ds/D0 0.23 0.29 0.2 0.25 ± 0.1Λc/D0 0.17 0.1 0.1 ± 0.05

Table 5.3: Electron branching ratio of charmed hadrons

Particles Ratio to D0 BR(c → e + X)D0 1 6.87 ± 0.28%D+ 0.45 ± 0.1 17.2 ± 1.9%Ds 0.25 ± 0.1 8 ± 5%Λc 0.1 ± 0.05 4.5 ± 1.7%

Total BR(c → e) 9.5 ± 0.4%

Table 5.4: The parameters obtained by fitting the PYTHIA electron spectrum.

a ((c/GeV)2) 0.00101594b ((GeV/c)) 1.4598c 4.11875d ((c/GeV)2) 1.2852

respectively. Here, the PYTHIA electron spectrum is shown as the invariant yield usingTpp = 1/σpp = 0.238 mb−1 (σpp = 42 mb).

This PYTHIA electron spectrum is parameterized as follows:

f(pT) = a ·(

1

1 + pT/b+ d · p2T

)c

, (5.5)

where the parameters a, b, c and d are obtained by fitting. The magenta curve in Fig.5.15 show the fit function. The obtained parameters are summarized in Tab. 5.4. Usingthe fit function, we obtained the conversion factor RpT

= 12.6%.Now all the extrapolation and the conversion factors are prepared. We can calcu-

late the total charm cross section step by step. In step 1, In order to calculate thedσe/dy(0.8 < pT < 4.0), the charm electron spectrum is scaled with TAA and fit withthe PYTHIA electron function within 0.8 < pT < 4.0 GeV/c. In this fitting, the spec-tral shape is fixed and the normalization is only a free parameter. Figure 5.16 shows thescaled invariant yields of charm electrons ( 1

TAA

12πpT

dNe

dydpT) with the fit function. The panels

from the top-left to the bottom-right correspond to minimum bias and the five centralityclasses, respectively.


(GeV/c)T

p0 1 2 3 4 5 6

)2 (

(c/G

eV)

TN

/dy

dp

2 d

T pπ

1/2

-1010

-910

-810

-710

-610

-510

-410

-310

-210PYTHIA calculation

e→charm

e→bottom

e→charm + bottom

e→fit c + b

Figure 5.15: Invariant yield of electrons from charm and bottom decays calculated byPYTHIA. The green and the blue points represent the charm and the bottom contribu-tions, and the red point is the combined contribution of charms and bottoms, respectively.

(GeV/c)Tp

0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5

T/d

ydp

e)d

NT

pπ)(

1/2

AA

(1/T

10-8

10-7

10-6

10-5

10-4

10-3

10-2

10-1

Min. Bias

(GeV/c)Tp

0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5

T/d

ydp

e)d

NT

pπ)(

1/2

AA

(1/T

10-8

10-7

10-6

10-5

10-4

10-3

10-2

10-1

0-10%

(GeV/c)Tp

0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5

T/d

ydp

e)d

NT

pπ)(

1/2

AA

(1/T

10-8

10-7

10-6

10-5

10-4

10-3

10-2

10-1

10-20%

(GeV/c)Tp

0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5

T/d

ydp

e)d

NT

pπ)(

1/2

AA

(1/T

10-8

10-7

10-6

10-5

10-4

10-3

10-2

10-1

20-40%

(GeV/c)Tp

0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5

T/d

ydp

e)d

NT

pπ)(

1/2

AA

(1/T

10-8

10-7

10-6

10-5

10-4

10-3

10-2

10-1

40-60%

(GeV/c)Tp

0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5

T/d

ydp

e)d

NT

pπ)(

1/2

AA

(1/T

10-8

10-7

10-6

10-5

10-4

10-3

10-2

10-1

60-92%60-92%

Figure 5.16: Scaled differential invariant yields 1TAA

12πpT

dσe

dydpTfor minimum bias and the

five centrality classes. The spectrum is fit by the PYTHIA electron function with 0.8 <pT < 4.0 GeV/c (blue)


Using the fit function, the integrated yield dσe/dy(0.8 < pT < 4.0) is calculated. Theresult is summarized in Tab. 5.5. Then, these integrated dσe/dy(0.8 < pT < 4.0) areconverted to the total charm cross section σcc per N +N collision.

Table 5.5: dσe/dy (0.8 < pT < 4.0 GeV/c) for minimum bias and the five centrality classes.The systematic uncertainty represents the combined uncertainty of the spectrum and TAA.

Centrality Ncoll dNe/dy (µb) stat.error sys.error(0.8 < pT < 4.0)

Min. bias 257.8 1.707 ± 0.157 ± 0.3730-10 % 955.4 1.638 ± 0.256 ± 0.35710-20 % 602.6 1.634 ± 0.316 ± 0.34920-40 % 296.8 2.006 ± 0.321 ± 0.44640-60 % 90.65 2.308 ± 0.562 ± 0.55960-92 % 14.83 1.382 ± 1.036 ± 0.486

To estimate the systematic uncertainty on the total charm cross section, we considerthe following sources of the uncertainty:

(1) the uncertainty in dNe/dy(0.8 < pT < 4.0) obtained from the pT distribution of heavyflavor electrons.

(2) the uncertainty in Ncoll.

(3) the uncertainty in the shape of the fit function.

(4) the uncertainty in the method to calculate the total charm cross section.

(5) the uncertainty in the pT range of the fitting.

The uncertainty of dNe/dy(0.8 < pT < 4.0) of heavy flavor electrons (source (1)) isestimated (21%) by refitting the electron spectrum at the higher and lower error band. Theuncertainties are summarized in Tab. 5.5. Here, the uncertainty in Tab. 5.5 representsthe combined uncertainty of the source (1) and (2).

For the source (3), we tested the two other methods to calculate dNe/dy(0.8 < pT <4.0). One is to count the bin entries in the pT range and the other is to fit the spectrumwith a power-law function. The uncertainty of the source (3) is summarized in Tab. 5.7.

For the source (4), we test the another conversion and the extrapolation factors , RpT

in step (2), BR(c → e) in step (3) and Ry (4). Since the intrinsic 〈kT〉 can modifythe shape of the pT spectrum, we test the two other 〈kT〉 values, 1.0 and 2.0 insteadof 1.5 GeV/c. Then, the RpT

= 11.5% and 14.1% is determined for 〈kT〉 = 1.0 and2.0 GeV/c, respectively. Thus, the RpT

uncertainty is assigned as 10.1%. The BR(c →e) uncertainty is estimated using the errors of the charmed hadron ratios in Tab. 5.2and assigned as 4.3%. The rapidity distribution of the PYTHIA charm calculation isstudied. Since the rapidity distribution is sensitive to the initial state gluon distribution,


rapidity y-4 -2 0 2 4 6 8

0

500

1000

1500

2000

2500

3000

3500

4000

45002x10

Rapidity y distributionCTEQ5L

dydn/all=23.00%

GRV98LO

dydn/all=21.52% ... -6.43 %

MRST_c_g

dydn/all=21.62% ... -5.98 %

Rapidity y distribution

Figure 5.17: Rapidity distribution of charmed hadrons.

Table 5.6: The systematic uncertainty of the source (4).

step uncertaintyRpT

step (2) 10.1%BR(c→ e) step (3) 4.3%Ry step (4) 6.2%Total for dσcc/dy step (2) and (3) 11.0%Total for σcc all 3 steps 12.6 %

we test the two other PDFs , GRV98LO and MRST(c-g) in stead of CTEQ5L. Fig. 5.17shows the rapidity distributions of the charmed hadron. The red, the green and the bluehistograms correspond to CTEQ5L, GRV98LO and MRST(c-g), respectively. From thesedistributions, The extrapolation factor Ry = 23.0%, 21.5% and 21.6% are determined forCTEQ5L, GRV98LO and MRST(c-g), respectively. The relative difference of these valuesis 6.2% which is assigned as the Ry uncertainty. Total uncertainty of the source (4) isestimated by adding these uncertainties in quadrature. The systematic uncertainty of thesource (4) is summarized in Tab. 5.6.

In order to estimate the uncertainty of the source (5), we test the pT range of 0.6 −4.0 GeV/c to fit the charm electron spectrum, then calculate the integrated invariantyield using same pT range dNe/dy(0.6 < pT < 4.0) and the total charm cross section. Theconversion factor RpT

for dNe/dy(0.6 < pT < 4.0) is 25.1% determined using the PYTHIAelectron spectrum. Then, these total charm cross sections with the different pT fit rangesare compared to estimate the uncertainty. The difference of these charm cross sectionsis assigned as the uncertainty of the source (5). Since all the systematic uncertainties


are independent each other, these uncertainties are added in quadrature to obtain thetotal systematic uncertainty. All the systematic uncertainties and total uncertainty aresummarized in Tab. 5.7.

Finally, we calculated the differential (dσcc/dy) and the total charm cross section (σcc).The total charm cross section in minimum bias Au + Au collisions is σcc = 622 ± 57 ±160 µb. The result for minimum bias and the five centrality classes are summarized inTab. 5.8.

Recently, the total charm cross section is updated using the new data with muchhigher statistics in Run 4 period. The updated total charm cross section in minimum biasAu + Au collisions at

√sNN = 200 GeV is σcc = 568 ± 8 ± 150 µb. The method and the

result of the total charm cross section in Run4 are described in detail in Appendix E.Even though the statistical error is much reduced, the systematic uncertainty in Run 4 issimilar compared to that in Run 2. The obtained cross section in Run 4 is consistent withthat in Run 2 within the statistical and the systematic uncertainty. The comparison ofthe total charm cross section in

√s = 200 GeV is described in next section.

Table 5.7: All the systematic uncertainties are summarized and the total uncertainty isobtained by adding these uncertainties in quadrature.

Min.bias 0-10% 10-20% 20-40% 40-60% 60-92%source (1) 20.6% 20.7% 20.4% 20.7% 21.0% 20.5%source (2) 7.3% 6.9% 7.0% 8.2% 12.0% 28.6%source (3) 3.8% 3.4% 4.6% 5.5% 0.3% 5.2%source (4) 11.0% for dσcc/dy and 12.6% for σcc

source (5) 2.5% 5.9 % 8.0% 7.7% 4.2% 0.6%Total for dσcc/dy 24.8% 24.6% 24.5% 25.4% 26.6% 37.2%Total for σcc 25.5% 25.4% 25.2% 26.1% 27.3% 37.7%

Table 5.8: Centrality class, Ncoll, nuclear overlap function (TAA), charm cross section(dσcc/dy) and total charm cross section (σcc) in Au + Au collisions at

√sNN = 200 GeV.

Centrality Ncoll TAA (mb−1) dσcc/dy (µb) σcc (µb)Min. bias 257.8±25.4 6.14±0.45 143±13±36 622± 57±160

0-10% 955.4±93.6 22.76±1.56 137±21±35 597± 93±15610-20% 602.6±59.3 14.35±1.00 137±26±35 596±115±15820-40% 296.8±31.1 7.07±0.58 168±27±45 731±117±19940-60% 90.7±11.8 2.16±0.26 193±47±52 841±205±23260-92% 14.5± 4.0 0.35±0.10 116±87±43 504±378±190


5.6.2 Comparison with Di-electron Measurement

There is an alternative measurement of the total charm cross section. This total charmcross section is measured by PHENIX using the mass spectrum of electron-positron pairsin p+ p collisions at

√s = 200 GeV [132]. This result allows us to cross-check the charm

cross section.

After the combinatorial and the correlated background are subtracted, the measuredmass spectrum of e+e− pairs is compared with a cocktail calculation of e+e− pairs fromlight meson decays and vector meson decays. Then, the cocktail calculation can describethe measured mass spectrum of e+e− pairs in the mass region below Me+e− < 1.1 GeV/c2

within the systematic uncertainty. On the other hand, except for the vector meson peak,the e+e− pair yield in Me+e− > 1.1 GeV/c2 is dominated by semi-leptonic decays ofcharmed hadrons correlated through flavor conservation. Figure 5.18 shows the measuredmass spectrum of e+e− pairs comparing the cocktail calculation. In addition to charmhadrons, the remaining contributions from bottom hadrons and Drell-Yan calculated byPYTHIA are also shown in Fig. 5.18. To determine the charm contribution, the pair yieldis integrated within 1.1 < Me+e− < 2.5 GeV/c2 after subtracting the cocktail of the lightneutral mesons and the vector mesons. The integrated pair yield is extrapolated to zeroe+e− mass using PYTHIA and translated to charm pair yield using the known branchingratio. Then, assuming the rapidity distribution from NLO pQCD, the charm cross sectionis obtained as σcc = 544 ± 39(stat) ± 142(sys) ± 200(model)µb.

The result in the e+e− pair analysis is compared with the charm electron measurementin p+ p and Au + Au collisions at

√sNN = 200 GeV. Figure 5.19 shows the comparison of

the total charm cross section measured by the single electron measurements and the e+e−

pair measurement. All these charm cross sections measured by PHENIX are consistenteach other within the statistical and systematic uncertainty. The STAR collaborationreported 2-3 times larger cross section in d+ Au collisions [133]. The charm cross sectionby the FONLL calculation is smaller but compatible with these data within its uncertainty.

5.6.3 Collision Energy Dependence

The obtained total charm cross section in Au + Au collisions at√sNN = 200 GeV is

compared with those measured in the other experiments at several beam energies. Thecharm cross sections are measured in the fixed target experiments at SPS/FNAL [126, 135,136, 137] and the collider experiments at SppS [134]. Figure 5.20 shows the total charmcross sections as a function of the collision energy

√s. The theoretical curves of the charm

cross section calculated by PYTHIA (solid) and NLO pQCD (dashed) are also shown.The shaded band in Fig. 5.20 represents the systematic uncertainty of the NLO pQCDcalculation. The input parameters for PYTHIA is chosen to reproduce the charm dataas described in section 5.6.1. For the NLO pQCD calculation, HVQMNR [130] is used tocalculate the charm cross section with Mc = 1.5 GeV/c2, µF = 2Mc, 0.5Mc < µR < 2Mc

and CTEQ5M PDF as input.

The data obtained in this thesis are consistent with the energy dependences of thetotal charm cross section calculated by both PYTHIA and NLO pQCD within theoretical


0 1 2 3 4 5 6 7 8

/GeV

) IN

PH

EN

IX A

CC

EP

TA

NC

E2

(c

ee d

N/d

mev

t1/

N

-1010

-910

-810

-710

-610

-510

-410

-310 = 200 GeVsp+p at DATA

|y| < 0.35 > 0.2 GeV/c

Tp

0 1 2 3 4 5 6 7 8

/GeV

) IN

PH

EN

IX A

CC

EP

TA

NC

E2

(c

ee d

N/d

mev

t1/

N

-1010

-910

-810

-710

-610

-510

-410

-310

)2 (GeV/ceem0 0.2 0.4 0.6 0.8 1

/GeV

) IN

PH

EN

IX A

CC

EP

TA

NC

E2

(c

ee d

N/d

mev

t1/

N

-710

-610

-510

-410

-310 eeγ → 0π

eeγ → η

eeγ →’ η

ee→ ρ

ee0π ee & → ω

eeη ee & → φ

ee→ ψJ/

ee→’ ψ

ee (PYTHIA)→ cc

ee (PYTHIA)→ bb

ee (PYTHIA)→DY

sum

)2 (GeV/ceem0 0.2 0.4 0.6 0.8 1

/GeV

) IN

PH

EN

IX A

CC

EP

TA

NC

E2

(c

ee d

N/d

mev

t1/

N

-710

-610

-510

-410

-310

)2 (GeV/ceem0 1 2 3 4 5 6 7 8

Dat

a/C

ockt

ail

0

0.5

1

1.5

Figure 5.18: (top) The invariant mass spectrum of e+e− pairs in p+ p collisions [132]. Thepoints show the measured data with the statistical(bar) and the systematic(shades) errors.The curves represent the cocktail calculation of the light neutral mesons and the vectormesons. The contributions from c and b quarks and Drell-Yan are also shown. (bottom)The ratio of the data to the cocktail.

0 1 2 3 4 5 6 7 8

b)

µ (c c σ

0

200

400

600

800

1000

1200

1400

1600

1800

2000

2200

FONLL

p+pRun3

p+pRun5

p+pee pair

Au+AuRun2

Au+AuRun4

STARd+Au

=200GeVNNs in p+p / Au+Au at ccσ

Figure 5.19: The comparison of the total charm cross sections from the single electronsmeasurements and the e+e− pair measurement.


(GeV)s210 310

b)

µ ( c cσ

1

10

210

310

410

Charm Cross SectionPHENIX Au+Au 130GeVPHENIX Au+Au 200GeV

630GeVpUA2 p+SPS/FNAL

PYTHIANLO pQCD

Figure 5.20: The total charm cross sections as a function of the collision energy. Thetheoretical curves by PYTHIA and NLO pQCD are also shown.

uncertainties. This indicates that the charm production is well reproduced by pQCD.Therefore, as expected, the most charm quarks are produced in the initial stage of thespace time evolution in the heavy ion collisions.

Chapter 6

Summary and Conclusions

We have studied the centrality dependence of charm production in Au + Au collisions at√sNN = 200 GeV. We have measured the single electrons from semi-leptonic decays of

heavy flavors (charms and bottoms) in Au + Au collisions during Run 2 period.

In Run 2, RHIC first operated with the full energy in Au + Au collisions. PHENIXmeasured inclusive electrons in Au + Au collisions at

√sNN = 200 GeV. A photon con-

verter was installed in a part of run period. By comparing the inclusive electron yield withand without the converter, signal electrons are extracted from large amount of photonicbackgrounds which are electrons from Dalitz decays of light neutral mesons and photonconversions. After subtracting the backgrounds of Ke3 and vector meson decays, only thesignificant source of non-photonic electrons is the semi-leptonic decays of heavy flavors,dominantly charm at low pT. Since the contributions from J/ψ decays and Drell-Yanprocess is small, they are neglected.

We have measured the invariant differential yields of heavy flavor electrons for minimumbias Au + Au collisions and the five centrality classes, 0-10%, 10-20%, 20-40%, 40-60% and60-92% centralities, within the pT range from 0.3 to 4.0 GeV/c. These yields are comparedwith that in p+ p collisions at

√s = 200 GeV scaled with the number of binary nucleon-

nucleon collisions Ncoll, and are found to be consistent within relatively large error. Sincethe cross section of heavy flavor electrons in p+ p collisions agrees with a NLO pQCDtheory (FONLL) with a normalization factor of 1.71 ± 0.02 ± 0.18, the invariant yieldscaled by Ncoll in Au + Au collisions also agrees with the FONLL pQCD theory within theerrors in the measured pT region.

To quantify the centrality dependence of the charm production, the integrated yieldsof charm electrons dNe/dy/Ncoll(0.8 < pT < 4.0 GeV/c) are calculated for minimum biasand 0-10%, 10-20%, 20-40%, 40-60% and 60-92% centralities, and these yields are fittedwith ANα

coll for the linearity test. We found α = 0.938 ± 0.075(stat) ± 0.018(sys). Thisshows that the centrality dependence of the total charm production is consistent with Ncoll

scaling. This result is confirmed by the new data with higher statistic measured in Run 4.The updated result of the linearity test is α = 1.0097 ± 0.0094(stat) ± 0.0403(sys). Thisresult is consistent with the finding of energy loss of heavy quarks in the dense mattercreated in high energy heavy ion collisions. Since the final state effects such as the energyloss affect only the momentum distribution of charm, they have little or no effect on thetotal charm yield. Therefore, the result indicates that most charm quarks are produced

125

126 CHAPTER 6. SUMMARY AND CONCLUSIONS

by point-like pQCD processes in the initial stage of space time evolution in high energyheavy ion collisions.

The total charm cross sections per nucleon-nucleon collision are calculated for minimumbias Au + Au collisions and the five centrality classes. The obtained charm cross sectionper N + N collision is σcc = 622 ± 57(stat) ± 160(sys) µb in minimum bias Au+Aucollisions. In addition, the cross section is updated in Run 4 analysis as σcc = 568±8(stat)±150(sys) µb. These cross sections are confirmed by a complementary measurement of e+e−

pairs. By comparing a NLO pQCD prediction, the obtained cross section is consistent witha NLO pQCD prediction within theoretical error.

The Ncoll scaling of the total charm production provides an important informationfor the initial nuclear effect, e.g. Cronin effect and nuclear shadowing effect. From ourresults, no significant initial nuclear effect is observed. However, the present data can notexclude a small initial nuclear effect or possible cancellation of these effects. We need moresystematic study of collision system, especially including data in d + Au collisions. Thiscan help to understand the initial nuclear effects in Au + Au collisions.

High pT suppression of heavy quarks has been observed in the RAA measurement ofheavy flavor electrons in central Au + Au collisions. It is a surprising discovery that themagnitude of heavy quark suppression is almost same with that of light quark at high pT.This observation is not well explained by recent theoretical models for the heavy quarkenergy loss. In addition, the measurement of (b → e)/(b → e + c → e) in p+ p collisionsby partial reconstruction of D meson shows the significant contribution of bottom quarkabove pT > 4 GeV/c. Although theoretical models predict that heavy quarks suffer smallerenergy loss than light quarks such as dead cone effect due to their large mass, this resultindicates that not only charms, but bottoms could suffer energy losses.

In order to understand the initial gluon density, the systematic study of the heavyflavor production in p+ p, d + Au and Au + Au collisions is necessary with a much largeamount of statistics. The d + Au data in Run 7 with 20 times large statistics is nowanalyzed for the systematic study. In addition, to understand better the energy loss ofheavy quarks in the dense matter, it is necessary to measure the RAA of D and B mesonsseparately. The PHENIX experiment plans to install a silicon vertex detector (VTX) inorder to directly measure D and B meson by reconstructing the secondary decay vertexof these mesons. Furthermore, high luminosity upgrade of RHIC facility is underway.Therefore, the upcoming experimental results with the new detector and much higherstatistics will be shown in the near future.

Acknowledgments

First of all, I would like to appreciate all the people who advise and support my thesis workand school days. I would like to thank my supervisor Prof. T. Sugitate. He gave me thecontinues instruction and supported my student life at Hiroshima University. Once I gaveup my thesis halfway and got a job. After the termination of my doctor course for a fewyears, I decided to restart my thesis work. At that time, he advised me to get a good startagain. I would like to express my gratitude to Dr. Y. Akiba for his continues support.This thesis would never be finished without him. His valuable advises and discussionsalways encouraged me to get better understandings of both the analysis, the heavy quarkproductions and the phenomena of heavy ion collisions. I am grateful to Dr. K. Hommafor his longstanding help. He taught me the way of thinking of experimental physics. Iappreciate Prof. K. Shigaki for his arrangement of activity both at Hiroshima and BNL.

I wish to acknowledge all the PHENIX collaboration. I am obliged to the spokespersonsof the PHENIX experiment, Prof. W. A. Zajc and Prof. B. V. Jacak for their arrangementof my research activity in PHENIX. I appreciate to the members of the paper preparationgroup and the internal review committee, Dr. V. Cianciolo, Prof. M. Rosati, Dr. S. Kelly,Prof. J.-C. Peng,　Dr. A. Bazilevsky and Dr. A. Denisov for their combined effort of thepublication. I would like to thank Dr. R. Averbeck and Dr. S. Butsyk for their advicesand comments on my electron analysis.

I would like to express my thank to the PHENIX-J group. I acknowledge Prof. H.Hamagaki for his work as a leader of the PHENIX-J group and the financial support. Iwould like to thank Dr. K. Ozawa for his advices and comments on both research and lifeat BNL. I appreciate the chief scientist, Dr. H. En’yo for his various advices and financialsupport. He also gave me a chance to restart my thesis work. I wish to express my thankto Prof. Y. Miake for his help. He kindly gave me a working place in his laboratory.

I would like to thank Dr. H. Ohnishi, Dr. T. Nakamura, Dr. A. Enokizono andDr. T. Kohama. We worked together for the maintenance and the operation of the BBCsubsystem. With their help, I could handle my task as a BBC expert. I also appreciate Dr.C. Y. Chi, Dr. C. Mickey and Dr. S. Belikov for their cooperation of the BBC operation.I would like to express my deep thank to Dr. M. Kaneta, Dr. T. Chujo, Dr. A. Taketani,Prof. N. Saito, Dr. Y. Goto, Prof. K. Kurita, Dr. H. Torii, Dr. A. Kiyomichi, Dr. K.Okada, Dr. S. Sato, Prof. S. Esumi and Prof. M. Inaba for their instructive advices anddiscussions. In particular, Dr. M. Kaneta gave me a lot of supports for both my researchand life at BNL. By discussed with Prof. S. Esumi and Dr. T. Chujo, I could know thelatest understanding of QGP and the most interesting topics on heavy ion collisions. Iwish to thank to the member of RIKEN-CCJ, Dr. T. Ichihara, Dr. Y. Watanabe andDr. S. Yokkaichi for their maintenance works of CCJ. The main part of my analysis was

127

128 CHAPTER 6. SUMMARY AND CONCLUSIONS

performed by CCJ.I spent most part of my graduate school days at BNL. I would like to thank all the

Japanese colleague who worked together at BNL, Mr. T. Matsumoto, Dr. S. Kametani,Dr. H. Masui, Dr. S. Sakai, Dr. T. Horaguchi, Dr. F. Kajihara, Dr. M. Konno, Dr.M. Shimomura, Dr. T. Gunji, Dr. T. Isobe, Dr. M. Togawa, Dr. Y. Fukao and Mr.Y. Nagata. I would also like to thank all the member of quark physics laboratory atHiroshima University, Mr. T. Uzumaki, Mr. N. Sugita, Mr. Y. Tsuchimoto, Mr. R.Kohara, Dr. H. Harada, Mr. K. Haruna, Mr. Y. Nakamiya, Ms. M. Ouchida and Mr.D. Watanabe. I enjoyed my school days with their discussions and conversations. I wishto thank all the member of high energy nuclear physics group at University of Tsukuba,Mr. Y. Ikeda, Mr. K. Watanabe, Mr. M. Sano, Mr. D. Sakata, Ms. M. Kajigaya, Ms.M. Kimura, Mr. H. Yokoyama, Mr. E. Hamada, Mr. T. Todoroki, Mr. Y. Sekine, Mr.T. Takeuchi, Mr. S. Mizuno, Mr. Y. Watanabe, Mr. Y. Kondo, Mr. S. Kubota, Ms. H.Sakai and Mr. S. Kato. After I started my thesis work again, I had a very good time todiscuss and talk with them. The journal club held in the Tsukuba laboratory was veryuseful for me to remember and understand the phenomena of heavy ion collisions.

I would like to express my greatest thanks to my parents, father Masao and motherMasami for their continues supports. With their helps, I could start again and finish mythesis work.

Appendix A

Simulation Condition

A.1 Comparison between Read Data and Simulation

The detector responses of the electron ID variables in the real data are compared withthat in the simulation for several momentum classes. The momentum is subdivided for 11classes which correspond to 0.2-0.3, 0.3-0.4, 0.4-0.5, 0.5-0.6, 0.6-0.8, 0.8-1.0, 1.0-1.5, 1.5-2.0, 2.0-3.0, 3.0-5.0 and 5.0-10.0 GeV/c, respectively. All the variables in the simulationare tuned and/or smeared to reproduce the real data whether the response is significantlydifferent.

Figure A.1 and A.2 show the emcsdphi e and emcsdz e distributions measured by thePbSc EMCal. The black and red histograms correspond to the real data and the sim-ulation. The left-top to the right-bottom panels correspond to the lower to the highermomentum classes. Although the statistics in the real data is too small for higher mo-mentum class, the response in the simulation is in good agreement with the real data .The other comparisons are emcsdphi e and emcsdz e for the PbGl EMCal, n0, chi2/npe0and disp for the RICH, dep for the PbSc and PbGl EMCal, and emcdt for the EMCal areshown in Fig. A.3, A.4, A.5, A.6 A.7, A.8, A.9 and A.10, respectively. These figures showsthat the responses in the simulation are in good agreement with the real data.

129

130 APPENDIX A. SIMULATION CONDITION

-5 -4 -3 -2 -1 0 1 2 3 4 50

200

400

600

800

1000

1200

pbsc east dphi sigma of e+e- 0.2<p<0.3pbsc east dphi sigma of e+e- 0.2<p<0.3

-5 -4 -3 -2 -1 0 1 2 3 4 50

200

400

600

800

1000


-5 -4 -3 -2 -1 0 1 2 3 4 50

100

200

300

400

500

600


-5 -4 -3 -2 -1 0 1 2 3 4 50

50

100

150

200

250

300

350

400


-5 -4 -3 -2 -1 0 1 2 3 4 50

100

200

300

400

500


-5 -4 -3 -2 -1 0 1 2 3 4 50

20

40

60

80

100

120

140

160

180

200

pbsc east dphi sigma of e+e- 0.8<p<1pbsc east dphi sigma of e+e- 0.8<p<1

-5 -4 -3 -2 -1 0 1 2 3 4 50

20

40

60

80

100

120

140

pbsc east dphi sigma of e+e- 1<p<1.5pbsc east dphi sigma of e+e- 1<p<1.5

-5 -4 -3 -2 -1 0 1 2 3 4 50

5

10

15

20

25

pbsc east dphi sigma of e+e- 1.5<p<2pbsc east dphi sigma of e+e- 1.5<p<2

-5 -4 -3 -2 -1 0 1 2 3 4 50

2

4

6

8

10

12

pbsc east dphi sigma of e+e- 2<p<3pbsc east dphi sigma of e+e- 2<p<3

-5 -4 -3 -2 -1 0 1 2 3 4 50

0.5

1

1.5

2

2.5

3


-5 -4 -3 -2 -1 0 1 2 3 4 50

0.2

0.4

0.6

0.8

1


Figure A.1: Comparisons of emcsdphi e measured at the PbSc EMCal for several mo-mentum classes. The black and the red histograms correspond to the real data and thesimulation, respectively.

-5 -4 -3 -2 -1 0 1 2 3 4 50

200

400

600

800

1000

1200

pbsc east dz sigma of e+e- 0.2<p<0.3pbsc east dz sigma of e+e- 0.2<p<0.3

-5 -4 -3 -2 -1 0 1 2 3 4 50

200

400

600

800

1000


-5 -4 -3 -2 -1 0 1 2 3 4 50

100

200

300

400

500

600


-5 -4 -3 -2 -1 0 1 2 3 4 50

50

100

150

200

250

300

350

400


-5 -4 -3 -2 -1 0 1 2 3 4 50

50

100

150

200

250

300

350

400

450


-5 -4 -3 -2 -1 0 1 2 3 4 50

20

40

60

80

100

120

140

160

180

200

pbsc east dz sigma of e+e- 0.8<p<1pbsc east dz sigma of e+e- 0.8<p<1

-5 -4 -3 -2 -1 0 1 2 3 4 50

20

40

60

80

100

120

140

pbsc east dz sigma of e+e- 1<p<1.5pbsc east dz sigma of e+e- 1<p<1.5

-5 -4 -3 -2 -1 0 1 2 3 4 50

5

10

15

20

25

pbsc east dz sigma of e+e- 1.5<p<2pbsc east dz sigma of e+e- 1.5<p<2

-5 -4 -3 -2 -1 0 1 2 3 4 50

1

2

3

4

5

6

7

8

9

pbsc east dz sigma of e+e- 2<p<3pbsc east dz sigma of e+e- 2<p<3

-5 -4 -3 -2 -1 0 1 2 3 4 50

0.2

0.4

0.6

0.8

1


-5 -4 -3 -2 -1 0 1 2 3 4 50

0.01

0.02

0.03

0.04

0.05

0.06


Figure A.2: Comparisons of emcsdz e measured at the PbSc EMCal for several momentumclasses. The black and the red histograms correspond to the real data and the simulation,respectively.

A.1. COMPARISON BETWEEN READ DATA AND SIMULATION 131

-5 -4 -3 -2 -1 0 1 2 3 4 50

200

400

600

800

1000

1200

1400

1600

1800

pbgl dphi sigma of e+e- 0.2<p<0.3pbgl dphi sigma of e+e- 0.2<p<0.3

-5 -4 -3 -2 -1 0 1 2 3 4 50

200

400

600

800

1000

1200


-5 -4 -3 -2 -1 0 1 2 3 4 50

100

200

300

400

500

600

700


-5 -4 -3 -2 -1 0 1 2 3 4 50

50

100

150

200

250

300

350

400

450


-5 -4 -3 -2 -1 0 1 2 3 4 50

50

100

150

200

250

300

350

400


-5 -4 -3 -2 -1 0 1 2 3 4 50

20

40

60

80

100

120

140

160

180

pbgl dphi sigma of e+e- 0.8<p<1pbgl dphi sigma of e+e- 0.8<p<1

-5 -4 -3 -2 -1 0 1 2 3 4 50

20

40

60

80

100

120

pbgl dphi sigma of e+e- 1<p<1.5pbgl dphi sigma of e+e- 1<p<1.5

-5 -4 -3 -2 -1 0 1 2 3 4 50

5

10

15

20

25

pbgl dphi sigma of e+e- 1.5<p<2pbgl dphi sigma of e+e- 1.5<p<2

-5 -4 -3 -2 -1 0 1 2 3 4 50

1

2

3

4

5

6

7

8

pbgl dphi sigma of e+e- 2<p<3pbgl dphi sigma of e+e- 2<p<3

-5 -4 -3 -2 -1 0 1 2 3 4 50

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8


-5 -4 -3 -2 -1 0 1 2 3 4 50

0.02

0.04

0.06

0.08

0.1

0.12

0.14

0.16

0.18


Figure A.3: Comparisons of emcsdphi e measured at the PbGl EMCal for several mo-mentum classes. The black and the red histograms correspond to the real data and thesimulation, respectively.

-5 -4 -3 -2 -1 0 1 2 3 4 50

200

400

600

800

1000

1200

1400

1600

1800

pbgl dz sigma of e+e- 0.2<p<0.3

-5 -4 -3 -2 -1 0 1 2 3 4 50

200

400

600

800

1000

1200


-5 -4 -3 -2 -1 0 1 2 3 4 50

100

200

300

400

500

600

700


-5 -4 -3 -2 -1 0 1 2 3 4 50

50

100

150

200

250

300

350

400

450


-5 -4 -3 -2 -1 0 1 2 3 4 50

50

100

150

200

250

300

350

400


-5 -4 -3 -2 -1 0 1 2 3 4 50

20

40

60

80

100

120

140

160

180

pbgl dz sigma of e+e- 0.8<p<1

-5 -4 -3 -2 -1 0 1 2 3 4 50

20

40

60

80

100

120

140

pbgl dz sigma of e+e- 1<p<1.5

-5 -4 -3 -2 -1 0 1 2 3 4 502

4

6

8

10

12

14

16

18

20

22

24

pbgl dz sigma of e+e- 1.5<p<2

-5 -4 -3 -2 -1 0 1 2 3 4 50

1

2

3

4

5

6

7

8

9

pbgl dz sigma of e+e- 2<p<3

-5 -4 -3 -2 -1 0 1 2 3 4 50

0.1

0.2

0.3

0.4

0.5

0.6


-5 -4 -3 -2 -1 0 1 2 3 4 50

0.02

0.04

0.06

0.08

0.1

0.12

0.14

0.16


Figure A.4: Comparisons of emcsdz e measured at the PbGl EMCal for several momentumclasses. The black and the red histograms correspond to the real data and the simulation,respectively.


0 2 4 6 8 10

0

5000

crk n0 east 0.2<p<0.3crk n0 east 0.2<p<0.3

0 2 4 6 8 10

0

5000


0 2 4 6 8 10

0

2000

4000

6000

8000


0 2 4 6 8 10

0

2000

4000

6000

8000


0 2 4 6 8 10

0

2000

4000

6000

8000


0 2 4 6 8 10

0

500

1000

1500

2000

2500

3000

3500

4000

4500

crk n0 east 0.8<p<1crk n0 east 0.8<p<1

0 2 4 6 8 10

0

500

1000

1500

2000

2500

3000

crk n0 east 1<p<1.5crk n0 east 1<p<1.5

0 2 4 6 8 10

0

100

200

300

400

500

crk n0 east 1.5<p<2crk n0 east 1.5<p<2

0 2 4 6 8 100

20

40

60

80

100

120

140

160

crk n0 east 2<p<3crk n0 east 2<p<3

0 2 4 6 8 100

2

4

6

8

10

12

14

16


0 2 4 6 8 100

0.5

1

1.5

2


Figure A.5: Comparisons of n0 for several momentum classes. The black and the redhistograms correspond to the real data and the simulation, respectively.

0 5 10 15 20 250

2000

4000

6000

8000

crk chi2/npe0 east 0.2<p<0.3crk chi2/npe0 east 0.2<p<0.3

0 5 10 15 20 250

2000

4000

6000

8000


0 5 10 15 20 250

1000

2000

3000

4000

5000

6000

7000

8000


0 5 10 15 20 250

1000

2000

3000

4000

5000


0 5 10 15 20 250

1000

2000

3000

4000

5000


0 5 10 15 20 250

200

400

600

800

1000

1200

1400

1600

1800

2000

2200

crk chi2/npe0 east 0.8<p<1crk chi2/npe0 east 0.8<p<1

0 5 10 15 20 250

200

400

600

800

1000

1200

1400

1600

crk chi2/npe0 east 1<p<1.5crk chi2/npe0 east 1<p<1.5

0 5 10 15 20 250

50

100

150

200

250

crk chi2/npe0 east 1.5<p<2crk chi2/npe0 east 1.5<p<2

0 5 10 15 20 250

10

20

30

40

50

60

70

80

90

crk chi2/npe0 east 2<p<3crk chi2/npe0 east 2<p<3

0 5 10 15 20 250

1

2

3

4

5

6

7

8


0 5 10 15 20 250

0.2

0.4

0.6

0.8

1

1.2


Figure A.6: Comparisons of chi2/npe0 for several momentum classes. The black and thered histograms correspond to the real data and the simulation, respectively.

A.1. COMPARISON BETWEEN READ DATA AND SIMULATION 133

0 2 4 6 8 10 12 140

2000

4000

6000

8000

crk disp east 0.2<p<0.3crk disp east 0.2<p<0.3

0 2 4 6 8 10 12 140

2000

4000

6000

8000


0 2 4 6 8 10 12 140

2000

4000

6000

8000


0 2 4 6 8 10 12 140

1000

2000

3000

4000

5000

6000


0 2 4 6 8 10 12 140

1000

2000

3000

4000

5000

6000


0 2 4 6 8 10 12 140

500

1000

1500

2000

2500

crk disp east 0.8<p<1crk disp east 0.8<p<1

0 2 4 6 8 10 12 140

200

400

600

800

1000

1200

1400

1600

1800

2000

crk disp east 1<p<1.5crk disp east 1<p<1.5

0 2 4 6 8 10 12 140

50

100

150

200

250

300

350

crk disp east 1.5<p<2crk disp east 1.5<p<2

0 2 4 6 8 10 12 140

20

40

60

80

100

crk disp east 2<p<3crk disp east 2<p<3

0 2 4 6 8 10 12 140

2

4

6

8

10


0 2 4 6 8 10 12 140

0.2

0.4

0.6

0.8

1

1.2

1.4

1.6


Figure A.7: Comparisons of disp for several momentum classes. The black and the redhistograms correspond to the real data and the simulation, respectively.

-4 -3 -2 -1 0 1 2 3 40

100

200

300

400

500

600

700

(E-p)/p/s pbsce n0>2&&ch2<10&&(emc||pc3)_acc<3sig 0.2<p<0.3(E-p)/p/s pbsce n0>2&&ch2<10&&(emc||pc3)_acc<3sig 0.2<p<0.3

-4 -3 -2 -1 0 1 2 3 40

100

200

300

400

500


-4 -3 -2 -1 0 1 2 3 40

50

100

150

200

250

300

350


-4 -3 -2 -1 0 1 2 3 40

50

100

150

200

250


-4 -3 -2 -1 0 1 2 3 40

50

100

150

200

250


-4 -3 -2 -1 0 1 2 3 40

20

40

60

80

100

120

(E-p)/p/s pbsce n0>2&&ch2<10&&(emc||pc3)_acc<3sig 0.8<p<1(E-p)/p/s pbsce n0>2&&ch2<10&&(emc||pc3)_acc<3sig 0.8<p<1

-4 -3 -2 -1 0 1 2 3 40

10

20

30

40

50

60

70

80

90

(E-p)/p/s pbsce n0>2&&ch2<10&&(emc||pc3)_acc<3sig 1<p<1.5(E-p)/p/s pbsce n0>2&&ch2<10&&(emc||pc3)_acc<3sig 1<p<1.5

-4 -3 -2 -1 0 1 2 3 40

2

4

6

8

10

12

14

16

18

20

(E-p)/p/s pbsce n0>2&&ch2<10&&(emc||pc3)_acc<3sig 1.5<p<2(E-p)/p/s pbsce n0>2&&ch2<10&&(emc||pc3)_acc<3sig 1.5<p<2

-4 -3 -2 -1 0 1 2 3 40

1

2

3

4

5

(E-p)/p/s pbsce n0>2&&ch2<10&&(emc||pc3)_acc<3sig 2<p<3(E-p)/p/s pbsce n0>2&&ch2<10&&(emc||pc3)_acc<3sig 2<p<3

-4 -3 -2 -1 0 1 2 3 40

0.1

0.2

0.3

0.4

0.5


-4 -3 -2 -1 0 1 2 3 40

0.01

0.02

0.03


Figure A.8: Comparisons of dep measured at the PbSc EMCal for several momentumclasses. The black and the red histograms correspond to the real data and the simulation,respectively.


-4 -3 -2 -1 0 1 2 3 40

200

400

600

800

1000

(E-p)/p/s pbgl n0>2&&ch2<10&&(emc||pc3)_acc<3sig 0.2<p<0.3

-4 -3 -2 -1 0 1 2 3 40

100

200

300

400

500

600

700


-4 -3 -2 -1 0 1 2 3 40

50

100

150

200

250

300

350

400


-4 -3 -2 -1 0 1 2 3 40

50

100

150

200

250

300


-4 -3 -2 -1 0 1 2 3 40

50

100

150

200

250


-4 -3 -2 -1 0 1 2 3 40

20

40

60

80

100

(E-p)/p/s pbgl n0>2&&ch2<10&&(emc||pc3)_acc<3sig 0.8<p<1

-4 -3 -2 -1 0 1 2 3 40

10

20

30

40

50

60

70

(E-p)/p/s pbgl n0>2&&ch2<10&&(emc||pc3)_acc<3sig 1<p<1.5

-4 -3 -2 -1 0 1 2 3 40

2

4

6

8

10

12

14

16

(E-p)/p/s pbgl n0>2&&ch2<10&&(emc||pc3)_acc<3sig 1.5<p<2

-4 -3 -2 -1 0 1 2 3 4-1

0

1

2

3

4

5

6

7

8

(E-p)/p/s pbgl n0>2&&ch2<10&&(emc||pc3)_acc<3sig 2<p<3

-4 -3 -2 -1 0 1 2 3 4-1

-0.5

0

0.5

1


-4 -3 -2 -1 0 1 2 3 40

0.2

0.4

0.6

0.8

1


Figure A.9: Comparisons of dep measured at the PbGl EMCal for several momentumclasses. The black and the red histograms correspond to the real data and the simulation,respectively.

-2 -1 0 1 2 3 40

500

1000

1500

2000

2500

PbSc dt east n0>2&&ch2<10&&(emc||pc3)_acc<3sig 0.2<p<0.3PbSc dt east n0>2&&ch2<10&&(emc||pc3)_acc<3sig 0.2<p<0.3

-2 -1 0 1 2 3 40

200

400

600

800

1000

1200

1400

1600

1800


-2 -1 0 1 2 3 40

200

400

600

800

1000

1200

1400


-2 -1 0 1 2 3 40

100

200

300

400

500

600

700

800

900


-2 -1 0 1 2 3 40

200

400

600

800

1000


-2 -1 0 1 2 3 40

100

200

300

400

500

PbSc dt east n0>2&&ch2<10&&(emc||pc3)_acc<3sig 0.8<p<1PbSc dt east n0>2&&ch2<10&&(emc||pc3)_acc<3sig 0.8<p<1

-2 -1 0 1 2 3 40

50

100

150

200

250

300

350

400

PbSc dt east n0>2&&ch2<10&&(emc||pc3)_acc<3sig 1<p<1.5PbSc dt east n0>2&&ch2<10&&(emc||pc3)_acc<3sig 1<p<1.5

-2 -1 0 1 2 3 40

10

20

30

40

50

60

70

80

90

PbSc dt east n0>2&&ch2<10&&(emc||pc3)_acc<3sig 1.5<p<2PbSc dt east n0>2&&ch2<10&&(emc||pc3)_acc<3sig 1.5<p<2

-2 -1 0 1 2 3 40

5

10

15

20

25

30

35

40

PbSc dt east n0>2&&ch2<10&&(emc||pc3)_acc<3sig 2<p<3PbSc dt east n0>2&&ch2<10&&(emc||pc3)_acc<3sig 2<p<3

-2 -1 0 1 2 3 40

2

4

6

8

10

12

14

16

18

20

22


-2 -1 0 1 2 3 40

5

10

15

20

25

30

35

40


Figure A.10: Comparisons of emcdt for several momentum classes. The black and the redhistograms correspond to the real data and the simulation, respectively.

A.2. PHOTON CONVERTER IN SIMULATION 135

A.2 Photon Converter in Simulation

The photon converter in the PISA simulation is implemented. The element compositionof the converter material (brass) is Cu (70%), Zn (29.88%), Fe (0.05%) and Pb (0.07%).The mass density is 8.5 (g/cm3). The tube shape with 29 cm radius, 60.96 cm height and0.0254 cm thickness is placed at center of the PHENIX along with beam axis.

The density and the thickness of the converter was checked by measuring the area andthe weight of a piece of the converter which was actually installed in the experiment. Themeasured area density is 0.215313 (g/cm3). The converter is a 10 mil (= 0.0254 cm) thickbrass sheet. The density of the converter is 8.4769 (g/cm3). Therefore, the differencebetween the measured value and the simulation value is less than 0.3% (8.4769/8.5 =0.9973).

In the real experiment, the converter is fixed around the MVD by a plastic tape witha overlap width at the bottom of the MVD. the radiation length of the plastic tapeis negligible relative to the converter. Therefore, the tape is not implemented in thesimulation. On the other hand, the overlap region with 6.985 cm overlap width at thebottom of the converter tube is also implemented.

Figure A.11 shows the photon converter wrapped the MVD in the simulation. The leftand the right figure correspond to the beam view and the side view, respectively. The redtube around the MVD is the converter. The blue part at the bottom of the MVD is theoverlap region of the converter.

Figure A.11: The photon converter wrapped the MVD in simulation. The panels showsthe beam view (left) and the side view (right), respectively. The blue part at the bottomof the MVD in the beam view (left) is the overlap width of the converter.

Appendix B

Cocktail Calculation for Centralities

The composition of photonic electrons in Au + Au collisions is modeled by the cocktailcalculation. The method for the cocktail calculation is described in detail in section 4.10.2.For the inputs of the cocktail calculation, the spectral shapes of pion for each centralityclasses are obtained by fitting the measured pT spectra of charged and neutral pionssimultaneously. The average of π+ and π− represents “charged pion”. In this analysis,the fittings are performed for six centrality classes corresponding minimum bias, 0-10%,10-20%, 20-40%, 40-60% and 60-92%, respectively. The fit functions are expressed as:

1

2πpT

dN

dydpT

= f(pT) (B.1)

= p0 · (1 + pT/p1)p2 + p3 · e−pT/p4 (for min. bias), (B.2)

= p0 · (1 + pT/p1)p2 · e−pT/p4 (for 0 − 10%), (B.3)

= p0 · (1 + pT/p1)p2 · ep3+p4·pT+p5·p2T (for 10 − 20%), (B.4)

= p0 · (1 + pT/p1)p2 + p3 · e−pT/p4 (for 20 − 40%), (B.5)

= p0 · (1 + pT/p1)p2 + p3 · e−pT/p4 (for 40 − 60%), (B.6)

= p0 · (1 + pT/p1)p2 (for 60 − 92%), (B.7)

where p1 − p6 are the parameters.

Table B.1: The parameters obtained by fitting the pion spectra for each centrality.

Centrality p0 p1 p2 p3 p4 p5(GeV/c) (GeV/c) ((c/GeV)2)

Min. bias 415.479 2.33895 -13.9435 6.71687 × 10−4 1.2504 -0-10% 1095.74 4.76572 -34.6169 - 0.554525 -

10-20% 716.026 2.5 -14.37 0.264 -0.231 0.04588(c/GeV)

20-40% 530.297 2.03543 -12.5758 1.11136 × 10−4 1.70697 -40-60% 253.29 1.55091 -10.7785 1.64298 0.297068 -60-92% 72.2085 1.29234 -9.99336 - - -

136

137

Figure B.1 shows the pT spectra of charged and neutral pions for minimum bias andfive centrality classes measured in PHENIX [105, 115]. The green, the blue points andthe red curves correspond to the charged and the neutral pions, and their fit functions,respectively. The bottom plots in each panels are the ratio of the spectrum and thefit function. This indicates that the fit works reasonable. The obtained parameters aresummarized in Tab. B.1.

)2 (

(c/G

eV)

Td

y d

pN2 d T

pπ2

1

-710

-610

-510

-410

-310

-210

-110

1

10

210)/2-π++π (

0π

fit

pion spectra min. bias

)2 (

(c/G

eV)

Td

y d

pN2 d T

pπ2

1

-710

-610

-510

-410

-310

-210

-110

1

10

210)/2-π++π (

0π

fit

pion spectra 0-10%

)2

((c

/GeV

)T

dy

dpN2

d T

pπ2

1

-710

-610

-510

-410

-310

-210

-110

1

10

210)/2-π++π (

0π

fit

pion spectra 10-20%

(GeV/c)p0 1 2 3 4 5 6 7 8 9 10

Rat

io

00.20.40.60.8

11.21.41.61.8

2

(GeV/c)p0 1 2 3 4 5 6 7 8 9 10

Rat

io

00.20.40.60.8

11.21.41.61.8

2

(GeV/c)p0 1 2 3 4 5 6 7 8 9 10

Rat

io

00.20.40.60.8

11.21.41.61.8

2

)2 (

(c/G

eV)

Td

y d

pN2 d T

pπ2

1

-710

-610

-510

-410

-310

-210

-110

1

10

210)/2-π++π (

0π

fit

pion spectra 20-40%

)2 (

(c/G

eV)

Td

y d

pN2 d T

pπ2

1

-710

-610

-510

-410

-310

-210

-110

1

10

210)/2-π++π (

0π

fit

pion spectra 40-60%

)2

((c

/GeV

)T

dy

dpN2

d T

pπ2

1

-710

-610

-510

-410

-310

-210

-110

1

10

210)/2-π++π (

0π

fit

pion spectra 60-92%

(GeV/c)p0 1 2 3 4 5 6 7 8 9 10

Rat

io

00.20.40.60.8

11.21.41.61.8

2

(GeV/c)p0 1 2 3 4 5 6 7 8 9 10

Rat

io

00.20.40.60.8

11.21.41.61.8

2

(GeV/c)p0 1 2 3 4 5 6 7 8 9 10

Rat

io

00.20.40.60.8

11.21.41.61.8

2

Figure B.1: The pion spectra for minimum bias and five centrality classes. The blue andthe green points are the charged and the neutral pion, respectively. The spectral shape isobtained by fitting.

Based on the obtained spectral shape, photonic background electrons are determinedby the cocktail calculation. The calculations are performed for centrality class by classindependently. Figure B.2 shows the cocktails of photonic electrons for minimum bias andfive centrality classes.

138 APPENDIX B. COCKTAIL CALCULATION FOR CENTRALITIES

(GeV/c)T

p0 1 2 3 4 5 6

)2

((c

/GeV

)T

dy

dpN

2 d T

pπ2

1

-1110

-1010

-910

-810

-710

-610

-510

-410

-310

-210

-110

1

10

210

Total eeγ → 0π

eeγ → η eeγ →’ η ee0π → ω

eeη → φ ee→ ω ee→ ρ ee→ φ

min. bias

(GeV/c)T

p0 1 2 3 4 5 6

)2

((c

/GeV

)T

dy

dpN

2 d T

pπ2

1

-1110

-1010

-910

-810

-710

-610

-510

-410

-310

-210

-110

1

10

210

Total eeγ → 0π



0-10%

(GeV/c)T

p0 1 2 3 4 5 6

)2

((c

/GeV

)T

dy

dpN

2 d T

pπ2

1-1110

-1010

-910

-810

-710

-610

-510

-410

-310

-210

-110

1

10

210

Total eeγ → 0π



10-20%

(GeV/c)T

p0 1 2 3 4 5 6

)2 (

(c/G

eV)

Td

y d

pN2 d

T pπ

2 1

-1110

-1010

-910

-810

-710

-610

-510

-410

-310

-210

-110

1

10

210

Total eeγ → 0π



20-40%

(GeV/c)T

p0 1 2 3 4 5 6

)2 (

(c/G

eV)

Td

y d

pN2 d

T pπ

2 1

-1110

-1010

-910

-810

-710

-610

-510

-410

-310

-210

-110

1

10

210

Total eeγ → 0π



40-60%

(GeV/c)T

p0 1 2 3 4 5 6

)2 (

(c/G

eV)

Td

y d

pN2 d

T pπ

2 1

-1110

-1010

-910

-810

-710

-610

-510

-410

-310

-210

-110

1

10

210

60-92%

Total eeγ → 0π



60-92%

Figure B.2: The pT spectra of photonic background electrons for minimum bias and fivecentrality classes. These contributions are determined by the cocktail calculation.

Appendix C

Data Table

C.1 Invariant Differential Yield of Heavy Flavor Elec-

trons

The invariant differential yield of heavy flavor electrons has been measured in minimumbias Au + Au collisions at

√sNN = 200 GeV and five centrality classes. These results are

shown in Fig. 5.1, 5.2, 5.3, 5.4, 5.5 and 5.6. These are results in Run 2 analysis. The datapoints are summarized in the following tables.

Table C.1: The invariant differential yield of heavy flavor electrons in minimum biasAu + Au collisions in

√sNN = 200 GeV.

pT (GeV/c) 12πpT

dNe

dydpTstat. error sys. error (+) sys. error (−) comment

0.347 5.1383e-02 1.0911e-02 2.5235e-02 2.6695e-020.446 3.4119e-02 6.3093e-03 1.2778e-02 1.3485e-020.546 1.8739e-02 3.6599e-03 7.0393e-03 7.4293e-030.684 1.3407e-02 1.4443e-03 3.1393e-03 3.2799e-030.885 4.8469e-03 6.9054e-04 1.0466e-03 1.0901e-031.086 1.8593e-03 3.5921e-04 3.8546e-04 4.0074e-041.288 7.5209e-04 2.1681e-04 1.6098e-04 1.6751e-041.611 2.4963e-04 4.9966e-05 4.0639e-05 4.1719e-052.320 2.1728e-05 9.3890e-06 3.6175e-06 3.7197e-063.358 7.1900e-07 1.6719e-06 1.5439e-07 7.1900e-07 shown as “→”

139

140 APPENDIX C. DATA TABLE

Table C.2: The invariant differential yield of heavy flavor electrons in 0-10% centralAu + Au collisions in

√sNN = 200 GeV.

pT (GeV/c) 12πpT

dNe




√sNN = 200 GeV.

pT (GeV/c) 12πpT

dNe

dydpTstat. error sys.error (+) sys.error (−) comment

0.347 2.5069e-01 5.1595e-02 1.2311e-01 1.3024e-010.446 8.1438e-02 2.8830e-02 3.0498e-02 3.2186e-020.546 1.5789e-02 1.6897e-02 5.9310e-03 1.5789e-02 shown as “→”0.684 2.2646e-02 6.5243e-03 5.3026e-03 5.5401e-030.885 1.0463e-02 3.1854e-03 2.2592e-03 2.3531e-031.086 5.6593e-03 1.7259e-03 1.1732e-03 1.2197e-031.288 1.7300e-03 1.0216e-03 3.7028e-04 3.8530e-041.611 4.1933e-04 2.3646e-04 6.8267e-05 7.0080e-052.320 9.4155e-06 4.3146e-05 1.5676e-06 9.4155e-06 shown as “→”3.358 1.1223e-05 8.5209e-06 2.4099e-06 2.5168e-06


√sNN = 200 GeV.

pT (GeV/c) 12πpT

dNe


0.347 2.0903e-02 2.4537e-02 1.0266e-02 2.0903e-02 shown as “→”0.446 4.9819e-02 1.4409e-02 1.8657e-02 1.9690e-020.546 2.2979e-02 8.2808e-03 8.6318e-03 9.1101e-030.684 2.0266e-02 3.2620e-03 4.7454e-03 4.9579e-030.885 7.5990e-03 1.6080e-03 1.6408e-03 1.7090e-031.086 1.9210e-03 8.3206e-04 3.9825e-04 4.1403e-041.288 1.0073e-03 5.0680e-04 2.1560e-04 2.2435e-041.611 3.9285e-04 1.2126e-04 6.3957e-05 6.5655e-052.320 1.7569e-05 2.2482e-05 2.9252e-06 1.7569e-05 shown as “→”

C.1. INVARIANT DIFFERENTIAL YIELD OF HEAVY FLAVOR ELECTRONS 141


√sNN = 200 GeV.

pT (GeV/c) 12πpT

dNe




√sNN = 200 GeV.

pT (GeV/c) 12πpT

dNe

dydpTstat. error sys. error (+) sys. error(−) comment

0.347 6.2483e-03 4.4116e-03 3.0686e-03 3.2462e-030.484 1.9156e-03 1.2876e-03 7.1720e-04 7.5697e-040.740 3.0672e-04 2.6902e-04 6.9811e-05 7.2860e-051.149 1.1195e-04 7.6627e-05 2.3400e-05 2.4331e-051.611 1.6400e-05 upper limit2.320 7.1337e-06 4.5662e-06 1.1877e-06 1.2212e-06

Table C.7: The Ncoll scaled integrated yield of heavy flavor electrons dNe/dyNcoll

(0.8 < pT <

4.0 GeV/c).

Centrality NcolldNe/dy

Ncoll(0.8<pT<4.0) stat.error sys.err(+) sys.err(−)

0-10% 955.4 ± 93.6 4.072e-05 6.522e-06 9.222e-06 9.514e-0610-20% 602.6 ± 59.3 4.155e-05 8.007e-06 9.533e-06 9.842e-0620-40% 296.8 ± 31.1 5.143e-05 8.122e-06 1.185e-05 1.222e-0540-60% 90.7 ± 11.8 5.437e-05 1.441e-05 1.304e-05 1.339e-0560-92% 14.5 ± 4.0 3.576e-05 2.595e-05 1.207e-05 1.224e-05

minimum bias 257.8 ± 25.4 4.269e-05 4.000e-06 9.700e-06 1.001e-05p + p 1.0 5.631e-05 1.247e-05

Appendix D

Centrality Dependence of CharmProduction

D.1 Systematic Uncertainty of α for Run 2 data

The systematic uncertainty of the parameter α is estimated for the linearity test of thecharm production in Run 2. Since the α influences the curvature of the fit function A·Nα−1

coll ,we only consider the following sources:

(1) Multiplicity dependent efficiency loss (7%),

(2) Extraction of non-photonic electrons (uncertainty of Rγ and RCN).

These two sources change the shape of the distribution of dNe/dy(0.8 < pT < 4.0)/Ncoll. Incontrast, the other sources of the systematic uncertainties (e.g. the electron ID efficiency)do not affect the α, but contribute the overall normalization of dNe/dy. Therefore, theother sources are neglected.

For the source (1), the systematic uncertainty is 7% as described in section 4.9.2. Ap-plying the higher (lower) limit of the uncertainty instead of the center value, we re-calculatethe dNe/dy(0.8 < pT < 4.0)/Ncoll, and then fit it by the function, A · Nα−1

coll . Figure D.1(left) shows the three kinds of dNe/dy(0.8 < pT < 4.0)/Ncoll which correspond to thecenter (gray) value, the higher (blue point) and the lower (red point) limit of the multi-plicity dependent efficiency applied, respectively. The curves represent the correspondingfit functions. The obtained α values are summarized in Tab. D.1. The difference of theα between the center and the higher (lower) is assigned as the higher (lower) systematicuncertainty.

Next, we study the systematic uncertainty of the source (2). We re-calculate the non-photonic electron spectrum by changing Rγ, RCN and the blocking effect with their higher(lower) limit, and then re-calculate dNe/dy/Ncoll. The combination of the higher (lower)Rγ, the lower (higher)RCN and the lower (higher) blocking effect provides a larger (smaller)non-photonic electron yield. Then, α is obtained for the higher and lower dNe/dy/Ncoll.Figure D.1 (right) shows the re-calculated dNe/dy(0.8 < pT < 4.0)/Ncoll using the uncer-tainty of the source (2). The gray, the blue and the red points correspond to the center

142

D.1. SYSTEMATIC UNCERTAINTY OF α FOR RUN 2 DATA 143

collN0 200 400 600 800 1000 1200

Co

ll<4

.0)

/ NT

/dy

(0.8

<pe

dN

0

0.01

0.02

0.03

0.04

0.05

0.06

0.07

0.08

0.09

0.1-3x10

collN0 200 400 600 800 1000 1200

Co

ll<4

.0)

/ NT

/dy

(0.8

<pe

dN

0

0.01

0.02

0.03

0.04

0.05

0.06

0.07

0.08

0.09

0.1-3x10

Figure D.1: The dNe/dy(0.8 < pT < 4.0)/Ncoll for the source (1) (left) and (2) (right).The gray, the blue and the red points correspond to the center value, the higher and thelower uncertainties, respectively. The curves represent the fit functions.

value, the higher and the lower uncertainties, respectively. The curves represent the fitfunctions. The obtained α is summarized in Tab. D.1.

In order to average the higher and the lower systematic uncertainties, we calculatea root mean square (RMS) of the higher and the lower errors for source (1) and (2),respectively. The total systematic uncertainty of α is calculated adding in quadraturethese RMS errors of source (1) and (2). The total systematic error is also shown in Tab.D.1.

Table D.1: The systematic uncertainty of α for the source (1) and (2). The total systematicerror is estimated adding these errors in quadrature.

Source of uncertainty α − 1 difference RMS (high and low)Center value -0.0623 - -Source 1 for higher -0.0496 0.0127 0.0130Source 1 for lower -0.0756 -0.0133Source 2 for higher -0.0779 -0.0156 0.0120Source 2 for lower -0.0558 0.0065total sys. error 0.0177 (0.0130 ⊕ 0.0120)

144 APPENDIX D. CENTRALITY DEPENDENCE OF CHARM PRODUCTION

D.2 Systematic Uncertainty of α for Run 4 and Run

5 data

The systematic uncertainty of the linearity parameter α is estimated in Run 4 data analysisas same as Run 2. Since the most of the systematic errors (the electron ID and theacceptance) are independent of centrality. These errors do not affect the linearity (α).The sources of the systematic uncertainties are considered as following:

(1) Multiplicity dependent efficiency loss,

(2) Rγ which is used to separate non-photonic and photonic electrons,

(3) Number of nucleon-nucleon collisions, Ncoll.

First, the systematic uncertainty from the source (1) is considered. The systematicuncertainty of the multiplicity dependent efficiency loss (εmult) is summarized in Tab.D.2 [122]. To estimate the systematic uncertainty of α from source (1), the data pointsare moved up (down) by the higher (lower) uncertainty of εmult for the correspondingcentrality, and then the moved data are fit by the function, A ·Nα−1

coll . Figure D.2 shows themoved-up (down) dNe/dy(pT > 0.4)/TAA as a function of Ncoll. The black, the magentaand the light-blue points correspond to the center, the moved data by the higher andthe lower systematic uncertainties, respectively. The curves are the fit functions for thecorresponding data. The fit are performed for Au + Au and for Au + Au and p+ p data.The fit results are shown in the left and the right panel of Fig. D.2.

Table D.2: Systematic uncertainty of the multiplicity dependent efficiency loss [122].

Centrality Sys. error of εmult

0-10% 4.43%10-20% 3.47%20-40% 2.50%40-60% 1.72%60-92% 1.26%

Min. bias 3.22%p+ p 0%

Second, the systematic uncertainty of α from source (2) is estimated. We, first, considerthe error propagation of Rγ to the non-photonic electron yield based on Eq. D.1:

Nnon−γe =

Rγ ·NConv−oute −NConv−in

e

Rγ − 1, (D.1)

D.2. SYSTEMATIC UNCERTAINTY OF α FOR RUN 4 AND RUN 5 DATA 145

collN-200 0 200 400 600 800 1000 1200

(u

b)

AA

dN

/dy(

pT

>0.4

) / T

4

4.5

5

5.5

6

6.5

7

7.5

8Fit with the move-up dataFit with the move-down data

, Au+AumultεSystematic Error of

collN-200 0 200 400 600 800 1000 1200

(u

b)

AA

dN

/dy(

pT

>0.4

) / T

4

4.5

5

5.5

6

6.5

7

7.5

8Fit with the move-up dataFit with the move-down data

, Au+Au + ppmult

εSystematic Error of

Figure D.2: dNe/dy(pT > 0.4)/TAA as a function of pT. The data points are moved-up(down) by the systematic uncertainty of εmult. The magenta and the light-blue pointscorrespond to the higher and the lower systematic uncertainties. The curves are the fitfunctions for the corresponding data. The fit are performed for the Au + Au (left) and forthe Au + Au and p+ p data (right) separately.

where the blocking effect is omitted for simplification. If Rγ has an error “e”, Rγ ismodified as Rγ → R

′γ = Rγ + e. Then, Eq. D.1 is transformed as follows:

Nnon−γ′

e =R

′γ ·NConv−out

e −NConv−ine

R′γ − 1

=(Rγ + e) ·NConv−out

e −NConv−ine

(Rγ + e) − 1(D.2)

= Nnon−γe ·

(1 +

e · (k − 1)

Rγ + e− 1

), (D.3)

where k is the ratio of inclusive to non-photonic electrons (k = N ince /Nnon−γ

e ). Therefore,the relative error of non-photonic electrons (δNnon−γ

e /Nnon−γe ) is expressed as:

δNnon−γe

Nnon−γe

=Nnon−γ′

e −Nnon−γe

Nnon−γe

=e · (k − 1)

Rγ + e− 1. (D.4)

Based on Eq. D.4, we obtain the relative errors of non-photonic electrons from the sys-tematic uncertainty of Rγ. Here, we use Rγ = 2.333 at pT = 0.5 GeV/c and e = 0.064(2.74% of Rγ) described in [122]. Figure D.3 shows the ratio k of inclusive to non-photonicelectrons as a function of centrality classes. The ratio k and the obtained relative errorsare summarized in Tab. D.4.

Using these relative errors, the data points are moved to the higher and the lower limitof systematic uncertainty for dNe/dy/TAA. Then, they are fit to obtain the limit of α.Figure D.4 shows the higher and the lower limit of dNe/dy/TAA and the center value. Thecurves represent the fit functions. The left and right panels correspond to the data inAu + Au and in Au + Au and p+ p collisions, respectively. The systematic uncertainty ofα is obtained by averaging in the same way. The obtained uncertainty are summarized inTab. D.5.


Table D.3: The systematic uncertainty of α for the source (1). The parameter α areobtained for the Au + Au data and for the Au + Au and p+ p data separately.

Au + Au dataα difference RMS of difference

center 1.00922 - -higher 1.01944 0.01022 0.01042lower 0.99860 -0.01062

Au + Au and p+ p datacenter 1.00971 - -higher 1.01826 0.00855 0.00864lower 1.00097 -0.00874

Table D.4: Summary of the ratio k and the relative error of non-photonic electrons in Run4 analysis.

Centrality k Relative Error0-10% 3.47 11.30%10-10% 3.86 13.08%20-10% 4.45 15.78%40-10% 4.95 18.07%60-10% 6.08 23.23%Min.bias 4.09 14.13%p+ p 6.942 27.18%

Third, the systematic uncertainty from the source (3) is estimated. The systematicuncertainty of Ncoll consists of two components. There is the correlated and the un-correlated term with respect to centrality. The un-correlated error only contributes to theα. The estimation of the un-correlated term of Ncoll is described in detail in AppendixD.3.

Using the higher and the lower limits of Ncoll systematic uncertainty, the higher and thelower limit of dNe/dy/Ncoll is calculated, and then they are fit to obtain the uncertaintyof α. The Ncoll value used in this analysis is summarized in Tab. D.6. The Ncoll valuein p+ p collisions is by definition 1.0 with no error. However, the relative value betweenp+ p data point and all five Au + Au data points are moved up/down due to the commonsystematic error of Ncoll. Therefore, we assign 7.3% to the p+ p point as this commonerror of Ncoll which is equivalent to the systematic uncertainty of TAA in minimum biascollisions.

Figure D.5 shows the higher and the lower limit of dNe/dy/TAA and the center value.These three curves represent the fit functions for these data points. The left and theright panels correspond to the data in Au + Au and in Au + Au and p+ p collisions,respectively. The systematic uncertainty of α is obtained by averaging in the same way.


collN0 200 400 600 800 1000

/dy

γn

on

-e

dN

/dy

inc

ed

N

3

4

5

6

7

8

9

Centrality Binned

Min.Bias

p+p

Figure D.3: Ratio k of inclusive to non-photonic electrons. Electrons are measured in Run4 Au + Au [121] and Run 5 p+ p collisions [39].

Table D.5: The systematic uncertainty of α for the source (2). The parameter α areobtained for Au + Au and for Au + Au and p+ p separately.


center 1.00922 - -higher 0.98326 -0.02596 0.03169lower 1.04576 0.03654

Au + Au and p+ p datacenter 1.00971 - -higher 0.98690 -0.02281 0.02881lower 1.04348 0.03377

The uncertainty is summarized in Tab. D.7.Finally, these three uncertainties are added in quadrature to obtain the total uncer-

tainty. Table D.8 shows these three and the total uncertainties. As the result, we obtainα = 1.0092±0.0120(stat)±0.0506(sys) in Au + Au collisions. If the p+ p point is included,α = 1.0097 ± 0.0094(stat) ± 0.0403(sys).


collN-200 0 200 400 600 800 1000 1200

AA

dN

/dy(

pT

>0.4

) / T

4

4.5

5

5.5

6

6.5

7

7.5

8Fit higher sys.errorFit lower sys.error

at Au+Auγ

Systematic Error of R

collN-200 0 200 400 600 800 1000 1200

AA

dN

/dy(

pT

>0.4

) / T

4

4.5

5

5.5

6

6.5

7

7.5


at Au+Au + ppγ

Systematic Error of R

Figure D.4: dNe/dy(pT > 0.4)/TAA as a function of pT. The data points are moved-up (down) by the systematic uncertainty of Rγ. The magenta and the light-blue pointscorrespond to the higher and the lower systematic uncertainties. The curves are the fitfunctions for corresponding data. The fit are performed for Au + Au (left) and for Au + Au+ p+ p data (right) separately.

Table D.6: Uncorrelated systematic uncertainty of Ncoll described in Appendix D.3.

Centrality Ncoll Sys. error Comment(Uncorrelated term only)

0-10% 955.4 9.710-20% 602.6 3.320-40% 296.8 7.840-60% 90.7 5.360-92% 14.5 3.6p+ p 1 18.9 7.3%. systematic error of

TAA in minimum bias


collN-200 0 200 400 600 800 1000 1200

AA

dN

/dy(

pT

>0.4

) / T

4

4.5

5

5.5

6

6.5

7

7.5


at Au+Aucoll

Systematic Error of N

collN-200 0 200 400 600 800 1000 1200

AA

dN

/dy(

pT

>0.4

) / T

4

4.5

5

5.5

6

6.5

7

7.5


at Au+Au + ppcoll

Systematic Error of N

Figure D.5: dNe/dy(pT > 0.4)/TAA as a function of pT. The data points are moved-up(down) by the systematic uncertainty of Ncoll. The magenta and the light-blue pointscorrespond to the higher and the lower systematic uncertainties. The curves represent thefit functions for the corresponding data. The fit are performed for Au + Au (left) and forAu + Au + p+ p data (right).

Table D.7: The systematic uncertainty of α for the source (3). The parameter α areobtained for Au + Au and for Au + Au and p+ p separately.


center 1.00922 - -higher 1.04051 0.03131 0.03800lower 0.96553 -0.04369

Au + Au and p+ p datacenter 1.00971 - -higher 1.03369 0.02398 0.02683lower 0.98031 -0.02940

Table D.8: Summary of the α systematic uncertainty from all three sources

Au + Au Au + Au and p+ pSource (1) 0.01042 0.00864Source (2) 0.03169 0.02881Source (3) 0.03800 0.02683

Total 0.05057 0.04031


D.3 Uncorrelated Systematic Uncertainty of Ncoll

We studied the systematic uncertainty of Ncoll. The uncertainty of Ncoll is subdivided tothe correlated and the uncorrelated term with respect to centrality. The uncertainty ismade from 14 components [106] listed below:

1. σpp = 39 mb, instead of σNN = 42 mb,

2. σpp = 45 mb, instead of σNN = 42 mb,

3. Different Woods-Saxon parameters (R = 6.65 fm, a = 0.55 fm),

4. Different Woods-Saxon parameters (R = 6.25 fm, a = 0.53 fm),

5. Different radius of hard-core. (Rcore = 0.4),

6. Different neutron loss in the ZDC,

7. Different BBC fluctuation,

8. Gray disk nucleon overlap function,

9. Gaussian nucleon overlap function,

10. Different origin of angles for the centrality selection,

11. Higher BBC trigger efficiency (92.2 + 2.5%),

12. Lower BBC trigger efficiency (92.2 - 3.0%),

13. Additional uncertainty of the centrality determination (more central),

14. Additional uncertainty of the centrality determination (less central).

The uncertainties calculated by these modified parameters are compared to that of thedefault parameters. The some components are the pair of the higher and the lower limit.Thus, the differences from the following pairs are averaged.

(i) σNN (average of 1 and 2)

(ii) Wood-Saxon parameter (average of 3 and 4)

(iii) Different radius of hard-core (5)

(iv) Different neutron loss in the ZDC (6)

(v) Different BBC fluctuation (7)

(vi) Overlap function ( average of 8 and 9)

(vii) Different origin of angles (10)

D.3. UNCORRELATED SYSTEMATIC UNCERTAINTY OF NCOLL 151

collN0 200 400 600 800 1000 1200

Co

rr E

rr (

%)

-0.15

-0.1

-0.05

0

0.05

0.1

0.15 = 39mbnnσ = 45mbnnσ

collN0 200 400 600 800 1000 1200

Co

rr E

rr (

%)

-0.15

-0.1

-0.05

0

0.05

0.1

0.15Woods-Saxon:R=6.65fm,d=0.55fm

Woods-Saxon:R=6.25fm,d=0.53fm

collN0 200 400 600 800 1000 1200

Co

rr E

rr (

%)

-0.15

-0.1

-0.05

0

0.05

0.1

0.15

=0.4fm)hard

Nucleons with hard core(r

collN0 200 400 600 800 1000 1200

Co

rr E

rr (

%)

-0.15

-0.1

-0.05

0

0.05

0.1

0.15

ZDC: different neutron loss function

collN0 200 400 600 800 1000 1200

Co

rr E

rr (

%)

-0.15

-0.1

-0.05

0

0.05

0.1

0.15

different BBC fluctuations

collN0 200 400 600 800 1000 1200

Co

rr E

rr (

%)

-0.15

-0.1

-0.05

0

0.05

0.1

0.15gray disk nn-overlap functiongaussian nn-overlap function

collN0 200 400 600 800 1000 1200

Co

rr E

rr (

%)

-0.25

-0.2

-0.15

-0.1

-0.05

0

0.05

0.1

0.15

0.2

0.25

different origin for angle calculation

collN0 200 400 600 800 1000 1200

Co

rr E

rr (

%)

-0.15

-0.1

-0.05

0

0.05

0.1

0.15BBC TrigEff 91.4+2.5%%: more central

BBC TrigEff 91.4-3.0%%: less central

collN0 200 400 600 800 1000 1200

Co

rr E

rr (

%)

-0.15

-0.1

-0.05

0

0.05

0.1

0.15exp. centrality selection : more central

exp. centrality selection : less central

Figure D.6: The correlated terms of the uncertainty as a function of centrality. Thesepanels from the top-left to the bottom-right correspond to the errors from components (i)to (ix), respectively. The line in each panel shows the weighted average.

(viii) BBC trigger efficiency (11 and 12)

(ix) Additional uncertainty (13 and 14)

The total error is calculated adding these averaged errors in quadrature (from i to ix).Each component contains both the correlated and uncorrelated terms. To separate theseterms, we define that the correlated term is the weighted average of these errors at thefive centrality bins and the uncorrelated term is the difference from the weighted average.Here, the Ncoll value in each centrality class is used as the weight factor. Figure D.6 showsthe errors from these components. These panels from the top-left to the bottom-rightcorrespond to the errors from the components (i) to (ix), respectively. The line in eachpanel shows the weighted average which is the correlated term for each component.

Figure D.7 shows the uncorrelated errors as a function of Ncoll for these components.In each panel, the error pairs (closed circles) listed above and the averages of these pairs(open circles) are shown. The averages of these pairs represent the uncorrelated errors.The obtained uncorrelated errors from these components (i to ix) are summarized in Tab.D.9.

Figure D.8 shows the systematic error of Ncoll as a function of Ncoll. The yellow and thegray band represent the total and the uncorrelated errors, respectively. The uncorrelated


collN0 200 400 600 800 1000 1200

Un

corr

Err

(%

)

-0.05

-0.04

-0.03

-0.02

-0.01

0

0.01

0.02

0.03

0.04

0.05 = 39, 45nnσAve of

= 39mbnnσ = 45mbnnσ

collN0 200 400 600 800 1000 1200

Un

corr

Err

(%

)

-0.05

-0.04

-0.03

-0.02

-0.01

0

0.01

0.02

0.03

0.04

0.05Ave of W/S_radius=6.25,6.65



collN0 200 400 600 800 1000 1200

Un

corr

Err

(%

)

-0.05

-0.04

-0.03

-0.02

-0.01

0

0.01

0.02

0.03

0.04

0.05=0.4fm)

hardNucleons with hard core(r

=0.4fm)hard


collN0 200 400 600 800 1000 1200

Un

corr

Err

(%

)

-0.05

-0.04

-0.03

-0.02

-0.01

0

0.01

0.02

0.03

0.04

0.05ZDC: different neutron loss function

ZDC: different neutron loss function

collN0 200 400 600 800 1000 1200

Un

corr

Err

(%

)

-0.1

-0.08

-0.06

-0.04

-0.02

0

0.02

0.04

0.06

0.08

0.1



collN0 200 400 600 800 1000 1200

Un

corr

Err

(%

)

-0.1

-0.08

-0.06

-0.04

-0.02

0

0.02

0.04

0.06

0.08

0.1Ave of disk overlap function

gray disk nn-overlap function

gaussian nn-overlap function

collN0 200 400 600 800 1000 1200

Un

corr

Err

(%

)

-0.25

-0.2

-0.15

-0.1

-0.05

0

0.05

0.1

0.15

0.2

0.25different origin for angle calculation

different origin for angle calculation

collN0 200 400 600 800 1000 1200

Un

corr

Err

(%

)

-0.1

-0.08

-0.06

-0.04

-0.02

0

0.02

0.04

0.06

0.08

0.1Ave of BBC TrigEff 91.4+2.5-3.0

BBC TrigEff 91.4+2.5%%: more central

BBC TrigEff 91.4-3.0%%: less central

collN0 200 400 600 800 1000 1200

Un

corr

Err

(%

)

-0.05

-0.04

-0.03

-0.02

-0.01

0

0.01

0.02

0.03

0.04

0.05Ave of exp. cent more/less

exp. centrality selection : more central

exp. centrality selection : less central

Figure D.7: The uncorrelated terms of the uncertainty as a function of centrality. In eachpanel, the closed circles are the errors from these components and the open circles are theaverages of these error pairs.

Table D.9: Summary of the uncorrelated Ncoll uncertainty for the five centrality classes.

Centrality 0-10% 10-20% 20-40% 40-60% 60-92%Component Uncorrelated error (average of pair)

i 0.33% 0.21% 0.58% 0.80% 2.56%ii 0.28% 0.07% 0.48% 1.64% 2.73%iii -0.15% -0.05% 0.45% 0.74% -1.93%iv -0.03% -0.04% 0.20% -0.08% 0.03%v 0.01% -0.22% 0.03% -0.33% 5.93%vi 0.60% 0.25% 1.21% 3.20% 6.63%vii 0.19% -0.37% -1.90% 3.22% 20.30%viii 0.06% 0.06% 0.06% 0.06% 7.33%ix 0.65% 0.02% 1.01% 2.91% 3.05%

Total 1.02% 0.55% 2.63% 5.75% 23.92%

D.3. UNCORRELATED SYSTEMATIC UNCERTAINTY OF NCOLL 153

collN0 200 400 600 800 1000 1200

coll

def

ault

) / N

coll

def

ault

- N

coll

X(N

-0.4

-0.3

-0.2

-0.1

-0

0.1

0.2

0.3

0.4 = 39mbnnσ = 45mbnnσ

Woods-Saxon:R=6.65fm,d=0.55fmWoods-Saxon:R=6.25fm,d=0.53fm

=0.4fm)hard


ZDC: different neutron loss functiondifferent BBC fluctuations

gray disk nn-overlap functiongaussian nn-overlap functiondifferent origin for angle calculationBBC TrigEff 91.4+2.5%%: more centralBBC TrigEff 91.4-3.0%%: less centralexp. centrality selection : more centralexp. centrality selection : less central

Figure D.8: Systematic uncertainty ofNcoll as a function of centrality. The points representthe difference of the calculated Ncoll compared to the default value. The yellow and graybands correspond to the total and the uncorrelated error of Ncoll.

errors are summarized in Tab. D.10.

Table D.10: Summary of the total and the uncorrelated systematic uncertainties for thefive centrality classes.

Centrality Ncoll Total error Uncorrelated error0-10% 955 86.0 9.710-20% 599.8 51.7 3.320-40% 295.4 24.9 7.840-60% 91.5 8.4 5.360-92% 14.9 3.6 3.6

Appendix E

Charm Cross Section in Run 4

The total charm cross section in Au + Au collisions at√sNN = 200 GeV is calculated

using the heavy flavor electron spectra with high statistics measured in Run 4 [121]. Themethod to calculate the cross section is slightly modified from that described in section5.6.1. This modified method is initially used to calculate the charm cross section in p+ pcollisions [39, 131]. The procedure is described as follow:

(1) The invariant integrated yield per nucleon-nucleon collision (dNe

dy/TAA (pT > pTlow))

above pT > pTlow is calculated. We denote dNe

dy/TAA (pT > pTlow) = dσe/dy (pT >

pTlow). The pTlow = 0.4 GeV/c is chosen, since the available pT range is extended tolower pT due to the higher statistics.

(2) The obtained integrated yield (dσe/dy (pT > pTlow) is extrapolated to the entire pT

range (dσe/dy). The extrapolation factor RpT= 0.54 is obtained from the FONLL

calculation instead of the PYTHIA calculation.

(3) The contributions from bottom decays (b→ e) and bottom cascade decays (b→ c→e) are subtracted. These contributions (b → e, b → c → e) is determined based on aFONLL calculation. These contributions are σb→e = 0.061 µb and σb→c→e = 0.053 µb,respectively.

(4) The integrated yield (dσe/dy) of electrons is converted to the charm yield (dσcc/dy)using a total branching ratio BR(c→ e) = 9.5 ± 1.0%.

dσcc

dy=

1

BR(c→ e)

1

Ce/D

dσe

dy(E.1)

Here, Ce/D = 0.935 is a correction factor to account for the difference of the rapiditydistribution between D mesons and their decayed electrons.

(5) dσcc/dy is extrapolated to the entire rapidity range to obtain the total cross section.

σcc =1

Ry

1

Cc/D

dσcc

dy(E.2)

154

155

The extrapolation factor Ry = 0.225 is determined using a HVQMNR calculation [130]instead of the PYTHIA calculation. Here, the factor Cc/D = 0.96 accounts for thedifference of rapidity distribution between D mesons and charm quarks.

For the step 1, we calculate dσe/dy(pT > pTlow) by adding the bin entries of heavyflavor electron spectra with pT > pTlow, instead of fitting by a PYTHIA electron functiondescribed in section 5.6.1. The bin counting method is more reliable, because the fittinguncertainty is not involved to the systematic uncertainty. The obtained integrated yield(dσe/dy(pT > pTlow)) is summarized in Tab. E.1. The systematic uncertainty is calculatedfrom the heavy flavor electron spectrum and TAA.

Table E.1: The summary of dσe/dy(pT > 0.4) calculated in the step 1.

Centrality TAA(mb−1) dNe/dy(pT > 0.4)(×10−3) dσe/dy(pT > 0.4)(µb)0-10% 22.8 148.0 ± 2.7 ± 21.1 6.50 ± 0.12 ± 0.9210-20% 14.4 90.7 ± 1.9 ± 13.9 6.30 ± 0.13 ± 0.9620-40% 7.07 43.5 ± 0.9 ± 7.6 6.15 ± 0.13 ± 1.0740-60% 2.16 14.3 ± 0.5 ± 2.8 6.64 ± 0.22 ± 1.2860-92% 0.35 2.10 ± 0.15 ± 0.59 5.98 ± 0.42 ± 1.68Min.bias 6.14 36.6 ± 0.5 ± 5.9 5.96 ± 0.08 ± 0.96p+ p [131] 5.95 ± 0.59 ± 1.59

For the step 2 to 5, the extrapolation factor RpTand Ry are determined based on the

FONLL and the HVQMNR calculation, respectively, instead of the PYTHIA calculation.Although PYTHIA contains only lowest order (LO) pQCD, FONLL and HVQMNR isNLO and beyond. Thus, FONLL and HVQMNR can provide a trustworthy calculation.In addition, the FONLL calculation agrees well with the data within the statistical andthe systematic uncertainty [39]. As a result, the factor RpT

= 0.54 ± 0.05(10%) andRy = 0.225 ± 0.034(15%) are obtained. The errors of these factors are assigned as thesystematic uncertainty.

HVQMNR calculates the rapidity distribution of bare charm quarks, but not charmedhadrons and/or charm electrons. We need to take into account the difference of rapiditydistribution between charm quarks, D mesons and charmed electrons. The factor Ce/D isused to correct the rapidity distribution of charm electrons to D mesons, and Cc/D alsocorrects the rapidity spread of charm quarks to D mesons. These factors Ce/D = 0.935and Cc/D = 0.96 are determined by PYTHIA.

The results in step 2 and 3 are summarized in Tab. E.2. In these steps, the invariantintegrated yield dσe/dy is calculated and then bottom contributions calculated by FONLLare subtracted.

After the subtraction of the b contribution, the charm contribution only remains inheavy flavor electrons. Therefore, we denote the electrons as charm electrons. In step 4and 5, the integrated yield of charm electrons is translated to the charms yield (dσcc/dy)using BR(c → e) = 9.5 ± 1.0%. Then, that is extrapolated to the total charm cross

156 APPENDIX E. CHARM CROSS SECTION IN RUN 4

Table E.2: Summary of dσe/dy and then b contribution is subtracted.

Centrality dσe/dy (µb) dσe/dy afterb subtraction (µb)

0-10% 12.04±0.22±2.04 11.93±0.22±2.0410-20% 11.66±0.24±2.09 11.55±0.24±2.0920-40% 11.38±0.23±2.24 11.27±0.23±2.2440-60% 12.30±0.40±2.62 12.18±0.40±2.6260-92% 11.08±0.77±2.38 10.97±0.77±2.38Min.bias 11.03±0.15±2.05 10.92±0.15±2.05p+ p [131] 11.02±1.10±3.12 10.90±1.10±3.12

section (σcc). The obtained dσcc/dy and σcc are summarized in Tab. E.3. The systematicuncertainties are determined by adding the errors in these steps in quadrature.

Table E.3: Summary of dσcc/dy and σcc in minimum bias Au+Au collisions and fivecentrality classes. The charm cross section in p+ p collisions is also shown [39].

Centrality TAA (mb−1) dσcc/dy (µb) σcc (µb)0-10% 22.8 134.3±2.4±27.0 620±11±15610-20% 14.4 130.0±2.7±27.3 600±12±15520-40% 7.07 126.8±2.6±28.6 586±12±15940-60% 2.16 137.1±4.6±32.9 633±21±18060-92% 0.35 123.4±8.7±39.2 570±40±201Min.bias 6.14 122.9±1.6±26.5 568± 8±150p+ p [131] 122.8±12.3±37.5 567±57±193

Bibliography

[1] F. Halzen and A. D. Martin, John Wiley & Sons (1984) [Quarks and Leptons].

[2] K. Yagi, T. Hatsuda and Y. Miake, Camb. Monogr. Part. Phys. Nucl. Phys. Cosmol.23, 1 (2005) [Quark-Gluon Plasma].

[3] E. V. Shuryak, Phys. Rep. 61, 71 1980.

[4] F. Karsch, Lect. Notes Phys. 583, 209 2002.

[5] K. Rajagopal, Nucl. Phys. A 661, 150 1999.

[6] K. Adcox et al., Nucl. Phys. A 757, 184 (2005) [PHENIX Collaboration].

[7] X. N. Wang, M. Gyalassy, Phys. Rev. Lett. 68, 1480 2002.

[8] X M. Gyalassy et al., nucl-th/0302077.

[9] S. S. Adler et al., Phys. Rev. Lett. 91, 072301 (2003) [PHENIX Collaboration].


[11] U. Heinz, nucl-th/0407067 (2004)

[12] K. H. Ackermann et al., Phys. Rev. Lett. 86, 402 (2001) [STAR Collaboration].


[14] J. Adams et al., Phys. Rev. C 72, 014904 (2005) [STAR Collaboration].

[15] P. Huovinen et al., Phys. Lett. B503, 58 (2001).

[16] D. Molnar and S. Voloshin, Phys. Rev. Lett. 91, 092301 (2003).

[17] J. A. Appel, Annu. Rev. Nucl. Part. Sci. 42, 367 (1992).

[18] B. Muller and X. N. Wang, Phys. Rev. Lett. 68, 2437 (1992).

[19] Y. L. Dokshitzer and D. E. Kharzeev, Phys. Lett. B 519, 199 (2001).

[20] R. Vogt, Heavy Ion Phys. 17, 75 (2003).

[21] F. W. Busser et al., Phys. Lett. B 53, 212 (1974).

157

158 BIBLIOGRAPHY

[22] F. W. Busser et al., Nucl. Phys. B 113, 189 (1976).

[23] P. Perez et al., Phys. Lett. B 112, 260 (1982).

[24] M. Basile et al., Nuovo Cimento A 65, 421 (1981).

[25] I. Hinchliffe and C. H. Llewellyn Smith, Phys. Lett. B 61, 472 (1976).

[26] E. Norrbin and T. Sjostrand, Eur. Phys. J. C 17, 137 (2000).

[27] C. Amsler et al., Phys. Lett. B 667, 1 (2008). [Particle Data Group] (URL:http://pdg.lbl.gov)

[28] T. Sjostrand, Comput. Phys. Commun. 82, 74 (1994).

[29] H. L. Lai et al. Eur. Phys. J. C 12, 375 (2000).

[30] A. D. Martin et al. Eur. Phys. J. C 35, 325 (2004).

[31] M. Gluck et al. Eur. Phys. J. C 53, 355 (2008).

[32] M. Cacciari et al. JHEP 9805, 007 (1998).

[33] M. Cacciari et al. Phys. Rev. Lett. 95, 122001 (2005).

[34] M. Cacciari et al. Phys. Rev. Lett. 89 , 122003 (2002).

[35] M. Cacciari et al. JHEP 0407, 033 (2004).

[36] M. Cacciari,http://www-rnc.lbl.gov/ esichter/HeavyQuark-2007/talks/Fri/cacciari.pdf

[37] A. Abulencia et al. Phys. Rev. D 75 , 012010 (2007) [CDF Collaboration].

[38] D. Acosta et al. Phys. Rev. D 71 , 032001 (2005) [CDF Collaboration].

[39] A. Adare et al. Phys. Rev. Lett. 97 , 252002 (2006) [PHENIX Collaboration].

[40] J. D. Bjorken, Phys. Rev. D 17, 171 (1978).

[41] C. Peterson et al., Phys. Rev. D 27, 105 (1983).

[42] V. G. Kartvelishvili et al., Phys. Lett. B 78, 615 (1978).

[43] P. Collins and T. Spiller, J. Phys. G 11, 1289 (1985).

[44] G. Colangelo and P. Nason, Phys. Lett. B 285, 167 (1992).

[45] M. G. Bowler, Z. Phys. C 11, 169 (1981).

[46] E. Braaten et al., Phys. Rev. D 51, 4819 (1995).

[47] P. Nason and C. Oleari, Nucl. Phys. B 565, 245 (2000).

BIBLIOGRAPHY 159

[48] M. Cacciari and P. Nason, Phys. Rev. Lett. 89, 12 (2002).

[49] M. Artuso et al., Phys. Rev. D 70, 112001 (2004) [CLEO Collaboration].

[50] R. Seuster et al., Phys. Rev. D 73, 032002 (2006) [BELLE Collaboration].

[51] A. Heister et al., Phys. Lett. B 512, 30 (2001) [ALEPH Collaboration].

[52] G. Abbiendi et al., Eur. Phys. J. C 29, 463 (2003) [OPAL Collaboration].

[53] K. Abe et al., Phys. Rev. D 65, 092006 (2002) [SLD Collaboration].

[54] R. K. Ellis, W. J. Stirling and B. R. Webber, Camb. Monogr. Part. Phys. Nucl. Phys.Cosmol. 8, 1 (1996) [QCD and Collider Physics].

[55] R. K. llis, 2004 CTEQ Summer School Document (2004).

[56] J. D. Bjorken, Phys. Rev. D 27, 140 (1983).

[57] P. F. Kolb, Heavy Ion Phys. 21, 234 (2004).

[58] L. Ahle et al., Phys. Lett. B 332, 258 (1994).

[59] T. Alber et al., Phys. Rev. Lett. 75, 3814 (1995).

[60] K. Adkox et al., Phys. Rev. Lett. 87, 052301 (2001).

[61] S. S. Adler et al., Phys. Rev. C 71, 034908 (2005).

[62] S. S. Adler et al., Phys. Rev. C 71, 049901 (2005).

[63] R. J. Glauber and G. Matthiae, Nucl. Phys. B 21, 135 (1970).

[64] M. J. Tannenbaum, PHENIX Analysis Note 294 (2004).

[65] P. Levai et al., Phys. Rev. C 51, 3326 (1995).

[66] J. W. Cronin et al., Phys. Rev. D 11, 3105 (1975).

[67] M. Lev and B. Petersson, Z. Phys. C 21, 155 (1983).

[68] S. S. Adler et al., Phys. Rev. C 74, 024904 (2006) [PHENIX Collaboration].

[69] C. N. Brown et al., Phys. Rev. C 54, 3195 (1996) [E789 Collaboration].

[70] J. J. Aubert et al., Phys. Lett. B 123, 275 (1985) [EMC Collaboration].

[71] M. Arneodo et al., Nucl. Phys. B 441, 12 (1995) [NMC Collaboration].

[72] H. Wohri and C. Lourenco, J. Phys. G 30, S315 (2004).

[73] K. K. Eskola et al., Eur. Phys. J. C 9, 61 (1999).

160 BIBLIOGRAPHY

[74] S. Kelly, J. Phys. G. 30, S1189 (2004).

[75] R. Baier et al., Nucl. Phys. B 483, 291 (1997) .

[76] M. Gyulassy et al., Nucl. Phys. A 661, 637c (1999) .

[77] M. Gyulassy et al., Phys. Rev. Lett. 85, 5535 (2000) .

[78] M. Gyulassy et al., Nucl. Phys. B 571, 197 (2000) .

[79] M. Gyulassy et al., Phys. Rev. Lett. 86, 2537 (2001) .

[80] M. G. Mustafa, Phys. Rev. C 72, 014905 (2005) .

[81] S. Wicks et al., Nucl. Phys. A 784, 426 (2007) .

[82] M. Djordjevic, Phys. Rev. C 74, 064907 (2006) .

[83] A. Adare et al., Phys. Rev. C 77, 064907 (2008) [PHENIX Collaboration].

[84] To be published. [PHENIX Collaboration].

[85] H. Hahn et al., Nucl. Inst. and Meth. A 499, 245 (2003) .

[86] K. Adcox et al., Nucl. Inst. and Meth. A 499, 469 (2003) .

[87] M. Allen et al., Nucl. Inst. and Meth. A 499, 549 (2003) .

[88] T. Hachiya et al., PHENIX Technical Note 393 (2002) .

[89] C. Adler et al., Nucl. Inst. and Meth. A 499, 433 (2003) .

[90] C. Adler et al., Nucl. Inst. and Meth. A 470, 488 (2001) .

[91] M. Chiu et al., Phys. Rev. Lett. 89, 012302 (2002) .

[92] S. H. Aronson et al., Nucl. Inst. and Meth. A 499, 480 (2003) .

[93] K. Adcox et al., Nucl. Inst. and Meth. A 499, 489 (2003) .

[94] M. Aizawa et al., Nucl. Inst. and Meth. A 499, 508 (2003) .

[95] M. Aizawa et al., Nucl. Inst. and Meth. A 433, 143 (1999) .

[96] L. Aphecetche et al., Nucl. Inst. and Meth. A 499, 521 (2003) .

[97] M. M. Aggarwal et al., Phys. Rev. Lett. 81, 4087 (1998) [WA98 collaboration].

[98] S. S. Adler et al., Nucl. Inst. and Meth. A 499, 560 (2003)

[99] GEANT 3.2.1, CERN program library.

[100] X. N. Wang and M. Gyulassy, Phys. Rev. D 44 3501 (1991).

BIBLIOGRAPHY 161

[101] T. Hachiya et al., PHENIX Analysis Note 107 (2002).

[102] J. Nagle et al., PHENIX Analysis Note 113 (2002).

[103] H. Appelshauser et al., Eur. Phys. J. A 2, 383 (1998) [NA49 collaboration].

[104] S. Kelly et al., PHENIX Analysis Note 33 (2000)

[105] S. S. Adler et al., Phys. Rev. C 69, 034909 (2004)

[106] K. Reygers, PHENIX Analysis Note 169 (2003)

[107] J. T. Mitchell et al., Nucl. Inst. and Meth. A 482, 491 (2002)

[108] J. Burward-Hoy et al., PHENIX Analysis Note 172 (2003)

[109] T. Chujo et al., PHENIX Analysis Note 187 (2003)

[110] S. S. Adler et al., Phys. Rev. C 69, 034910 (2004)

[111] R. Averbeck for the PHENIX collaboration, Nucl. Phys. A 715, 695c (2003)

[112] A. Lebedev et al., PHENIX Analysis Note 137

[113] Y. Akiba et al., PHENIX Analysis Note 124

[114] K. Adcox et al., Phys. Rev. Lett. 88, 192303 (2002)

[115] D. d’Enterria for the PHENIX collaboration, Nucl. Phys. A 715, 749c (2003)

[116] M. Kaufman and D. d’Enterria, PHENIX Analysis Note 337 (2004)

[117] T. Hachiya and Y. Akiba, PHENIX Analysis Note 340 (2004)

[118] R. Averbeck et al., PHENIX Analysis Note 89 (2001)

[119] R. Seto for the PHENIX collaboration, J. Phys. G 30, S1017 (2004)

[120] S. S. Adler et al., Phys. Rev. Lett. 96, 032001 (2006)

[121] A. Adare et al., Phys. Rev. Lett. 98, 172301 (2007)

[122] Y. Akiba et al., PHENIX Analysis Note 557 (2007)

[123] A. Adare et al., Phys. Rev. Lett. 103, 082002 (2009)

[124] M. Leitch,http://www.int.washington.edu/talks/WorkShops/int 09 42W/People/Leitch M/Leitch.pdf

[125] S. Brodsky,http://indico.cern.ch/getFile.py/access?contribId=6&resId=1&materialId=slides&confId=72423

162 BIBLIOGRAPHY

[126] G. A. Alves et al., Phys. Rev. Lett. 77, 2388 (1996) [E769 Collaboration].

[127] M. Adamovich et al., Nucl. Phys. B 495, 3 (1997) [WA92 Collaboration].

[128] D. Acosta et al., Phys. Rev. Lett. 91, 241804 (2003) [CDF Collaboration].

[129] A. Tai for the STAR Collaboration, J. Phys. G 30, S809 (2004)

[130] M. Mangano et al., Nucl. Phys. B 405, 507 (1993)

[131] Y. Akiba et al., PHENIX Analysis Note 509.

[132] A. Adare et al., Phys. Lett. B 670, 313 (2009) [PHENIX Collaboration]

[133] J. Adams et al., Phys. Rev. Lett. 94, 062301 (2005) [STAR Collaboration]

[134] O. Botner et al., Phys. Lett. B236, 488 (1990) [UA2 Collaboration]

[135] M. Aguilar-Benitez et al., Z. Phys. C 40, 321 (1988) [NA27 Collaboration]

[136] K. Kodama et al., Phys. Lett. B 263, 573 (1991) [E653 Collaboration]

[137] R. Ammar et al., Phys. Rev. Lett. 61, 2185 (1988) [E653 Collaboration]

公表論文

(1) Centrality Dependence of Charm Production from a Measurement of Single Electrons in Au+Au Collisions at = 200 GeV

S. S. Adler et al., (別紙、共著者リスト①参照) Physical Review Letters 94, 082301 (2005).

NNs

PRL 94, 082301 (2005) P H Y S I C A L R E V I E W L E T T E R S week ending4 MARCH 2005

Centrality Dependence of Charm Production from a Measurementof Single Electrons in Au�Au Collisions at

��sNN

p� 200 GeV

S. S. Adler,5 S. Afanasiev,17 C. Aidala,5 N. N. Ajitanand,43 Y. Akiba,20,38 J. Alexander,43 R. Amirikas,12 L. Aphecetche,45

S. H. Aronson,5 R. Averbeck,44 T. C. Awes,35 R. Azmoun,44 V. Babintsev,15 A. Baldisseri,10 K. N. Barish,6 P. D. Barnes,27

B. Bassalleck,33 S. Bathe,30 S. Batsouli,9 V. Baublis,37 A. Bazilevsky,39,15 S. Belikov,16,15 Y. Berdnikov,40

S. Bhagavatula,16 J. G. Boissevain,27 H. Borel,10 S. Borenstein,25 M. L. Brooks,27 D. S. Brown,34 N. Bruner,33 D. Bucher,30

H. Buesching,30 V. Bumazhnov,15 G. Bunce,5,39 J. M. Burward-Hoy,26,44 S. Butsyk,44 X. Camard,45 J.-S. Chai,18 P. Chand,4

W. C. Chang,2 S. Chernichenko,15 C. Y. Chi,9 J. Chiba,20 M. Chiu,9 I. J. Choi,52 J. Choi,19 R. K. Choudhury,4 T. Chujo,5

V. Cianciolo,35 Y. Cobigo,10 B. A. Cole,9 P. Constantin,16 D. G. d’Enterria,45 G. David,5 H. Delagrange,45 A. Denisov,15

A. Deshpande,39 E. J. Desmond,5 A. Devismes,44 O. Dietzsch,41 O. Drapier,25 A. Drees,44 R. du Rietz,29 A. Durum,15

D. Dutta,4 Y. V. Efremenko,35 K. El Chenawi,49 A. Enokizono,14 H. En’yo,38,39 S. Esumi,48 L. Ewell,5 D. E. Fields,33,39

F. Fleuret,25 S. L. Fokin,23 B. D. Fox,39 Z. Fraenkel,51 J. E. Frantz,9 A. Franz,5 A. D. Frawley,12 S.-Y. Fung,6

S. Garpman,29,* T. K. Ghosh,49 A. Glenn,46 G. Gogiberidze,46 M. Gonin,25 J. Gosset,10 Y. Goto,39

R. Granier de Cassagnac,25 N. Grau,16 S. V. Greene,49 M. Grosse Perdekamp,39 W. Guryn,5 H.-A. Gustafsson,29

T. Hachiya,14 J. S. Haggerty,5 H. Hamagaki,8 A. G. Hansen,27 E. P. Hartouni,26 M. Harvey,5 R. Hayano,8 N. Hayashi,38

X. He,13 M. Heffner,26 T. K. Hemmick,44 J. M. Heuser,44 M. Hibino,50 J. C. Hill,16 W. Holzmann,43 K. Homma,14

B. Hong,22 A. Hoover,34 T. Ichihara,38,39 V. V. Ikonnikov,23 K. Imai,24,38 D. Isenhower,1 M. Ishihara,38 M. Issah,43

A. Isupov,17 B. V. Jacak,44 W. Y. Jang,22 Y. Jeong,19 J. Jia,44 O. Jinnouchi,38 B. M. Johnson,5 S. C. Johnson,26 K. S. Joo,31

D. Jouan,36 S. Kametani,8,50 N. Kamihara,47,38 J. H. Kang,52 S. S. Kapoor,4 K. Katou,50 S. Kelly,9 B. Khachaturov,51

A. Khanzadeev,37 J. Kikuchi,50 D. H. Kim,31 D. J. Kim,52 D. W. Kim,19 E. Kim,42 G.-B. Kim,25 H. J. Kim,52 E. Kistenev,5

A. Kiyomichi,48 K. Kiyoyama,32 C. Klein-Boesing,30 H. Kobayashi,38,39 L. Kochenda,37 V. Kochetkov,15 D. Koehler,33

T. Kohama,14 M. Kopytine,44 D. Kotchetkov,6 A. Kozlov,51 P. J. Kroon,5 C. H. Kuberg,1,27 K. Kurita,39 Y. Kuroki,48

M. J. Kweon,22 Y. Kwon,52 G. S. Kyle,34 R. Lacey,43 V. Ladygin,17 J. G. Lajoie,16 A. Lebedev,16,23 S. Leckey,44

D. M. Lee,27 S. Lee,19 M. J. Leitch,27 X. H. Li,6 H. Lim,42 A. Litvinenko,17 M. X. Liu,27 Y. Liu,36 C. F. Maguire,49

Y. I. Makdisi,5 A. Malakhov,17 V. I. Manko,23 Y. Mao,7,38 G. Martinez,45 M. D. Marx,44 H. Masui,48 F. Matathias,44

T. Matsumoto,8,50 P. L. McGaughey,27 E. Melnikov,15 F. Messer,44 Y. Miake,48 J. Milan,43 T. E. Miller,49 A. Milov,44,51

S. Mioduszewski,5 R. E. Mischke,27 G. C. Mishra,13 J. T. Mitchell,5 A. K. Mohanty,4 D. P. Morrison,5 J. M. Moss,27

F. Muhlbacher,44 D. Mukhopadhyay,51 M. Muniruzzaman,6 J. Murata,38,39 S. Nagamiya,20 J. L. Nagle,9 T. Nakamura,14

B. K. Nandi,6 M. Nara,48 J. Newby,46 P. Nilsson,29 A. S. Nyanin,23 J. Nystrand,29 E. O’Brien,5 C. A. Ogilvie,16

H. Ohnishi,5,38 I. D. Ojha,49,3 K. Okada,38 M. Ono,48 V. Onuchin,15 A. Oskarsson,29 I. Otterlund,29 K. Oyama,8 K. Ozawa,8

D. Pal,51 A. P. T. Palounek,27 V. S. Pantuev,44 V. Papavassiliou,34 J. Park,42 A. Parmar,33 S. F. Pate,34 T. Peitzmann,30

J.-C. Peng,27 V. Peresedov,17 C. Pinkenburg,5 R. P. Pisani,5 F. Plasil,35 M. L. Purschke,5 A. K. Purwar,44 J. Rak,16

I. Ravinovich,51 K. F. Read,35,46 M. Reuter,44 K. Reygers,30 V. Riabov,37,40 Y. Riabov,37 G. Roche,28 A. Romana,25

M. Rosati,16 P. Rosnet,28 S. S. Ryu,52 M. E. Sadler,1 N. Saito,38,39 T. Sakaguchi,8,50 M. Sakai,32 S. Sakai,48 V. Samsonov,37

L. Sanfratello,33 R. Santo,30 H. D. Sato,24,38 S. Sato,5,48 S. Sawada,20 Y. Schutz,45 V. Semenov,15 R. Seto,6 M. R. Shaw,1,27

T. K. Shea,5 T.-A. Shibata,47,38 K. Shigaki,14,20 T. Shiina,27 C. L. Silva,41 D. Silvermyr,27,29 K. S. Sim,22 C. P. Singh,3

V. Singh,3 M. Sivertz,5 A. Soldatov,15 R. A. Soltz,26 W. E. Sondheim,27 S. P. Sorensen,46 I. V. Sourikova,5 F. Staley,10

P. W. Stankus,35 E. Stenlund,29 M. Stepanov,34 A. Ster,21 S. P. Stoll,5 T. Sugitate,14 J. P. Sullivan,27 E. M. Takagui,41

A. Taketani,38,39 M. Tamai,50 K. H. Tanaka,20 Y. Tanaka,32 K. Tanida,38 M. J. Tannenbaum,5 P. Tarjan,11 J. D. Tepe,1,27

T. L. Thomas,33 J. Tojo,24,38 H. Torii,24,38 R. S. Towell,1 I. Tserruya,51 H. Tsuruoka,48 S. K. Tuli,3 H. Tydesjo,29

N. Tyurin,15 H. W. van Hecke,27 J. Velkovska,5,44 M. Velkovsky,44 V. Veszpremi,11 L. Villatte,46 A. A. Vinogradov,23

M. A. Volkov,23 E. Vznuzdaev,37 X. R. Wang,13 Y. Watanabe,38,39 S. N. White,5 F. K. Wohn,16 C. L. Woody,5 W. Xie,6

Y. Yang,7 A. Yanovich,15 S. Yokkaichi,38,39 G. R. Young,35 I. E. Yushmanov,23 W. A. Zajc,9,† C. Zhang,9 S. Zhou,7

S. J. Zhou,51 and L. Zolin17

(PHENIX Collaboration)

1Abilene Christian University, Abilene, Texas 79699, USA2Institute of Physics, Academia Sinica, Taipei 11529, Taiwan

3Department of Physics, Banaras Hindu University, Varanasi 221005, India

0031-9007=05=94(8)=082301(6)$23.00 082301-1 2005 The American Physical Society


4Bhabha Atomic Research Centre, Bombay 400 085, India5Brookhaven National Laboratory, Upton, New York 11973-5000, USA

6University of California at Riverside, Riverside, California 92521, USA7China Institute of Atomic Energy (CIAE), Beijing, People’s Republic of China

8Center for Nuclear Study, Graduate School of Science, University of Tokyo, 7-3-1 Hongo, Bunkyo, Tokyo 113-0033, Japan9Columbia University, New York, New York 10027, USAand Nevis Laboratories, Irvington, New York 10533, USA10Dapnia, CEA Saclay, F-91191, Gif-sur-Yvette, France

11Debrecen University, H-4010 Debrecen, Egyetem ter 1, Hungary12Florida State University, Tallahassee, Florida 32306, USA

13Georgia State University, Atlanta, Georgia 30303, USA14Hiroshima University, Kagamiyama, Higashi-Hiroshima 739-8526, Japan

15Institute for High Energy Physics (IHEP), Protvino, Russia16Iowa State University, Ames, Iowa 50011, USA

17Joint Institute for Nuclear Research, 141980 Dubna, Moscow Region, Russia18KAERI, Cyclotron Application Laboratory, Seoul, Korea

19Kangnung National University, Kangnung 210-702, Korea20KEK, High Energy Accelerator Research Organization, Tsukuba-shi, Ibaraki-ken 305-0801, Japan

21KFKI Research Institute for Particle and Nuclear Physics (RMKI), H-1525 Budapest 114, P.O. Box 49, Hungary22Korea University, Seoul, 136-701, Korea

23Russian Research Center ‘‘Kurchatov Institute,’’ Moscow, Russia24Kyoto University, Kyoto 606, Japan

25Laboratoire Leprince-Ringuet, Ecole Polytechnique, CNRS-IN2P3, Route de Saclay, F-91128, Palaiseau, France26Lawrence Livermore National Laboratory, Livermore, California 94550, USA

27Los Alamos National Laboratory, Los Alamos, New Mexico 87545, USA28LPC, Universite Blaise Pascal, CNRS-IN2P3, Clermont-Fd, 63177 Aubiere CEDEX, France

29Department of Physics, Lund University, Box 118, SE-221 00 Lund, Sweden30Institut fur Kernphysik, University of Muenster, D-48149 Muenster, Germany

31Myongji University, Yongin, Kyonggido 449-728, Korea32Nagasaki Institute of Applied Science, Nagasaki-shi, Nagasaki 851-0193, Japan

33University of New Mexico, Albuquerque, New Mexico 87131, USA34New Mexico State University, Las Cruces, New Mexico 88003, USA35Oak Ridge National Laboratory, Oak Ridge, Tennessee 37831, USA

36IPN-Orsay, Universite Paris Sud, CNRS-IN2P3, BP 1, F-91406, Orsay, France37PNPI, Petersburg Nuclear Physics Institute, Gatchina, Russia

38RIKEN (The Institute of Physical and Chemical Research), Wako, Saitama 351-0198, Japan39RIKEN BNL Research Center, Brookhaven National Laboratory, Upton, New York 11973-5000, USA

40St. Petersburg State Technical University, St. Petersburg, Russia41Universidade de Sao Paulo, Instituto de Fısica, Caixa Postal 66318, Sao Paulo CEP05315-970, Brazil

42System Electronics Laboratory, Seoul National University, Seoul, Korea43Chemistry Department, Stony Brook University, SUNY, Stony Brook, New York 11794-3400, USA

44Department of Physics and Astronomy, Stony Brook University, SUNY, Stony Brook, New York 11794, USA45SUBATECH (Ecole des Mines de Nantes, CNRS-IN2P3, Universite de Nantes) BP 20722-44307, Nantes, France

46University of Tennessee, Knoxville, Tennessee 37996, USA47Department of Physics, Tokyo Institute of Technology, Tokyo 152-8551, Japan

48Institute of Physics, University of Tsukuba, Tsukuba, Ibaraki 305, Japan49Vanderbilt University, Nashville, Tennessee 37235, USA

50Waseda University, Advanced Research Institute for Science and Engineering, 17 Kikui-cho, Shinjuku-ku, Tokyo 162-0044, Japan51Weizmann Institute, Rehovot 76100, Israel

52Yonsei University, IPAP, Seoul 120-749, Korea(Received 28 September 2004; published 2 March 2005)

The PHENIX experiment has measured midrapidity transverse momentum spectra (0:4< pT <4:0 GeV=c) of single electrons as a function of centrality in Au� Au collisions at

��sNN

p� 200 GeV.

Contributions from photon conversions and Dalitz decays of light neutral mesons are measured byintroducing a thin (1.7% X0) converter into the PHENIX acceptance and are statistically removed. Thesubtracted nonphotonic electron spectra are primarily due to the semileptonic decays of hadronscontaining heavy quarks, mainly charm at lower pT . For all centralities, the charm production crosssection is found to scale with the nuclear overlap function, TAA. For minimum-bias collisions the charmcross section per binary collision is Ncc=TAA � 622� 57�stat� � 160�syst� �b.

082301-2


DOI: 10.1103/PhysRevLett.94.082301 PACS numbers: 25.75.Dw

In central Au� Au collisions at��sNN

p� 200 GeV neu-

tral pions and charged hadrons are strongly suppressed athigh transverse momentum (pT) [1–3]. In contrast, a mod-est high-pT enhancement is observed in d� Au collisionsat the same energy [4,5]. Taken together, these observa-tions indicate that the suppression in Au� Au collisions iscaused by final-state effects (e.g., parton energy loss in adense medium produced in the reaction).

Heavy quarks (charm and bottom) are complementaryprobes of the hot and dense matter produced in high energyheavy ion collisions. Because of their large masses, charmand bottom cross sections are calculable via perturbativequantum chromodynamics (pQCD) and their yield is sen-sitive to the initial gluon density [6]. It has been predictedthat heavy quarks suffer less energy loss than light quarkswhile traversing partonic matter due to the ‘‘dead cone’’effect [7–9]. This can be studied through systematic mea-surements of the pT spectra of open heavy flavor. Inaddition, the open-charm yield is an important baselinefor understanding J= production which has been pre-dicted to be either suppressed [10] or enhanced [11] indeconfined partonic matter.

The PHENIX experiment observed that inclusive singleelectrons in central and minimum-bias Au� Au collisionsat

��sNN

p� 130 GeV were produced in excess of purely

‘‘photonic’’ contributions (primarily due to �0 Dalitz de-cays and conversion of�0 photons in the detector material)[12]. This excess is consistent with the expected charmproduction, assuming that it scales with the number ofbinary nucleon-nucleon collisions (Ncoll) or, equivalently,with the nuclear overlap function, TAA. In this Letter, wepresent results on the single electron measurement in��sNN

p� 200 GeV Au� Au collisions. Our measurement

is in a pT range sensitive only to charm production. Thenew data have higher statistics and smaller systematicerrors than the 130 GeV data, allowing us to measurecharm production as a function of collision centrality.

The data used in this analysis were collected by thePHENIX detector [13] during the 2001 run period of theRelativistic Heavy Ion Collider. A coincidence of thebeam-beam counters (BBC), a pair of detector arrayscovering 2� in azimuth and � � ��3:0–3:9�, and thezero degree calorimeters (ZDC) provides the minimum-bias trigger (92:2�2:5

�3:0% of the 6:8� 0:5 barn Au� Auinelastic cross section). The centrality is determined bythe correlation between the multiplicity measured by theBBC and the energy of spectator neutrons measured by theZDC. The BBC also measures the collision vertex, z, withresolution � � 0:7 cm. We require jzj< 20 cm to elimi-nate electrons originating from the central magnet.

Charged particles are measured by the PHENIX east-arm spectrometer (j�j< 0:35, �� =2) with resolution�p=p ’ 0:7% 1:0%p�GeV=c�. Tracks are reconstructed

08230

with the drift chamber and the first layer of pad chambersand confirmed by requiring an electromagnetic calorimeter(EMCal) matching hit within 2� in position. Electroncandidates are required to have at least three associatedhits in the ring imaging Cerenkov detector (RICH) thatpass a ring shape cut, and are required to pass a timing cutin either the EMCal or the time-of-flight detector. Afterthese cuts, a clear electron signal is observed as a narrowpeak at E=p � 1. By requiring �2�< �E� p�=p < 3�,background from hadrons, which deposit only a fraction oftheir energy in the EMCal, and nonvertex electrons, whichhave misreconstructed momenta, is further reduced.Remaining background in the electron sample, due toaccidental coincidences between RICH hits and hadrontracks, is estimated ( � 10%) and subtracted by an event-mixing method.

Inclusive electrons contain two components: (i) ‘‘non-photonic’’—primarily semileptonic decays of mesonscontaining heavy quarks, and (ii) photonic—Dalitz decaysof light neutral mesons (�0,�,�0, �,!, and�) and photonconversions in the detector material. To separate these twocomponents, a photon converter (a thin brass tube of 1.7%radiation length surrounding the beam pipe at r � 29 cm)was installed.

We analyzed 2.2 M (2.5 M) events with the converter in(out). The corresponding raw electron pT spectra forminimum-bias collisions are shown in Fig. 1(a). The pho-ton converter multiplies the photonic contribution to theelectron yield by a factor R�:

NConv-oute � N�

e � Nnon-�e ; (1)

NConv-ine � R�N

�e � �1� ��Nnon-�

e : (2)

Here NConv-ine (NConv-out

e ) is the measured electron yieldwith (without) the converter; N�

e (Nnon-�e ) is the electron

yield due to the photonic (nonphotonic) component; and�� 2:1%� represents a small loss of electrons due to theconverter. We next define RCN as the ratio of the rawelectron yield with and without the converter. DividingEq. (2) by Eq. (1) and defining RNP Nnon-�

e =N�e , one has

RCN NConv-ine

NConv-oute

�R� � �1� ��RNP

1� RNP: (3)

If there were no contribution from the nonphotonic com-ponent (RNP � 0), then RCN � R�.

The photonic electron yield per photon is approximatelygiven by Y / � 7

9 t, where is the Dalitz branching ratioper � relative to 2� (for �0, �, and �0) or 1� (for �,!, and�) decay, and t is the thickness of the conversion materialin radiation length (X0). The factor 7

9 is the approximateprobability for a � to convert in one X0. Plugging in �

0�

0:6%, t � 1:1% (t � 2:8%) for converter out (in) we findR�

0

� � YConv-in=YConv-out � 1:9. There is some pT depen-

1-3

(ar

b. u

nit

)

eN

10-6

10-5

10-4

10-3

10-2 - + e

+Raw yield of incl. e

Conv-outeN

Conv-ineN

(a)

γ o

r R

CN

R

1

1.2

1.4

1.6

1.8

2

2.2

with error bandCNR

with error bandγR

(b)

(GeV/c)Tp0 0.5 1 1.5 2 2.5 3 3.5 4

NP

R

-0.5

0

0.5

1

1.5 Systematic uncertainty e+X→K

(c)

FIG. 1. Shown vs pT (a) raw e� spectra measured with theconverter in (open circles) and out (closed circles). (b) Ratio ofthe converter in/out e� yields (RCN , points) and ratio of photonice� yield with/without the converter (R�, line and shaded band).(c) Ratio of nonphotonic to photonic e� yields (RNP, points) andcontribution from kaon decays (dashed line).


dence in the complete formula for Y and the value of isspecies dependent ( � � 0:8%), so we perform a fullGEANT [14] simulation with and without the converter tocalculate R�. We determine R� for�0 and� separately. Weuse the �0 spectrum measured by PHENIX [1] as the inputfor the �0 simulation and assume mT scaling (pT!��p2T�M

2��M

2�

q, normalized at high pT to �=�0�

0:45�0:1) to obtain the input for the � simulation.Contributions from other mesons which undergo Dalitzdecay (�0; �; !;�) are small (6% at pT � 3 GeV=c, andsmaller at lower pT). Since they have � �, we assignthem R� � R�� . When calculating the combined R�, weuse the particle ratios at high pT (�0=�0�0:25�0:13,�=�0 � !=�0 � 1� 0:5, �=�0�0:4�0:2). The �=�0

ratio used here is consistent with our �0 and � measure-ment [15]. The uncertainties in the particle ratios areincluded in the systematic uncertainties of R�. For thismethod it is essential that the amount of material is accu-rately modeled in the simulation. We compared the yield ofidentified photon conversion pairs in the data and in thesimulation and conclude that the simulation reproduces R�within �2:0%. This uncertainty is included in the overallsystematic uncertainty.

Figure 1(b) shows that RCN gradually decreases withincreasing pT , while R� slightly increases with pT . Thedifference between RCN and R� indicates the existence of

08230

nonphotonic electrons. Figure 1(c) shows RNP obtainedfrom R� and RCN using Eq. (3). RNP increases with pTand is more than 30% for pT > 0:6 GeV=c. The smallamount of conversion material in the PHENIX detectorallows a sensitive measurement of RCN .

Background from kaon decays (K ! �e&) and dielec-tron decays of �, !, and � remain in the nonphotonicelectron yield. The background from kaon decays isestimated with a GEANT simulation using the kaon pTspectrum measured by PHENIX [16] as input. The contri-bution of kaon decays to the nonphotonic yield, shown inFig. 1(c), is 18% at pT � 0:4 GeV=c and decreases rapidlyto less than 6% for pT > 1 GeV=c. To calculate back-ground from the e�e� decays of �, !, and �, we firstgenerate spectra by applyingmT scaling to the PHENIX�0

spectrum, as described above. The contribution of thesedecays to the nonphotonic electrons is <3% for all pT .Background from J= ! e�e� decays and from Drell-Yan pairs is negligible. Possible enhancement of low massdileptons through �� ! �! e�e�, as reported inPb� Pb collisions at the Super Proton Synchrotron [17],would contribute to the nonphotonic electrons. However,this is neglected since the estimated � contribution in theabsence of enhancement is only � 0:6% over all pT .

After these backgrounds are subtracted the only othersignificant source of nonphotonic electrons is the semi-leptonic decay of heavy flavor, overwhelmingly charm. Wedenote the remaining electrons as charm electrons. The rawspectrum of charm electrons is corrected for geometricalacceptance (�geo), track reconstruction efficiency (�rec),and electron identification efficiency (�eID) determinedby GEANT simulation. The efficiency �geo � �rec is about11% of dNe=dy, and �eID is about 65% as confirmed withelectrons identified through photon conversion. Correctionof multiplicity dependent efficiency losses, estimated byembedding simulated electron tracks into real events, is pTindependent and increases from 5% to 26% from periph-eral to central collisions. The 1� systematic uncertainty ofthese corrections is 11.8%. Fully corrected charm electronspectra are shown in Fig. 2 for minimum-bias collisionsand for five centrality bins. We determined photonic elec-tron spectra and found them to be in good agreement withthe background calculated from the measured �0 spectrausing the method described in [12].

We also measured the charm electron spectrum in p� pcollisions at

��sNN

p� 200 GeV [18]. The lines in Fig. 2

show the best fit curve of this spectrum, scaled by TAA foreach Au� Au centrality bin. Here, TAA is the nuclearoverlap function calculated by a Glauber model [1](Table I). The Au� Au data points are in reasonableagreement with the p� p fit in all centrality bins.

To quantify the centrality dependence of charm produc-tion, we calculated the integrated yield dNe=dy (0:8<pT < 4:0 GeV=c) and fit it to AN(

coll, where A is constant.In the absence of medium effects ( � 1 is the expectation

1-4

collN0 200 400 600 800 1000 1200

Co

ll/d

y

/ N

ed

N

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

-4x10

(m

b)

AA

/dy

/ Te

dN

0

0.05

0.1

0.15

0.2

0.25

0.3

0.35-2x10

centrality binned

min-bias

= 200 GeVsp + p at

<4.0

)T

(0.8

<p

<4.0

)T

(0.8

<p

FIG. 3. Charm electron yield (0:8< pT < 4:0 GeV=c) mea-sured in Au� Au collisions at 200 GeV scaled by Ncoll as afunction of Ncoll (left-hand scale). Normalizing instead by thenuclear overlap function we obtain charm electron cross sectionper N � N collision (right-hand scale).

(GeV/c)Tp0 0.5 1 1.5 2 2.5 3 3.5 4

]2d

y [(

c/G

eV)

T)d

N/d

pT

pπ(1

/2

10-8

10-7

10-6

10-5

10-4

10-3

10-2

10-1

1

10

102

103

=200GeVNNs)/2 + X @ -+e+

(e→Au+Au

min. bias3 10× 0% - 10% 2 10× 10% - 20% 1 10× 20% - 40% -1 10× 40% - 60% -2 10× 60% - 92%

Best fit curve of pp

FIG. 2. Fully corrected charm electron pT spectra for differentAu� Au centralities scaled by successive factors of 10 forclarity. Error bars (brackets) correspond to statistical (system-atic) uncertainties. Curves are described in the text.


in the absence of medium effects. In this comparison,most of the systematic effects will cancel. Figure 3shows dNe=dy�0:8< pT < 4:0�=Ncoll vs Ncoll for mini-mum bias and five centrality bins in Au� Au collisionsand p� p collisions. We find ( � 0:938� 0:075�stat� �0:018�syst�. If p� p data are included, ( � 0:958�0:035�stat�. This shows that the total yield of charm elec-trons for all centralities is consistent with Ncoll scaling.

For each centrality bin we scale the charm electron spec-trum (pT > 0:8 GeV=c) by TAA and fit it with a PYTHIAcalculation of the electron spectrum resulting from leadingorder charm and bottom production. We used PYTHIA6.205 with a modified set of parameters (described in[12]) and CTEQ5L parton distribution functions [19].Based on experimental input [20,21] we modified thePYTHIA default charm ratios, using instead D�=D0�0:45�0:1, Ds=D0�0:25�0:1, and $c=D0 � 0:1� 0:05.

TABLE I. Centrality bin, number of NN collisions, nuclear overlapmultiplicity per NN collision, in

��sNN

p� 200 GeV Au� Au reacti

Centrality (%) Ncoll TAA (mb�1)

Minimum bias 258� 25 6:14� 0:450–10 955� 94 22:8� 1:6

10–20 603� 59 14:4� 1:020–40 297� 31 7:07� 0:5840–60 91� 12 2:16� 0:2660–92 14:5� 4:0 0:35� 0:10

08230

This gives a c! e total branching ratio of 9:5� 0:4%.Charmed hadron ratios from the statistical model [22]gives a similar branching ratio (9.3%). The scaled charmand bottom cross sections are treated as fit parameters,although our data are restricted to pT values which aresensitive only to charm production. We evaluated the sys-tematic error due to background subtraction (�21%) byrefitting to the electron spectrum at the minimum andmaximum of its 1� systematic error band. The change ofthe pT range for fitting the charm electron spectrum gives3% systematic error for minimum-bias collisions. The sys-tematic error due to the PYTHIA spectral shape (�11%) isdominated by the uncertainty in hkTi�1:5�0:5GeV=c.Different parton distribution functions yield a systematicerror of 6.2% for the rapidity-integrated cross section.These systematic errors are added in quadrature to givethe overall systematic error on the charm cross section. Forminimum-bias collisions we obtain 1

TAA

dNccdy jy�0�143�

13�stat��36�syst��b and Ncc=TAA�622�57�stat��160�syst��b. Results for all centrality bins are shown inTable I. The STAR collaboration reports a somewhat largercharm cross section (�NNcc �1:3�0:2�0:4mb) in p�pand d�Au collision at

��sNN

p� 200 GeV [23]. Next-to

leading order pQCD calculations of the charm productioncross section have large associated uncertainties, withtypical values between 300 and 450 �b [24].

function, charm cross section per NN collision, and total charmons.

1TAA

dNccdy jy�0 (�b) Ncc=TAA (�b)

143� 13� 36 622� 57� 160137� 21� 35 597� 93� 156137� 26� 35 596� 115� 158168� 27� 45 731� 117� 199193� 47� 52 841� 205� 232116� 87� 43 504� 378� 190

1-5


We note that final-state effects influence only the mo-mentum distribution of charm; they have little or no effecton the total open-charm yield. Therefore, our results in-dicate Ncoll scaling of the initial charm production, asexpected for pointlike pQCD processes. pQCD calcula-tions without charm quark energy loss and hydrodynamiccalculations assuming complete thermalization of charmquarks predict very similar charm electron spectra forpT < 2 GeV=c [25]. Differentiating between these oppo-site physical pictures is possible only for pT > 2:5 GeV=c,where statistics of the current analysis are limited.

In conclusion, we have measured single electrons fromcharm decays in Au� Au collisions at

��sNN

p� 200 GeV.

We observe that the centrality dependence of charm quarkproduction is consistent with Ncoll scaling, as expected forhard processes. The much larger Au� Au data set col-lected by PHENIX in the 2003-04 run will allow us moredetailed exploration of medium effects on charm produc-tion, both through deviations of the charm electron spec-trum fromNcoll scaling, and also through a measurement ofcharm quark flow.

We thank the staff of the Collider-Accelerator andPhysics Departments at BNL for their vital contributions.We acknowledge support from the Department of Energyand NSF (USA), MEXT and JSPS (Japan), CNPq andFAPESP (Brazil), NSFC (China), CNRS-IN2P3 and CEA(France), BMBF, DAAD, and AvH (Germany), OTKA(Hungary), DAE and DST (India), ISF (Israel), KRFand CHEP (Korea), RMIST, RAS, and RMAE (Russia),VR and KAW (Sweden), U.S. CRDF for the FSU,U.S.-Hungarian NSF-OTKA-MTA, and U.S.-Israel BSF.

08230

*Deceased.†PHENIX Spokesperson.Electronic address: [email protected]

[1] S. S. Adler et al., Phys. Rev. Lett. 91, 072301 (2003).[2] J. Adams et al., Phys. Rev. Lett. 91, 172302 (2003).[3] S. S. Adler et al., Phys. Rev. C 69, 034910 (2004).[4] S. S. Adler et al., Phys. Rev. Lett. 91, 072303 (2003).[5] J. Adams et al., Phys. Rev. Lett. 91, 072304 (2003).[6] J. A. Appel, Annu. Rev. Nucl. Part. Sci. 42, 367 (1992).[7] Y. L. Dokshitzer and D. E. Kharzeev, Phys. Lett. B 519,

199 (2001).[8] M. Djordjevic and M. Gyulassy, Phys. Lett. B 560, 37

(2003).[9] B. W. Zhang, E. Wang, and X.-N. Wang, Phys. Rev. Lett.

93, 072301 (2004).[10] T. Matsui and H. Satz, Phys. Lett. B 178, 416 (1986).[11] R. L. Thews, M. Schroedter, and J. Rafelski, Phys. Rev. C

63, 054905 (2001).[12] K. Adcox et al., Phys. Rev. Lett. 88, 192303 (2002).[13] K. Adcox et al., Nucl. Instrum. Methods Phys. Res.,

Sect. A 499, 469 (2003).[14] GEANT 3.2.1, CERN program library.[15] R. Seto et al., J. Phys. G 30, S1017 (2004).[16] S. S. Adler et al., Phys. Rev. C 69, 034909 (2004).[17] G. Agakichiev et al., Phys. Lett. B 422, 405 (1998).[18] S. Kelly et al., J. Phys. G 30, S1189 (2004).[19] H. L. Lai et al., Eur. Phys. J. C 12, 375 (2000).[20] D. Acosta et al., Phys. Rev. Lett. 91, 241804 (2003).[21] K. Hagiwara et al., Phys. Rev. D 66, 010001 (2002).[22] A. Andronic et al., Phys. Lett. B 571, 36 (2003).[23] J. Adams et al., nucl-ex/0407006.[24] R. Vogt, hep-ph/0203151.[25] S. Batsouli et al., Phys. Lett. B 557, 26 (2003).

1-6

参考論文

(1) Measurement of Single Electrons and Implications for Charm Production in Au+Au Collisions at = 130 GeV

K. Adcox et al., (別紙、共著者リスト②参照) Physical Review Letters 88, 192303 (2002).

(2) Measurement of Bottom Versus Charm as a Function of Transverse Momentum

with Electron-Hadron Correlations in p + p Collisions at = 200 GeV A. Adare et al., (別紙、共著者リスト③参照) Physical Review Letters 103, 082002 (2009).

(3) PHENIX inner detectors

M. Allen et al., (別紙, 共著者リスト④参照) Nuclear Instruments and Methods in Physics Research A 499, 549 (2003).

NNs

s

VOLUME 88, NUMBER 19 P H Y S I C A L R E V I E W L E T T E R S 13 MAY 2002

Measurement of Single Electrons and Implications for Charm Productionin Au 1 Au Collisions at

psNN 5 130 GeV

K. Adcox,40 S. S. Adler,3 N. N. Ajitanand,27 Y. Akiba,14 J. Alexander,27 L. Aphecetche,34 Y. Arai,14 S. H. Aronson,3

R. Averbeck,28 T. C. Awes,29 K. N. Barish,5 P. D. Barnes,19 J. Barrette,21 B. Bassalleck,25 S. Bathe,22 V. Baublis,30

A. Bazilevsky,12,32 S. Belikov,12,13 F. G. Bellaiche,29 S. T. Belyaev,16 M. J. Bennett,19 Y. Berdnikov,35 S. Botelho,33

M. L. Brooks,19 D. S. Brown,26 N. Bruner,25 D. Bucher,22 H. Buesching,22 V. Bumazhnov,12 G. Bunce,3,32

J. Burward-Hoy,28 S. Butsyk,28,30 T. A. Carey,19 P. Chand,2 J. Chang,5 W. C. Chang,1 L. L. Chavez,25

S. Chernichenko,12 C. Y. Chi,8 J. Chiba,14 M. Chiu,8 R. K. Choudhury,2 T. Christ,28 T. Chujo,3,39 M. S. Chung,15,19

P. Chung,27 V. Cianciolo,29 B. A. Cole,8 D. G. D’Enterria,34 G. David,3 H. Delagrange,34 A. Denisov,12 A. Deshpande,32

E. J. Desmond,3 O. Dietzsch,33 B. V. Dinesh,2 A. Drees,28 A. Durum,12 D. Dutta,2 K. Ebisu,24 Y. V. Efremenko,29

K. El Chenawi,40 H. En’yo,17,31 S. Esumi,39 L. Ewell,3 T. Ferdousi,5 D. E. Fields,25 S. L. Fokin,16 Z. Fraenkel,42

A. Franz,3 A. D. Frawley,9 S.-Y. Fung,5 S. Garpman,20,* T. K. Ghosh,40 A. Glenn,36 A. L. Godoi,33 Y. Goto,32

S. V. Greene,40 M. Grosse Perdekamp,32 S. K. Gupta,2 W. Guryn,3 H.-Å. Gustafsson,20 T. Hachiya,11 J. S. Haggerty,3

H. Hamagaki,7 A. G. Hansen,19 H. Hara,24 E. P. Hartouni,18 R. Hayano,38 N. Hayashi,31 X. He,10 T. K. Hemmick,28

J. M. Heuser,28 M. Hibino,41 J. C. Hill,13 D. S. Ho,43 K. Homma,11 B. Hong,15 A. Hoover,26 T. Ichihara,31,32

K. Imai,17,31 M. S. Ippolitov,16 M. Ishihara,31,32 B. V. Jacak,28,32 W. Y. Jang,15 J. Jia,28 B. M. Johnson,3

S. C. Johnson,18,28 K. S. Joo,23 S. Kametani,41 J. H. Kang,43 M. Kann,30 S. S. Kapoor,2 S. Kelly,8 B. Khachaturov,42

A. Khanzadeev,30 J. Kikuchi,41 D. J. Kim,43 H. J. Kim,43 S. Y. Kim,43 Y. G. Kim,43 W. W. Kinnison,19 E. Kistenev,3

A. Kiyomichi,39 C. Klein-Boesing,22 S. Klinksiek,25 L. Kochenda,30 V. Kochetkov,12 D. Koehler,25 T. Kohama,11

D. Kotchetkov,5 A. Kozlov,42 P. J. Kroon,3 K. Kurita,31,32 M. J. Kweon,15 Y. Kwon,43 G. S. Kyle,26 R. Lacey,27

J. G. Lajoie,13 J. Lauret,27 A. Lebedev,13,16 D. M. Lee,19 M. J. Leitch,19 X. H. Li,5 Z. Li,6,31 D. J. Lim,43 M. X. Liu,19

X. Liu,6 Z. Liu,6 C. F. Maguire,40 J. Mahon,3 Y. I. Makdisi,3 V. I. Manko,16 Y. Mao,6,31 S. K. Mark,21 S. Markacs,8

G. Martinez,34 M. D. Marx,28 A. Masaike,17 F. Matathias,28 T. Matsumoto,7,41 P. L. McGaughey,19 E. Melnikov,12

M. Merschmeyer,22 F. Messer,28 M. Messer,3 Y. Miake,39 T. E. Miller,40 A. Milov,42 S. Mioduszewski,3,36

R. E. Mischke,19 G. C. Mishra,10 J. T. Mitchell,3 A. K. Mohanty,2 D. P. Morrison,3 J. M. Moss,19 F. Mühlbacher,28

M. Muniruzzaman,5 J. Murata,31 S. Nagamiya,14 Y. Nagasaka,24 J. L. Nagle,8 Y. Nakada,17 B. K. Nandi,5 J. Newby,36

L. Nikkinen,21 P. Nilsson,20 S. Nishimura,7 A. S. Nyanin,16 J. Nystrand,20 E. O’Brien,3 C. A. Ogilvie,13 H. Ohnishi,3,11

I. D. Ojha,4,40 M. Ono,39 V. Onuchin,12 A. Oskarsson,20 L. Österman,20 I. Otterlund,20 K. Oyama,7,38 L. Paffrath,3,*A. P. T. Palounek,19 V. S. Pantuev,28 V. Papavassiliou,26 S. F. Pate,26 T. Peitzmann,22 A. N. Petridis,13 C. Pinkenburg,3,27

R. P. Pisani,3 P. Pitukhin,12 F. Plasil,29 M. Pollack,28,36 K. Pope,36 M. L. Purschke,3 I. Ravinovich,42 K. F. Read,29,36

K. Reygers,22 V. Riabov,30,35 Y. Riabov,30 M. Rosati,13 A. A. Rose,40 S. S. Ryu,43 N. Saito,31,32 A. Sakaguchi,11

T. Sakaguchi,7,41 H. Sako,39 T. Sakuma,31,37 V. Samsonov,30 T. C. Sangster,18 R. Santo,22 H. D. Sato,17,31 S. Sato,39

S. Sawada,14 B. R. Schlei,19 Y. Schutz,34 V. Semenov,12 R. Seto,5 T. K. Shea,3 I. Shein,12 T.-A. Shibata,31,37

K. Shigaki,14 T. Shiina,19 Y. H. Shin,43 I. G. Sibiriak,16 D. Silvermyr,20 K. S. Sim,15 J. Simon-Gillo,19 C. P. Singh,4

V. Singh,4 M. Sivertz,3 A. Soldatov,12 R. A. Soltz,18 S. Sorensen,29,36 P. W. Stankus,29 N. Starinsky,21 P. Steinberg,8

E. Stenlund,20 A. Ster,44 S. P. Stoll,3 M. Sugioka,31,37 T. Sugitate,11 J. P. Sullivan,19 Y. Sumi,11 Z. Sun,6 M. Suzuki,39

E. M. Takagui,33 A. Taketani,31 M. Tamai,41 K. H. Tanaka,14 Y. Tanaka,24 E. Taniguchi,31,37 M. J. Tannenbaum,3

J. Thomas,28 J. H. Thomas,18 T. L. Thomas,25 W. Tian,6,36 J. Tojo,17,31 H. Torii,17,31 R. S. Towell,19 I. Tserruya,42

H. Tsuruoka,39 A. A. Tsvetkov,16 S. K. Tuli,4 H. Tydesjö,20 N. Tyurin,12 T. Ushiroda,24 H. W. van Hecke,19

C. Velissaris,26 J. Velkovska,28 M. Velkovsky,28 A. A. Vinogradov,16 M. A. Volkov,16 A. Vorobyov,30 E. Vznuzdaev,30

H. Wang,5 Y. Watanabe,31,32 S. N. White,3 C. Witzig,3 F. K. Wohn,13 C. L. Woody,3 W. Xie,5,42 K. Yagi,39

S. Yokkaichi,31 G. R. Young,29 I. E. Yushmanov,16 W. A. Zajc,8 Z. Zhang,28 and S. Zhou6

(PHENIX Collaboration)1Institute of Physics, Academia Sinica, Taipei 11529, Taiwan

2Bhabha Atomic Research Centre, Bombay 400 085, India3Brookhaven National Laboratory, Upton, New York 11973-5000

4Department of Physics, Banaras Hindu University, Varanasi 221005, India5University of California–Riverside, Riverside, California 92521

6China Institute of Atomic Energy (CIAE), Beijing, People’s Republic of China7Center for Nuclear Study, Graduate School of Science, University of Tokyo, 7-3-1 Hongo, Bunkyo, Tokyo 113-0033, Japan

192303-1 0031-9007�02�88(19)�192303(6)$20.00 © 2002 The American Physical Society 192303-1


8Columbia University, New York, New York 10027and Nevis Laboratories, Irvington, New York 10533

9Florida State University, Tallahassee, Florida 3230610Georgia State University, Atlanta, Georgia 30303

11Hiroshima University, Kagamiyama, Higashi-Hiroshima 739-8526, Japan12Institute for High Energy Physics (IHEP), Protvino, Russia

13Iowa State University, Ames, Iowa 5001114KEK, High Energy Accelerator Research Organization, Tsukuba-shi, Ibaraki-ken 305-0801, Japan

15Korea University, Seoul, 136-701, Korea16Russian Research Center “Kurchatov Institute,” Moscow, Russia

17Kyoto University, Kyoto 606, Japan18Lawrence Livermore National Laboratory, Livermore, California 94550

19Los Alamos National Laboratory, Los Alamos, New Mexico 8754520Department of Physics, Lund University, Box 118, SE-221 00 Lund, Sweden

21McGill University, Montreal, Quebec, Canada H3A 2T822Institut für Kernphysik, University of Münster, D-48149 Münster, Germany


25University of New Mexico, Albuquerque, New Mexico 8713126New Mexico State University, Las Cruces, New Mexico 88003

27Chemistry Department, State University of New York–Stony Brook, Stony Brook, New York 1179428Department of Physics and Astronomy, State University of New York–Stony Brook, Stony Brook, New York 11794

29Oak Ridge National Laboratory, Oak Ridge, Tennessee 3783130PNPI, Petersburg Nuclear Physics Institute, Gatchina, Russia

31RIKEN (The Institute of Physical and Chemical Research), Wako, Saitama 351-0198, Japan32RIKEN BNL Research Center, Brookhaven National Laboratory, Upton, New York 11973-5000

33Universidade de São Paulo, Instituto de Física, Caixa Postal 66318, São Paulo CEP05315-970, Brazil34SUBATECH (Ecole des Mines de Nantes, INP3/CNRS, Universite de Nantes), BP 20722-44307, Nantes-Cedex 3, France

35St. Petersburg State Technical University, St. Petersburg, Russia36University of Tennessee, Knoxville, Tennessee 37996

37Department of Physics, Tokyo Institute of Technology, Tokyo, 152-8551, Japan38University of Tokyo, Tokyo, Japan

39Institute of Physics, University of Tsukuba, Tsukuba, Ibaraki 305, Japan40Vanderbilt University, Nashville, Tennessee 37235

41Waseda University, Advanced Research Institute for Science and Engineering, 17 Kikui-cho, Shinjuku-ku, Tokyo 162-0044, Japan42Weizmann Institute, Rehovot 76100, Israel

43Yonsei University, IPAP, Seoul 120-749, Korea44KFKI Research Institute for Particle and Nuclear Physics (RMKI), Budapest, Hungary†

(Received 5 February 2002; published 30 April 2002)

Transverse momentum spectra of electrons from Au 1 Au collisions atp

sNN � 130 GeV have beenmeasured at midrapidity by the PHENIX experiment at the Relativistic Heavy Ion Collider. The spectrashow an excess above the background from photon conversions and light hadron decays. The electronsignal is consistent with that expected from semileptonic decays of charm. The yield of the elec-tron signal dNe�dy for pT . 0.8 GeV�c is 0.025 6 0.004�stat� 6 0.010�syst� in central collisions, andthe corresponding charm cross section is 380 6 60�stat� 6 200�syst� mb per binary nucleon-nucleoncollision.

DOI: 10.1103/PhysRevLett.88.192303 PACS numbers: 25.75.Dw

In this Letter, we report the first measurement of singleelectron spectra, �e1 1 e2��2, in Au 1 Au collisions atp

sNN � 130 GeV at the Relativistic Heavy Ion Collider(RHIC). The measurement of single leptons at high trans-verse momentum �pT * 1 GeV�c� is a useful way tostudy heavy-quark production, an important probe of hotand dense matter created in high energy heavy ion col-lisions. Charm production is sensitive to the initial stategluon density [1,2]. Nuclear and medium effects, suchas shadowing and charm quark energy loss [3,4], can bestudied by comparison of charm production in AA, pA,and pp collisions. Measurement of charm is important

192303-2

for understanding J�c suppression (a proposed signal ofthe deconfinement phase transition [5,6]) and the dileptonmass distribution in 1 , Ml1l2 , 3 GeV, where leptonpairs from charm make significant contributions [7]. In ppcollisions at the CERN Intersecting Storage Rings (ISR)�p

s � 30 63 GeV�, production of single electrons wasobserved �e�p � 1024� for pT . 1 GeV�c [8–11], andinterpreted as evidence of open charm production [12]. Inpp collisions at RHIC energies, the signal level is expectedto be higher, since charm production increases with

psNN

faster than pion production. We recently observed suppres-sion of high pT pion production in Au 1 Au collisions at

192303-2


RHIC relative to binary nucleon-nucleon �NN � collisionscaling [13]. If charm production scales with NN col-lisions, as expected in the absence of nuclear effects, thee�p ratio will be even higher in Au 1 Au collisions atRHIC.

Data used for this analysis were recorded by thePHENIX west-arm spectrometer [14] (Df � 90± inazimuth, jhj , 0.35 in pseudorapidity), which consistedof a drift chamber (DC), a layer of pad chambers (PC1),a ring imaging Cerenkov detector (RICH), and a lead-scintillator electromagnetic calorimeter (EMCAL). Thetrigger was provided by beam-beam counters (BBC)and zero-degree calorimeters (ZDC). ZDC and BBCsignals were combined to select centrality: central(0%–10%), peripheral (60%–80%), and minimum bias(0%–92%) [15].

The analysis uses 1.23M minimum bias events withvertex position jzj , 30 cm. Charged particle tracks arereconstructed by the DC and the PC1 with a momentumresolution dp�p � 0.6% © 3.6% p �GeV�c�. Tracksare confirmed by a matching hit in the EMCAL, whichmeasures the energy E deposited with a resolution of8.2%�

pE�GeV� © 1.9% for test beam electrons. Elec-

tron identification is performed using the RICH and theEMCAL [14]. The RICH is filled with 1 atm CO2 anddetects on average 10.8 photoelectrons per electron track,while a pion with p , 4.7 GeV�c produces no signal.It is required that at least three RICH hits are associatedwith the track and that their hit pattern is consistentwith that of an electron track. After these cuts, a clearelectron signal is observed as a narrow peak centeredat E�p � 1.0. We select tracks in the peak as electroncandidates. The E�p cut reduces hadron background andremoves conversion electrons created far from the vertex.A hadron deposits only a fraction of its energy in theEMCAL, and the momentum of an off-vertex conversionelectron is reconstructed incorrectly. The remainingbackground, about 10% of the electron candidates, iscaused by accidental association of RICH hits with hadrontracks. The background level is measured statistically byan event mixing method, and is subtracted from the yield.

The electron acceptance (�7.4% of dN�dy) and effi-ciency ��60%� are determined using a detailed GEANT [16]simulation, which satisfactorily reproduces the detector re-sponse. Additionally, a multiplicity dependent efficiencyloss due to detector occupancy is evaluated by embeddingsimulated electrons into real events. This efficiency loss is27 6 4% �4 6 2%� for central (peripheral) collisions andhas no significant pT dependence.

Figure 1 shows the pT distributions of electrons inPHENIX for central, minimum bias, and peripheralcollisions. Errors in the figure are statistical. The overallsystematic uncertainty, which is the quadratic sum ofseveral few percent effects, is about 11%. Expectedsources of electrons are (1) Dalitz decays of p0, h,h0, v, and f, (2) dielectron decays of r, v, and f,(3) photon conversions, (4) kaon decays �K0,6 ! pen�,

192303-3

(GeV/c)TP0 0.5 1 1.5 2 2.5 3

-2d

y (G

eV/c

)T

/dp

2 d

NT

1/p

π1/

2ev

t1/

N

10-6

10-5

10-4

10-3

10-2

10-1

1

10=130 GeV)NNs)/2 + X (-+e

+Au+Au --> (e

Central (0-10%)

Peripheral (60-80%)

min. bias (0-92%)

FIG. 1. Transverse momentum spectra of electrons in PHENIXfrom Au 1 Au collisions at

psNN � 130 GeV.

(5) semileptonic decay of charm, and (6) other contribu-tions such as bottom decays and thermal dileptons. In thisanalysis, sources (1)–(4) are considered to be background.

We have calculated the contributions from Dalitz and di-electron decays with a hadron decay generator. PHENIXhas measured the pT distributions of p6 in 0.2 , pT ,2.2 GeV�c [17] and of p0 in 1 , pT , 4 GeV�c [13].Since the p6 and p0 data are consistent in the overlap-ping region, we fit a power law function to the combineddata sets to determine the input p0 spectrum for the de-cay generator. The pT distribution of any other hadron his obtained from the p0 spectrum by replacing pT withq

p2T 1 m2

h 2 m2p0 . The shapes of the resulting pT spec-

tra of K6, p, and p agree with the PHENIX measure-ments [17] within 20%. In this parametrization h�p0

ratios approach constants at high pT . We assume thefollowing asymptotic ratios to fix the relative normaliza-tions: h�p0 � 0.55, h0�p0 � 0.25, r�p0 � v�p0 �1.0, f�p0 � 0.40. Except for the f, these ratios aretaken from proton beam data of CERN Super Proton Syn-chrotron (SPS), FNAL, and ISR experiments [18,19]. Theh�p0 ratio is consistent with a measurement in Pb 1 Pbcollisions at SPS [20]. The f�p0 ratio is based on theintegrated ratio f�h2 � 0.02 in Au 1 Au collisions atp

sNN � 130 GeV [21]. We assign to each ratio a conser-vative systematic uncertainty of 50%.

Photon conversions are evaluated using a combinationof the GEANT simulation and the hadron decay generator.Since pT spectra of externally converted electrons are simi-lar to those from Dalitz decay, the conversion spectra canbe approximated by scaling the Dalitz decay spectra byan experiment specific factor, Rconv � conversion�Dalitz.Rconv is evaluated using the GEANT simulation and is cross-checked by comparing the relative yield of reconstructedDalitz and conversion pairs in the simulation and in thedata. The simulation shows that Rconv has only a weakpT dependence, primarily due to the energy dependence

192303-3


of the pair creation cross section. Rconv is parametrized as�1.9 6 0.2� 3 �1 2 0.0718 3 p20.76

T �.Background from kaon decays is also evaluated using

the GEANT simulation and is found to be negligible.The upper panel of Fig. 2 shows the ratio of the mea-

sured electrons to the calculated background versus pT forminimum bias events. The shaded region is the quadraticsum of systematic errors in the electron measurement andin the background. The latter includes uncertainties in thenormalization and the shape of the p0 spectrum, in theh�p0 ratios, and in Rconv. A significant electron excessabove the background is observed for pT . 0.6 GeV�c.Central collisions show a similar excess. The peripheralcollision data sample lacks sufficient statistics to reveal asignal in this analysis.

Fractional contributions to the background are shown inthe lower panel of Fig. 2. More than 80% of the back-ground is from p0 decay, directly from the Dalitz decayor indirectly from photon conversion. The p0 spectrum iswell constrained by the PHENIX measurement. The nextmost important background source is h decay. Given theassigned systematic error, the upper limit of the high pT

asymptotic h�p0 ratio is 0.83. Since this ratio, correctedfor feed-down, would imply that the primary h�p0 � 1,this provides a conservative limit on contributions fromh’s. Contributions from all other hadrons combined areonly a few percent of the total.

Background-subtracted electron spectra are shown inFig. 3. The error bars on the data points represent the

rati

o d

ata/

bac

kgro

un

d

1

1.5

2

2.5

3

3.5

= 130 GeVNNsAu+Au

Minimum Bias Collisions

[GeV/c]Tp0 0.5 1 1.5 2 2.5 3re

lati

ve c

on

trib

uti

on

to

bac

kgro

un

d

10-4

10-3

10-2

10-1

1 total conv.γ Dalitz0π

Dalitzη

’ Dalitzηρ

ωφ

e3K

FIG. 2. Ratio of the electron data to the calculated backgroundas a function of pT in minimum bias collisions (upper panel) andrelative contributions to the background from various sources(lower panel). The curves for v and f show the sum of theDalitz and the dielectron decay modes.

192303-4

statistical errors, while the systematic error due to the back-ground subtraction is indicated by brackets. The integratedyield of the electron signal dNe�dy for pT . 0.8 GeV�cis 0.025 6 0.004�stat� 6 0.010�syst� for central collisionsand is 0.0079 6 0.0006�stat� 6 0.0034�syst� for mini-mum bias collisions.

Semileptonic decay of charmed hadrons is an expectedsource of the electron signal. We use the event generatorPYTHIA [22] to estimate electron spectra from charm decay.We tuned the parameters [23] of PYTHIA such that charmproduction data at SPS and FNAL [24] and single electrondata at the ISR [9–11] are well reproduced. The charmproduction cross section in pp collisions from this PYTHIAcalculation is scc � 330 mb at

ps � 130 GeV. The elec-

tron spectrum in Au 1 Au collisions is then calculatedas EdNe�dp3 � TAA 3 Edse�dp3, where Edse�dp3 isthe electron spectrum from charm decay calculated withPYTHIA, and TAA (listed in Table I) is the nuclear overlapintegral calculated from a Glauber model [13]. The cal-culated electron spectra shown in Fig. 3 are in reasonableagreement with the data.

Before attributing the entire electron signal to opencharm decays, it is necessary to quantify contributionsfrom other possible sources. An analogous PYTHIA esti-mate of the bottom decay contribution is shown in Fig. 3.It becomes significant only above the measured pT range.Expected contributions from J�C and Drell-Yan are neg-ligible. In Pb 1 Pb collisions at SPS, direct photons [20]and an enhanced yield of low mass dileptons [25] havebeen reported. If these are due to thermal radiation fromhot matter, an even larger production is expected at RHICenergies and can contribute to the electron signal. Sincer ! e1e2 contributes less than 1% to the calculatedbackground as shown in Fig. 2, and since the dominant

(GeV/c)TP0 0.5 1 1.5 2 2.5 3

-2d

y (G

eV/c

)T

/dp

2 d

NT

1/p

π1/

2ev

t1/

N

10-6

10-5

10-4

10-3

10-2

10-1

1

10

102

=130 GeV)NNs)/2 + X (-+e+

(e→Au+Au min. bias (x100)central

e (central,min.bias)→c e (central)→b

e (central)→" γ"direct

FIG. 3. The background-subtracted electron spectra for mini-mum bias (0%–92%) (scaled up by a factor of 100) and central(0%–10%) collisions compared with the expected contributionsfrom open charm decays. Also shown, for central collisionsonly, are the expected contribution from bottom decays (dashedline) and the conversion electron spectrum from a direct photonprediction (dotted line).

192303-4


TABLE I. Charm cross section per NN collision derived from the single electron data forcentral (0%–10%) and minimum bias (0%–92%) collisions. The first and second errors arestatistical and systematic, respectively.

Centrality TAA �mb21� dscc�dyjy�0 �mb� scc �mb�

0%–10% 22.6 6 1.6(syst) 97 6 13 6 49 380 6 60 6 2000%–92% 6.2 6 0.4(syst) 107 6 8 6 63 420 6 33 6 250

source of thermal dileptons is p 1 p ! r ! e1e2

[26], a significant contribution from thermal dileptons isunlikely. There are several predictions for direct photonsat RHIC energies [27,28]. The conversion electron spec-trum calculated from a prediction in Ref. [27] is shown inFig. 3 for central collisions. It could explain 10%–20%of the signal, with large theoretical uncertainties.

Neglecting these other possible sources and assumingthat all the electron signal is from charm, we derive thecharm cross section corresponding to the electron data.We fit the charm electron spectrum from PYTHIA to thedata for pT . 0.8 GeV�c and obtain the rapidity den-sity dNcc�dyjy�0 and the total yield Ncc of open charm.They are then converted to cross sections per NN col-lision: dscc�dy � �dNcc�dy��TAA and scc � Ncc�TAA.Results are shown in Table I. The systematic error is aquadratic sum of many sources. For central collisions,they are background subtraction �644%�, uncertainties inthe PYTHIA calculation (611% from �kT � � 1.5 6 0.5,613% from D1�D0 � 0.65 6 0.35, 68% from PDFs),fit range �618%�, and TAA �67%�. Note that any finitecontribution from neglected sources would reduce the de-rived charm cross section. Without nuclear or medium ef-fects in charm production, scc per NN collision should beindependent of centrality. Within uncertainties, our data

(GeV)s30 40 50 60 70 80 90100 200

)b

inar

yb

/Nµ

( ccσ

10-1

1

10

102

103

)b

inar

y/d

y (n

b/N

eσd

1

10

102

103

104

ccσ

<1.4 GeV/cT1.0<p

>1.4 GeV/cTp

ccσ

PHENIXSPS/FNAL

/dyeσdPHENIXCCRSBasile

e→c

FIG. 4. Single electron cross sections dse�dyjy�0 of thismeasurement and ISR experiments [9,11,30] are displayed(bottom, right-hand scale) with charm decay contributionscalculated with PYTHIA. Open and filled symbols are for1.0 , pT , 1.4 GeV�c and pT . 1.4 GeV�c, respectively.The derived charm cross section of this measurement is com-pared with charm cross sections from SPS/FNAL experiments(top, left-hand scale). The thick curve and the shaded bandrepresent the charm cross section in the PYTHIA model and in aNLO pQCD calculation [31], respectively.

192303-5

are consistent with this expectation, in possible contrast tothe attribution of increased charm production as the sourceof enhanced dimuon production reported in Pb 1 Pb col-lisions at SPS [29].

The single electron signal yield (divided by TAA to givethe cross section per NN collision) and the derived charmcross section are compared with single electron data ofISR experiments and charm data of fixed target experi-ments [24] in Fig. 4. Cross section curves calculated withPYTHIA, which has been tuned to the charm data and theISR electron data, and a charm cross section curve froma next-to-leading order (NLO) pQCD calculation [31] arealso shown in the figure. Our data are consistent with bothof the calculations within large uncertainties.

In conclusion, we have observed single electrons abovethe expected background from decays of light hadronsand photon conversion in Au 1 Au collisions at

psNN �

130 GeV. The observed signal is consistent with semilep-tonic decay of charm. The forthcoming high statisticsAu 1 Au data and pp comparison data at full RHIC en-ergy �

psNN � 200 GeV� will be useful to clarify the na-

ture of the single electron signal and to better determineheavy-quark production in Au 1 Au collisions at RHIC.

We thank the staff of the Collider-Accelerator andPhysics Departments at BNL for their vital contributions.We acknowledge support from the Department of Energyand NSF (U.S.A.), MEXT and JSPS (Japan), RAS,RMAE, and RMS (Russia), BMBF, DAAD, and AvH(Germany), VR and KAW (Sweden), MIST and NSERC(Canada), CNPq and FAPESP (Brazil), IN2P3/CNRS(France), DAE and DST (India), KRF and CHEP (Korea),the U.S. CRDF for the FSU, and the U.S.–Israel BSF.

*Deceased.†Not a participating institution.

[1] J. A. Appel, Annu. Rev. Nucl. Part. Sci. 42, 367 (1992).[2] B. Müller and X. N. Wang, Phys. Rev. Lett. 68, 2437

(1992).[3] Z. Lin and M. Gyulassy, Phys. Rev. Lett. 77, 1222 (1996).[4] Y. L. Dokshitzer and D. E. Kharzeev, Phys. Lett. B 519,

199 (2001).[5] T. Matsui and H. Satz, Phys. Lett. B 178, 416 (1986).[6] M. C. Abreu et al., Phys. Lett. B 447, 28 (2000).[7] R. Vogt et al., Phys. Rev. D 49, 3345 (1994).[8] F. W. Büsser et al., Phys. Lett. 53B, 212 (1974).[9] F. W. Büsser et al., Nucl. Phys. B113, 189 (1976).

[10] P. Perez et al., Phys. Lett. 112B, 260 (1982).[11] M. Basile et al., Nuovo Cimento Soc. Ital. Fis. 65A, 421

(1981).

192303-5


[12] I. Hinchliffe and C. H. Llewellyn Smith, Phys. Lett. 61B,472 (1976); M. Bourquin and J.-M. Gaillard, Nucl. Phys.B114, 334 (1976).

[13] K. Adcox et al., Phys. Rev. Lett. 88, 022301 (2002).[14] H. Hamagaki et al., Nucl. Phys. A698, 412 (2002).[15] K. Adcox et al., Phys. Rev. Lett. 86, 3500 (2001).[16] GEANT Users Guide, 3.15, CERN Program Library.[17] K. Adcox et al., nucl-ex/0112006.[18] R. Albrecht et al., Phys. Lett. B 361, 14 (1995); G. Aga-

kichiev et al., Eur. Phys. J. C 4, 249 (1998).[19] M. Diakonou et al., Phys. Lett. 89B, 432 (1980).[20] M. M. Aggarwal et al., nucl-ex/0006007; M. M. Aggarwal

et al., Phys. Rev. Lett. 85, 3595 (2000).[21] C. Adler et al., Phys. Rev. C 65, R041901 (2002).[22] T. Sjostrand, Comput. Phys. Commun. 82, 74 (1994).[23] We used PYTHIA 6.152 with CTEQ5L PDF [H. L. Lai et al.,

Eur. Phys. J. C 12, 375 (2000)]. Modified PYTHIA parame-ters are PARP�91� � 1.5 ��kt��, PMAS�4, 1� � 1.25 �mc�,

192303-6

PARP�31� � 3.5 (K factor), MSTP�33� � 1, MSTP�32� �4 (Q2 scale).

[24] G. A. Alves et al., Phys. Rev. Lett. 77, 2388 (1996).[25] G. Agakichiev et al., Phys. Lett. B 422, 405 (1998).[26] R. Rapp, Phys. Rev. C 63, 054907 (2001).[27] J. Alam et al., Phys. Rev. C 63, 021901 (2001).[28] F. D. Steffen and M. H. Thoma, Phys. Lett. B 510, 98

(2001); D. K. Srivastava, nucl-th/0103023.[29] M. C. Abreu et al., Eur. Phys. J. C 14, 443 (2000).[30] The 1.0 1.4 GeV�c point of CCRS is calculated from the

e�p ratio in Ref. [9] and the pion cross section in B. Alperet al., Nucl. Phys. B100, 237 (1975).

[31] M. Mangano, P. Nason, and G. Ridolfi, Nucl. Phys. B405,507 (1993). Their program HVQMNR is used with CTEQ5MPDF to calculate scc in Fig. 4 with mc � 1.5 GeV�c2,mF � 2mc , and 0.5mc , mR , 2mc.

192303-6

Measurement of Bottom Versus Charm as a Function of Transverse Momentumwith Electron-Hadron Correlations in pþ p Collisions at

ffiffiffis

p ¼ 200 GeV

A. Adare,11 S. Afanasiev,25 C. Aidala,12,36 N.N. Ajitanand,53 Y. Akiba,47,48 H. Al-Bataineh,42 J. Alexander,53 K. Aoki,30,47

L. Aphecetche,55 R. Armendariz,42 S. H. Aronson,6 J. Asai,47,48 E. T. Atomssa,31 R. Averbeck,54 T. C. Awes,43

B. Azmoun,6 V. Babintsev,21 M. Bai,5 G. Baksay,17 L. Baksay,17 A. Baldisseri,14 K. N. Barish,7 P. D. Barnes,33

B. Bassalleck,41 A. T. Basye,1 S. Bathe,7 S. Batsouli,43 V. Baublis,46 C. Baumann,37 A. Bazilevsky,6 S. Belikov,6,*

R. Bennett,54 A. Berdnikov,50 Y. Berdnikov,50 A.A. Bickley,11 J. G. Boissevain,33 H. Borel,14 K. Boyle,54 M. L. Brooks,33

H. Buesching,6 V. Bumazhnov,21 G. Bunce,6,48 S. Butsyk,33,54 C.M. Camacho,33 S. Campbell,54 B. S. Chang,62

W.C. Chang,2 J.-L. Charvet,14 S. Chernichenko,21 J. Chiba,26 C.Y. Chi,12 M. Chiu,22 I. J. Choi,62 R. K. Choudhury,4

T. Chujo,58,59 P. Chung,53 A. Churyn,21 V. Cianciolo,43 Z. Citron,54 C. R. Cleven,19 B. A. Cole,12 M. P. Comets,44

P. Constantin,33 M. Csanad,16 T. Csorgo,27 T. Dahms,54 S. Dairaku,30,47 K. Das,18 G. David,6 M. B. Deaton,1 K. Dehmelt,17

H. Delagrange,55 A. Denisov,21 D. d’Enterria,12,31 A. Deshpande,48,54 E. J. Desmond,6 O. Dietzsch,51 A. Dion,54

M. Donadelli,51 O. Drapier,31 A. Drees,54 K. A. Drees,5 A.K. Dubey,61 A. Durum,21 D. Dutta,4 V. Dzhordzhadze,7

Y. V. Efremenko,43 J. Egdemir,54 F. Ellinghaus,11 W. S. Emam,7 T. Engelmore,12 A. Enokizono,32 H. En’yo,47,48

S. Esumi,58 K.O. Eyser,7 B. Fadem,38 D. E. Fields,41,48 M. Finger, Jr.,8,25 M. Finger,8,25 F. Fleuret,31 S. L. Fokin,29

Z. Fraenkel,61,* J. E. Frantz,54 A. Franz,6 A.D. Frawley,18 K. Fujiwara,47 Y. Fukao,30,47 T. Fusayasu,40 S. Gadrat,34

I. Garishvili,56 A. Glenn,11 H. Gong,54 M. Gonin,31 J. Gosset,14 Y. Goto,47,48 R. Granier de Cassagnac,31 N. Grau,12,24

S. V. Greene,59 M. Grosse Perdekamp,22,48 T. Gunji,10 H.-A. Gustafsson,35 T. Hachiya,20 A. Hadj Henni,55

C. Haegemann,41 J. S. Haggerty,6 H. Hamagaki,10 R. Han,45 H. Harada,20 E. P. Hartouni,32 K. Haruna,20 E. Haslum,35

R. Hayano,10 M. Heffner,32 T. K. Hemmick,54 T. Hester,7 X. He,19 H. Hiejima,22 J. C. Hill,24 R. Hobbs,41 M. Hohlmann,17

W. Holzmann,53 K. Homma,20 B. Hong,28 T. Horaguchi,10,47,57 D. Hornback,56 S. Huang,59 T. Ichihara,47,48 R. Ichimiya,47

Y. Ikeda,58 K. Imai,30,47 J. Imrek,15 M. Inaba,58 Y. Inoue,49,47 D. Isenhower,1 L. Isenhower,1 M. Ishihara,47 T. Isobe,10

M. Issah,53 A. Isupov,25 D. Ivanischev,46 B.V. Jacak,54,† J. Jia,12 J. Jin,12 O. Jinnouchi,48 B.M. Johnson,6 K. S. Joo,39

D. Jouan,44 F. Kajihara,10 S. Kametani,10,47,60 N. Kamihara,47,48 J. Kamin,54 M. Kaneta,48 J. H. Kang,62 H. Kanou,47,57

J. Kapustinsky,33 D. Kawall,36,48 A. V. Kazantsev,29 T. Kempel,24 A. Khanzadeev,46 K.M. Kijima,20 J. Kikuchi,60

B. I. Kim,28 D.H. Kim,39 D. J. Kim,62 E. Kim,52 S. H. Kim,62 E. Kinney,11 K. Kiriluk,11 A. Kiss,16 E. Kistenev,6

A. Kiyomichi,47 J. Klay,32 C. Klein-Boesing,37 L. Kochenda,46 V. Kochetkov,21 B. Komkov,46 M. Konno,58 J. Koster,22

D. Kotchetkov,7 A. Kozlov,61 A. Kral,13 A. Kravitz,12 J. Kubart,8,23 G. J. Kunde,33 N. Kurihara,10 K. Kurita,49,47

M. Kurosawa,47 M. J. Kweon,28 Y. Kwon,56,62 G. S. Kyle,42 R. Lacey,53 Y.-S. Lai,12 Y. S. Lai,12 J. G. Lajoie,24 D. Layton,22

A. Lebedev,24 D.M. Lee,33 K. B. Lee,28 M.K. Lee,62 T. Lee,52 M. J. Leitch,33 M.A. L. Leite,51 B. Lenzi,51 P. Liebing,48

T. Liska,13 A. Litvinenko,25 H. Liu,42 M.X. Liu,33 X. Li,9 B. Love,59 D. Lynch,6 C. F. Maguire,59 Y. I. Makdisi,5

A. Malakhov,25 M.D. Malik,41 V. I. Manko,29 E. Mannel,12 Y. Mao,45,47 L. Masek,8,23 H. Masui,58 F. Matathias,12

M. McCumber,54 P. L. McGaughey,33 N. Means,54 B. Meredith,22 Y. Miake,58 P. Mikes,8,23 K. Miki,58 T. E. Miller,59

A. Milov,6,54 S. Mioduszewski,6 M. Mishra,3 J. T. Mitchell,6 M. Mitrovski,53 A.K. Mohanty,4 Y. Morino,10 A. Morreale,7

D. P. Morrison,6 T. V. Moukhanova,29 D. Mukhopadhyay,59 J. Murata,49,47 S. Nagamiya,26 Y. Nagata,58 J. L. Nagle,11

M. Naglis,61 M. I. Nagy,16 I. Nakagawa,47,48 Y. Nakamiya,20 T. Nakamura,20 K. Nakano,47,57 J. Newby,32 M. Nguyen,54

T. Niita,58 B. E. Norman,33 R. Nouicer,6 A. S. Nyanin,29 E. O’Brien,6 S. X. Oda,10 C. A. Ogilvie,24 H. Ohnishi,47

H. Okada,30,47 K. Okada,48 M. Oka,58 O.O. Omiwade,1 Y. Onuki,47 A. Oskarsson,35 M. Ouchida,20 K. Ozawa,10 R. Pak,6

D. Pal,59 A. P. T. Palounek,33 V. Pantuev,54 V. Papavassiliou,42 J. Park,52 W. J. Park,28 S. F. Pate,42 H. Pei,24 J.-C. Peng,22

H. Pereira,14 V. Peresedov,25 D.Yu. Peressounko,29 C. Pinkenburg,6 M. L. Purschke,6 A.K. Purwar,33 H. Qu,19 J. Rak,41

A. Rakotozafindrabe,31 I. Ravinovich,61 K. F. Read,43,56 S. Rembeczki,17 M. Reuter,54 K. Reygers,37 V. Riabov,46

Y. Riabov,46 D. Roach,59 G. Roche,34 S. D. Rolnick,7 A. Romana,31,* M. Rosati,24 S. S. E. Rosendahl,35 P. Rosnet,34

P. Rukoyatkin,25 P. Ruzicka,23 V. L. Rykov,47 B. Sahlmueller,37 N. Saito,30,47,48 T. Sakaguchi,6 S. Sakai,58

K. Sakashita,47,57 H. Sakata,20 V. Samsonov,46 S. Sato,26 T. Sato,58 S. Sawada,26 K. Sedgwick,7 J. Seele,11 R. Seidl,22

A.Yu. Semenov,24 V. Semenov,21 R. Seto,7 D. Sharma,61 I. Shein,21 A. Shevel,46,53 T.-A. Shibata,47,57 K. Shigaki,20

M. Shimomura,58 K. Shoji,30,47 P. Shukla,4 A. Sickles,6,54 C. L. Silva,51 D. Silvermyr,43 C. Silvestre,14 K. S. Sim,28

B. K. Singh,3 C. P. Singh,3 V. Singh,3 S. Skutnik,24 M. Slunecka,8,25 A. Soldatov,21 R.A. Soltz,32 W. E. Sondheim,33

S. P. Sorensen,56 I. V. Sourikova,6 F. Staley,14 P.W. Stankus,43 E. Stenlund,35 M. Stepanov,42 A. Ster,27 S. P. Stoll,6

T. Sugitate,20 C. Suire,44 A. Sukhanov,6 J. Sziklai,27 T. Tabaru,48 S. Takagi,58 E.M. Takagui,51 A. Taketani,47,48

PRL 103, 082002 (2009) P HY S I CA L R EV I EW LE T T E R Sweek ending

21 AUGUST 2009

0031-9007=09=103(8)=082002(7) 082002-1 � 2009 The American Physical Society

R. Tanabe,58 Y. Tanaka,40 K. Tanida,47,48 M. J. Tannenbaum,6 A. Taranenko,53 P. Tarjan,15 H. Themann,54 T. L. Thomas,41

M. Togawa,30,47 A. Toia,54 J. Tojo,47 L. Tomasek,23 Y. Tomita,58 H. Torii,20,47 R. S. Towell,1 V-N. Tram,31 I. Tserruya,61

Y. Tsuchimoto,20 C. Vale,24 H. Valle,59 H.W. van Hecke,33 A. Veicht,22 J. Velkovska,59 R. Vertesi,15 A.A. Vinogradov,29

M. Virius,13 V. Vrba,23 E. Vznuzdaev,46 M. Wagner,30,47 D. Walker,54 X. R. Wang,42 Y. Watanabe,47,48 F. Wei,24

J. Wessels,37 S. N. White,6 D. Winter,12 C. L. Woody,6 M. Wysocki,11 W. Xie,48 Y. L. Yamaguchi,60 K. Yamaura,20

R. Yang,22 A. Yanovich,21 Z. Yasin,7 J. Ying,19 S. Yokkaichi,47,48 G. R. Young,43 I. Younus,41 I. E. Yushmanov,29

W.A. Zajc,12 O. Zaudtke,37 C. Zhang,43 S. Zhou,9 J. Zimanyi,27,* and L. Zolin25

(PHENIX Collaboration)

1Abilene Christian University, Abilene, Texas 79699, USA2Institute of Physics, Academia Sinica, Taipei 11529, Taiwan

3Department of Physics, Banaras Hindu University, Varanasi 221005, India4Bhabha Atomic Research Centre, Bombay 400 085, India

5Collider-Accelerator Department, Brookhaven National Laboratory, Upton, New York 11973-5000, USA6Physics Department, Brookhaven National Laboratory, Upton, New York 11973-5000, USA

7University of California - Riverside, Riverside, California 92521, USA8Charles University, Ovocny trh 5, Praha 1, 116 36, Prague, Czech Republic9China Institute of Atomic Energy (CIAE), Beijing, People’s Republic of China

10Center for Nuclear Study, Graduate School of Science, University of Tokyo, 7-3-1 Hongo, Bunkyo, Tokyo 113-0033, Japan11University of Colorado, Boulder, Colorado 80309, USA

12Columbia University, New York, New York 10027 and Nevis Laboratories, Irvington, New York 10533, USA13Czech Technical University, Zikova 4, 166 36 Prague 6, Czech Republic

14Dapnia, CEA Saclay, F-91191, Gif-sur-Yvette, France15Debrecen University, H-4010 Debrecen, Egyetem ter 1, Hungary

16ELTE, Eotvos Lorand University, H - 1117 Budapest, Pazmany P. s. 1/A, Hungary17Florida Institute of Technology, Melbourne, Florida 32901, USA

18Florida State University, Tallahassee, Florida 32306, USA19Georgia State University, Atlanta, Georgia 30303, USA

20Hiroshima University, Kagamiyama, Higashi-Hiroshima 739-8526, Japan21IHEP Protvino, State Research Center of Russian Federation, Institute for High Energy Physics, Protvino, 142281, Russia

22University of Illinois at Urbana-Champaign, Urbana, Illinois 61801, USA23Institute of Physics, Academy of Sciences of the Czech Republic, Na Slovance 2, 182 21 Prague 8, Czech Republic

24Iowa State University, Ames, Iowa 50011, USA25Joint Institute for Nuclear Research, 141980 Dubna, Moscow Region, Russia

26KEK, High Energy Accelerator Research Organization, Tsukuba, Ibaraki 305-0801, Japan27KFKI Research Institute for Particle and Nuclear Physics of the Hungarian Academy of Sciences (MTA KFKI RMKI),

H-1525 Budapest 114, Post Office Box 49, Budapest, Hungary28Korea University, Seoul, 136-701, Korea

29Russian Research Center ‘‘Kurchatov Institute’’, Moscow, Russia30Kyoto University, Kyoto 606-8502, Japan

31Laboratoire Leprince-Ringuet, Ecole Polytechnique, CNRS-IN2P3, Route de Saclay, F-91128, Palaiseau, France32Lawrence Livermore National Laboratory, Livermore, California 94550, USA

33Los Alamos National Laboratory, Los Alamos, New Mexico 87545, USA34LPC, Universite Blaise Pascal, CNRS-IN2P3, Clermont-Fd, 63177 Aubiere Cedex, France

35Department of Physics, Lund University, Box 118, SE-221 00 Lund, Sweden36Department of Physics, University of Massachusetts, Amherst, Massachusetts 01003-9337, USA

37Institut fur Kernphysik, University of Muenster, D-48149 Muenster, Germany38Muhlenberg College, Allentown, Pennsylvania 18104-5586, USA


41University of New Mexico, Albuquerque, New Mexico 87131, USA42New Mexico State University, Las Cruces, New Mexico 88003, USA43Oak Ridge National Laboratory, Oak Ridge, Tennessee 37831, USA

44IPN-Orsay, Universite Paris Sud, CNRS-IN2P3, BP1, F-91406, Orsay, France45Peking University, Beijing, People’s Republic of China

46PNPI, Petersburg Nuclear Physics Institute, Gatchina, Leningrad region, 188300, Russia47RIKEN Nishina Center for Accelerator-Based Science, Wako, Saitama 351-0198, Japan

48RIKEN BNL Research Center, Brookhaven National Laboratory, Upton, New York 11973-5000, USA


21 AUGUST 2009

082002-2

49Physics Department, Rikkyo University, 3-34-1 Nishi-Ikebukuro, Toshima, Tokyo 171-8501, Japan50Saint Petersburg State Polytechnic University, St. Petersburg, Russia

51Universidade de Sao Paulo, Instituto de Fısica, Caixa Postal 66318, Sao Paulo CEP05315-970, Brazil52System Electronics Laboratory, Seoul National University, Seoul, Korea

53Chemistry Department, Stony Brook University, Stony Brook, SUNY, New York 11794-3400, USA54Department of Physics and Astronomy, Stony Brook University, SUNY, Stony Brook, New York 11794, USA

55SUBATECH (Ecole des Mines de Nantes, CNRS-IN2P3, Universite de Nantes) BP 20722 - 44307, Nantes, France56University of Tennessee, Knoxville, Tennessee 37996, USA

57Department of Physics, Tokyo Institute of Technology, Oh-okayama, Meguro, Tokyo 152-8551, Japan58Institute of Physics, University of Tsukuba, Tsukuba, Ibaraki 305, Japan

59Vanderbilt University, Nashville, Tennessee 37235, USA60Waseda University, Advanced Research Institute for Science and Engineering, 17 Kikui-cho, Shinjuku-ku, Tokyo 162-0044, Japan

61Weizmann Institute, Rehovot 76100, Israel62Yonsei University, IPAP, Seoul 120-749, Korea

(Received 29 March 2009; published 18 August 2009)

The momentum distribution of electrons from semileptonic decays of charm and bottom quarks for

midrapidity jyj< 0:35 in pþ p collisions atffiffiffis

p ¼ 200 GeV is measured by the PHENIX experiment at

the Relativistic Heavy Ion Collider over the transverse momentum range 2< pT < 7 GeV=c. The ratio of

the yield of electrons from bottom to that from charm is presented. The ratio is determined using partial

D= �D ! e�K�X (K unidentified) reconstruction. It is found that the yield of electrons from bottom

becomes significant above 4 GeV=c in pT . A fixed-order-plus-next-to-leading-log perturbative quantum

chromodynamics calculation agrees with the data within the theoretical and experimental uncertainties.

The extracted total bottom production cross section at this energy is �b �b ¼ 3:2þ1:2�1:1ðstatÞþ1:4

�1:3ðsystÞ�b.

DOI: 10.1103/PhysRevLett.103.082002 PACS numbers: 13.85.Qk, 13.20.Fc, 13.20.He, 25.75.Dw

Measurements of heavy flavor production (charm andbottom) in pþ p collisions present stringent tests forperturbative quantum chromodynamics (pQCD) calcula-tions. For instance, while bottom production at theTevatron is well described by next-to-leading order(NLO) pQCD [1], the cross section for charm productionat high pT , though compatible within the theoretical un-certainties, is higher than the preferred theoretical value by�50% [2]. Measurement of heavy flavor in pþ p colli-sions also provides an important baseline for study of themedium created in relativistic heavy-ion collisions. ThePHENIX experiment at the Relativistic Heavy Ion Collider(RHIC) has measured single electrons from the semilep-tonic decay of heavy flavor at midrapidity in pþ p andAuþ Au collisions at

ffiffiffiffiffiffiffiffisNN

p ¼ 200 GeV [3,4]. Strong sup-

pression of the single electron yield at high pT , whichincludes contributions from both charm and bottom de-cays, was observed in central Auþ Au collisions [4]. Thiseffect is conventionally attributed to energy loss by theparent parton in the medium [5]; one also expects theenergy loss suffered by bottom quarks to be significantlyless than that suffered by charm quarks due to the differ-ence in their masses [6,7]. Clearly, for both pQCD com-parisons and the heavy-ion reference, one wants to dis-entangle the yields of charm and bottom at RHIC energies.

In this Letter, we present the yield ratio of single elec-trons from bottom to those from heavy flavor at midrapid-ity in pþ p collisions at

ffiffiffis

p ¼ 200 GeV, using partialD= �D ! e�K�X (K unidentified) reconstruction. Thedata were collected with the PHENIX detector [8] in the2005 and 2006 RHIC runs using its two central arm spec-

trometers. Each spectrometer covers j�j< 0:35 in pseu-dorapidity and �� ¼ �=2 in azimuth. The arms includedrift chambers (DC) and pad chambers (PC1,2,3) forcharged particle tracking, a ring imaging Cerenkov detec-tor (RICH) and an electromagnetic calorimeter (EMCal)for electron identification and triggering. Beam-beamcounters (BBCs), covering pseudorapidity 3:1< j�j<3:9, measure the position of the collision vertex along thebeam (zvtx) and provide the interaction trigger. In the 2005run, helium bags were placed in the space between thebeam pipe and DC to reduce photon conversions. The bagswere removed in 2006.Two data sets are used for the analysis: (i) a minimum

bias (MB) data set recorded with the BBC trigger, and(ii) an electron enriched sample, recorded with a level-1‘‘ERT’’ trigger requiring a combination of EMCal andRICH information in coincidence with the BBC trigger.The BBC trigger cross section is 23:0� 2:2 mb [9]. TheBBC trigger selects �53% of inelastic pþ p collisionsand ð79� 2Þ% of hard scattering events, such as those withhigh-pT particles at midrapidity. The latter efficiency isapproximately pT and particle independent, which wasverified by the observed ratio of high-pT pion and etayields with and without the BBC trigger, and confirmedwith a Monte Carlo simulation [9,10]. After selection ofgood runs and a vertex cut of jzvtxj< 25 cm, an integratedluminosity (

RLdt) in the ERT data of 1:77 pb�1 in the

2005 run and 4:22 pb�1 in the 2006 run are used for thisanalysis.Charged particle tracks are reconstructed using the

DC and PC1. The momentum resolution is �1% at


21 AUGUST 2009

082002-3

http://dx.doi.org/10.1103/PhysRevLett.103.082002

pT � 1 GeV=c, and the momentum scale is calibratedwithin 1%. Electron identification (eID) is performed usingthe RICH and EMCal. The purity of the electron sample isbetter than 99% for 1 <pT < 5 GeV=c [3]. Our previousmeasurement [3] determined the spectrum of the singleelectrons from heavy flavor in the 2005 run. Inclusiveelectron spectra from the 2005 run and the 2006 run areconsistent within 5% after taking into account a contribu-tion from the increased photon conversion due to theabsence of the helium bags.

The spectrum of the single electrons from heavy flavor isdetermined using the ‘‘cocktail method’’ [3,4]. The elec-tron spectrum from all known sources except semileptonicdecay of heavy flavor is calculated using a Monte Carlosimulation and subtracted from the inclusive spectrum inthe cocktail method. The dominant source of background isthe �0 Dalitz decay. The cocktail also includes contribu-tions from quarkonium (J=c , �) and the Drell-Yan pro-cess, which were neglected in our previous measurements[3,4]. These contributions are negligible (smaller than 1%in background) for pT < 1 GeV=c, but become significantat high pT (above 10% for pT > 2:5 GeV=c) [11]. Thesignal to background ratio increases with increasing pT ,approaching unity for pT � 3 GeV=c [3].

The systematic uncertainties of the inclusive electronspectrum includes the uncertainty in luminosity (9.6%),geometrical acceptance (3%), eID efficiency (2%), andthe ERT trigger efficiency (4% at pT > 2 GeV=c). Theuncertainty in the cocktail method is pT dependent (3%at pT � 2 GeV=c, increasing to 9% at 9 GeV=c).

The ratio of (b ! e) to ½ðc ! eÞ þ ðb ! eÞ� is extractedfrom the correlation between the heavy flavor electrons andassociated hadrons [12,13]. The extraction is based onpartial reconstruction of the D= �D ! e�K�X decay. Theinvariant mass of unlike charge-sign electron-hadron pairs

reveals a correlated signal below the D meson mass of�1:9 GeV=c2, because of the charge correlation in the Ddecays. Pairs are formed between a trigger electron (2:0<pT < 7:0 GeV=c) and an oppositely charged hadron(0:4< pT < 5:0 GeV=c). The hadron pT is required tobe less than 5 GeV=c, because pions also emit Cerenkovphotons in the RICH above 5 GeV=c. The acceptances ofpositive and negative charged particles are forced to beidentical by a geometrical acceptance cut. Since the mo-mentum range of good charged kaon identification is lim-ited, K identification is not performed but the mass of allreconstructed hadrons is set to be that of the K. Most eþe�pairs are then removed by an electron veto cut for thehadrons. The reconstructed mass of eþe� pairs has a clearpeak at low mass. The remaining background eþe� pairsare removed by requiring Mee > 80 MeV=c2, where thepair mass is calculated assuming both particles in the pairare electrons.Depending on the origin of the trigger electrons, the

inclusive reconstructed electron-hadron pairs are(i) unlike-sign pairs from charm, (ii) unlike-sign pairsfrom bottom, (iii) combinatorial background where theelectron is a background electron and (iv) backgroundfrom unlike-sign hadron-hadron pairs due to hadron con-tamination in the electrons. The main background source isthe combinatorial background (iii) and almost all back-ground electrons are from eþe� pair creation. Like-signelectron-hadron pairs are used to subtract this background.Since electrons from eþe� pair creation and hadrons donot contribute to charge correlated signals, subtractionusing like-sign pairs cancels out completely the combina-torial background where the trigger electron is from eþe�pair creation (iii). Only the negligibly small (<1%) con-tribution from K0

e3 decay is not canceled out by the sub-

traction in the background (iii). The contribution from

)2(GeV/ceKM0.5 1 1.5 2 2.5 3 3.5 4 4.5

GeV2 c

eKM∆

pai

rN∆

e(H

F)

N1

0

0.02

0.04

0.06

0.08

0.1

0.12e-K(NO PID) reconstruction

(a) 3~4Gev/cT

data electron p

simulation(charm)

simulation(bottom)

simulation(charm+bottom at

e))→e+b→e/(c→measured b

)2(GeV/ceKM

0.5 1 1.5 2 2.5 3 3.5 4 4.5

0.4~5Gev/cT

hadron p

(b) 4~5Gev/cT

data electron p

simulation (combination of

e+hadron from jet)→c

FIG. 1 (color online). Comparison of data to a PYTHIA and EVTGEN [14,15] simulation of the invariant mass distributions in PHENIXacceptance for the reconstructed signal in the 2006 run. The electron pT range is 3:0–4:0 GeV=c (a) and 4:0–5:0 GeV=c (b). Theratios, ðb ! eÞ=½ðc ! eÞ þ ðb ! eÞ�, in solid lines are 0.26 (a) and 0.63 (b). Error bars (boxes) indicate statistical (systematic)uncertainties.


21 AUGUST 2009

082002-4

hadron contamination (iv) is also less than a 1% effect dueto the excellent electron identification. After the subtrac-tion, the reconstructed pairs include a contribution frombottom (ii) due to not identifying K. The contribution frombottom (ii) is much smaller than that from charm (i) due tothe bottom decay modes and kinematics. The reconstructedpairs also contain a signal from partial reconstruction ofheavy flavor hadrons and a contribution from a combina-tion of heavy flavor electrons and hadrons from jet frag-mentation. The ratio of the yield of unlike-sign pairs to thatof like-sign pairs is about 1.1 for invariant masses (MeK)below 1:9 GeV=c2.

The fraction of bottom contribution to the electrons fromheavy flavor is obtained as follows:

b ! e

ðc ! eÞ þ ðb ! eÞ ¼ �c � �data�c � �b

; (1)

where �data is the tagging efficiency in real data and �cðbÞ isthe tagging efficiency for charm (bottom) production.These are defined as

�data �Npair

NeðHFÞ¼ Nc!tag þ Nb!tag

ðc ! eÞ þ ðb ! eÞ ; (2)

�c �Nc!tag

c ! e; �b �

Nb!tag

b ! e; (3)

where NeðHFÞ is the number of measured heavy flavor

electrons. Npair is the number of background subtracted

unlike-sign electron-hadron pairs for invariant mass within0:4<MeK < 1:9 GeV=c2, which corresponds to the massrange of charmed hadrons. Here, NcðbÞ!tag is the number of

reconstructed signals within 0:4<MeK < 1:9 GeV=c2 forcharm (bottom) production.

Figure 1 shows theMeK distribution of the reconstructedsignals, which is normalized by the yield of heavy flavorelectrons (NeðHFÞ) in the range 3<pT < 4 GeV=c [panel

(a)] and 4< pT < 5 GeV=c [panel (b)]. The tagging effi-ciency in real data, �data, is determined by the integration ofthe MeK distribution in Fig. 1 from MeK ¼ 0:4 to1:9 GeV=c2 as a function of electron pT .The tagging efficiencies for charm and bottom produc-

tion, �c and �b, are calculated with the combination ofPYTHIA and EVTGEN [14,15]. PYTHIA is used to simulate

charm and bottom production in pþ p collisions atffiffiffis

p ¼200 GeV and is tuned to reproduce heavy flavor hadronratios: Dþ=D0 ¼ 0:45� 0:10, Ds=D0 ¼ 0:25� 0:10,�c=D

0 ¼ 0:10� 0:05, Bþ=B0 ¼ 0:50, Bs=B0 ¼ 0:40�0:20, and Bbaryon=B

0 ¼ 0:20� 0:15 [11,16–19]. The

Monte Carlo simulation EVTGEN, which is suited for de-cays of D and B hadrons, is used to simulate the semi-leptonic decays. The dashed (dotted) lines in Fig. 1 showthe MeK distributions of the reconstructed signal for thesimulated charm (bottom) production for an electron3<pT < 4 GeV=c (panel a) and 4<pT < 5 GeV=c[panel (b)]. Some fluctuations in the simulated curves inFig. 1 come from the limited statistics in the simulation,but the statistical uncertainties in the simulation are negli-gible compared to that of the data. �cðbÞ is determined in the

same way as �data from the MeK distribution for charm(bottom) production. Since about 85% of the extracted

(GeV/c)T

Electron p2 3 4 5 6 7

e)

→ e

+b→

e/(

c→b

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1PHENIX |y| <0.35

FONLL y=0

FONLL error band y=0

=200 GeVs p+p at

90% C.L.

90% C.L.

FIG. 2 (color online). ðb ! eÞ=½ðc ! eÞ þ ðb ! eÞ� as a func-tion of electron pT compared to a FONLL calculation [20]. Thepoints show the experimental result. Vertical arrows are used toindicate upper and lower limits. The solid line is a FONLLprediction and the dotted lines represent the uncertainty of thisFONLL prediction.

)3 c-2

(m

b G

eV3

/dp

σ3E

d

-910

-810

-710

-610

-510

90% C.L.

90% C.L.

90% C.L.

90% C.L.

=200 GeVs)/2 + X at - + e+ (e→ p+p (a)

e)→ e + b →PHENIX data(c e)→PHENIX data(c e)→PHENIX data((b

e)→ e + b →FONLL(c e)→FONLL(c e)→FONLL(b

98765432

Dat

a/F

ON

LL

1

2

3 (b) e)→(c90% C.L.

90% C.L.

(GeV/c)T

p

2 3 4 5 6 7 8 9

1

2

3 (c) e)→(b

90% C.L.

90% C.L.

FIG. 3 (color online). (a) Invariant cross sections of electronsfrom charm and bottom with the FONLL calculation [20].(b) and (c) The ratios of data points over the FONLL predictionas a function of electron pT for charm and bottom. The shadedarea shows the uncertainty in the FONLL prediction.


21 AUGUST 2009

082002-5

signal comes from partial reconstruction of heavy flavorhadrons, the tagging efficiency is determined largely bydecay kinematics and �cðbÞ can be determined with good

precision. The dot-dash lines in Fig. 1 show the contribu-tion from the combination of an electron from charm andhadrons from jet fragmentation for charm production. Thesolid line in Fig. 1 shows the sum of the MeK distributionsfor charm and bottom in the simulation with the ratio, ðb !eÞ=½ðc ! eÞ þ ðb ! eÞ�, obtained with Eq. (1).

Systematic uncertainties are categorized into two partsrelated to (i) �data in the real data analysis and (ii) �c and�b in the simulation study. The dominant uncertainty in�data is the uncertainty in the number of heavy flavorelectrons (�10%). Uncertainty in �data also includes abackground subtraction uncertainty (1%–10%, pT depen-dent). Category (2) includes the uncertainties in geometri-cal acceptance (3%) and the event generator (�8% forcharm and �9% for bottom). The event generator uncer-tainty is based on uncertainties, which are known in theproduction ratios of heavy flavor hadrons (Dþ=D0,Ds=D

0,etc.), known in the branching ratios [16–19], estimated inthe momentum distribution of heavy flavor hadrons andestimated in the PYTHIA parameters.

Figure 2 shows the resulting bottom fraction, ðb !eÞ=½ðc ! eÞ þ ðb ! eÞ� as a function of electron pT com-pared to a fixed-order-plus-next-to-leading-log perturba-tive QCD calculation (FONLL) [20]. In this figure, thepoints show the measured ðb ! eÞ=½ðc ! eÞ þ ðb ! eÞ�.For the bins with electron pT ranges 2<pT < 3 and 5<pT < 7 GeV=c, 90% C.L. and mean values are shown. Thesolid line shows the central value of the FONLL predictionand the dotted lines show its uncertainty.

In Fig. 3, the single electron spectra for charm andbottom are measured from the ratio, ðb ! eÞ=½ðc ! eÞ þðb ! eÞ�, and the spectrum of the electrons from heavyflavor decays. The top panel shows the resulting singleelectron spectra from charm (triangles) and bottom(squares) compared to the FONLL predictions [20]. Themeasured spectrum of single electrons (circles) is alsoshown for reference. The middle (bottom) panel showsthe ratio of the measured cross sections to the FONLLcalculation for charm (bottom) production. The shadedarea shows the uncertainty in the FONLL prediction. Thelarger mass makes this uncertainty smaller in the case ofbottom quarks. These calculations agree with the data forbottom production within the large theoretical and experi-mental uncertainty. The same is true for charm within thetheoretical uncertainty with a ratio of data/FONLL of �2.A similar tendency was obtained at the Tevatron [1,2].

The electron spectrum from bottom shown in Fig. 3is integrated from pT ¼ 3 to 5 GeV=c and gives4:8þ1:8

�1:6ðstatÞþ1:9�1:8ðsystÞ nb. This spectrum is then extrapo-

lated to pT ¼ 0 using the shape predicted by pQCD.PYTHIA with varying intrinsic kT (1:5< kT < 10 GeV=c)and FONLL with varying factorization (�F) and renor-

malization (�R) scales (0:5<�F;R=ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffim2 þ p2

T

q< 2) are

used to evaluate the systematic uncertainty (12%) tothis extrapolation. The extrapolation results in a bot-tom cross section at midrapidity of d�b �b=dy jy¼0¼0:92þ0:34

�0:31ðstatÞþ0:39�0:36ðsystÞ�b, using a b ! e total branching

ratio of 10� 1%, calculated using the heavy flavor hadronratios described above. Using HVQMNR [21] with CTEQ5M

[22] parton distribution functions (PDF’s) to integrate overrapidity, the total bottom cross section is determined to be�b �b ¼ 3:2þ1:2�1:1ðstatÞþ1:4�1:3ðsystÞ�b. Various PDF’s and bot-

tom mass values are used to evaluate the systematic un-certainty (8%) of the rapidity extrapolation. This result isconsistent with our result from the dielectron spectrum,which gave �b �b ¼ 3:9� 2:5ðstatÞþ3

�2ðsystÞ�b [23].

FONLL predicts �b �b ¼ 1:87þ0:99�0:67�b, in agreement with

both these experimental results.The fraction of bottom in heavy flavor electrons is found

to be larger than 0.33 with 90% confidence level at pT >5 GeV=c. Furthermore, the assumption of no bottom sup-pression directly leads to a lower limit on the nuclearmodification factor of single electrons, RAA, of greaterthan 0.33 with the same confidence level. However, accord-ing to our measurements, RAA is �0:25� 0:05ðstatÞ �0:05ðsystÞ at 5<pT < 6 GeV=c [4] in the 0–10% centralAuþ Au collisions. At the same time the current level ofuncertainty in the measurement precludes us from placingsignificant limits on the possible energy loss of bottomquarks.In conclusion, the ratio of the yield of electrons from

bottom to that from charm has been measured in pþ pcollisions at

ffiffiffis

p ¼ 200 GeV. The ratio provides the firstmeasurement of the spectrum of electrons from bottom atRHIC. FONLL calculations [20] agree with this result,which provides an important baseline for the study ofheavy quark production in the hot and dense matter createdin Auþ Au collisions.We thank the staff of the Collider-Accelerator and

Physics Departments at BNL for their vital contributions.We acknowledge support from the Office of NuclearPhysics in DOE Office of Science, NSF and a sponsoredresearch grant from Renaissance Technologies (USA),MEXT and JSPS (Japan), CNPq and FAPESP (Brazil),NSFC (China), MSMT (Czech Republic), IN2P3/CNRS,and CEA (France), BMBF, DAAD, and AvH (Germany),OTKA (Hungary), DAE (India), ISF (Israel), KRF andKOSEF (Korea), MES, RAS, and FAAE (Russia), VRand KAW (Sweden), U.S. CRDF for the FSU, US-Hungary Fulbright, and US-Israel BSF.

*Deceased.†PHENIX [email protected]

[1] D. Acosta et al., Phys. Rev. D 71, 032001 (2005).


21 AUGUST 2009

082002-6

[2] D. Acosta et al., Phys. Rev. Lett. 91, 241804 (2003).[3] A. Adare et al., Phys. Rev. Lett. 97, 252002 (2006).[4] S. S. Adler et al., Phys. Rev. Lett. 98, 172301 (2007).[5] M. Gyulassy and M. Plumer, Phys. Lett. B 243, 432

(1990).[6] Y. L. Dokshitzer and D. E. Kharzeev, Phys. Lett. B 519,

199 (2001).[7] H. van Hees, M. Mannarelli, V. Greco, and R. Rapp, Phys.

Rev. Lett. 100, 192301 (2008).[8] K. Adcox et al., Nucl. Instrum. Methods Phys. Res., Sect.

A 499, 469 (2003).[9] S. S. Adler et al., Phys. Rev. C 75, 024909 (2007).[10] S. S. Adler et al., Phys. Rev. Lett. 95, 202001 (2005).[11] Y. Morino, arXiv:0903.3504.

[12] C. Albajar et al., Phys. Lett. B 186, 237 (1987).[13] A. Mischke et al., J. Phys. G 35, 104117 (2008).[14] T. Sjostrand, Comput. Phys. Commun. 82, 74 (1994).[15] D. Lange, Nucl. Instrum. Methods Phys. Res., Sect. A 462,

152 (2001).[16] D. Acosta et al., Phys. Rev. Lett. 91, 241804 (2003).[17] L. Gladilin, arXiv:hep-ex/9912064.[18] G. Alves et al., Phys. Rev. Lett. 77, 2388 (1996).[19] W.M. Yao et al., J. Phys. G 33, 1 (2006).[20] M. Cacciari, P. Nason, and R. Vogt, Phys. Rev. Lett. 95,

122001 (2005); (private communication).[21] M. L. Mangano et al., Nucl. Phys. B 405, 507 (1993).[22] H. L. Lai et al., Eur. Phys. J. C 12, 375 (2000).[23] A. Adare et al., Phys. Lett. B 670, 313 (2009).


21 AUGUST 2009

082002-7

Nuclear Instruments and Methods in Physics Research A 499 (2003) 549–559

PHENIX inner detectors

M. Allena, M.J. Bennettb, M. Bobrekc, J.B. Boissevainb, S. Boosed, E. Boszee,C. Brittonc, J. Change, C.Y. Chif, M. Chiuf, R. Conwayb, R. Cunninghamb,

A. Denisovg, A. Deshpandeh, M.S. Emeryc, A. Enokizonoi, N. Ericsonc, B. Foxh,S.-Y. Funge, P. Giannottid, T. Hachiyai, A.G. Hansenb, K. Hommai, B.V. Jacakb,D. Jaffec, J.H. Kangj, J. Kapustinskyb, S.Y. Kimj, Y.G. Kimj, T. Kohamai,P.J. Kroond, W. Lenzd, N. Longbothamk, M. Musrockc, T. Nakamurai,H. Ohnishid, S.S. Ryuj, A. Sakaguchii, R. Setoe, T. Shiinal, M. Simpsonc,

J. Simon-Gillob, W.E. Sondheimb, T. Sugitatei, J.P. Sullivanb,*, H.W. van Heckeb,J.W. Walkerc, S.N. Whited, P. Willisk, N. Xua

aUniversity of Tennessee, Knoxville, TN 37996, USAbLos Alamos National Laboratory, Los Alamos, CA 87545, USA

cOak Ridge National Lab, Oak Ridge, TN 37831, USAdBrookhaven National Laboratory, Upton, NY 11973-5000, USA

eUniversity of California, Riverside, CA 92521, USAfColumbia University, New York, NY 10027 and Nevis Laboratories, Irvington, NY 10533, USA

g Institute for High Energy Physics (IHEP), Protvino, RussiahRIKEN BNL Research Center, Brookhaven National Laboratory, Upton, NY 11973-5000, USA

iHiroshima University, Kagamiyama, Higashi-Hiroshima 739-8526, JapanjYonsei University, Seoul 120-749, South Korea

kAbilene Christian University, Abilene, TX 79699, USAlUniversity of Alabama, Huntsville, AL 35899, USA

The PHENIX Collaboration

Abstract

The timing, location and particle multiplicity of a PHENIX collision are determined by the Beam–Beam Counters

(BBC), the Multiplicity/Vertex Detector (MVD) and the Zero-Degree Calorimeters (ZDC). The BBCs provide both the

time of interaction and position of a collision from the flight time of prompt particles. The MVD provides a measure of

event particle multiplicity, collision vertex position and fluctuations in charged particle distributions. The ZDCs provide

information on the most grazing collisions. A Normalization Trigger Counter (NTC) is used to obtain absolute cross-

section measurements for p–p collisions. The BBC, MVD and NTC are described below.

r 2002 Published by Elsevier Science B.V.

PACS: 25.75.�q; 29.40.Ka; 29.40.Mc; 29.40.Gx

Keywords: Relativistic heavy-ion collisions; Cherenkov detectors; Scintillation detectors; Tracking and position sensitive detectors

*Corresponding author. Tel.: +1-505-665-5963; fax: +1-505-665-7920.

E-mail address: [email protected] (J.P. Sullivan).

0168-9002/03/$ - see front matter r 2002 Published by Elsevier Science B.V.

doi:10.1016/S0168-9002(02)01956-3

1. Introduction

The PHENIX detector [1] at the RelativisticHeavy Ion Collider (RHIC) is designed to performa broad study of A–A, p–A and p–p collisions toinvestigate nuclear matter under extreme condi-tions. The characteristics of the interactions ofprojectiles at the PHENIX interaction diamondare determined by the Beam–Beam Counters(BBCs) and the Multiplicity/Vertex Detector(MVD). The BBCs [2] are composed of two arraysof Cherenkov counters with quartz radiators andphotomultiplier readout each located just outsidethe poles of the central magnet. The BBC providesboth the time of interaction and its position fromthe flight time of prompt particles. The time ofinteraction is used as a start time for the time-of-flight (ToF) measurement and a signal for theLevel-1 (LVL1) trigger to limit the vertex regionwithin the PHENIX acceptance. Typical BBCresults from the first year of PHENIX running aredescribed. The MVD provides a measure of eventmultiplicity, collision vertex position and fluctua-tions in charged particle distributions. It consistsof Si strip detectors in concentric barrels aroundthe beam pipe and endcaps of Si pad detectors.The NTC is used to enhance beam normalizationmeasurements during polarized proton running.The ZDC [3] is described elsewhere in this volume.

2. Beam–beam counters

The main role for PHENIX BBC is to providethe time of beam–beam collisions for the ToF [4]measurement, to produce a signal for the PHE-NIX LVL1 trigger and to measure the collisionvertex point (ZVTX) along the beam axis. Sincethe longitudinal size of the beam bunch at RHICfor Au–Au collisions is designed to be 25 cmRMS, the time spread of nuclear collisions couldbe as much as 2 ns: Charged hadron identificationby the ToF system requires its resolution to be lessthan 100 ps for the whole ToF system, thusallowing pion–kaon separation up to 2:4 GeV=c

with a 4s standard deviation. The precise mea-surement of the ToF start time determined by theBBC is critical for the identification of hadrons.

2.1. BBC requirements and detector configuration

The BBC consists of two identical sets ofcounters installed on both sides of the collisionpoint along the beam axis, one on the North sideand the other on the South side of the PHENIXinteraction region where each counter is namedBBCN and BBCS, respectively. The BBCs areplaced 144 cm from the center of the interactiondiamond and surround the beam pipe. Thiscorresponds to a pseudorapidity range from 3.0to 3.9 over the full azimuth.The BBC needs to satisfy the following require-

ments:

(i) The number of charged particles generatedby Au–Au collisions at

ffiffiffiffiffiffiffiffisNN

p¼ 200 GeV

within the BBC acceptance is as much as afew thousands for central collisions. More-over the BBC also needs to operate with p–pcollisions up to

ffiffis

p¼ 500 GeV: The BBC

must thus have a capability to function overa large dynamic range from 1 MIP to30 MIPs:

(ii) Since the BBC is placed in a very high-levelradiation area around the beam pipe near theinteraction point, it is required to be radia-tion hard.

(iii) The BBC is installed just behind the PHE-NIX central spectrometer magnet. The ex-pected magnetic field along the beam axisaround this location is 3 kG: Therefore theBBC needs to work in a high magnetic fieldenvironment.

After studying the possibilities carefully, wechose the following configuration to satisfy theabove requirements. Each counter is composed of64 1-inch diameter mesh-dynode photomultipliertubes (Hamamatsu R6178) equipped with 3 cmquartz on the head of the PMT as a Cherenkovradiator. Fig. 1(a) and (b) show individual BBCelements and the BBC array mounted on the BBCmechanical frame, respectively. The outer dia-meter of the BBC is 30 cm and the inner diameteris 10 cm with clearance between the beam pipe andthe BBC of 1 cm: For this configuration theexpected number of charged particles for a central

M. Allen et al. / Nuclear Instruments and Methods in Physics Research A 499 (2003) 549–559550

Au–Au collision atffiffiffiffiffiffiffiffisNN

p¼ 200 GeV is expected

to be 15 particles per BBC element. A laser signalwhich is used by the ToF and ElectroMagneticCalorimeter (EMCal) [5] calibration systems isalso delivered to individual BBC elements tomonitor and calibrate the drift of the timing forthe ToF measurement. The BBC system has eighthigh voltage channels for each side. A single highvoltage supply operates at least eight BBCelements. To equalize the PMT gains the BBCelements are divided into 16 groups of tubes withsimilar gain based on cosmic-ray measurements.Operational high voltage is set to obtain 75 pCoutput charge for a MIP particle from each BBCelement.The temperature inside the BBC is a problem for

BBC operation since 64 PMTs are confined to avery small space. With no cooling the temperaturecan exceed 751C which may cause electric shorts ofthe cable inside the bleeder circuit. To avoid thisproblem an air flow of 200 l=min is used to coolthe BBC. Moreover, 32 thermocouples are in-stalled inside the BBC to monitor the temperature.Flow meters are also installed to measured the airflow rate. The information from the thermocou-ples and flow meters is connected to the BBC highvoltage interlock system so that if the temperature

inside the BBC exceeds 501C; or no air flow isdetected in the BBC cooling system, the highvoltage system for the BBC is shutdown immedi-ately by the interlock system. Fig. 1(c) shows theBBC installed in the mounting fixture on thecentral spectrometer magnet. Fig. 2 in the PHE-NIX Overview article [1] gives a view of the BBCin relation to the rest of the PHENIX detector.

2.2. BBC readout electronics and PMT gain

adjustment

The BBC readout electronics chain consists ofdiscriminators, shaping amplifiers, time-to-voltageconverters (TVC) and flash ADCs (FADC). Thetime and pulse height information is digitized inreal time and is stored in digital buffer memories.The BBC provides the LVL1 trigger with an inputsignal thus the timing and pulse height of BBCelements are digitized during each beam crossingby the TVC and FADC and stored at the sametime in the buffer memory. A detailed descriptionof the front-end electronics used in the PHENIXdetector is found elsewhere in this volume [6].The gains of the PMTs were adjusted using the

MIP peak in the pulse height distribution. Fig. 2shows a typical pulse height distribution for a BBC

Fig. 1. (a) Single BBC consisting of 1 in mesh dynode photomultiplier tubes mounted on a 3 cm quartz radiator; (b) A BBC array

comprising 64 BBC elements; and (c) The BBC is shown mounted on the PHENIX detector. The beam pipe is seen in the middle of the

picture. The BBC is installed on the mounting structure just behind the central spectrometer magnet.

M. Allen et al. / Nuclear Instruments and Methods in Physics Research A 499 (2003) 549–559 551

element. The hatched area in this figure shows thesame distribution for low multiplicity events whereless than 10 valid hits out of 64 BBC elements arerequired. The MIP peaks are visible at approxi-mately 75 pC on the x-axis. The variation of thePMT gains was 20% RMS.

2.3. Calibration and time resolution

Each BBC measures arrival times of leadingcharged particles from beam collisions. The arrivaltime for each BBC ðTS and TNÞ is defined as theaverage of the hit time in each BBC. Thesystematic shift of timing of individual BBCelements from the average hit time is mostlycaused by the time walk by the discriminator andtime offset in individual BBC elements. Thefollowing empirical formula was used for the timewalk and time offset correction to get a precisetime:

Tcorr ¼ Traw � Toffset � C=ffiffiffiffiQ

pð1Þ

here, Traw is the uncorrected arrival time, Toffset isthe time offset in each individual BBC element, Q

is the charge of the signal as measured with theFADC and C is the coefficient of the time walk.

Toffset and C are determined by minimizing theresidual time dti ¼ TS or N � ti for all BBC ele-ments. The search for the calibration parameters isperformed iteratively and the parameters convergewithin three iterations.After application of the time calibration to all

BBC elements, the time deviation from the averagehit time is mostly caused by the time resolution ofthe particular BBC element, since the average hittimes for the north and south BBC arrays arestatistically well determined. Fig. 3(a) shows thedistribution of timing deviations from the BBCaverage time for a typical BBC element. Thestandard deviation of this distribution is defined asthe time resolution of each BBC element. Fig. 3(b)shows the distribution of time resolution overall BBC elements. The time resolution of a singleBBC element is 5274 ps (rms) under real experi-mental conditions. Fig. 4 shows the ToF distribu-tion where the start time is defined by the BBC.The ToF resolution is 96 ps which allows theseparation of pions and kaons up to particlemomenta of 2:4 GeV=c: The detailed descriptionfor the ToF detector is found elsewhere in thisvolume [4].

2.4. Z Vertex measurement and performance with

LVL1 trigger

Fig. 5(a) shows a correlation of ZVTX calcu-lated by BBC and ZDC, and Fig. 5(b) shows theprojection of Fig. 5(a) on to the axis of BBCZVTX. The peak at ZVTX ¼ 7144 cm inFig. 5(b) corresponds to beam collisions outsideof the BBC.The ZVTX position and number of hit PMTs in

each BBC are also calculated online and sent to theLVL1 trigger. jZVTX jo20 cm (from the onlinecalculation) and two or more PMTs fired in eachBBC are required as the LVL1 trigger condition.The hatched area in Fig. 5(b) corresponds to theevents selected by the LVL1 trigger. The triggerefficiency with respect to inelastic Au–Au colli-sions is evaluated by a PHENIX detector simula-tion with the HIJING [12] event generator toproduce Au–Au collisions as input, and was foundto be 9272%:

0 50 100 150 200

1

10

102

103

104

Cou

nts

pulse height distribution [pC]

MIP peak

Fig. 2. Typical pulse height distribution for a BBC element.

The hatched area shows the same distribution for low-multi-

plicity events where less than 10 valid hits on the 64 BBC

elements are required. The peak at the high pulse height end is

due to overflow.


2.5. Centrality measurement

The BBC charge sum has a monotonic correla-tion with collision centrality. In the offlineanalysis, the combined information on spectatorneutrons measured by the ZDC placed 18 m fromthe beam crossing point and the charged suminformation measured by the BBC is used to definethe collision centrality. The correlation betweenthe BBC and ZDC measurements is shown inFig. 6. The lines shown in Fig. 6 bin the eventswith centrality beginning on the far right with the

0–5% bin and going to the left at intervals of 5%up to 50%.

2.6. Performance

The BBC is fully implemented and installed inthe PHENIX detector complex. The performance

-200-150-100 -50 0 50 100 150 2000

200

400

600

800

1000

1200

1400

1600

1800

(a) [ps]average, N/S - TiT

Eve

nts

10 20 30 40 50 60 70 80 900

2

4

6

8

10

12

14

16

18

20

22

24

(b) Time resolution [ps]

num

ber

of P

MT

s

Fig. 3. (a) Distribution of timing deviation for a typical BBC element from BBC averaged hit timing and (b) profile of timing

resolution for each BBC element.

TOF - L/c(beta) [ns]

-1 0 1 2 3 40

10

20

30

40

50

60

~ 96psσ

π

K p

1.4< p <1.8GeV/c

Fig. 4. ToF distribution measured between the BBC and ToF

detectors. The start time of the ToF measurement is provided

by the BBC and the ToF detector defines the stop time. The

ToF in ns is given on the horizontal axis. The ToF is

determined by assuming the pion mass for each particle. A

detailed description of the ToF measurement is given in Ref. [4].

-300

-200

-100

0

100

200

300

Z V

erte

x fr

om Z

DC

[cm

](a)

-150 -100 -50 0 50 100 1500

50

100

150

200

250

Z Vertex from BBC [cm]

(b)

Eve

nts

Fig. 5. (a) Correlation of determined ZVTX between BBC and

ZDC and (b) ZVTX distribution from BBC. Hatched area

corresponds to the events satisfying the PHENIX Local LVL1

trigger condition.


of the BBC was evaluated usingffiffiffiffiffiffiffiffisNN

p¼ 130 GeV

Au–Au collisions and the arrival time of secondaryparticles generated by beam interactions, whichtravel nearly at the speed of light. The collisiontime (T0) is defined from the average arrival timefor the two BBC counters. The timing differencebetween each counter provides the ZVTX alongthe beam axis.All the detector elements and readout electronic

channels worked well during the first year ofRHIC operation. The timing achieved for a singleelement was 5274 ps: The BBC is used as a startcounter for the ToF measurement and a ToFresolution of 96 ps was obtained. The collisionvertex determined by the BBC is used not only foroffline analysis but also for the LVL1 trigger tolimit the vertex position. Combined informationfrom the BBC and ZDC is used to define thecollision centrality in the offline analysis.

3. Multiplicity vertex detector

The PHENIX Multiplicity Vertex Detector(MVD) provides event characterization and selec-tion, measures fluctuations in charged particleproduction and provides a collision vertex posi-

tion. The design criteria included large rapidityand good azimuthal coverage and granularitywhile also minimizing costs and material in theelectron arm acceptance. A discussion of thesimulation of the performance of the MVD hasbeen published [7].

3.1. Design of MVD

The MVD consists of two concentric barrels ofsilicon strip detectors (300 mm thick and 200 mmpitch) around the beampipe and two disk-shapedendcaps of silicon pad detectors at zE735 cm;where z refers to the beam axis. The length of theactive part of the silicon strip barrels is approxi-mately 64 cm: The basic mechanical element of thebarrels is the ‘‘C-cage’’. A sketch of a C-cage isshown in Fig. 7.A C-cage is made of light-weight rigid rohacell

foam (density 0:075 gm=cm3) with Si detectorsglued to the inner and outer surfaces. Flexiblekapton cables, with 50 mm thick copper traces,carry the signals from the detectors to Multi-ChipModules (MCMs) which are the first stage of thefront-end electronics. The barrel is constructed of12 of these C-cages stacked end-to-end to form 1=2of the barrel. The inner barrel consists of thedetectors glued to the inner surface of the rohacell.It is 5 cm from the beam line. The outer barrel is7:5 cm from the beam line. A completed half of theMVD, including some of the other mechanical andelectronic parts is shown in Fig. 8. The second halfis a mirror image of the first. The two halves areconnected by a hinge at the bottom. They ‘‘clam-shell’’ open in order that they lift up and then closearound the beam pipe. The inner barrel has 72positions (6 azimuthal times 12 longitudinal)populated. The outer barrel with R ¼ 7:5 cm willbe only partially populated with Si to reduce theamount of material in the electron arm acceptance.The top 2=3 of the detector will omit the detectorsin the eight C-cages nearest the center of theMVD.A few of the pad detectors can be seen in Fig. 8.

Most are hidden from view by the barrel but partsof several wedges can be seen just below the farend of Fig. 8 and to the left of the barrel. Eachendcap of the MVD contains 12 wedge-shaped pad

maxBBC/QBBCQ

0 0.2 0.4 0.6 0.8 1

max

ZD

C/E

ZD

CE

0

0.2

0.4

0.6

0.8

1ZDC vs BBC analog response

Fig. 6. Correlation between BBC charge sum and energy

deposit in ZDC. The lines define bins containing intervals

covering 5% centrality. The bin to the far right represents

events ranging from 0% to 5% centrality. The central line

shown in the figure is the centroid of the BBC–ZDC

distribution with the lines defining the centrality bins drawn

perpendicular to it.


detectors. Each pad detector is divided into 12azimuthal ðfÞ segments and 21 radial segments.The radial segments increase from 2.5 to 5 mmgoing from the innermost radial position ð5 cmÞ tothe outermost ð12 cmÞ: This is done to keep thepseudorapidity ðdZÞ coverage of each pad roughlyconstant. A diagram of one of the pad detectors isshown in Fig. 9.The single-event coverage about midrapidity is

five units in Z for the inner barrel layer and fourunits in Z for the outer barrel layer. The Z coveragefor event-averaged quantities is extended due tothe variation of the vertex position along thedirection of the colliding beams. The disk-shapedendcaps cover 1:79ojZjo2:64 for an event at z ¼0: The vertex position can be located using hits inthe barrel for z positions between �40 and�40 cm; which covers the majority of the lengthof the interaction region ðs ¼ 22 cmÞ:

3.2. MVD readout

Both the MVD pad and strip detectors are readout with identical MCMs, which are optimized forminimum size and power dissipation. The MCMs[8] are constructed by the High Density Inter-connect (HDI) process at Lockheed-Martin Cor-poration. They are approximately 43� 48�1:5 mm3 with a mass of 13:0 g: The substrate isalumina. Each MCM contains eight TGV-32preamplifier/discriminator chips [9] and eight

Fig. 8. A diagram of one half of the MVD barrel. Some of the

mechanical support structure has been removed from the

diagram to make the interior visible.

Fig. 9. A diagram of a MVD pad detector. The trapezodial-

shaped readout pads are shown.

Fig. 7. A diagram of an MVD ‘‘C-cage’’. Si detectors are glued

to a rohacell foam cage. Kapton cables connect the Si to MCMs

which are enclosed in a cooling plenum, also made of rohacell.


analog memory unit plus ADC chips (AMU/ADC) [10]. The TGV-32 and AMU/ADC chipseach have 32 channels. In addition, the MCMscontain two Xilinx 4010E Field ProgrammableGate Arrays (FPGAs), a temperature sensor(AD590) and an op-amp (CLC426). One of theFPGAs is called the Address List Manager (ALM)and is used to control the read and write addressesto/from the 64-deep AMUs in the AMU/ADCchips. The second FPGA is called the HeapManager (HM) and is used to control most ofthe other logic signals on the MCM. After a level-1accept is received, it collects all 256 ADC valuesfrom the MCM, constructs the serial data packetcontaining the ADC values and then sends it offfrom the MCM.A kapton cable connects up to six MCMs in the

barrel to a power/communications board. Theseboards can be seen at the bottom of the barrel inFig. 8. In the complete MVD, the north and southhalf each contain 12 power/communicationsboards. These, in turn, are connected in groupsof six to ‘‘mother boards’’. The mother boards areapproximately the shape of one half of a disk andare in the endcaps of the MVD. One ring-shaped‘‘daughter board’’ is also attached to each motherboard. The daughter board has 6 MCMs attachedto it. Each of these six MCMs is wire-bondeddirectly to a Si pad detector. In Fig. 8 the Si paddetectors can be seen, but the MCMs and thedaughter boards are obscured by a cooling plenumwhich covers them. The data packets are sent outof the MVD via cables connected to the motherboards. The mother boards are also connected totiming and control cables which bring in the 9.5and 38 MHz clocks for the MCMs as well asvarious control signals including the level-1 accept.There are also serial-control cables for eachpower/communications and daughter board con-nected to the mother board. These are used to loadthe Xilinx programs in the MCMs, and to set andread back various serial control bits in the MCMs,the AMU/ADCs and the TGV32s.The data packets coming off of the mother

boards are connected via 7 m shielded ribboncables to Data Collection Interface Modules(DCIMs) [11] in a rack below the intersectionregion. One DCIM module is associated with each

power/communications board or daughter board.The DCIMs serve as buffers for the data (they canhold up to 7 data packets in FIFOs), slightlymodify the data packets and also convert theincoming Low Voltage Differential Signals(LVDS) to a Glink serial output format on asingle optical fiber. These fibers go to the PHENIXdata acquisition system.The MCMs and the chips on them were

designed to minimize power dissipation. TheMCMs associated with the Si strips are containedin a rohacell plenum, which can be seen in Figs. 7and 8. Each MCM dissipates 1:8 W (about 7 mWper input channel), which is small enough to allowthe MCMs to be cooled by chilled air blownthrough the plenum. The MCMs associated withthe pad detectors are in a separate plenum, but arealso air cooled. Each mother board contains 38voltage regulators. The relatively localized heatgenerated by these regulators is removed by watercooling lines glued to the back of the mother-board. Temperature, air flow and water flowsensors throughout the cooling system are con-nected to a set of custom circuit boards whichautomatically shut off power to the affected partof the MVD when a problem arises.

3.3. Low mass construction of MVD

The MVD design minimizes the amount ofmaterial ‘‘seen’’ by the electron arms. The purposeis to reduce the background from photon conver-sion to eþe� pairs in the material in the MVD. Aspart of this effort, most of the mass is concentratedat the endcaps or at the bottom of the MVD. Ingeneral, only the lower energy electronsðu200 MeV=cÞ produced in the material at thebottom of the MVD (mainly the MCMs and thepower/communications boards) are bent into thecentral arm acceptance by the field of the centralmagnet. If only the material on a straight line pathfrom the MVD to the central arms is considered,the inner shell represents 0.17% of a radiationlength, the rohacell in the C-cages represents anaverage thickness of 0.074% of a radiation length,the outer shell 0.18% of a radiation length, eachkapton cable represents 0.042% of a radiationlength and each layer of Si represents 0.32% of a


radiation length. The amount of Si and thenumber of layers of kapton varies with positionin the barrel, but the minimum is 0.42% (no Si, nokapton), the maximum is 1.23% (two Si and fourkapton layers) and 0.79% is ‘‘typical’’ (one Si, onekapton).

3.4. MVD performance

The second year of RHIC running will be thefirst year of data taking for the MVD. As anexample of the expected performance Fig. 10shows a bench-top test of a single Si-strip channelwith a 106Ru source. The source passed through aSi-strip detector into a trigger scintillator. Thereadout was through the MVD readout hardware.The trigger scintillator was large ð22� 22 mm2Þcompared to a single Si strip, so most events are inthe ‘‘pedestal’’ of the ADC distribution shown inFig. 10. However, the few percent of all eventswhich actually hit this single strip are clearlyseparated from the pedestal. The signal to noiseratio is 14=1: In this example, common mode noise(i.e. noise common to all 256 channels) wasremoved in the analysis. We expect to be able to

do this for the real detector setup also, either bybetter grounding and shielding (preferable) or inthe offline analysis. Before removing the commonmode noise the signal to noise was 7:4=1 which wasonly adequate.

4. Normalization trigger counters

A Normalization Trigger Counter (NTC) hasbeen installed in the vertex region of the PHENIXdetector. This additional counter is necessary toenhance normalization measurements during p–prunning using polarized protons. The NTC con-sists of two identical fiber-readout scintillationcounters, each situated between the endplate of theMVD and the brass nosecone in the vertex regionon the north and south sides, respectively, of thePHENIX detector.For p–p collisions where the average multiplicity

is much less than for heavy ion collisions anormalization device with a greater coverage ofthe cross-section than that provided by the BBCsis needed. The latter sees approximately 60% ofthe cross-section but the addition of the NTCsincreases this number to approximately 85%. As aresult of this increased coverage, the error in thetotal p–p cross-section is reduced significantly,thereby improving the ability to use the p-p data tocompare with that from heavy ion running. Inaddition, the knowledge of the relative luminosityof different bunches is critical for the spinprogram. The NTC aids this effort by providinganother measurement to compare with the ZDCand BBC coincidence rates, thereby assuring thatthis critical quantity is measured correctly.The NTCs are also helpful in a number of other

ways. They provide a means for separating beamgas from beam–beam events by forming a coin-cidence between the NTC counters on each end ofthe PHENIX detector. In order to fulfill thisrequirement the NTC time resolution is in thesubnanosecond range. The NTCs also give anefficient and clean trigger with a different bias thanthe other ‘‘standard’’ PHENIX triggers, therebyaiding in understanding the bias of events collectedduring p–p running. They can also provide a

Fig. 10. ADC distribution for a 106Ru b� source in a single Si-strip channel. The pedestal is much larger than the signal

because the trigger scintillator ð22� 22 mm2Þ was much largerthan the strip ð5� 0:02 cm2Þ shown in this sample plot.


means of rejecting the unacceptable rate of cosmic-ray events in the muon arms.The NTC scintillators are situated in the gap

between the MVD and the nosecone and cover theradial area between the beam pipe and the outerring of the MVD. The fibers from the scintillatorsrun along the surface of the nosecone up to thephototubes. The phototubes, along with thecoupling enclosure for the fibers, are placed in analuminum structure which is affixed to thenosecone.Fig. 11 is a drawing for a single counter. One is

installed on either end of the PHENIX detector.The counter is constructed from four 40�40 cm2 � 5:1 mm thick pieces of BC-404 (poly-vinyltoluene-base) plastic scintillator into which80 cm long, 1 mm thick Bicron BCF-92-WLSwave-shifting fibers have been embedded ingrooves cut into the surface of the scintillator.The fibers are glued into place using BC-600optical epoxy. The upper quadrants are glued tothe lower quadrants to form halves that are lightsealed using a wrapping made of a combination of

BC-638 Black (0:2 mm thick) Tape, BC-642 PTFEReflector (0:08 mm thick) Tape and BC-620RTiO2 Reflector Paint. Each half is encased by0:25 mm thick stainless steel to protect the fibersand the surface. The thickness of the counter is lessthan 9:5 mm: Each NTC detector half is attachedto the MVD endplate at three points along theouter diameter via aluminum brackets and threepoints on the inner diameter via set screwsattached to an aluminum extension to the innerdiameter framework of the MVD.The unattached ends of fibers from each

quadrant of the NTC counter are bundled togetherand optically coupled to a Hamamatsu H6614-01phototube operating at a maximum voltage of�2500 V: Each phototube is in a light-tightenclosure in an aluminum tube which is mountedon the brass nosecone. The phototubes are near aHe bag and He degrades their performance.Therefore air from the BBC compressed gassystem flows through the gap between the photo-tube enclosure and the outer aluminum tube. Thesignals from the 8 phototubes are fed into the

Fig. 11. Schematic diagram of one of the two identical NTC scintillators. In the left half of the figure the pattern of wave-shifting fibers

is shown.


PHENIX data-acquisition system via 8 m Belden8214 cables and the HV is fed to the detectors viaBelden RG-59 high-voltage cables.The NTC was operational for the p–p run

during the second year of PHENIX data taking.The efficiency of the NTC will be mapped usingelectrons from the National Synchrotron LightSource at Brookhaven to determine any non-uniformities. The timing resolution was deter-mined in place in the PHENIX beamline to beabout 600 ps using precise timing from the BBCs.This resolution was sufficient to allow inclusion ofthe NTC in the interaction trigger. The datacollected using this interaction trigger was acritical factor in obtaining accurate cross-sectionmeasurements during the p–p run.

Acknowledgements

We acknowledge support from the Departmentof Energy (USA) and MEXT (Japan).

References

[1] K. Adcox, et al., PHENIX detector overview, Nucl. Instr.

and Meth. A, (2003) this volume.

[2] K. Ikematsu, et al., Nucl. Instr. and Meth. A 411 (1998)

238.

[3] C. Adler, et al., Nucl. Instr. and Meth. A 470 (2001)

488.

[4] M. Aizawa, et al., PHENIX central arm particle

ID detectors, Nucl. Instr. and Meth. A, (2003) this

volume.

[5] L. Aphecethe, et al., PHENIX calorimeter, Nucl. Instr.


[6] S.S. Adler, et al., PHENIX on-line systems, Nucl. Instr.


[7] M.J. Bennett, et al., IEEE Trans. Nucl. Sci. NS-46 (1999)

2022.

[8] S.F. Hahn, et al., IEEE Trans. Nucl. Sci. NS-47 (2000)

802.

[9] C.L. Britton, et al., Rev. Sci. Instrum. NS-70 (1999)

1684.

[10] M.S. Emery, et al., IEEE Trans. Nucl. Sci. NS-44 (1997)

374.

[11] M.N. Ericson, et al., IEEE Trans. Nucl. Sci. NS-45 (1998)

833.

[12] X.N. Wang, M. Gyulassy, Phys. Rev. D 44 (1991) 351.


study of charm production from the measurement …...目 次 1．主論文 study of charm production...

Documents

study of charm production from the measurement …...目次 1．主論文 study of charm production...