2-2 atom optics and atom lithography - nict - トップ...

1 Introduction

With the recent development of single-mode tunable laser technologies, research anddevelopment is underway on the use of thescattering force and dipole force of light tocontrol the motion of atoms. The techniquesinvolved in these studies have developed intoa new field of research referred to as “atomoptics,” which differs significantly from con-ventional optics, in which light is controlledwith matter. In particular, atom lithography[1]-[4] uses laser light to manipulate velocity-selected atoms to form a desired microstruc-ture as these atoms are deposited on a sub-strate. This technology has enabled high-pre-cision, non-contact generation of fine periodicatomic patterns on the scale of the wavelengthof light. This method has drawn a great dealof attention as a new means of nanofabrica-tion, one that differs fundamentally from con-ventional lithography techniques. Thismethod holds significant potential as new

material processing technology applicable tothe development of future info-communica-tions devices, including three-dimensionalphotonic crystals. However, present atomlithography experiments have revealed limita-tions in the species of atoms that may be usedand in the contrast of atomic patterns; theseproblems must be overcome prior to practicalapplication of atom lithography.

In an attempt to overcome the limitationsof conventional atom lithography, we per-formed experiments using a one-dimensionalmagneto-optical trap (1D-MOT) [5] to guideand deflect the atomic beam - a fundamentaltechnology for controlling atom source that isindispensable in high-performance atomlithography. With the velocity-selected beamthus obtained, we were able to achieve high-resolution atomic channeling [6]. Further, weconducted the first experiments to date inatom lithography using ytterbium (Yb) atoms,successfully producing intended atomic pat-terns [7]. This paper reports on the results of

OHMUKAI Ryuzo and WATANABE Masayoshi 17

2-2 Atom Optics and Atom Lithography

OHMUKAI Ryuzo and WATANABE Masayoshi

High-resolution atomic channeling using velocity-selected atoms may be able to over-come precision limitations of the conventional atom lithography. We have experimentallyclarified the dependence of line width and contrast of atomic patterns in the channelingregion on the velocity spread of the atomic source for the first time. Thermal or velocity-selected atomic beams prepared with a one-dimensional magneto-optical trap wereemployed as the atomic sources. We can show that narrower line width and higher contrastatomic patterns are obtained as the velocity spread becomes narrower. We also successful-ly produced periodic ytterbium (Yb) narrow lines on a substrate using a near-resonant laserlight and the direct-write atom lithography technique. We clearly observed a grating patternof Yb atoms fabricated on a substrate with a line separation of approximately 200 nm afterexamining the surface of the substrate with an atomic force microscopy. This is the firstdemonstration of nanofabrication using the atom-optical approach with Yb atoms.

Keywords Atom optics, Atom lithography, Magneto-optical trap, Optical molasses, Channeling

18

these experiments.

2 High-resolution atomic chan-neling

2.1 Experimental principles and setupTo prepare for the atom lithography, we

first confined the 87Rb atoms within a regionthe size of a wavelength of light using thedipole force potential induced by an opticalstanding wave; the channeling experiment wasthen conducted within the half-wavelengthperiod of the given light wave [8]. We per-formed the channeling experiments usingatomic beams at various velocity distributionwidths. Our goal in this experiment was toclarify the effect of the velocity distributionwidth of the atomic source on the characteris-tics of the channeled atomic pattern and toimprove its resolution.

When an atom enters the optical standingwave, the light imposes a dipole force on theatom. The direction of the force is determinedby the sign of the detuning of the light thatgenerates the optical standing wave, and themagnitude of the force is proportional to thegradient of the light intensity. As the spatialintensity distribution of the optical standingwave features a period corresponding to halfthe wavelength of the light, this standing waveacts as nanometer-scale arrays of cylindricallenses to direct and focus the atoms. Thus,when an atomic beam with approximately uni-form spatial distribution passes through theoptical standing wave, the atoms are focusedand channeled to the nodes or the antinodes ofthe standing wave corresponding to half of theperiod of the light wave.

Figure 1 shows the experimental setup.The apparatus consists of three componentsections. The first section generates the atom-ic beam. In this experiment we used a rubidi-um (Rb) beam; the goal was therefore to chan-nel 85Rb, using the transition 5S1/2 F=3→5P3/2

F=4. A thermal Rb atomic beam generatedusing an oven and two pinholes served as theatomic source. The divergence angle of thebeam was 10 mrad.

The second section of the apparatus per-forms velocity selection. The velocity selectoremployed uses a one-dimensional magneto-optical trap (1D-MOT). This velocity selectorperforms simultaneous transverse cooling(collimation) and velocity selection of theatomic beam. This configuration representsthe first reported use of a 1D-MOT scheme forthe atomic beam guide and velocity selector.The 1D-MOT (interaction length 15 cm) istilted at an angle θ relative to the axis of thethermal atomic beam and is constructed offour copper rods, four rectangular mirrors, andtwo laser beams.

Two external-cavity diode lasers serve aslight sources: one is used to deflect the atomicbeam and the other is for repumping atoms,with intensities of 10 mW/cm2 and 1 mW/cm2,respectively. These two laser beams areshaped to 15 mm×3 mm, superposed, set toσ＋-polarization, and introduced into the 1D-MOT. The laser light is multi-reflected fromthe four rectangular mirrors, as shown inFig.1. The σ＋-polarized light beam and theσ－-polarized light beam are irradiated to theatomic beam from opposite directions. Thefour copper rods are placed parallel to thedeflection axis and spaced 1.0 cm apart. Elec-tric currents are passed through each of theseadjacent rods in alternating directions, gener-ating the quadrupole magnetic field.

As the atoms entering the 1D-MOT regionexperience strong restoring and dampingforces toward the 1D-MOT axis, this tiltedaxis becomes the deflection axis. As a result,this apparatus can guide atoms within a widevelocity range along this deflection axis,change their directions of propagation, andextract them as a deflected beam. Thismethod can involve the use of transverse lasercooling to collimate the deflected beam suffi-ciently along the deflection axis while at thesame time compressing the beam diameter.This is expected to increase the atomic densityof the deflected beam. Further, precise controlof the laser frequency enables isotope selec-tion within the extracted atomic beam.

The third part is the channeling section. A

Journal of the National Institute of Information and Communications Technology Vol.51 Nos.1/2 2004

Ti:sapphire laser beam is focused to 1 mmwith a lens, reflected by a mirror, and allowedto interact with the atomic beam to generatethe optical standing wave that will be used tochannel the atoms. To generate the standingwave, the laser light is detuned to +200 MHz;optical power is set to 50 mW. Channelingproperties are measured from the frequencyshifts of atomic absorption by the channeledRb atoms. The absorption frequency of anatom at a position x in a strong optical stand-ing wave shows a shift from the resonant fre-quency（ν0）without the optical standing wave,which is expressed as the following equation [8]:

Here,δis the detuning of the light andΩ（x）　

isthe Rabi frequency at position x, which is pro-portional to the intensity of the optical stand-ing wave. In other words, the position of theatom in the optical standing wave can beprobed through the frequency shifts of theabsorption lines of the atom, and the atomicdensity at this position can be obtained fromthe absorption intensity. Scanning the fre-quency, the probe beam for the frequency shiftmeasurement is introduced through the chan-neling region, approximately perpendicular to

the atomic beam. The beam diameter of theprobe beam is set to 1 mm or less. The photo-diode receives the transmitted light to measurethe atomic absorption.

2.2 Experimental resultsFirst, we will present the results of veloci-

ty selection using the 1D-MOT. The resonantlaser light was irradiated perpendicular to theatomic beam 35 cm downstream from the 1D-MOT, and the spatial density distribution ofthe deflected atoms was then obtained fromthe spatial intensity distribution of the laser-induced fluorescence. Figures 2 (a) - (c) showthe results for the deflection angles (θ) of 4, 3,and 1.5 degrees, respectively. Two fluores-cence spots were observed in the CCD cameraimages in all cases. The upper spots weregenerated by atoms at high velocities propa-gating in a straight line, with little interactionwith the 1D-MOT. The lower spots were gen-erated by the deflected atomic beam. Themeasured deflection displacement agreed withvalues calculated from the deflection angleand the flight distance. Similar experimentswere conducted at various probe-beam fre-quencies, with results indicating that thedeflected beam consists only of 85Rb atoms.

The velocity distribution of the deflectedatomic beam guided by the 1D-MOT was


Experimental configuration for high-resolution atomic channelingFig.1

20

measured, revealing that as the deflectionangle is reduced, the upper limit velocity ofatoms that can be guided along the deflectionaxis increases, along with an increase in thedensity of deflected atoms. Forθ = 4°, onlyatoms with velocities of approximately 120m/s or less were guided along the deflectionaxis. Atθ = 3°, the deflected beam consistedof atoms with velocities of 180 m/s or less.When the angle was reduced toθ = 1.5°, theupper limit velocity increased to 250 m/s, withdeflection efficiency also increasing, to 40 %.

An atomic beam with a smaller θ valuefeatures a lower velocity component perpendi-cular to the deflection axis, and longer interac-tion time. This is considered to increase theupper limit velocity and the deflection effi-ciency of the atoms successfully guided. Itshould be noted that Fig. 2 indicates that thedeflected beam was collimated along thedeflection axis as intended; the divergence ofthe deflected beam was 1.5 mrad. The diame-ter of the deflected beam was also measuredand revealed to be smaller than that of thethermal atomic beam in all cases. Forθ= 3°,the diameter of the deflected beam was 2 mm,while that of the thermal beam was 8 mm.Thus, the extracted beam simultaneouslyunderwent deflection, collimation, and com-pression to form a high-density atomic beam.The high collimation and high density of theresultant atomic beam renders this beam suit-able as an atomic source in atom lithography.

Next, channeling experiments were per-formed using the deflected and velocity-selected beam obtained via the methoddescribed above as the atomic source. First,the thermal atomic beam, which features thelargest velocity spread, was used for channel-ing; the characteristics of the resultant chan-neling patterns were then measured. The Rbthermal atomic beam featured a velocity dis-tribution width corresponding to 270 m/s.Figure 3 (a) shows the measured absorptionspectrum for the channeling atoms.

As this experiment did not require velocityselection, one-dimensional optical molasseswas used to collimate an atom beam with a

divergence angle of 1 mrad, instead of the 1D-MOT. A broad absorption spectrum can beseen in Fig.3 (a). This spectrum differsmarkedly from the Doppler-free absorptionspectrum. Absorption intensities with no fre-quency shift correspond to atoms in the nodesof the optical standing wave, while absorp-tions with a larger frequency shift correspondto atoms closer to the antinodes of the opticalstanding wave. The absorption at a frequencyof －120 MHz corresponds to the atoms in theantinodes.

Figure 4 (a) shows the spatial density dis-tribution of the channeled atoms deduced fromthe frequency shift and absorption intensity.These results show that a grating pattern witha line width (full width at half intensity maxi-mum, or “FWHM”) of 100 nm was success-fully formed. The contrast of the pattern wasapproximately 2.6, and the pattern showed abackground atomic layer, probably due toabsorption by atoms in the probing regionwith little or no channeling. Figures 3 (b) and(c) show the results of similar experimentsusing atomic sources with narrower velocitydistribution widths. The atomic sources con-sisted of the deflected beams obtained whenthe deflection angles (θ) of the 1D-MOT were1.5 and 3 degrees, respectively, and the corre-sponding velocity distribution widths of theatomic sources were 100 m/s and 50 m/s,respectively. When the velocity distributionwidth was smaller and closer to a mono-veloc-ity beam, the absorption increased in the


Results of deflected beam measure-ment

Fig.2

regions of low frequency shift, while adecreasing absorption intensity gradientbecame conspicuous toward the regions oflarger frequency shift.

Figures 4 (b) and (c) show the spatial den-sity distribution of the atoms during channel-ing, obtained from the spectra of Fig.3 (b) and(c), respectively. The distributions of the 85Rbatoms in the optical standing wave are steeperthan that shown in Fig.4 (a). In Figure 4 (b), apattern with a line width of 83 nm is observed,and in Fig.4 (c), the line width is even narrow-er, at 57 nm. This line width value (resolu-tion) is finer than the resolution achieved inconventional optical lithography technology(100 nm). At the same time, pattern contrastis improved, to 3.0 (Fig.4 (b)) and 3.5 (Fig.4(c)), respectively. These results offer the firstexperimental evidence for the effects of veloc-ity distribution of the atomic source on chan-neling characteristics, clarifying that a narrow-er velocity distribution width in the atomicsource leads to higher resolution in atomicchanneling. It is clear that high-resolutionatomic patterning (atomic lithography) willrequire high-precision atomic manipulation,and we believe that the foregoing results willprove quite useful in establishing such a tech-nique.

3 Atom lithography using ytter-bium atoms

3.1 Experimental principles and setupBased on the technique developed as dis-

cussed above, we then proceeded to performexperiments to construct microstructures onthe substrate through deposition of the chan-neled atoms. We chose ytterbium (Yb) as thetarget atom, which had never been used inatom lithography experiments. High-densityYb thermal atomic beams can be obtained atrelatively low temperatures (approximately800 K), and structures constructed with Yb arestable outside of the vacuum chamber (i.e.,exposed to air). The laser wavelength formanipulating Yb atoms is 399 nm, which canbe generated using commercially available

laser sources. Further, Yb is one of a fewatomic species with which Bose-Einstein con-


Absorption spectra of channelingatoms; spectrum for (a) is superposedon (b) and (c) for comparison

Fig.3

Channeling pattern resultsFig.4

Frequency (MHz)

22

densation of atomic gas is achieved [9]. Thiselement also has potential in the developmentof lithographic holography technology usingan atom laser. Together these characteristicsmake this element an extremely significantexperimental target.

Figure 5 shows the experimental configu-ration. A frequency-doubled Ti:sapphire laser,pumped by an Ar-ion laser, was chosen as thelight source, providing single-mode (spectralline width of 1 MHz or less) 399-nm lightwith an output power of 40 mW. Two sets ofthese laser light sources were used: one forcollimation of the atomic beam, and the otherfor atomic channeling. A thermal atomicbeam was generated from an Yb atomic beamwith a divergence of 4 mrad, using an ovenand two pinholes. An Yb thermal atomicbeam consists of 7 isotopes, and in this experi-ment, the dominant isotope, 174Yb, served asthe target of fabrication. The oven tempera-ture was maintained at 800 K during deposi-tion (approximately 30 minutes), and thedegree of vacuum in the vacuum chamber wasmaintained at 1.0×10-9 Torr. The generated Ybthermal atomic beam was collimated with theone-dimensional optical molasses to a diver-gence angle of 1.5 mrad, and this collimatedbeam was then used as the atomic source for

atomic patterning. The 399-nm light used for the optical

molasses was adjusted to an intensity of 45mW/cm2 and an interaction length of 15 mm.The laser frequency was detuned by －30MHz from the resonant frequency of 174Yb,and then stabilized with the temperature-stabi-lized reference cavity. The collimated beamwas then channeled into the optical standingwave. The intensity of this optical standingwave was 40 W/cm2, and the beam diameterwas 0.5 mm. The standing wave was generat-ed by using the method described in Section2.1, through reflection of the 399-nm light bya mirror; the frequency detuning was +1.5GHz.

The silicon substrate was placed directlyafter the optical standing wave for depositionof the channeled Yb atoms. After the atomswere deposited, the substrate was removedfrom the vacuum chamber for investigatingthe characteristics of the resultant surfacestructures with an atomic force microscope(AFM).

3.2 Experimental results and discus-sion

Figure 6 shows the results of the Yb atomlithography experiment. Figure 6 (a) is the


Experimental configuration for ytterbium atom lithographyFig.5

AMF image of the substrate surface with thedeposited Yb atoms. The scan range was 10×10 micrometers. Color brightness correspondsto the height of the Yb atomic structure. Thisimage shows that a regular and highly-parallelYb atomic wire structure was successfullyformed. The heights and widths of the wireswere also uniform. The period of the Ybstructure was estimated to be 200 nm, whichapproximately agrees with the intensity periodof the optical standing wave (199.5 nm) usedin atomic channeling. Figure 6 (b) shows thetypical height distribution of the Yb structurein the range of 3 micrometers. Analysis of theheight distribution results reveals that the peri-od of the wire structure is 205 ±9 nm. Thisvalue agrees with the value estimated above(199.5 nm) within an acceptable margin oferror. The height of the structure is approxi-mately 10 nm, and the line width (FWHM) is93 ±5 nm. The period and the shape of thepattern agree with the predicted results. Theseactual results lead to the conclusion that suc-cessful manipulation of atoms enabled the fab-rication of the intended Yb atomic pattern.

Next, we simulated the Yb atomic chan-neling process to analyze the line width(approximately 93 nm) obtained in the experi-ment. For simplicity, the intensity distributionof the optical standing wave is assumed uni-formity along the direction of atomic propaga-

tion, and also assumed that it can be describedalong the axis (in the x direction) of the opticalstanding wave with the following equation:

Here,λ is the wavelength of the light used forthe optical standing wave and I0 is the intensi-ty of the light at the antinodes of the opticalstanding wave. Taking the atomic velocity,the incident position, and the divergence angleas parameters, the channeling processes werecalculated for approximately 400,000 atoms toobtain the expected Yb atomic pattern. Figure7 shows the calculation results. Figure 7 (a)shows the results obtained under the experi-mental conditions described above. The hori-zontal axis represents displacement from anode of the optical standing wave, and the ver-tical axis represents the height of the generat-ed structure. Even under the experimentalconditions, an Yb pattern is expected with aline width of 56 nm. The difference betweenthe calculated and the measured value(approximately 40 nm) is probably attributableto factors such as the diffusion of Yb atoms onthe substrate, jitter in the intensity and fre-quency of the optical standing wave, andvibration of the substrate. The actual linewidth of 56 nm may have been due to imper-fect collimation and the velocity distributionof the Yb thermal atomic beam, and the shape


Experimental results for ytterbium atom lithographyFig.6

24

of the dipole potential of the optical standingwave (i.e., deviation from the harmonic poten-tial).

To determine the factors that must beimproved to narrow pattern line width signifi-cantly, as well as the extent to which it can benarrowed, we continued our calculations withthe collimation of the atomic beam as theparameter. Figure 7 (b) shows the Yb atomicpattern calculated with an atomic beam diver-gence angle of 1.1 mrad. This pattern clearlyindicates that an improvement in atomic beamdivergence reduces the channeling line widthto 42 nm, 14 nm narrower than before theimprovement in collimation. This divergenceangle (1.1 mrad) corresponds to the Dopplercooling limit of the Yb transition resonating ata wavelength of 399 nm. We believe that thisdivergence angle can be achieved by increas-ing the laser intensity of the one-dimensionaloptical molasses shown in Fig.5. The Ybatomic pattern was calculated assuming thatchanneling is performed using an atomic beamwith further improved divergence (0.2 mrad)as the atomic source. Figure 7 (c) shows thecalculated results. The line width of the wirewas 20 nm (the narrowest so far), 36 nm nar-rower than that of Fig.7 (a), representing anapprox. 2/3 reduction. These simulationresults show that improvement in the collima-tion of the atomic source can lead to extensivenarrowing of line width of the atomic pattern

and a corresponding improvement in resolu-tion.

With Yb atoms, an atomic beam with adivergence angle of 0.2 mrad can be obtainedusing Doppler cooling through the forbiddentransition (1S0 - 3P1) of the Yb atom [10]. One-dimensional optical molasses employing theconventional 399-nm transition and this for-bidden transition can thus contribute to experi-mental applications of the atomic beam.

4 Summary

Guidance and velocity selection of an Rbthermal atomic beam were performed usingthe 1D-MOT, leading to the first experimentaldemonstration of high-resolution Rb atomicchanneling. Using channeled Yb atoms, aperiodic Yb atomic wire structure was suc-cessfully constructed on a substrate, in thefirst demonstration of atom lithography usingYb atoms. The technique developed in thisstudy is not limited to the Rb and Yb usedhere; in principle it can be applied to atomssuch as Cr, Al, and In, which are of significantpractical importance in engineering applica-tions. In the future we plan to employ theseatoms, as well as mono-velocity atomicbeams, in order to promote the developmentof techniques of atomic patterning offeringhigher contrast and featuring finer line widths.


Simulated results of the ytterbium atomic channeling patternFig.7


References1 G. Timp, R. E. Behringer, D. M. Tennant, and J. E. Cunningham: Phys. Rev. Lett. 69, 1636, 1992.

2 J. J. McClellant, R. E. Scholten, E. C. Palm, and R. J. Celotta: Science 262, 877, 1993.

3 R. W. McGowan, D. M. Giltner, and S. A. Lee: Opt. Lett. 20, 2535, 1995.

4 U. Dorodofsky, J. Stuhler, Th. Schulze, M. Drewsen, B. Brezger, T. Pfau, and J. Mlynek: Appl. Phys. B 65,

755, 1997.

5 R. Ohmukai, S. Urabe, and M. Watanabe: Appl. Phys. B 69, 123, 1999.



8 C. Salomon, J. Dalibard, A. Aspect, H. Metcalf, and C. Cohen-Tannoudji: Phys. Rev. Lett. 59, 1659, 1987.

9 Y. Takasu, K. Maki, K. Komori, T. Takano, K. Honda, M. Kumakura, T. Yabuzaki, and Y. Takahashi: Phys.

Rev. Lett. 91, 040404, 2003.

10 T. Kuwamoto, K. Honda, Y. Takahashi, and T. Yabuzaki: Phys. Rev. A 60, R745, 1999.

OHMUKAI Ryuzo, Ph. D.

Senior Researcher, Advanced LaserScience Group, Basic and AdvancedResearch Department

Atom Optics, Quantum Electronics

WATANABE Masayoshi, Ph. D.

Research Center Supervisor, KansaiAdvanced Research Center, and GroupLeader, Advanced Laser ScienceGroup, Basic and Advanced ResearchDepartment

Laser Engineering, Atom Optics

26 Journal of the National Institute of Information and Communications Technology Vol.51 Nos.1/2 2004

2-2 atom optics and atom lithography - nict - トップ...

Documents