Three-dimensional recording by femtosecond pulses in ... · Figure 2. CCD images of the interference region at the tip of Fresnel biprism (a, d) with intensity cross sections across

Three-dimensional recording by femtosecondpulses in dielectrics

Saulius Juodkazisa,b Toshiaki Kondo, Vygantas Mizeikisa,c,Shigeki Matsuo, Hiroyuki Muratab, and Hiroaki Misawa,aCore Research for Evolution Science & Technology (CREST),

Japan Science & Technology Corporation (JST),bDepartment of Ecosystem Engineering, University of Tokushirna,

2-1 Minamijosanjirna, Tokushima 770-8506, JapanC5atellite Venture Business Laboratory, University of Tokushima,

2-1 Minarnijosanjirna, Tokushirna 770-8506, Japan

ABSTRACTFabricatioii aiid characteristics of two—dimensional (2D) and three—dimensional (3D) periodic structures, recordedill the bulk of SU8 photoresist film by multiple-beam iiiterfererice is described. Multiple beams (up to iiine) weregenerate(I by a diffractive 1)eam splitter. Recording was performed by ultrashort laser pulses with temporal wi(Ithof 140 fs (FV\VM) and central wavelength of 800 nrn, derived from a Ti:sapphire laser. Intensity-depeiideiitphotoiiiodificatioii of the photoresist was due to si11g1e—)hoton as well as multi—photon (two and three) absorp—tion. After the development, the exposed resist films contained free-standing 2D and 3D periodic dielectricstructures with uiiexposed exposed regions removed by the development . Detailed examinatioii of the sampleshas revealed close resemblance between their structure and the light intensity distributions in the multiple-beaniinterference fields, expected from the numerical calculations. Quality of the sanhl)les recorded by a single-photonabsorption was lower than that of that of other samnples, in particular due to poor developrnemit quality. Themriicrofabrication method used in this work appears to be a suitable for obtaining photonic crystal temni)lates.

Keywords: Photonic crystals, holographic recording

1. INTRODUCTIONDevelopment of photonics is progressing at a steady pace, and currently is capable of delivering sophisticatedoptoelectronic devices, imnportance of which matches that of microelectronic devices.1 2 Amnong the main tasksof photonics is spectral and sl)atial manipulation of electromnagnetic radiation. Multi-dimnensional diffractiongratings riiay be useful in realizing these tasks via dispersive and photonic band gap (PBG) effects.3 Var-jolts techniques for the diffraction grating fabrication exist, but two-dimnensional (2D) and three-dimensional( 3D) gratings can be most cOnveniemitly recorded holographically in appropriate photosensitive media, such ascrystalline material,4' liquid crystals,6' 7 or plioto-thermo-refractive (PTR) glass.8

Most of the earlier reports on holographic fabricatiomi have concentrated on 1D structures, while the fabrica-tion of 2D or 3D periodic structures has received somewhat less attention. -In 1990, Buriis et al.9 demonstratedan interference technique, which allowed direct visualization of the 2D optical field intensity and a periodicarrangement of micrometer-sized polystyrene particles in the water assisted by optical forces. In this work,the periodically arranged particles were termed "optical crystals" . Berger et al. have fabricated 2D hexagonallattice iii a thin layer of photoresist by interference of three beamns at A =325 nni.1 The resist layer wassubsequently used as the mask for reactive ion etching, and fabrication of 2D PhC in GaAs with the depthof 3 tm. The lattice period was 2.66 /tnl, but in principle it could be downscaled to = 217 nmn. Similartechniques have been also used by others."

Further author information: (Send correspondence to H. M.)H. M.: E-mail: mnisawa©eco.tokushimna-u.ac.jp, Telephone: (+81 88) 656 7389, Fax: (+81 88) 656 7598.

Invited Paper

Photon Processing in Microelectronics and Photonics II, A. Piqué, K. Sugioka, P. R. Herman,J. Fieret, F. G. Bachmann, J. J. Dubowski, W. Hoving, K. Washio, D. B. Geohegan, F. Träger,

K. Murakami, Editors, Proceedings of SPIE Vol. 4977 (2003) © 2003 SPIE · 0277-786X/03/$15.00

94

Downloaded from SPIE Digital Library on 01 Mar 2012 to 136.186.72.84. Terms of Use: http://spiedl.org/terms

a) b)

Figure 1. Interference of two cohereiit plane waves (a), use of diffractive beam splitter for obtaining multiple planewaves (b).

Fabrication of 1)hOtOlliC crystal (PhC) structures by laser interference was first reported by Campbell etal. .12 By usiiig 355 mn nanoSecond laser pulses, they obtuiied 3D fcc structures with a lattice constant of922 iiin in pliotoresist having a thickness of several tens of micrometers. 'ftigoiial structures were also obtainedfor different angles between the beams. By using the fabricated structure as a template, they obtained aninverse periodic structure in Ti02, which is known to a have higher refractive index thaii the PhotoreSiSt.Holographic fabrication of PliCs was also perforriied in photopolyrnerizirig resins ,13 where light irradiationleads to solidification of liquid polymers. Five overlapping waves were used to form 3D triangular lattice withhexagonal structure. The same authors also fabricated 2D and 3D PhC by laser interference using a simpleSetup l)ased on a diffractive beam splitter (DBS). The initial laser beam was split by a DBS, arid selected beamswere gathered by two lenses. This setup is very simple and ensures equal optical path lengths of interferingbeams, which is helpful whemi working with short laser pulses.

The present work focuses on the holographic recordiiig of 2D arid 3D gratings in photoresist films usinginterference of ultrashort laser pulses, obtaiiied from a single pulse by a DBS, arid employing non-linear absorp-tion mechanisms. The experimental setup used is based on the DBS in the course of experiments in photoresist,performed with very similar optical setup. This paper is organized as follows. The principles on which lasermicrofabrication by holographic (or interference) recording are based, are described in Sect. 2. Numerical sun-ulations of interference patterns created by five laser beams are given, amid applicability of the holographictechiiique for the fabrication of 3D photonic crystal templates is examined using numerical modeling iii Sect. 3.Sect. 4 describes the experimental setup and details of the fabrication process. The main results which describegeneration of various light interference fields, amid recor(ling of four-beam and five-beam interference patternsin photoresist films are presented in (Sec. 5).

2. PRINCIPLES OF RECORDING BY INTERFERENCEThe basic prilicif)les of holographic laser fabrication technique are illustrated by the sketch in Fig. 1 .Interferenceof two coherent plane waves 1 and 2 creates one-dimensionally (1D) periodic intensity pattern, 1(x), whichdepends on the coordimiate x as

1(x) o Ii + 12 + 2I . 12 C05+12] , (1)

A =2sin(O/2) ' (2)

where Ii and 12 are the intensities of the two waves, and A is the spatial period of the pattern, which dependson the laser wavelength, A, and the angle between the wave propagation directions, 0. At intermediate angles0 71/4, A becomes comparable to the laser wavelength, reaching its upper limit A = A/2 at 0 = r (counter-propagating beams). The phase of the modulation pattern is comitrolled by the phase difference between the

Proc. of SPIE Vol. 4977 95


Cl)

../•:::. •.1 :

-B

I.: :

: .. . . cI): : •

a)

4

Frequency (l4tm)-2 0 2

Figure 2. CCD images of the interference region at the tip of Fresnel biprism (a, d) with intensity cross sections acrossthe fringes (b, e), and corresponding spatial frequency analysis by FFT given in (c, f) . The duration of pulse incidenton the biprism (with 1100 angle at the tip) was about 130 fs (a-c) and 70 fs (d-f), respectively.

two waves, 012. By placing a photosensitive medium in the interference field, structure with periodicallymodulated optical density or refractive index can be recorded, arid later used as a diffraction grating. It ishelpful to note here, that if the recorded grating is subsequently illuminated by, for example, wave 2, one ofits first-order diffracted waves will reproduce the wave 1, including the information about its phase. Recordingof 1D diffraction gratings by two plane wave interference is therefore the simplest kind of holography. Thissimple method can be extended to produce much more complex periodic 2D and 3D patterns, if many wavespropagating along carefully selected directions with different phases, are allowed to interfere. The setup whichenables to achieve interference between multiple waves is sketched in Fig. 1, arid will be described in ixioredetail in Section 4. Such arrangement uses a diffractive beam splitter (DBS) which produces multiple, mutuallycoherent laser beams from a single input beam. The beams are subsequently focused by a leiis into the bulkof photosensitive material. All of the beams deliver to the focal spot region of the focusing optics nearlyplane waves, )ropagating at different directions. Thus DBS together with focusing optics automatically ensuresvanishing difference between the optical paths of the beanis, which is especially important if ultra.sliort laserpulses are used for the recording.14 Indeed, an overlap of two ultra-short coherent pulses is determined bytheir longitudinal dimension, l = er,, which is about 30 tni for a 100 Es pulse. Hence, the width of thespatial overlap is w = cr/ sin(O/2) for 2-beam interference. In contrary, for diffracted pulses the I)ulse froiit(an intensity front) is riot perpendicular to direction of propagation (the wave front), i.e. , the pulses are tilted,arid the region of overlap is given w = d/cos(O/2) , where d is the cross-section diameter of the beam. Thus,sub-mm-scale patterns can be routinely produced using even a tight focusing o1)tics (with numerical aperture,NA > 0.5).

Figure 2(a) demonstrates interference patterns imaged on CCD camera for two interfering pulses, whichwere produced by passing a pulse via Fresnel bi-prism. Hence, this corresponds to interference of two pulseswhose wave fronts (and pulse fronts) are perpendicular to the propagation direction, i.e. not tilted. The widthof the light intensity distribution (b) was obviously dependent on the pulse duration. FFT analysis of intensitypatterns (c) was used to determine the period, which was found A = 1.06 tni and 1.11 pin for 150 and 50 fs

Distance (Jim)

-2 0 2

Frequency (1/Jim)

96 Proc. of SPIE Vol. 4977


pulses. This is very close to the expected value of 1.094 pan defined byrefractive iiidex 1.453 at 800 urnwavelength of 1 10° bi-prisin.

Use of short puise exCitatli)11 together with beani focusiiig allows to achieve power densities, sufficient for in—ducmg effective photornodificatiori of initial material, optically transparent iii the linear regime, via higher orderiioii—linear absorption effects. As in the two—wave case explained al)ove, i)(riods of multiple—wave interferencel)atter15 (lepend crucially on their mutual convergence aiigles, 19 . The latter can be controlled by changing theinter—beam distance prior to the focusing, or the focal length, which in turn requires to use lenses with variousfocal lengths and numerical apertures (NA) . Axial depth of the iiiterference region, and hence of the recor(ledstructure. is proportioiial to the waist size of the interfering beanis and inversely proportional to the aiigle 0.For short i)ulses, i.e. , when l becomes smaller than the diameter of the beams, lateral size of the interferencearea may beconie soniewhat reduced. Iii practice, structures with modulation periods below one inicronieter.and containing tens of periods along all three directions are routinely achieved using fairly tight focusing opticswith NA> 0.5. Coiriplexity of the interference patterii depends on the iiumber of participating waves, whichcan be changed by o)ening an(l closing various beams with a mask, as illustrated in Fig. 1 (b) . Depth of theintensity modulation can be controlled by adjusting the intensities of the interfering l)eanls. Differeiit kinds ofphotoseiisitive materials can be u5ed for the structure recording.

3. MULTIPLE-BEAM INTERFERENCE PATTERNS3.1. Numerical simulationsIii this sectioii we will use simple numerical calculations to deirionstrate the patterns that can be generated bymultiple—beam interference, and to outline the main expectatiolis for the experiments described later.

Iii general, spatial light intensity (listribution in the multiple bearri interference field can be exj)ressed asfollows:

n 2

), (3)

where E , k , are amplitude, wavevector and phase of the of the i-tii field (beani), w is the frequency, r theradius—vector, n. the iiurnber of beams, arid t is the time.

Figure 3 shows isosurfaces of the 3D light intensity distributions created by five interfering plane waves,calculated according to Eqn. 3. Since the structure recording will be done through a non-linear absorptiomi,which usually occurs above well-pronounced threshold intensity, one can expect that at a certain irradiationintensity the isosurfaces del)icted in the figure can be transformed into the boundaries between exj)Ose(l andunexposed regions. The beam configuration used to obtain these results is illustrated inn the sketch shown inFig. 1 (b). At the entrance to the lens four of them are aligned on the corners of a square, while the fifth beamis located in its center. We have found that for aclueving maximum modulation amplitude in this configurationthe intensities of all four cornier beams must be equal. At the same time, intensity of the central beam must beequal to the total intensity of the corner beams, i.e. , four times higher. The ninnerical calculations have alsorevealed profound role of the phiaes of the waves on the structure of the interference fields, which is illustratedby cases (a,b,c) in Fig. 3. For the illustration purpose we show the case where phases of two beams locatedon the opposite corners of the square, begin to differ from the phases of the other beams. When phases of allbeams are equal (Fig. 3 (a)), the periodic structure has body-centered tetragonal (bct) lattice with a single"atom" (a high-intensity parallipeped, extending along in the z-axis). To obtain a body-centered cubic (bcc)lattice, usually, a sharper focusing is needed. When phase difference up to ir/2 is introduced for two beams(b,c), the periodic structure exhibits a smooth change, whereby the same bcc (or bet) lattice is retained, butan atomic basis, consisting of two high-intensity ellipsoids is formed. After a more detaile(1 inspection suchstructure can be identified as a diamond structure (if bcc structure is formed by all equal phases beams). Thetransformation towards the diamond structure is interesting, since the latter is a highly promising candidatefor the application in photonic crystals (PhC), and is difficult to obtain by most of the currently known PhCfabrication techniques. It is also obvious from the figure, that diamond structure will have a framework of cross-linked unexposed regions, and a framework of cross-linked exposed regions. This layout provides full access into



00

X(X) 30

NI

-2 -2 X (A.)Y (04

90

jsoI°I°90

80

lit

I°

-2 -1 0 1 2x (oj

Y (04 -2 -2 X (A)

90

Sc.

0

so

I°

70

65

60

55

(4c)2

d.cN

NY (A) -2 -2 X (A)

90

70

60

55

—

160 2a -2 -1 0 1 2I X(A)55

Figure 3. Spatial light intensity distributions generated by interference of five plane waves- The complex amplitudesof the central beam in all plots is 4Ioe°, while corner beams have different amplitudes: (a) all Ioe'°; (b) Ioe4,be'; Joe; IocC°) ; (c) be1, Ioe, I0cC°, Ioe°. The plots show iso-surfaces drawn at different intensities:27 for (a), 45 for (b), and 52 for (c). The coordinates are given in units, normalized to the wavelength of the recordinglight, A (note the difference between the scale of z-axis and x,y axes.). Maximum beam convergence angle °ajr 600was assunmed for the calculations.

the structure for the developer solution. Moreover, after the development such structure is rnechauically stable,arid can be used as a template for the infiltration by high refractive index materials.

The above described example shows only a few simple possibilities existing among great variety of the latticetypes that can be generated by multiple-beam interference. More examples will be given in the following sectionsdevoted to the description arid discussion of the experimental results.

3.2. Suitability for photonic crystals applicationsIii PhCs periodic modulation of the dielectric constant leads to the formation of photonic baudgaps PBG, orspectral ranges in which light waves cannot penetrate the PhC.3 Since iu general the PBG central wavelength



a) fcc lattice of dielectric ellipsoidsn=3.5 in air (n=1)

x

b) diamond lattice of dielectric ellipsoidsn=3.5 in air (n=l)

Figure 4. Calculated PhOtonic baud diagrams for silicon ellipsoids in air, forming (a) simple fcc, and (b) diamondlattices. The frequency axis is given in normalized frequency units wa/(2rc), where ' is the frequency, a is the latticeperiod, and c is velocity of light. Elongation factor of the ellipsoids along tIme z-axis is 1.5.

is comparable to the modulation period, PhCs must have dielectric lattices with sub-micrometric periods andcontain dielectric features of even smaller size. With these requirements, fabrication of 3D PhC structures mayprove too difficult. Therefore much attention has been focused recently omi search for novel PhC fabricationtechniques which would enable easier fabrication of 3D periodic dielectric rnicrostructures from bulk materials.As noted in the previous Section, in this regard holographic recording may provide interesting capabilities dueto the possibility to obtain field distributions that have fcc amid diamond type structures. The latter one isespecially interesting, since it favors a large PBG, which opens at a refractive index contrast as low as 2 to 1.15In this section we shall briefly examine possibilities to use holographic mnicrofabrication techniques in PhCfabrication. Intensity patterns shown in Fig. 3 (b,c), successfully transferred to SU8 photoresist, will create fccand diamond lattices with a single ellipsoid (for fcc) or two ellipsoids (for diamnoiid), filled by air, at each latticesite. The longer axis of ellipsoid will be oriented along the z-axis direction, the elongation factor will be about1.5-2.0, depending on the beam convergence angle at the focal spot. In the x —y plane ellipsoids will touch eachother, such that empty regions (as well as unexposed regions) will be cross-linked. \'Ve used this description fornumerical modeling of the photonic band diagrams of the fcc ari(l diamond lattices.

It must be noted iii advance, that bare framework of unexposed photoresist regions in air can riot exhibitcoml)lete (direction-independent) PBG under any circumstances due to its relatively low refractive index contrast(1.5 to 1 ). Despite this circumstance, it is interesting to investigate whether or not such structure can be used asa template for infiltration by high refractive index materials. This option allows one to raise the refractive indexcontrast in the initial structure amid has attracted much interest in relation with infiltration of self-organizingsystems, like opals.16 Semiconductor growth and processing technology provide the procedures required foreffective infiltration of temj)lates by high refractive index semiconductors, e.g. ,chermcal vapor dpo on17If the framework of infiltrated material is mechanically stable and self-supporting it is possible to remove theoriginal template, thus further raising the refractive index contrast.

We assume that all these techniques are in principle applicable for holographically fabricated structures,

00.5

0

0N

—

I

UL FXWK XU L TXWKwavevector wave vector



Figure 5. GRENOUILLE (time x spectrum) traces of fs-laser (Tsunami oscillator) pulses: at the output of oscillator(a), just before oil-immersion objective lens (b), and after the focus, which was placed inside cover glass (c). Pulsedurations at FWHM retrieved by FROG algorithm are shown on each image. Horizontal size of the images correspondsto 1779 fs; the vertical -33.7 nm.

and that air VOi(IS in the above described fcc aIlil diaIflOn(I teiriplates can be filled by sihcoii (n 3.4). aiidsubsequently removed from the structure. Photonic band diagrams for these cases were calculated using a freelyavailable software package (MIT Pliotonic Bands), that implements a targeted iterative eigeiisolver of Maxwell'sequations on a plane-wave basis.18 Two representative results of the calculations are shown in Fig. 4 for an fccand a diamond cases. Perhaps the most important diflèrence between these two plots is absence of the PBG inthe first case and presence of the PBG in the second case. Both results are not very surprising, since it is known,that fcc (or opal) lattice compoSed of spherical dielectric elenierits can not opeii a PBG even at the refractiveindex contrast as high as 3.5 to 1. On the other hand, similar diamond lattice ha.s a complete PBG (about 8% ofits central frequency) when the refractive index contrast exceeds relatively low value of 2 to 1.15 However, in thecase regarded here dielectric elements are not spheres, hut ellipsoids. Thus the result shown in Fig. 4 illustratesthat while fcc lattice does not acquire additional robustness at these lattice parameters, the diamond latticeretains its capability of opening a complete PBG. This indicates that holographically recorded structures inprinciple can be used as templates for the PIiC fabrication. However, much broader investigations are reqUire(Ifor full assessment of their suitability. For exaniple, we have found, that even for diamond structure, PBGtends to disappear when ellipsoid elongation factor approaches 2, hence the recording beams must converge atlarge angles, which may not always be possible to achieve. On the other hand, fcc lattice may still prove usefulat different orientation of the ellipsoids , for example when their elongation is in the (1 1 1) direction, and thestructure becomes similar to the diamond structure.

4. EXPERIMENTAL DETAILSThe layout of the optical setup used for the generation of multiple beam interference patterns and recordingof periodic structures is described in detail in the literature.19 Below we give brief descriptioiis of the niainfeatures of the optical system used, and of the samples and their J)rocessing.

4.1. Femtosecond light sourceThe light source for the experiments was a femtosecond Ti:Sapphire laser (Tsunami) with a regenerative amplifier(Spitfire) , from Spectra Physics, which was tuned to the 800 nm wavelength, anid produced pulses of T =130 10 fs temporal length (FWHM) at the pulse repetition rate of 1 kHz. Prior to the experiments, temporalpulse spreading after passage through various dispersive elements (lenses, DBS) was evaluated. The temporalpulse length at different locations along the beanmi optical path and at the focus point was mnmeasured by theGRENOUILLE apparatus20 froni Swamp Optics, which recorded a convolution of temporal and spectral pulseshapes, and the pulse length was later retrieved using the frequency-resolved optical gating (FROG) algorithnnfrom Femtosecond Technologies. For the measurements of the pulse duratiomi at the focal spot , aim additionalsolid immersion lens, not present during the structure recording, was employed in order to couple the stronglydivergent beam into the GRENOUILLE. Figure 5 shows an example of the initial composite pulse imageobtained by the GRENOULLE, and the subsequently extracted pulse durations at the output of the annplifier,



3sL Lers resistRetarder

Figure 6. Optical setup for holographic recor(hng of lflUlti(IifllellsiOflal periodic structures. Cover glass plates, tilted atvarial)le angle () were used as J)hase retarders of the selected beanis.

after passing through the optical aligririierit and other elements before the eritraiice to the microscope objectivelens, arid after the oil—immersion objective lens with NA = 1.3. Ii all cases FROG error was lower than 2%.Using this al)l)roach we have lneasure(1 the pulse duratiou to be 160 15 fs for NA =0.75 objective leiis used inthe structure recording. The lenses and other parts of the optical setup introduce a pulse temporal stretching(ille to dispersion, since pulse (I1.lration at the outpnt of the regenerative amplifier is 130 fs.

4.2. Generation and control of the interference fieldsThe key ideas aIX)lit operation of the optical setup have been already outlined in Section 2. Its more detailedsclieniatic I)icture is depicted in Fig. 6. The DBS divides a single iIiI)Ut laser beam into several components,which are collected amid focused into the material by a pair of lenses. The nUmfll)er of interfering waves arid theirrelative phases are controlled by an aperture and phase retarder Plates placed between the lenses. The aperturehas a mmumnber of openings, which select time required beams arid block the mniwarited ones. Time phase retarderI)lates are lieces of microscope cover glass, introduced llit() the beanis at variable orientation angles a. A smallpart of the incident beams is split off by a larger cover glass plate, and is focused on a CCD for observation ana monitor. For inspection of 3D interference patterns the CCD camera was moved along the optical axis, anddifferent cross sectional views of the structure were examined. The lens used for focusing the of the beams intotime sample is a dry objective lens (Olympus UApo) with 2O magnification and numerical aperture NA = 0.75.

4.3. Samples and their exposureFilms of negative photoresist SU-8 (Microlitliography Chemical Corp.), which is optically transparermt for wave-lengths A > 400 nm, were used for the recordirmg. Time I)hOtOresiSt was spimi-coated on cover-glass plates, andimad thickrmess of about of 5 — 6 /mm. The samples were pre-baked prior to time optical exposure. After the expo-sure, the samnl)les were post-baked in order to enhance photo-initiated crosslirmkirmg reaction. The developmentprocedure resulted in removal of the exposed regiomms fromrm time sample. Typical exposure pararmieters use(I inthe experimermts were: average laser power before the DBS 0.21 W, at the focal spot 1.67 mW (in the case offour beamrms). For the 150 rnm diameter spot and 'r = 150 fs, this corresponds to time 2.4 mnJ/crn2 pulse energydensity, and 1.6 x 1010 W/cnm2 power dermsity. The exposure timne was varied within time 5 —90 secomids imiterval.

The fabricated samples were inspected by a scanning electron microscopy (SEM), for which a timirm layer of Au(< 20 nm) was coated by sputtering on time samrmple.



Figure 7. Two-dimensional CCD images of the interference patterns created by seventeen interferiiig beams, arrangedinto a cross at the entrance to the focusing lens. Phases of nine beams belonging to one bar of the cross were smoothlychanged and corresponding intensity distributions are shown on the top row of the images. For a scale reference,interference patterns of four beams with no phase difference among them are also shown in the bottom row of theimages.

(P1,27t/2 (34O (21.2.3.4 misaligned•*****,.•4***,* 0 **•#*i** *

*q..,

, *4(b) (c) (d)

Figure 8. Illustration of phase control of the interfering beams by tilting cover glass plates (thickness 0.2 mm, refractiveindex 1.5). (a) shows the dependence of the beam phase on the glass tilt angle, the experimental configuration usedis depicted in the inset, (b-d) CCD images of interference fields for various beam phases: (b) phases of all beams areequal, (c) phases of beams 1 and 2 shifted by ir/2 with respect to those of beams 3 and 4, (d) when the optical pathdifference between the beam pairs (1,2) and (3,4) exceeds the coherence length. All CCD images were taken at the samemagnification. Insets in (b) and (c) shows calculated intensity patterns for those cases.

5. RESULTS AND DISCUSSION5.1. Manipulation and monitoring of interference fieldsThis section presents experimental examples of interference fields with 3D intensity distributions more complexthan those inferred from the calculations in Section 3.1. Figure 7 shows interference patterns of 17 beams,aligned along the diagonals of the square shown in Fig. 1 (b) . If along one bar of the cross the phases ofthe beams are smoothly and simultaneously varied, while along the other bar they remnain fixed, time intensitydistribution exhibits smooth transformations. The images shown in the figure were captured with a CCDcamera, and phases of the beams were changed using the cover glass retarder elemnent. Time same figure alsoshows niuch simnpler patterns generated by four-beam interference. All of the displayed patterns have distinct

;*:I

. is, n U,mm?*:

ri;n::cE:i :i:rIJ:u:i:rh

' " " A' " " a"

iu*fr: *tU flUU

(P1.2.3,4°

(I)a)

C)a)09.

C)C

.C0a)(1)(U

0

* . 'I::: ::i•

•••* ::: • ::: ::c•::.::

••:: ••• • '4'¶ :.4 •.. : ___0*00000•0 0

Angle of incidence, a, (degrees)



, (c)Cl) 200C

I b0:., :d,

Distance (tm)

Figure 9. Calculated intensity profile (a) and fabricated (b) patterns of 6-beams interference. Scale in (a) is in wavelengthfor a 0/2 = 69°. (c) Cross-section intensity profile along a dashed line in (b). The width of the intensity spikes is 220 am(FWHM).

Figure 10. Perspective and top-view SEM images of periodic structures in SU-8 resist, (a) hole structure recorded bytwo consecutive exposures (beam pairs consisted of beams located on each diagonal of the square), (b) rod structurerecorded by a single four-beam exposure. Focusing corresponded to a Oair/2 32° angle. Scale bars are 2 tm.

periodic structure containing fine features, and can be recorded in a photoresist or other material. Four-beamconfiguration is actually more suitable for revealing the importance of phase control. This is illustrated by dataand hnages shown in Fig. 8 (see explanations in the figure caption).

One may ask whether or not the images of interference fields taken by the CCD (Fig. 6) resemble the intensitydistribution at the focus of the lens, where actual fabrication takes place. Of course, both distributions are notexactly the same; in the CCD image smallest features will be smeared by diffraction and therefore not visible.Nevertheless, larger features can be correctly imaged on the CCD, and for them one can expect close matchingbetween the CCD image and time intensity distribution at the focus. Fig. 9 shows comparison between thecalculated light intensity pattern created by six-beam interference and time pattern recorded in time photoresistafter time development . As is evident from the figure, the size (about 220 nm) and shape of the features in bothimages match closely. Time recording light wavelength was 800 urn, and the objective lens used had NA =0.25.

5.1.1. Recording of periodic structures in photoresistIn holographic fabrication experiments we start from simpler 2D structures, that can be recorded by four beams,arranged on the corners of a square. By using four beams, two opposite pattern types can be recorded. One

A



Figure 11. (a) Calculated light intensity distribution for four-beams interference at the mutual convergence angle0 = 69° , (b-c) SEM images of the SU8 photoresist structures recorded by four beams (scale bars are 3 and 1 .5 m,respectively). The inset in (c) shows the geometry of the resist illumination.

of them contains periodically arranged holes in the resist film, while the second one consists of periodicallyarraiiged rods. Figure 10 shows the examples of two such I)atterllS. The hole i)atterfl was recorded by acombination of two consecutive two-beairi exposures, and the rod pattern was recorded by a single four-beamexposure. The structures exhibit high degree of order and are reminiscent of some 2D PhC structures reportedill various i)1lb1iCati01lS. Despite the fact that low refractive index of the pliotoresist will prevent occurrence ofsignificant PBG effects, these structures may still be useful as templates for processing of high refractive indexmaterials. It is helpful to rermnd here, that various lattices can generated by carefully controlling the phases ofinterfering beams. For 2D interfereiice lattices these possibilities are considered in detail in the literature,2' aridfor 3D structures some of them were outlined in Sect. 3. In the case of the 2D samples shown in Fig. 10, theirstructure, size and shapes were found to be very close to the ones expected from the calculations. For example,the lattice period (1 ,am) was found to be exactly equal to the expected value of A/('s/sifl(Oajr/2)) 1.02 jim.In fact, it should be possible to record the structure shown in Fig. 10 even if the phase difference is larger thanthe coherence length (see Fig. 8 (d)).

Figure 11 illustrates further the good agreement between the light intensity distribution and the experimen-tally obtained structures, and demonstrates excellent quality of the latter. The calculated intensity distributionshown in (a) clearly reveals rectangular arrays of nearly cylindrical high intensity regions. After time exposureand development only the array of rods , corresponding to exposed regions is left (b,c). Close-up SEM imagesshown iii the figure also clearly reveal wavy side surfaces of the rods. These features are result of interfer-ence between the incident and reflected waves during the fabrication and their period is exactly given by theA/(2n) D 270 nnm, where refractive index of resist is about n = 1.5.

Turning to time case of 3D holographic structures, we remind here that material in Sect. 3 demonstratedtheoretically time possibilities to tune the structure type from simple fcc to diamond by controlling the phasesof the recording waves. Jim practice it is not easy to demonstrate such transition because in fact phases ofthe recording beams are not constant over the entire lateral cross section of time irradiated area. Despitethese difficulties we have succeeded in recording of dianmond-like structures. This result is shown in Fig. 12. Theinmages were taken using optical cormfocal nmicroscopy. Unlike imm the previous experimrmermts, this tirrme the recordingwas done with fenmtosecond pulses having 400 nm central wavelength. At 400 nm single-photon absorption isalready significant, and due to this reason it was difficult to develop the exposed structures evenly. A 400 nmwavelength had been chosemm for exposure imm order to ensure a better cross-linking of the structure, however timeexperimermt had proven that, in fact, a fummdamental absorption caused too extensive cross-linking. Either oneor two-photon exposures are used time dosage and intensity as well as time resist spin-coating and developmentprocedures simould be optimnized for a better fabrication.

Unlike the dianmond structure discussed above, most of the holographically fabricated mnicrostructures arerecorded via two-photon or multi-photomm nonlinear absorption processes. Time mmommlinearity allows to expectsuccessful recordimmg of structures with smallest feature size below the "irmterferenmce-limnited" resolution of time

C.5 -V() .1 .1



(a)

Figure 12. Recording of diamond-like structure by five-beam interference at the wavelength of 400 nni, where single-photon absorption is significant (a) geometry of the experiment, (b) cross-sectional views of the structure taken by opticalconfocal microscopy. The top image depicts cross-section along the z-axis direction, and the bottom images show x —yplane cross-sectional images taken at z coordinates which correspond to different atomic planes.

multiple-beam optical field. Indeed, such resolution increase is a common consequence of absorption nonlinearity,arid is often encomitered in laser microfabrication of transparent dielectric materials.22 Experiments, whichexplore this possibility are currently being I)erformed in our laboratories, and the results will be reported indetail later. Preliminary confirmation of this possibility has been already obtained using four arid six-beamrecording of 2D patterns. Using time beam configuration, shown in Figure 8 we expect to be able produceinterference patterns similar to those shown in Figure 7. For example, the bottom row of images in Figure 7produced 2D interference patterns with period 1 ILmn. Six-beam pattern shown in Fig. 9(c) demoiistrates areproducible features with size of 0.22 im way below the limit of diffraction for a given optics.

To conclude this section, we note that holographic recording can be useful not just for microstructuring ofphotoresist , but also for a broad range of other materials. Recently, we have found that thick phase gratingscan be recorded by ferntosecond pulses at 800 mu wavelength in photo-thermo-refractive (PTR) gl&ss23 by four-beam interference patterns.24 Time mechanism, responsible for the photosensitivity of time PTR glass (whicimhas an Ce3+ absorption at time wavelength of 308 nm), is most probably related to the nonlinear absorption andionization of time glass mimatrix, which provides electrons, necessary for a latent image formation.25 Structures,recorded in PTR glass amid SU-8 resist film, were practically identical. Readout of the holograms, recorded inglass indicated that they are phase holograms.24 Their estimated parameters (refractive index modulation,period, amid thickness) were consistent with time values calculated from time geometry of time experiment and fronmKogelmmik's coupled wave theory.26

6. CONCLUSIONSWe have demonstrated holograpimic recording of 2D and 3D periodic dielectric structures in the filnms of SU-8 plmotoresist usimmg femntosecond laser pulses amid both limmear and mmon-linear absorptiomi nmecimamiismns. Thistechnique enables sinmple recording of structures in a single procedure, which does riot require sample or laserbeam translation. We have also denmonstrated that with mmmultiple laser beams/pulses, it is possible to createvery complex field patterns. Sonic of these patterns appear to be promising canmdidates for the muicrofabricationmof photonic crystal templates, since their structure can be relatively easily tuned from 3D fcc to diamond typeby adjusting time phases of commstituemit beanms. We have demonstrated simple implementation of time facilitieswhich allowed to achieve sufficient degree of control over the number of interfering waves, their amplitudes,phases, convergence angles arid other pararmmeters, amid enabled successful recording of diammiond structure.

We have also denmon.strated holographic fabrication of 2D periodic structures, which consisted of circular holesor rods, amid had very high structural quality. Time holographic microfabricatiomi technique appears to be versatiletool for bulk microstructuring of photoresist as well as other materials. In addition, it also promises to besuitable for surface mnicrostructuring, highly demnanded in the applications based on high-harmonics generation,

x

Y



local field erthanceinent, electron field emissioii al)I)liCatiOllS, etc. Furthermore, is spatial resolution call beenhaiiced by absorption noulmearities and (in )l1otopolymerizi11g niaterials) by rOper choice of polymerizationconditions. Recent reports indicate, that pattern auto—amplification effects occur under rOier conditions.27 Infuture the interference teclniique may gain further advantages if interactions between the recording waves andtheir conhl)onents diffracted for the fabricated structure caii be exploited. Such transient feedback effects mayfurther decrease sizes of the peak intensity regions.

6. 1 . AcknowledgmentsThis work was partially supl)orted by the Satellite Venture Business Laboratory of the University of Tokusliiina.Also, we a grateful to Dr. 0. Efimov for fruitful discussions.

REFERENCES1. E. Yablonovitch, "Inhibited spontaneous emission in solid-state physics and electronics," Phys. Rev. Lett. 58(20),

pp. 2059—2062, 1987.2. N. Bloembergen, "From nanosecond to femtosecond science," Rev. Mod. Phys. 71(2), pp. S283—S287, 1999.3. E. Yablonovitch, "Inhibited spontaneous emission in solid-state physics and electronics," Phys. Rev. Lett. 58,

pp. 2059—2062, 1987.4. N. Korneev, H. Veenhuis, K. Buse, and E. Krätzig, "Thermal fixing of holograms and their electrically assisted

development in barium calcium titanate crystals," J. Opt. Soc. Am. B 18(11), pp. 1570—1577, 2001.5. M. A. Noginov, C. B. Loutts, K. Ross, T. Grandy, N. Noginova, B. D. Lucas, and T. Mapp, "Role of traps in

photocoloration of Mn:YA1O3," J. Opt. Soc. Am. B 18(7), pp. 931—941, 2001.6. C. M. Titus, J. R. Kelly, E. C. Gartland, S. V. Shiyanovski, and J. A. Anderson, "Asymmetric transmissive behavior

of liquid-crystal diffraction gTatings," Opt. Lett. 26(15), pp. 1188—1190, 2001.7. V. P. Tondiglia, L. V. Natarajan, R. L. Sutherland, D. Tomlin, and T. J. Bunning, "Holographic formation of

electro-optical polymer-liquid crystal photonic ctrystals," Adv. Mater. 14(3), pp. 187—191, 2002.8. 0. M. Efimov, L. B. Clebov, and V. I. Sniirnov, "High-frequency bragg gratings in a photothermorefractive glass,"

Opt. Lett. 25(23), pp. 1693—1695, 2000.9. M. Burns, J.-M. Fournier, and J. Golovchenko, "Optical matter: Crystallization and binding in intense optical

fields," Science 249, PP 749—754, 1990.10. V. Berger, O.Cauthier-Lafaye, and E. Costard, "Photonic band gaps and holography," J. Appi. Phys. 82, pp. 60—64,

1997.11. P. Visconti, C. Turco, R. Rinaldi, and R. Cingolani, "Nanopatterning of organic and inorganic materials by holo-

graj)hic lithography and J)lasma etching," Microelectron. Eng. 53, pp. 391—394, 2000.12. M. Campbell, D. N. Sharp, M. T. Harrison, R. C. Deiming, and A. J. Thrberfield, "Fabrication of J)hotonic crystals

for the visible spectrum by holographic lithography," Nature 404, pp. 53—56, 2000.13. S. Shoji and S. Kawata, "Photofabrication of three-dimensional photonic crystals by multibeam laser interference

into a photopolymerizable resin," Appl.Phys. Lett. 76, pp. 2668—2670, 2000.14. A. A. Maznev, T. F. Crimmins, and K. A. Nelson, "How to make femtosecondpulses overlap," Opt. Lett. 23,

pp. 1378—1380, 17 1998.

15. K. Ho, C. Chan, and C. Sokoulis, "Existence of photonic gaps in periodic dielectric structures," Phys. Rev. Lett.65, p. 3152, 1990.

16. K. Busch and S. Joim, "Photonic band-gap formation in certain self-organizing systems," Phys. Rev. E 58, pp. 3896—3908, 1998.

17. M. Razeghi, The MOCVD G'hallenge, Adam Huger, Bristol, 1989.18. S. Johnson and J. D. Joannopoulos, "Block-iterative frequency-domain methods for maxwell's equations in a

planewave basis," Optics Express 8, pp. 173—190, 2001.19. T. Kondo, S. Matsuo, S. Juodkazis, and H. Misawa, "Femtosecond laser interference technique with diffractive beam

sl)litter for fabrication of three-dimensional photonic crystals," Appi. Phys. Lett. 79, pp. 725—727, 2001.20. P. O'Shea, M. Kinimel, X. Cu, and R. Trebino, "Highly simplified device for ultrashort-pulse measurement," Opt.

Lett. 26(12), pp. 932—934, 2001.21. S. Matsuo, T. Kondo, S. Juodkazis, V. Mizeikis, and H. Misawa, "Fabrication of 3D photonic crystals by femtosecond

laser interference," in SPIE Proc. Photonic bandgap materials and devices (OPTO 5OO2) vol. 4655, A. Adibi,A. Scherer, and S.-Y. Lin, eds., pp. 327—334, 2002.

22. T. Tanaka, H-B. Sun, and S. Kawata, "Rapid sub-diffraction-limit laser micro/nanoprocessing in a thresholdmaterial system," Appl. Phys. Lett. 80, pp. 312 — 314, 2 2002.

23. S. D. Stookey, "Photosensitive glass (a new photographic medium)," Industrial and Engineering Jhem. 41,pp. 856—861, 1949.



24. S. Juodkazis, V. Sniirnov, T. Kondo, S. Matsuo, L. B. Glebov, L. N. Glebova, and H. Misawa, "Three-dimensionalholographic recording in photo-thermo-refractive glass by femtosecond pulses," Adv. Mat. , 2002 (In press).

25. S. Juodkazis, "Optical properties of femtosecond laser irradiated photo-thermo-refractive glass," Lithuanian Journalof Physics 42, pp. 119—126, 2 2002.

26. H. Kogelnik The Bell System Technical Journal 48, p. 2909, 1969.27. R. Sigel, G. Fytas, N. Vainos, S. Pispas, amid N. Hadjichristidis, "Pattern formation in homogeneous polymer

solutions induced by a continous-wave visible laser," Science 297, pp. 67—69, 2002.



Documents

Three-dimensional recording by femtosecond pulses in ... · Figure 2. CCD images of the interference region at the tip of Fresnel biprism (a, d) with intensity cross sections across