Two-Color Holography in Reduced Near-Stoichiometric Lithium Niobate

fsctaes

Two-color holographyin reduced near-stoichiometric lithium niobate

Harald Guenther, Roger Macfarlane, Yasunori Furukawa, Kenji Kitamura, andRatnakar Neurgaonkar

We explored a number of factors affecting the properties relevant to holographic optical data storage byusing a two-color recording scheme in reduced, near-stoichiometric lithium niobate. Two-color, orphoton-gated, recording is achieved by use of 852-nm information-carrying beams and 488-nm gatinglight. Readout at 852 nm is nondestructive, with a gating ratio of ;104. Recording sensitivity, gatingratio, dynamic range, and dark decay were measured for crystals of differing stoichiometry, degree ofreduction, wavelength of the gating light, temperature, and optical power density. The two-color sen-sitivity per incident photon is still somewhat less than that of the one-color process at 488 nm for ;1Wycm2 of gating light but is essentially the same in terms of absorbed photons. Two-color recording isan attractive way of achieving nondestructive readout in a read–write material, and it allows selectiveoptical erasure. © 1998 Optical Society of America

OCIS codes: 090.2900, 160.5320, 210.2860, 210.4810, 190.4180.

et

tftinaio

1. Introduction

Since the discovery of photorefractivity in lithiumniobate,1,2 this material has remained importantor demonstrations of holographic optical datatorage.3–6 In all these cases Fe-doped material ofongruent composition has been used because it hashe best combination of sensitivity, dynamic range,nd optical quality for single-color recording. How-ver, there are some increases in sensitivity andpeed in material with higher stoichiometry.7The basic photorefractive mechanism is one in

which an Fe21 center absorbs a single photon, result-ing in photoionization followed by charge transportand retrapping ~Fig. 1!, to produce a spatially varyingcharge density. A one-center picture provides a use-ful model at low writing intensities, but at higher in-tensities it has been shown that a two-level picture isnecessary.8 The resulting electric field, through the

H. Guenther and R. Macfarlane [email protected]! arewith the IBM Almaden Research Center, IBM Corporation, 650Harry Road, San Jose, California 95120. Y. Furukawa and K.Kitamura are with the National Institute for Research in InorganicMaterials, 1-1 Namiki, Tsukuba 305, Japan. R. Neurgaonkar iswith the Rockwell International Science Center, 1049 Camino DosRios, Thousand Oaks, California 91360.

Received 29 April 1998; revised manuscript received 20 July1998.

0003-6935y98y07611-13$15.00y0© 1998 Optical Society of America

1

electro-optic effect, produces a refractive-index grat-ing. The photorefractive effect generally requiresonly very low concentrations of active centers, in therange of 1–100 ppm ~parts in 106!, and these can beimpurities or intrinsic defects. Characterizing thephotorefractive effect in lithium niobate is of interestnot only for holographic storage applications but alsofrom the standpoint of photorefractive optical damagein integrated-optics applications. Two-color damagecan occur, for example, in frequency-doubling periodi-cally poled lithium niobate structures.

One of the most important problems in the imple-mentation of holographic information storage is thatof fixing the stored information. The establishedtechniques for achieving this are thermal fixing,which is well known with lithium niobate,9–13 andlectrical fixing, which was first shown in bariumitanate14 and later in strontium barium niobate.15,16

Both of these fixing schemes increase resistance toerasure during readout and in some cases also in-crease the dark decay time. In thermal fixing, ionic~usually proton! mobility is thermally activated, andhe mobile ions compensate the electronic gratingormed by the initial writing process. Followinghis, the electronic grating is erased optically and theonic grating revealed. Because the information isow stored as a spatial pattern of ions rather than asspatially inhomogeneous distribution of filled traps,

t is not erased by the reading light. Disadvantagesf thermal fixing include the complexity of in situ

0 November 1998 y Vol. 37, No. 32 y APPLIED OPTICS 7611

aeaf

uct

n

cw

t

gt

7

processing, the need to write the entire memory be-fore fixing, and the dead time required for the fixingprocess. Electrical fixing involves ferroelectric-domain reorientation near the Curie temperature,which considerably restricts its generality, and thereare serious limitations on the spatial frequencies sup-ported by the material because of minimum domainsizes.16

Other fixing mechanisms address just the issue ofnondestructive readout, and these generally fall intothe category of two-color processes. One uses a dif-ferent wavelength of light for recording and readingthe holograms.17–19 However, for non-plane-wavepplications such as storage, the inability to achievexact phase matching leads to increased cross talknd a serious loss of efficiency for certain spatialrequencies.

A promising path to nondestructive readout is these of photon-gated or two-color holography in whichharge generation occurs by means of a two-step pho-oionization with two colors of light,20 as illustrated

schematically in Fig. 1. One color ~l1! carries theinformation signal, which is borne on coherent objectand reference beams, and the second color ~l2! ~orband of colors, as incoherent light can be used21! pro-vides the gating light. The gating light can bethought of as sensitizing the material for the writingprocess or as providing the ionization step for a grat-ing temporarily stored in a metastable level. Theformer mechanism applies to the lithium niobate sys-tem described in this paper.

Figure 2 shows the qualitative difference betweenone-color and two-color grating formation and read-out. In the one-color case the same photoionizationprocess that is used to write causes erasure duringreading. For two-color writing the gating light aidsthe writing process but is not at the writing wave-length, i.e., the information is contained on the long-

Fig. 1. Schematic diagram of the two-color and one-color photore-fractive effects. CB, conduction band; VB, valence band. Level 1is attributed to a Nb bipolaron state or Fe21yFe31 state, level 2 toa NbLi antisite polaron, and level 3 to an Fe31 trap. The single-enter model for one-color recording is appropriate to low-power cwriting.

612 APPLIED OPTICS y Vol. 37, No. 32 y 10 November 1998

wavelength beam. Reading at the longerwavelength in the absence of gating light is thennondestructive. A material in which erasure occursonly in the presence of two colors would be even bet-ter. Some advantages of optical fixing, in addition tonondestructive readout, are that l1 can be chosen inthe near IR for which suitable diode lasers are readilyavailable, the gating light can be incoherent andbroad band,21 no postprocessing is required, and im-plementation can be accomplished in a read–writematerial. Further, selective and rapid erasure of in-dividual data stacks is possible. Another advantageyet to be explored is that two-color writing permitsthe use of wider-band-gap materials because ioniza-tion occurs in two steps. These materials can havedeeper trap levels, hence exhibiting longer dark de-cays, and are more likely candidates for two-colorerasure.

The earliest research on two-color photorefractivematerials20 used excitation by high-power 10-ps pulsesfrom a mode-locked Nd:glass laser. In lithium nio-bate charge generation was attributed to two-photonexcitation of intrinsic band-to-band transitions and inLiNbO3:Cu to two-step, two-color excitation by use offundamental ~1064-nm! and second-harmonic ~532-

m! radiation. Subsequently, von der Linde et al.22

showed photon-gated holography in LiTaO3:Cr thatthey assigned to two-step ionization of Cr31. In

Fig. 2. Typical write–read–erase curves for holographic gratingsin lithium niobate crystals: ~a! One-color scheme in which anAr1 laser at 488 nm and 1 Wycm2 is used for both writing ~twobeams! and reading ~one beam!. ~b! Two-color scheme in which adistributed Bragg reflector laser diode at 852 nm and 4 Wycm2 ishe total intensity used for writing ~equally divided into two

beams! and an Ar1 laser at 488 nm and 1 Wycm2 is used for theating step. Nondestructive reading was carried out with one ofhe unattenuated writing beams ~2 Wycm2! and erasing with the

gating light.

23

1M

gmian

ds

Ls

rw

33

rMobeett

wwu

LiNbO3:Fe, Buse et al., by using ;20-ns pulses of064- and 532-nm light and peak intensities of ;20Wycm2, showed that the energy required to write a

two-color hologram is comparable with that for theone-photon process, which uses cw light. In a moredetailed analysis of the two-color writing process, Buseet al.24 showed that two-color writing required thereen light to precede the IR light, and this require-ent was interpreted in terms of a two-center picture

n which a shallow trap was filled by the green lightnd writing from this trap was accomplished by theear-IR light.The general picture described above contains the

essential ingredients of the model that also describesobservations at low powers with cw lasers. Changesthat have been made in more recent research21,25–27

have come from the realizations that the intermediatelevel in the two-color recording process can have a longlifetime of milliseconds to seconds, that high peak pow-ers are not required, and that cw lasers that provide;1 Wycm2 are adequate and practical writing andgating sources. The importance of stoichiometry forthe sensitivity of nonlinear writing processes waspointed out by Jermann et al.25 They showed inLiNbO3:Fe that, for low light intensities of ;1 Wycm2,stoichiometric lithium niobate has a higher sensitivitythan does congruent material, whereas at high lightintensities ~greater than 100 Wycm2! the congruentmaterial becomes more sensitive.

Underlying the studies of photorefractive gratingshave been steady advances in understanding the roleof intrinsic defects. In nonstoichiometric ~Li-

eficient! lithium niobate, the defect structure is de-cribed by the Li-vacancy model,28 which is expressed

by the formula @Li1–5xNbxh4x#@Nb#O3. In this modeli vacancies, designated by the box ~h!, are compen-ated by NbLi

51 antisite defects that reach a concen-tration of 1% in the congruent composition.Reduction results in a loss of oxygen and an in diffu-sion of the excess Li and Nb into Li vacancies,29 withelectrons being trapped in Nb51 ions. This producesa visible absorption band commonly stated to be at2.5 eV,30–34 but this conclusion is complicated by thefact that reduced-Fe impurities also produce absorp-tion around 2.5 eV, so the shape of the visible absorp-tion band depends to some degree on the number ofFe impurities. Irradiation in the visible producesanother band at 1.6 eV that is stable at low temper-atures and metastable at room temperature and isassociated with the appearance of an EPR ~electronparamagnetic resonance! spectrum showing thecharacteristic hyperfine structure of Nb41.35 Theintrinsic visible absorption was originally assigned toO22 vacancies and the 1.6-eV band to an electrontrapped in an O22 vacancy.31 Subsequently, it wasestablished that the O22 sites remain fully occupiedin reduced lithium niobate.36 Arizmendi et al.32 firstassociated a Nb center with the 1.6-eV band, andsubsequent research made a firm connection to theNbLi antisite.33,37,38 The visible absorption that isesponsible for the sensitization step of the two-colorriting described here is now attributed to bipolaron

1

absorption in which the bipolarons are electronstrapped at adjacent NbLi and NbNb sites34 and are thestable polaronic entity at room temperature. The1.6-eV band, which results principally from photodis-sociation of the bipolaron, is due to the small polaronformed at the NbLi

41 antisite defect and is the ab-sorption band used to write the two-color holograms.

The free small polaron has a lifetime that variesfrom milliseconds to seconds, depending on the crys-tal stoichiometry, the degree of reduction, the tem-perature, and the presence of other dopants such asFe. This lifetime is an important quantity becauseit determines, in large part, the sensitivity throughthe population that can be achieved under cw pump-ing. It also defines the time for which the materialremains sensitive after the gating light is switchedoff. In many of the early studies on two-color holog-raphy in lithium niobate by use of high-peak-powerlasers, the role of specific transition-metal dopantswas emphasized, but it appears likely that the basicprocess of two-color holography operates largelythrough intrinsic defects.

The key material parameters for holographic stor-age are recording sensitivity, resistance to erasureduring readout ~here, the gating ratio!, the dynamicange expressed by the saturation efficiency and the# ~see Subsection 5.C!, the dark decay time, and

ptical quality. We investigated these properties foroth undoped and Fe-doped lithium niobate of differ-nt stoichiometries and degrees of reduction. It isssential to monitor all these parameters becausehey cannot be optimized independently and somerade-offs need to be made.

2. Experiment

Measurements were made by use of the plane-wavetransmission geometry shown in Fig. 3. A colli-mated 852-nm distributed Bragg reflector diode laserwith a coherence length of approximately 100 m waspassed through a polarization rotator and a beamsplitter to provide two beams of up to 30 mW. Thebeams had a beam diameter at the sample of 1.3 mmand an external crossing angle of 20°. For mostmeasurements the two writing beams had compara-ble intensities, yielding a modulation index nearunity, but neutral-density filters could be inserted tovary this index over a wide range. We checked themechanical stability of the setup by time-averagingthe interference pattern produced by the writingbeams over a 1-h period during which the fringe con-trast remained better than 50:1. Poled single-domain lithium niobate crystals in the form of cubesthat were 8 mm to 1 cm on a side were oriented withthe c axis normal to one face. Measurements weremade with the two 852-nm writing beams incidentupon the horizontal plane on a face containing the caxis. They were polarized parallel to the c axis,

hich was also horizontal, resulting in a gratingave vector along the c direction. This arrangementses the r33 electro-optic coefficient that has a value

of 32 pmyV. For E perpendicular to c, r13 5 10pmyV is the appropriate electro-optic coefficient, so in


lmbppttn

adCSl

pT6oia

ai~df

7

this geometry the sensitivity is reduced by a factor of3. The diffraction efficiency as a function of timewas monitored with a 1-mW 650-nm diode laser thatproduced 3-ms pulses and operated at a duty cycle of0.01%–1%. The monitor beam was Bragg matchedat the appropriate angle and counterpropagated withthe writing beams. The linearity of the detectionwas checked over three decades. Absolute diffrac-tion efficiencies were calibrated by use of one stronglyattenuated writing beam. Gating light was usuallyprovided by an Ar1 laser at 488 nm that copropa-gated with, and at a small angle to, the probe laser.For some experiments, however, filtered white lightor GaN LED’s were used as the gating-light source.21

Absorption spectra were measured on a Perkin-Elmer Model Lambda-9 spectrophotometer at ambi-ent temperature. Both the excited-state absorption~ESA! at the writing wavelength of 852 nm and theifetime of the metastable level were measured by

eans of monitoring the transmission of one writingeam after excitation to the steady state with a longulse of 488-nm gating light at 1 Wycm2. Manyroperties were measured as functions of tempera-ure, for which a thermoelectric cooler operating be-ween 15 °C and 80 °C was used. The results for aumber of crystals are summarized in Table 1.

3. Materials and Processing

Crystals grown by two different methods were usedin this study. They are denoted CLN when theirstoichiometry, expressed as the ratio cLi 5 @Li#y~@Li#1 @Nb#!, has a congruent-composition value of 48.4%nd SLN when it deviates from this value in theirection of the stoichiometric composition of 50%.rystals SLN1 ~cLi 5 49.7%!, SLN2 ~cLi 5 49.7%!, andLN5 ~cLi 5 49.4%! were grown as described in ear-

ier reports39,7 by use of a double-crucible techniquein which the stoichiometric crystal is in equilibriumwith a Li-rich melt and a powder of stoichiometriccomposition is added to the outer crucible. CrystalsSLN3, SLN4, CLN1, and CLN2 were grown by theconventional Czochralski method but with different

Fig. 3. Experimental setup for writing and reading two-color gateanamorphic prism pair; POL, polarization rotator; BS, beam spligratings; DBR, distributed Bragg reflector.


amounts of Li enrichment in the starting melt.40

The highest stoichiometry achieved in this way wasobtained with a melt composition of cLi,melt 5 56%,resulting in a crystal with cLi 5 49.4%.

The stoichiometry was measured from the phase-matching temperature for second-harmonic genera-tion41 of a 1.047-mm Q-switched Nd:YLF laser. Thedata were fitted to a curve calculated from the gen-eralized Sellmeier equation, which describes thetemperature- and the wavelength-dependent disper-sion of the refractive indices of lithium niobate.42

We carried out the calculation for 1.047 mm and com-ared it with that for 1.064 mm, as shown in Fig. 4.he phase-matching temperature was measured to1 °C, which corresponds to a stoichiometry change

f 60.006%. The homogeneity of the stoichiometrys higher perpendicular to the c axis than along the cxis, which is the direction of growth. In all cases cLi

was constant to better than 0.02% over the measuredvolume of ;100 mm3. Although this did not affectthe photorefractive sensitivity of the materials stud-ied here, compositional inhomogeneities do producerefractive-index changes that affect the overall crys-tal quality.43 We found that, in general, the opticalquality of congruent material was higher than that ofmaterial with a more stoichiometric composition.This was measured ~i! by the quality of the images of

64-kbit Cr-on-glass data mask made with crystalsn the light path and recorded by a CCD camera andii! by scattered light levels. The crystals were re-uced in a 2-in.-diameter ~5.08-cm-diameter! Al tubeurnace at 950 °C for 1 h at values of O2 partial pres-

sure between 0.5 3 1023 and 11 3 1023 mbar. Dur-ing reduction, Li and Nb ions diffuse to fill Livacancies, resulting in an increase in the concentra-tion of NbLi antisite defects.29 The sample wasmaintained at the maximum temperature for 1 h toallow ionic diffusion to take place. The pressure wasmeasured with a capacitance manometer at the inputto the furnace tube. We reset the crystals to a stan-dard degree of oxidation before each reduction run by

ne-wave holograms. L1, collimating lens for the diode laser; AP,PM, photomultiplier; LD2, 650-nm laser diode used to probe the

d platter;

Table 1. Summary of Data and Comparison of Two-Color and One-Color Results

heating in air at 950 °C for 1 h. Following reduction,the crystal was cooled at a rate of ;20 °Cymin.

4. Spectroscopy

Congruent undoped lithium niobate has an absorp-tion edge that is shifted toward lower energy by ap-proximately 0.2 eV relative to that of stoichiometricmaterial.39 In our samples the wavelengths atwhich the absorption coefficient reached 10 cm21

were 330 nm ~congruent! and 309 nm ~stoichiometric,

ai

1

with cLi 5 49.7%!. The reduction of congruent lith-ium niobate induces a broad, rather unstructured,band in the visible30 that is subsequently assigned tobipolaron absorption.33

Data taken from our samples are shown in Fig. 5.Note that Fe, which is readily reduced to Fe21, alsohas absorption in the same spectral region and thatunintentional Fe impurities can be present in suffi-cient numbers to contribute to the visible absorption.In addition, at least some of the long-wavelength ab-sorption appears to be associated with the presence ofpopulated shallow traps because it is seen in samplesthat exhibit a short dark decay. In Fe-doped crys-tals reduction produces the well-known Fe21 band atapproximately 2.5 eV with very little absorption atlonger wavelengths of approximately 850 nm, wherethe undoped reduced crystals show appreciable ab-sorption ~Fig. 5!. Thus one of the roles of Fe, in

ddition to providing centers for charge generation,s to provide deep traps in the form of Fe31 that

bring the Fermi level closer to midgap. In more stoi-chiometric undoped crystals, reduction induces a vis-ible spectrum similar to that of the congruentundoped case but with less absorption at 850 nm @Fig.5~c!#, which is consistent with their showing a longerdark decay time. This is despite the fact that themore stoichiometric material is more resistant to re-duction and is processed at lower O2 partial pressures~see Table 1!.

The absorption induced at 852 nm by means ofpumping at the gating wavelength of 488 nm was

Fig. 4. Calculated phase-matching temperature curve for the1047-nm wavelength used as compared with that for 1064 nm.The insert shows a temperature scan of the harmonic-generationefficiency of a 1-cm crystal.

Characteristic

Material

SLN1 SLN2 SLN3 SLN4 SLN5 CLN1 CLN2

Storage performancea

Recording scheme Two color852 and488 nm

Two color852 and488 nm





One color488 nm

103 Sh2 ~incident! ~cmyJ! 8 9 0.3 5 8 0.02 100103 Sh1 ~absorbed! ~cmyJ! 160 150 ;150 180 160 .20 170M# 0.8 0.5 0.2 0.4 0.2 0.05 24Metastable lifetime ~s! 16 0.15 NMb 3.6 1.5 0.1 NyAGating ratio at 852 nm 1600 10,000 NM 7800 2500 NM NyADark decay td ~years! 0.07 0.03 NM 0.32 0.22 0.32 ;0.310

Material propertiesParameters Stoichio-

metricReduced

Stoichio-metric

Reduced

Stoichio-metric

UnreducedPr 0.2%

Stoichio-metric

ReducedPr 0.2%

Stoichio-metric

Reduced

CongruentUnreduced

CongruentReduced

cLi ~%! 49.7 49.7 49.4 49.4 49.4 48.4 48.4Thickness ~cm! 0.98 0.95 0.80 0.80 1.00 0.89 0.50Fe concentration ~ppm! 1.0 100 NM NM 5 NM 200Reduction p~O2! ~1023 mbar! 1 2 NyA 11 10 NyA NMProton concentration ~31017ycm3! 4.2 4.2 1.5 1.9 5.9 32 8.4a ~cm21! ~gate at 488 nm! 0.35 0.95 0.12 0.32 0.62 0.08 NyAa ~cm21! ~write at 852 nm! 0.05 0.06 ;0.002 0.028 0.050 ,0.001 1.8c

ESA depth ~eV! 0.75 0.66 NM 0.54 0.70 NM NyA

aIw 5 4 Wycm2 at 852 nm, Ig 5 1 Wycm2 at 488 nm, L 5 6 mm, and E parallel to the c axis.bNM denotes not measured.cWriting and reading wavelength of 488 nm.


m7co

4d

7

measured for 1 Wycm2 of pump light. The two-colorphotorefractive sensitivity is proportional to this in-duced absorption ~Table 1!. Only the more stoichio-metric material showed transient induced absorptionat levels above our sensitivity of approximately 0.001cm21 in the absorption coefficient and approximately10 ms in lifetime.

Transient absorption becomes persistent at lowtemperatures at which it appears as a well-known1.6-eV absorption assigned to a small polaron at theNbLi antisite.44 This absorption is associated with acharacteristic Nb41 EPR spectrum.35 It appearsthat at room temperature the lifetime of this smallpolaron in the congruent crystals, whether undopedor Fe doped, is so short that no significant population

Fig. 5. Absorption spectra for different lithium niobate crystalsbefore and after reduction. Note that the peak of the inducedabsorption is at approximately 3.3 eV rather than 2.5 eV, as inFe-doped crystals: ~a! congruent lithium niobate before and afterreduction ~note the significant absorption at 800 nm in the reducedcrystal!, ~b! the same for a crystal with stoichiometry of cLi 59.7%, and ~c! absorption of a stoichiometric crystal ~cLi 5 49.7%!oped with 100 ppm Fe.


builds up and no transient absorption is seen. Asthe stoichiometry increases, the number of antisitedefects decreases and the small-polaron lifetime in-creases.

The lifetime of the ESA was measured following a488-nm pulse that was long enough to achieve asteady-state population of the metastable level forthe given light intensity. Figure 6 shows the ESAdecay time as a function of inverse temperature forthree stoichiometric crystals. We determined thedecay times by fitting the ESA decay curves to single-exponential functions, which generally provides areasonable approximation. The depth of the meta-stable level of ;0.7 eV was obtained from the activa-tion energy of the ESA decay rate. The Fe-dopedand Pr-doped samples had slightly lower activationenergies, which suggests that the small-polaron en-ergies might be perturbed by impurities to which thepolarons bind.

Stoichiometry and the degree of reduction alsostrongly affect the IR absorption spectrum of theOH2 vibrational modes that are observed as a resultof the presence of H in the crystal.45 This influenceis shown in Fig. 7. The congruent material shows abroad ~;30-cm21! absorption at 3484 cm21, while inmore stoichiometric material two lines ~at 3466 and3480 cm21! are frequency shifted and an order of

agnitude narrower ~they are slit broadened in Fig.!, reflecting the lower defect density. The protononcentration, which is presumed to be equal to thatf OH2, was calculated from the integrated absorp-

tion by use of known oscillator strengths46 and isgiven in Table 1. It is assumed that the oscillatorstrength does not change with stoichiometry.

5. Material Properties

The key photorefractive-material properties of inter-est for holographic data storage are sensitivity, gat-ing ratio, M# or dynamic range, dark decay, andoptical quality. Table 1 summarizes our measure-ments of all but the last of these properties for crys-tals of different stoichiometries and in reduced andunreduced states. The degree of reduction is ex-pressed by the O2 partial pressure, and in the case ofundoped material this correlates with the absorptioncoefficient at the gating wavelength. In the case ofFe-doped material the bipolaron and the Fe21 ab-sorption contributions are difficult to separate.

A. Sensitivity

Photorefractive sensitivity for two-color recording isdefined for a gating-light intensity of Ig 5 1 Wycm2 at488 nm. It is necessary to specify a gating conditionbecause the sensitivity is linear in Ig at only lowvalues of Ig.21 This chosen standard intensity ispractical and reasonably easily achieved. Thegrowth of the square root of the diffraction efficiencyduring writing can be written as A0@1 2 exp~2tytr!#or ~A0ytr!t at short times, where tr is the recordingtime constant. The definition of sensitivity usedhere is that for one-color holography47 and makes

t

comparison with that case easy ~see Table 1!. Interms of incident intensities:

Sh2 5 A0ytr lIw, Ig 5 1 Wycm2, (1)

where Ig is the gating intensity, Iw is the total writingintensity in the two beams, l is the crystal length, andA0ytr is the writing slope or the initial time derivativeof the square root of the diffraction efficiency, ]=hy

Fig. 6. Temperature dependence measured from the decay of theThe activation energy is 0.7 eV.

Fig. 7. OH2 vibrational spectra of lithium niobate. The uppercurve represents congruent material with cLi 5 48.4%; the threelower curves represent a stoichiometry of cLi 5 49.7% ~see Table 1for details of the crystals!. SLN2# is a less reduced state of SLN2hat contains 1.7 3 1016 protonsycm3.

1

]tut50. The sensitivity can, of course, be increased byan increase in the amount of gating light, but as isshown in Fig. 8 saturation sets in above approxi-mately 1 Wycm2. There is considerable variation inthe onset of saturation in different samples; it is ear-lier in heavily reduced crystals or in those lacking Fe.In SLN1, for example, the sensitivity is high, butsaturation as a function of gating intensity sets in atless than 0.5 Wycm2. This material is more suitedfor gating with blue LED’s. The maximum diffrac-tion efficiency achievable for a given gating intensityis higher at higher writing intensities, suggestingthat saturation occurs because of competition be-tween writing and erasing. Recall that erasure re-sults from the gating light itself.

It is also useful to consider the sensitivity as de-fined in terms of absorbed power,47 which is Sh1 5Sh2ya, where a is the absorption coefficient at thewriting wavelength. In terms of this sensitivity allmaterials studied, including the single-photon Fe-doped material written at 488 nm, are almost equallysensitive. This result suggests that the transportand the trapping parameters might be similar andthat the sensitivity is determined by the amount oflight that can be absorbed at the writing wavelengthand is effective in charge generation. This is still aweak point in the two-color writing process at thepresent pump intensities of ;1 Wycm2; the maxi-mum absorption coefficient at the writing wavelengthis approximately 0.06 cm21, and it appears to satu-rate near this level under our conditions.

We return now to the sensitivity Sh2. This param-eter depends on a number of factors, with the approx-imate gains ~with respect to undoped congruent

of the intermediate-state lifetime for three stoichiometric crystals.
ESA

o

i

dstiwspgsrt s

w

7

materials! available from each being crystal stoichi-metry ~153!, degree of reduction27 ~203!, tempera-

ture ~53 for cooling from 20 °C to 0 °C!, gatingwavelength ~103 for a reduction in lgate from 520 to400 nm!,21 and gating intensity ~which is linear at lowintensity and starts to saturate above approximately1 Wycm2!. We did study the dependence of sensitiv-ty on stoichiometry for many values of cLi, but it

appears to be not very sensitive to stoichiometry forcLi greater than approximately 49.4%. Thus crys-tals with cLi in the range 49.4%–49.7% have similarsensitivity. Separating stoichiometry from the de-gree of reduction is not easy because the effects ofreduction also depend on stoichiometry—the higherthe stoichiometry, the more reduction is needed toinduce a given amount of absorption. Most of ourresults were obtained with crystals with a stoichiom-etry of 49.7%.

As was stated in Section 4, the effect of crystalreduction is to induce a broad visible absorptionband. Irradiation within this band sensitizes thematerial for writing. Figure 9 shows a typical writ-ing and erasing curve before and after reduction.Reduction was found to produce a sensitivity increaseof a factor of ;20, but the saturation efficiency or

ynamic range is essentially unchanged. The sen-itivity increase comes partly from increased absorp-ion at the gating wavelength and partly from anncrease in the lifetime of the metastable state fromhich writing occurs, which is shown in Table 1 for

ome of the samples studied. Because we use cwumping, the population in the metastable level isenerally proportional to this lifetime. In the mosttrongly reduced and stoichiometric crystal, SLN1, iteaches 16 s at 20 °C, with a corresponding absorp-ion at the writing wavelength of 5% for 1 Wycm2 of

gating light. It clearly would be desirable to in-

Fig. 8. Dependence of the sensitivity Sh2 on the gating intensitydepends strongly on the sample.


crease this absorption by material design rather thanby an increase of the gating-light intensity. Al-though some increase in lifetime is needed to increasethe sensitivity, there is an adverse effect when thelifetime gets too long because the intermediate levelthen competes with the ultimate deep trap and partof the grating is written to this level. This leads to afast component of dark decay, which can be seen fromFig. 9 for the reduced sample. Another consequenceis that the material remains sensitized, and thiscould lead to some erasure if readout occurs beforethe intermediate population has decayed. As we seebelow, Fe doping reduces this lifetime while main-taining the sensitivity.

8 nm for three stoichiometric crystals. The saturation behavior

Fig. 9. Write–dark–erase curves for unreduced ~SLN3! and re-duced ~SLN4! crystals with a stoichiometry of cLi 5 49.4%. Theaturation efficiency is essentially unchanged by reduction,hereas the sensitivity is greatly increased.

at 48

ib

if

ms

The third factor controlling sensitivity is the gatingwavelength. This is to be expected from the bandshape of the induced absorption shown in Fig. 5. Wemeasured this wavelength dependence by using awhite light source and a series of narrow-band filters,with the results shown in Fig. 10. A sensitivity in-crease of a factor of 10 was found for wavelengthsbetween 514 and 400 nm. The dependence of thegating efficiency on the photon energy does not followthe induced absorption closely. A similar result wasfound by Sweeney and Halliburton31 for the bleach-ng efficiency of the visible band to form the 1.6-eVand.Several factors could contribute to this difference

n dependence. The visible band has componentsrom both bipolarons and Fe21, with each resulting in

a different efficiency for populating the NbLi41 inter-

mediate state. For wavelengths approaching thebandgap, direct band-to-band transitions will con-tribute to charge generation, producing a steep rise atapproximately 3 eV. This makes GaN or otherLED’s attractive gating sources,21 especially in theform of high-density arrays or broad-area emitters.Ultimately, the transparency of lithium niobate atthe gating wavelength limits how short the gatingwavelength can be in practice.

Finally, the sensitivity is temperature dependentbecause of thermal depopulation of the metastablelevel. Measurement of this temperature depen-dence ~Fig. 11! shows not only the typical sensitivityincrease that can be expected on cooling ~53 between20 °C and 0 °C! but also a measure of the activationenergy or the depth of the metastable level below theconduction band. For crystal SLN3 this was 0.66eV. Because the sensitivity is a complex propertyinvolving ionization and transport, the trap depth isperhaps better obtained from a property such as thetemperature dependence of the induced absorption,as shown in Fig. 6, which gives a similar depth of 0.7eV. Most of the factors affecting sensitivity can becombined to produce a material with the product ofthe individual sensitivity-enhancement factors. For

Fig. 10. Dependence of the gating efficiency on the gating wave-length. Also shown is ~1! the absorption in the vicinity of theintrinsic band edge and ~2! the absorption band induced by reduc-tion.

1

example, we took measurements of SLN5 at 13 °Cand with 457.9-nm gating light and obtained a factorof 3.2 increase in sensitivity to 0.026 cmyJ comparedwith 20 °C and 488-nm gating. This sensitivity in-crease is close to that expected from Figs. 10 and 11,i.e., a factor of 2.0 for wavelength and of 1.9 for tem-perature for a total of a factor 3.8.

Some of the crystals contained Pr at a concentra-tion of ;0.2% because, in the early stages of thisproject, it was thought that Pr could contribute di-rectly to the photorefractive effect by a valencechange on either ionization or trapping.48 We foundsome small increases in sensitivity associated withthe presence of Pr, but none were large enough to besignificant. It seems most likely that they arose be-cause of the effect of Pr31’s occupying a Li site andreducing adjacent Nb sites through the need forcharge compensation. This is an alternative point ofview to that of Bai et al.,49 who proposed that absorp-tion into the 4f5d band of Pr dominates the two-color

echanism and that their material contained somepecial Pr sites that were photorefractively active.

B. Gating Ratio

The gating-ratio parameter expresses the essence oftwo-color holography, i.e., the resistance to erasureduring readout. Rather than measuring the gatingratio by a small change in the diffraction efficiencywith reading time, we preferred to use the zero-background measurement of the ratio of the sensitiv-ities in the presence of gating light to that in theabsence of gating light. The writing wavelength of852 nm that we chose is close to the peak of the 1.6-eV~induced! absorption band.

Typical values of the gating ratio in undoped re-duced crystals of a stoichiometry of 49.7% are in therange 2000–5000, whereas Fe-doped samples show

Fig. 11. Temperature dependence of the two-color sensitivity inan Fe-doped stoichiometric crystal showing an activation energy of0.66 eV, which is in good agreement with the depth of the writinglevel obtained from the ESA decay ~Fig. 6!.


v

stmtcl

Aitrobtapists

7

values as high as 10,000. This reflects the ability ofFe to reduce the occupation of levels contributing toabsorption at 852 nm and with it the single-photonred sensitivity. At shorter writing wavelengths thegating ratio decreases as one-color and self-gated con-tributions become appreciable. For example, at 670nm the gating ratio is only 5.

C. Dynamic Range

The dynamic range is a very important parameterbecause it determines how many holograms can bemultiplexed in a single volume. A useful parameterthat expresses dynamic range is the M#, introducedby Mok et al.50 and defined, in terms of the diffractionefficiency for the Mth hologram stored in a commonolume, as hM 5 ~M#yM!2. This term can be written

as a product of the erasure time te and the writingslope as M# 5 te ]=hy]tut50.

In a one-color ~one-photon! material, the writingslope is proportional to the product of the amplitudesof the two writing beams, i.e., ~I1I2!1y2, and the era-ure time constant is inversely proportional to theotal intensity. Thus the M# is proportional to theodulation index, m 5 2~I1I2!1y2y~I1 1 I2!. In a

wo-color material, on the other hand, erasure isaused by the gating light and not by the writingight. For low-intensity gating light Ig, the writing

slope is proportional to ~I1I2!1y2Ig and the erasuretime is inversely proportional to Ig. This results inan M# proportional to ~I1I2!1y2. Hence for equal-intensity object and reference beams the M# is pro-portional to the writing intensity ~see Fig. 12!.

lthough this relation provides a very nice way ofncreasing the dynamic range in a two-color material,he writing-power requirements in the present mate-ial system become rather high to gain, say, an orderf magnitude in dynamic range. For unequal object-eam and reference-beam intensities, fixing one in-ensity yields a square-root dependence on the other,s shown in Fig. 13. This square-root dependenceroduces a different dependence on the modulationndex than that for the one-color case. The M#hows a strong temperature dependence ~Fig. 14!hat is expected because the sensitivity is much moretrongly temperature dependent than is the erasure.

D. Dark Decay

The time decay of the holograms in the dark deter-mines the lifetime of the stored information. Wemeasured the dark decay times for a number of crys-tals of different origin, stoichiometry, and degree ofreduction. The dark decay is determined by theelectrical conductivity s through the relation47 td 5ee0ys. Congruent lithium niobate, in addition toshowing low sensitivity for two-color writing, alsoshows rapid conductivity increases on reduction, sothe dark decay of strongly reduced congruent sam-ples is often found to be only a few minutes or evenseconds. Presumably this is due to the presence ofshallow electron traps. Stoichiometric material, on


the other hand, shows dark decay times that areorders of magnitude longer.

The temperature dependence of the dark decayyields the activation energy for the grating decay, asshown in Fig. 15, and extrapolation to room temper-ature yields values of the order of several months ~seeTable 1!. This is the time scale over which it isexpected that thermal fixing occurs at room temper-ature. For the range of samples studied, we ob-tained a family of almost parallel curves that showedactivation energies between 0.93 and 1.29 eV. Thisrange is associated with the formation of ionic grat-ings arising from proton diffusion because these en-ergies correspond well with those obtained fromthermal fixing of single-color holograms in Fe-dopedmaterial.11,12

Although the activation energies shown by differ-ent crystals were similar, the dark decay times variedconsiderably. For a given stoichiometry, td de-

Fig. 12. Dependence of the M# or the dynamic range on thewriting intensity. This is in contrast to one-color holography inwhich the M# depends on the modulation index but not on thewriting intensity. The crystal length was 9.8 mm.

Fig. 13. M# as a function of one of the writing-beam intensities~I2!, while the other ~I1! is kept fixed. In contrast, the M# forone-color holograms is proportional to the modulation index.

2

ei

creases with increasing reduction and OH content.This trend was noted earlier for the congruent com-position.12 Without regard to stoichiometry, how-ever, there is no obvious relation between the darkdecay and the total OH2 content. The most likelylectronic deep trap in which the gratings are storednitially is the Fe21yFe31 level, which has been

shown to have a slightly higher activation energythan do the protons.

6. Role of Iron

Fe impurities play an interesting and important role,even in nominally undoped lithium niobate in which

toalmppdtsoit

1

residual impurity levels are typically 1–5 ppm. Asnoted in Subsection 5.D, the presence of some Fedoping is believed to be necessary to provide a deeptrap for the initial storage of the holograms. Nom-inally undoped, reduced stoichiometric materialsshowed long decay times, indicating that for this casea few parts in 106 of Fe is sufficient to provide thenecessary deep traps. For congruent material,which is not of interest in the intensity regime stud-ied here because of its low sensitivity, lengtheningthe dark decay requires a higher concentration of Fe.

In stoichiometric crystals the trapping effects of Feare also felt in the shortening of the lifetime of theintermediate level from which writing occurs. Thiseffect is shown in Fig. 16 for the addition of 60 ppm ofFe, which reduces the lifetime from 3.5 to 0.18 s.This quenching of the intermediate-state lifetimeeliminates the effect of the short-term dark decaythat otherwise occurs in reduced, undoped stoichio-metric crystals as a result of part of the grating’sbeing written in the intermediate level. The short-term dark decay constant measured in SLN4 was 4 s,which is in reasonable agreement with the ESA decaytime. Figure 17 shows recording and erasing curvesfor a higher stoichiometric sample under the effect ofFe doping. A similar quenching behavior is ob-served at 30 and 60 ppm of Fe. The transfer ofelectrons from the NbLi

41 small polaron to Fe31 ishus seen to be rather efficient. The concentrationf Fe should not be so high, however, that the Febsorption becomes strong at the gating wave-ength, because that limits the usable thickness of the

aterial and also causes undesirable heating andhotovoltaic damage effects. An interesting and im-ortant result is that, although the short-term darkecay is quenched by means of Fe doping, the absorp-ion coefficient at the writing wavelength, hence theensitivity, is maintained, as is shown in Table 1. Inther words, the shortening of the lifetime of thentermediate state by more than an order of magni-ude does not significantly reduce its population. A

Fig. 14. Temperature dependence of the M#, reflecting the factthat, at elevated temperatures, the sensitivity drops much fasterthan the erasing efficiency. The lengths of the crystals are givenin Table 1.

Fig. 15. Dark decays for several crystals of different doping, stoi-chiometry, and degrees of reduction, showing a range of activationenergies from 0.93 to 1.29 eV that is attributed to proton diffusion.SLN2# is a less reduced state of SLN2 that contains 1.7 3 1016

protonsycm3.

Fig. 16. Effect of Fe doping on the decay of the population fromthe intermediate level. The long decay is from a sample withoutintentional Fe doping ~SLN4!, and the fast decay is from a compa-rable sample doped with 60 ppm Fe ~SLN49!.


sRMfcmBs~Hs

7

possible reason for this lack of population decrease isthat the filling rate increases because of the popula-tion of the small polaron by direct photoexcitation ofFe, which absorbs efficiently at the gating wave-length.8

7. Conclusion

Two-color, photon-gated holography provides a prom-ising solution to the long-standing problem of de-structive readout in read–write digital holographicstorage. The basic process is based on two-step pho-toionization by use of a metastable ~small-polaron!level, which is the charge-generating step in the pho-torefractive process. This scheme has been studiedfor a number of years, mostly by use of high-peak-power lasers. We and others21,26,27 have investi-gated two-color holographic grating formation in themore practical low-power regime by using cw lasers.

Here we have used diode lasers as writing sourcesand a gating light of up to approximately 1 Wycm2 tostudy the issues affecting the main parameters ofinterest for holographic storage, i.e., sensitivity, gat-ing ratio, dynamic range, and dark decay, for a num-ber of doped and undoped lithium niobate crystals ofdifferent stoichiometries and degrees of reduction.In agreement with some earlier studies, the photore-fractively active centers are assigned mainly as in-trinsic defects such as bipolarons and NbLi antisitesmall polarons, but Fe dopants remain importantcontributing species to the overall performance, prin-cipally in providing the initial deep trap for hologramstorage.

The two-color sensitivity per incident photon is stillsomewhat less than that of the one-photon process inFe-doped lithium niobate at 488 nm, although thesensitivity per absorbed photon is approximately the

Fig. 17. Write–dark–erase curves for high-stoichiometry crys-tals: ~a! SLN1 ~undoped! and ~b! SLN2 ~doped with 100 ppm Fe!showing the effect of Fe in eliminating the fast component of thedark decay. This occurs because the presence of Fe shortens thelifetime of the intermediate level, as shown in Fig. 16 for a lower-stoichiometry sample.


same. This result focuses attention on ways to en-hance absorption at the writing wavelength. Fac-tors that determine the sensitivity have beeninvestigated, and optimization requires control of thestoichiometry, the degree of reduction, the tempera-ture, the gating wavelength, and the gating intensity.Two-color materials differ fundamentally from one-color materials in that the dynamic range or M# canbe increased by use of a higher writing intensity andthe sensitivity can be increased with a higher gatingintensity. Another route to increasing the M# wouldbe to find a material that exhibits a two-color erasureprocess. Substantial progress has been made in re-cent years in the field of two-color holography, andfurther progress can be expected on this complex andchallenging problem.

This research was carried out within the Photore-fractive Information Storage Materials ~PRISM! con-ortium, funded in part by the Defense Advancedesearch Projects Agency ~DARPA! under agreementDA972-94-2-0008. We acknowledge numerous

ruitful discussions with colleagues at IBM, espe-ially with G. Burr and R. Shelby, as well as withembers of the PRISM consortium, particularly Y.ai @Stanford Research International ~SRI!#, L. Hes-elink ~Stanford University and Optitek!, R. KachruSRI!, S. Orlov ~Optitek! and R. Schwartz ~Hughes!.. Guenther acknowledges support from the Deut-

che Forschungsgemeinschaft.

References1. A. Ashkin, G. D. Boyd, J. M. Dziedzic, R. G. Smith, A. A.

Ballmann, J. J. Levinstein, and K. Nassau, “Optically-inducedrefractive index inhomogeneities in LiNbO3 and LiTaO3,”Appl. Phys. Lett. 9, 72–74 ~1966!.

2. F. S. Chen, J. T. LaMacchia, and D. B. Fraser, “Holographicstorage in lithium niobate,” Appl. Phys. Lett. 13, 223–225~1968!.

3. G. A. Alphonse and W. Phillips, “Read–write holographic mem-ory with iron-doped lithium niobate,” Ferroelectrics 11, 397–401 ~1976!.

4. D. Psaltis and F. Mok, “Holographic memories,” Sci. Am. 273~5!, 70–76 ~1995!.

5. J. Heanue, M. Bashaw, and L. Hesselink, “Volume holographicstorage and retrieval of digital data,” Science 265, 749–752~1994!.

6. M.-P. Bernal, H. Coufal, R. K. Grygier, J. A. Hoffnagle, C. M.Jefferson, R. M. Macfarlane, R. M. Shelby, G. T. Sincerbox, P.Wimmer, and G. Wittmann, “A precision tester for studies ofholographic optical storage materials and recording physics,”Appl. Opt. 35, 2360–2374 ~1996!.

7. K. Kitamura, Y. Furukawa, Y. Ji, M. Zgonik, C. Medrano, G.Montemezzani, and P. Guenter, “Photorefractive effect inLiNbO3 crystals enhanced by stoichiometry control,” J. Appl.Phys. 82, 1006–1009 ~1997!.

8. F. Jermann and J. Otten, “Light-induced charge transport inLiNbO3:Fe at high light intensities,” J. Opt. Soc. Am. B 10,2085–2092 ~1993!.

9. J. J. Amodei and D. L. Staebler, “Holographic pattern fixing inelectro-optic crystals,” Appl. Phys. Lett. 18, 540–542 ~1971!.

10. H. Vormann, G. Weber, S. Kapphan, and E. Kratzig, “Hydro-gen as origin of thermal fixing in LiNbO3:Fe,” Solid StateCommun. 40, 543–545 ~1981!.

11. R. Mueller, L. Arizmendi, M. Carrascosa, and J. M. Cabrera,

“Time evolution of grating decay during photorefractive fixing

2

2

2

32. L. Arizmendi, J. M. Cabrera, and F. Agullo-Lopez, “Defects

3

3

3

3

4

4

4

5

processes,” J. Appl. Phys. 77, 308–312 ~1995!.12. A. Yariv, S. S. Orlov, and G. A. Rakuljic, “Holographic storage

dynamics in lithium niobate: theory and experiment,” J. Opt.Soc. Am. B 13, 2513–2523 ~1996!.

13. M. Bashaw and J. Heanue, “Quasi-stabilized ionic gratings inphotorefractive media for multiplex holography,” J. Opt. Soc.Am. B 14, 2024–2042 ~1997!.

14. F. Micheron and G. Bismuth, “Electrical control of fixation anderasure of holographic patterns in ferroelectric materials,”Appl. Phys. Lett. 20, 79–81 ~1972!.

15. F. Micheron and G. Bismuth, “Field and time thresholds forthe electrical fixation of holograms in ~Sr0.75Ba0.25!Nb2O6 crys-tals,” Appl. Phys. Lett. 23, 71–72 ~1973!; J. B. Thaxter and M.Kestigian, “Unique properties of SBN and their use in a lay-ered optical memory,” Appl. Opt. 13, 913–924 ~1974!.

16. Y. Qiao, S. Orlov, D. Psaltis, and R. R. Neurgaonkar, “Electri-cal fixing of photorefractive holograms in Sr0.75Ba0.25Nb2O6,”Opt. Lett. 18, 1004–1006 ~1993!.

17. H. C. Kuelich, “A new approach to read volume holograms atdifferent wavelengths,” Opt. Commun. 64, 407–411 ~1987!.

18. D. Psaltis, F. Mok, and H. S. Li, “Nonvolatile storage in pho-torefractive crystals,” Opt. Lett. 19, 210–212 ~1994!.

19. E. S. Bjornson, M. C. Bashaw, and L. Hesselink, “Digital quasi-phase-matched two-color nonvolatile holographic storage,”Appl. Opt. 36, 3090–3106 ~1997!.

20. D. von der Linde, A. M. Glass, and K. F. Rodgers, “Multiphotonphotorefractive processes for optical storage in LiNbO3,” Appl.Phys. Lett. 25, 155–157 ~1974!.

21. H. Guenther, G. Wittmann, R. M. Macfarlane, and R. R. Neur-gaonkar, “Intensity dependence and white light gating of two-color photorefractive gratings in LiNbO3,” Opt. Lett. 22, 1305–1307 ~1997!.

2. D. von der Linde, A. M. Glass, and K. F. Rodgers, “Opticalstorage using refractive index changes induced by two-stepexcitation,” J. Appl. Phys. 47, 217–220 ~1976!.

3. K. Buse, F. Jermann, and E. Kratzig, “Two-step photorefrac-tive hologram recording in LiNbO3:Fe,” Ferroelectrics 141,197–205 ~1993!.

4. K. Buse, F. Jermann, and E. Kratzig, “Infrared holographicrecording in LiNbO3:Fe and LiNbO3:Cu,” Opt. Mater. 4, 237–240 ~1995!.

25. F. Jermann, M. Simon, and E. Kratzig, “Photorefractive prop-erties of congruent and stoichiometric lithium niobate at highlight intensities,” J. Opt. Soc. Am. B 12, 2066–2070 ~1995!.

26. Y. S. Bai and R. Kachru, “Nonvolatile holographic storage withtwo-step recording in lithium niobate using cw lasers,” Phys.Rev. Lett. 78, 2944–2947 ~1997!.

27. D. Lande, S. S. Orlov, A. Akella, L. Hesselink, and R. R. Neur-gaonkar, “Digital holographic storage system incorporating op-tical fixing,” Opt. Lett. 22, 1722–1724 ~1997!.

28. N. Iyi, K. Kitamura, F. Izumi, J. K. Yamamoto, T. Hayashi, H.Asano, and S. Kimura, “Comparitive study of defect structuresin lithium niobate with different compositions,” J. Solid StateChem. 101, 340–352 ~1992!.

29. D. M. Smyth, “Defects and transport in LiNbO3,” Ferroelec-trics 50, 93–102 ~1983!.

30. M. G. Clark, F. J. DiSalvo, A. M. Glass, and G. E. Peterson,“Electronic structure and optical index damage of iron-dopedlithium niobate,” J. Chem. Phys. 59, 6209–6219 ~1973!.

31. K. L. Sweeney and L. E. Halliburton, “Oxygen vacancies inlithium niobate,” Appl. Phys. Lett. 43, 336–338 ~1983!.

1

induced in pure and doped LiNbO3 by irradiation and thermalreduction,” J. Phys. C 17, 515–529 ~1984!.

33. O. F. Schirmer, S. Juppe, and J. Koppitz, “ESR—optical andphotovoltaic studies of reduced LiNbO3,” Cryst. Latt. Def.Amorph. Mater. 16, 353–357 ~1987!.

34. O. F. Schirmer, O. Thiemann, and M. Woehlecke, “Defects inLiNbO3—experimental aspects,” J. Phys. Chem. Solids 52,185–200 ~1991!.

5. J. L. Ketchum, K. L. Sweeney, L. E. Halliburton, and A. F.Armington, “Vacuum annealing effects in lithium niobate,”Phys. Lett. 94A, 450–453 ~1983!.

6. P. Lerner, C. Legras, and J. P. Duman, “Stoechiometrie desmonocristaux de metaniobate de lithium,” J. Cryst. Growth3y4, 231–235 ~1968!.

37. D. Dutt, F. J. Feigl, and G. G. DeLeo, “Optical absorption andelectron paramagnetic resonance studies of chemically re-duced lithium niobate,” J. Phys. Chem. Solids 51, 407–415~1990!.

8. H.-J. Reyher, R. Schulz, and O. Thiemann, “Investigation ofthe optical absorption bands of Nb41 and Ti31 in lithium nio-bate using magnetic circular dichroism and optically detectedmagnetic-resonance techniques,” Phys. Rev. B 50, 3609–3619~1994!.

9. Y. Furukawa, M. Sato, K. Kitamura, and F. Nitanda, “Growthand characterization of off-congruent LiNbO3 single crystalsgrown by the double crucible method,” J. Cryst. Growth 128,909–914 ~1993!.

40. B. C. Grabmaier and F. Otto, “Growth and investigation ofMgO doped LiNbO3,” J. Cryst. Growth 79, 682–688 ~1986!.

41. R. L. Byer, J. F. Young, and R. S. Feigelson, “Growth of high-quality LiNbO3 crystals from the congruent melt,” J. Appl.Phys. 41, 2320–2325 ~1970!.

42. U. Schlarb and K. Betzler, “Refractive indices of lithium nio-bate as a function of temperature, wavelength and composi-tion: a generalized fit,” Phys. Rev. B 48, 15,613–15,620~1993!.

43. I. Baumann, P. Rudolph, D. Krabe, and R. Schalge, “Ortho-scopic investigation of the axial optical and compositional ho-mogeneity of Czochralski grown LiNbO3 crystals,” J. Cryst.Growth 128, 903–908 ~1993!.

44. O. F. Schirmer and D. von der Linde, “Two-photon and x-ray-induced Nb41 and O2 small polarons in LiNbO3,” Appl. Phys.Lett. 33, 35–38 ~1978!.

45. A. Grone and S. Kapphan, “Combination bands of libration 1vibration of OHyOD centres in ABO3 crystals,” J. Phys. Cond.Matter 7, 3051–3061 ~1995!.

46. R. Richter, T. Bremer, P. Hertel, and E. Kratzig, “Refractiveindex and concentration profiles of proton-exchanged LiNbO3

waveguides,” Phys. Stat. Solids 114, 765–774 ~1989!.7. P. Gunter and J.-P. Huignard, “Photorefractive materials and

their applications,” in Vol. 61 of Topics in Applied Physics~Springer-Verlag, Berlin, 1988!, pp. 47–52.

8. Y. S. Bai, R. Kachru, L. Hesselink, and R. M. Macfarlane,“Gated recording of holograms using rare-earth doped ferro-electric materials,” U.S. patent 5,665,493 ~9 September 1997!.

9. Y. S. Bai, R. R. Neurgaonkar, and R. Kachru, “High-efficiencynonvolatile holographic storage with two-step recording inpraseodymium-doped lithium niobate by use of continuous-wave lasers,” Opt. Lett. 22, 334–336 ~1997!.

0. F. H. Mok, G. W. Burr, and D. Psaltis, “A system metric forholographic memory systems,” Opt. Lett. 21, 896–898 ~1996!.


Documents

Two-Color Holography in Reduced Near-Stoichiometric Lithium Niobate