a applications nonlinear optical spectroscopyvibrational spectroscopy is capable of providing most...

Proc. Natl. Acad. Sci. USAVol. 93, pp. 12104-12111, October 1996Applied Physical Sciences

This contribution is part of a special series of Inaugural Articles by members of the National Academy of Scienceselected on April 25, 1995.

A few selected applications of surface nonlinearoptical spectroscopyY. R. SHENDepartment of Physics, University of California, and Materials Sciences Division, Lawrence Berkeley National Laboratory, Berkeley, CA 94720

Contributed by Y R. Shen, August 5, 1996

ABSTRACT As a second-order nonlinear optical process,sum-frequency generation is highly surface-specific and ac-cordingly has been developed into a very powerful and ver-satile surface spectroscopic tool. It has found many uniqueapplications in different disciplines and thus provided manyexciting new research opportunities in surface and surface-related science. Selected examples are discussed here toillustrate the power of the technique.

In recent years, we have been developing and applying modernoptical techniques to surface and interface studies. The onethat has attracted much attention is surface optical secondharmonic generation (SHG) and sum-frequency generation(SFG) (1). This is because the technique has very high surfacespecificity and wide applicability. SFG is a nonlinear opticalprocess in which two laser beams at frequencies wi and w2impinging on a medium mix and generate a sum-frequencyoutput at ws = Wl + W2. SHG is a special case of SFG with wi= w2. It is well known that a second-order nonlinear opticalprocess like SHG and SFG is forbidden in a medium withinversion symmetry (2). At a surface or interface, however, theinversion symmetry is necessarily broken. This then leads tothe surface specificity of SHG and SFG. More formally, SFG(or SHG) originates from a nonlinear polarization P(2)(ws)induced in a medium by laser wave mixing (2):

p(2)(W) = X(2):E()E(W2) [1]

where E(wi) is the laser field at wi and X(2) is a tensor describingthe nonlinear response of the medium to the fields. Inversionsymmetry of the medium yields72) = _.i2), and therefore x(2)must vanish.

Second-order nonlinear optical effects are naturally muchweaker than the linear ones. Typically, x(2) 10-15 esu for asurface molecular monolayer. To generate 100 sum-frequencyphotons per exciting laser pulse from such a monolayer, asignal that can be readily detected, we need to have laser pulseswith an energy of 10 pJ/pulse focused to an area of 0.1 mm2on the surface assuming a 10-ps pulsewidth, or 1 mJ/pulsefocused to an area of 1 mm2 assuming a 10-ns pulsewidth (2,3). Focusing of the laser beams is usually limited by laserdamage of the surface. Since both picosecond and nanosecondpulsed lasers with the required energies per pulse are easilyavailable nowadays in a laboratory, SHG or SFG as a surfaceprobe can fairly routinely function with a monolayer or sub-monolayer sensitivity. The SHG or SFG output from a surfaceis coherent and highly directional in both transmitted andreflected directions as required by the boundary condition thatthe input and output wavevector components along the surfacemust be matched (4). In usual experiments, detection of theoutput in the reflected direction is preferred.The SFG Lor SHG) measurement on a surface allows the

deduction of X(2)(Ws = Wl + W2), which is a tensor characteristic

of the surface (3). For example, the symmetry of 2) providesinformation about the surface structural symmetry. The spatialvariation of x42) over the surface directly reflects the spatialvariation of the surface structure. The time variation of92)indicates how the surface structure changes with time. Ofparticular importance is that 2) also carries spectroscopicinformation that enables us to learn microscopic details aboutthe surface. As shown in Fig. 1, if w'(w2) or wi + &o2 or bothare at resonances, then'x2) will be resonantly enhanced. Theresonant enhancement in x(2) versus wi and W2 naturally yieldsthe spectral information (3). However, it should be pointed outthat unlike ordinary absorption and emission spectroscopies,the SFG signalS is proportional to IX(2)12 and has contributionsfrom both real and imaginary parts of x(2). Even away fromresonances, S may be finite because of the presence of anonresonant background in X(2). Consequently, the spectralshape of a resonance may not look like that in an absorptionor emission spectrum. Consider for example the case of wi nearresonances. The SFG signal versus wi is given by (3):

2

S(c01)o x5+ 2W1- + i [2]n

where X7 is the nonresonant background in x(2), and n refersto the nth resonance characterized by the frequency wn,damping yn, and amplitude An. Interferences of the variousterms in Eq. 2 can severely distort the lineshapes. Only when2 is negligibly small and resonances sufficiently far apart canwe find resonant peaks similar to those in absorption spectra.However, even if an observed SFG spectrum shows unusualresonant lineshapes, it is possible to use Eq. 2 to fit thespectrum and deduce An, wn, and yn for the resonances.SFG (SHG) as a surface analytical tool has a number of clear

advantages (3). Being an opti-cal technique, it can be used toprobe all types of interfaces as long as they can be assessed bylight. The unusually high surface specificity together with thesubmonolayer sensitivity permits spectroscopic study of aninterface even if the associated bulk absorbs in the samespectral region. Because the SFG output is highly directional,spatial filtering can be employed in addition to spectral filteringto suppress the unwanted luminescence and scattering back-ground noise. The technique can therefore be used for in situremote sensing study of a surface even in a very hostile environ-ment. Finally, the spatial and spectral resolutions of the techniqueare only limited by spatial and spectral mode quality of theexciting lasers. The time resolution is limited by the laser pulsewidth which can be in the picosecond or femtosecond range.

Abbreviations: SHG, second harmonic generation; SFG, sum-frequency generation; OTS, octadecyltrichlorosilane; DOAC, diocta-decyldimethyl ammonium chloride; (s, s, p), (s-output, s-visible, p-infrared); UHV, ultrahigh vacuum.

12104

Dow

nloa

ded

by g

uest

on

June

25,

202

0

Proc. Natl. Acad. Sci. USA 93 (1996) 12105

Visible

02 0)2~~~~~~

et<|1 tc<c11(mIl ~<0l <01<o

(a) (b) (c)

FIG. 1. Schematic diagrams of resonant SFG processes: (a) wl atresonance, (b) Wl plus W2 at resonance, and (c) both cwt and w, plus c2= co at resonance.

These advantages have rendered SFG (SHG) an extremelypowerful and versatile tool for surface and interface studies.Applications that have already been successfully demonstratedappear in almost all areas of surface science, ranging fromphysics to chemistry, biology, and materials science. Many ofthem are highly unique, providing good opportunities toexplore new areas of research in various disciplines. In the caseof SFG, these include, for example, studies of buried interfaces(5-9), interfaces of pure liquids (10-13), electrochemicalprocesses (14-17), catalytic reactions in real atmosphere (17),surface monolayer microscopy (18), ultrafast surface dynamics(19-23), and so on. Quite a few review articles have alreadybeen written on the subject (1, 3, 24-25), a number of whichfocus on specific areas of applications (26-28).

In this paper, I shall discuss only a selected few experimentswe have performed in our laboratory with SFG surfacespectroscopy. They can be taken as examples to illustrate thepower and versatility of the technique. For the sake ofcompleteness, I shall begin with a brief discussion of theexperimental arrangement.

EXPERIMENTALFig. 2 shows a schematic diagram of a typical experimentalsetup for SFG surface spectroscopy. Two exciting laser (orlaser-like) beams at frequencies w, and 02, respectively, areneeded. While one can be of fixed frequency, the other mustbe tunable. The most convenient laser-like tunable sourcenowadays is the optical parametric systems (29). Together withsum- and difference-frequency generation stages, they canhave a continuous tuning range from 0.2 ,um to 10 Am or wider.In the longer wavelength region, however, free electron lasersare most appropriate since they can provide picosecond lJpulses readily tunable from 3 ,um to 100 ,um (30, 31, *). Fornonlinear optical experiments, shorter laser pulses often yieldlarger signals. In the case of surface SFG, the output signal isproportional to EIE2/AT (3), where E1 and E2 are the pulseenergies of the two exciting laser pulses impinging on thesurface, A is the beam spot size on the surface, and T is thelaser pulsewidth. For the same pulsed laser energies, the signalis inversely proportional to the pulsewidth. For this reason,although even continuous wave (cw) lasers could be used insome surface studies by SHG or SFG, ultrashort pulsed lasersare generally preferred. We also notice that the signal isinversely proportional to the beam spot size on the surface. Forgiven laser pulse energies, the minimum beam spot size is,however, limited by laser damage of the surface. For example,the damage threshold for metal surfaces is of the order ofmJ/cm2, but for liquid surfaces, it can be of the order of J/cm2.

Depicted in Fig. 2 is the usual geometry adopted for SFGmeasurements. The SFG output in the reflected direction isdetected and recorded. Other geometries can of course be

IR

InterferencePolanzer tilters

PMTHalf waveplate

Fresnelrhombhalfwaveplate Spacial filters

FIG. 2. A typical experimental setup for SFG.

used, depending on the practical situation. As discussed ear-lier, the output signal is rather weak, often of the order of 100photons per pulse or weaker. With the exciting laser energiesat ' 100 pJ/pulse (--1015 photons/pulse), discrimination of thesignal against laser scattering, and fluorescence background isnot so trivial, especially if the output sum frequency is not farfrom the input laser frequencies. Fortunately, the SFG outputis highly directional and is in a direction different from thoseof the reflected laser beams. Spatial filtering can therefore beused to suppress the unwanted background by many orders ofmagnitude. Together with spectral filtering by filters and/ormonochromators, it makes the SFG output readily detectableeven if the sum frequency and the laser frequency are only fewhundred cm-' (hundredths of eV) apart. This latter casehappens when SFG is used to probe surface excitations in themid-IR range (cot scanned over mid-IR resonances). To mea-sure various components of the x(2) tensor, the input laserbeams must be selectively polarized and the output polariza-tion of SFG analyzed.The fixed laser frequency (@2) is normally chosen to place

the output sum frequency (ws) in the visible so that the SFGoutput can be detected by a sensitive photomultiplier. Thesignal is then recorded by a gated integrator connected to acomputer. For small signal levels, a photon counting mode ofoperation is preferred.

SURFACE SPECTROSCOPIC STUDIESNonlinear optical surface studies were first developed withSHG requiring only a single fixed-frequency laser (32). Thefirst spectroscopic SHG measurement was performed on halfmonolayers of rhodamine dye adsorbed on quartz using atunable dye laser (33). The electronic resonances of theadsorbed rhodamine molecules were probed. Both SHG andSFG spectroscopy have now been used to study electronictransitions involving surface or interface states of metals andsemiconductors (5-9), in particular, buried semiconductorinterfaces which are difficult to investigate by other nonde-structive techniques.

In many areas of surface science and technology, surfacevibrational spectroscopy is capable of providing most usefulinformation. Known as finger prints of molecules, vibrationalspectra permit identification of molecular species, their ori-entations and conformations, and possibly their interactionswith neighbors. Surface SFG vibrational spectroscopy (Fig. lawith ctw scanned over vibrational resonances in the IR) hasindeed been very successful in finding meaningful applicationsto various subfields of surface science. In particular, the uniquecapabilities of SFG have enabled us to explore some importantareas that have been very much neglected because of lack ofsuitable experimental probes. A number of them that has beenworked out recently in our laboratory will be discussed below.

Surfactant Monolayers. Surfactants play a key role in manyimportant industrial processes-e.g., detergency, oil recovery,lubrication, painting, cosmetics, etc. (34, 35). Their formationof micelles and lamillar layers is closely related to biologicalmembranes. Chain orientations and conformations of surfac-tant monolayers adsorbed at interfaces have been a subject ofgreat interest to many researchers because they can directlyaffect the interfacial properties. IR and Raman spectroscopies

*Eliel, E. R., van der Ham, E. W. M., Vrehen, Q. H. F., t'Hooft, G.W., Auerhammer, J. M., van der Meer, A. F. G., Braun, R. & Bain,C. D., Proceedings of the 15th International Conference on Coherentand Nonlinear Optics, June 1995, St. Petersburg, Russia.

Applied Physical Sciences: Shen

Dow

nloa

ded

by g

uest

on

June

25,

202

0

12106 Applied Physical Sciences: Shen

(36-38) have long been used to probe vibrational spectra ofsurfactant monolayers. In most favorable cases, they can befree from spectral interference from the bulk, but are neces-sarily dominated by the CH2 modes (for surfactants with longhydrocarbon chains) from the large number of CH2 groups onthe chains (36-38). SFG vibrational spectroscopy is verydifferent. On a straight hydrocarbon chain, the distribution ofCH2 groups is nearly centrosymmetric. Consequently, the CH2modes nearly vanish, leaving the modes of the terminal CH3group appearing dominant in the spectrum. This is shown inFig. 3 lc and 2c for a compact monolayer of pentadecanoic acidon water (33). The peaks at 2875, 2940, and 2960 cm-1correspond to the CH3 symmetric stretch, antisymmetricstretch, and Fermi resonance between symmetric stretch andbending modes, respectively. If trans-gauche defects on thechain set in, then the CH2 modes will become prominent in thespectrum. Examples are seen in Fig. 3 la, 2a, lb, and 2b for thepentadecanoic acid monolayer densities significantly reducedfrom the fully packed one. The peaks at 2850 and 2925 cm-'can now be assigned to CH2 symmetric and antisymmetricstretches, respectively. This shows that the SFG vibrationalspectra of the hydrocarbon chains can give a clear indicationof whether the chains possess defects or not. An excessivenumber of defects, rendering the chains to contract into ahighly compact conformation, will cause all the spectral fea-tures to diminish as a result of the establishment of a nearinversion symmetry (39).

Unlike IR and Raman spectroscopies, the very high surfacespecificity of SFG makes the surface SFG spectra nearly freeof bulk contribution even in the presence of strong bulkabsorption in the same spectral region. Fig. 4 presents a clearexample (40). For an octadecyltrichlorosilane (OTS) surfac-tant monolayer adsorbed at two different liquid/quartz inter-faces (buried interfaces), CCl4/OTS/quartz and hexadecane/OTS/quartz, the spectrum is completely dominated by theOTS monolayer. Only the CH3 symmetric stretch, antisym-metric stretch, and Fermi resonance modes at 2875, 2940, and2960 cm-1, respectively, show up in the spectrum, indicatingthat the OTS chains are straight with little defects. In thehexadecane/OTS/quartz case, even though hexadecane liquidhas very strong bulk absorption in the CH stretch regionbecause of the large number of CH2 groups on the molecules,hardly any spectral features from hexadecane can be perceivedin the spectrum of Fig. 4. As a further proof, the SFG spectrum

1

D

<0

zi

0

zuJ

0)

L1

0k

22A2 /MOL 1 c)

34A 2 /MOL 1 b)

47A2 /MOLa)

02950 2850 2950 2850

INFRARED FREQUENCY (CM-1)

FIG. 3. SFG spectra of pentadecanoic acid at different surfacecoverages normalized per molecules. (la-ic) Data were taken with the(s-output, s-visible, p-infrared) polarization combination. (2a-2c)Data were taken with the (s, p, s) polarization combination: (a) 47A2/molecule, (b) 34 A2/molecule, and (c) 22 A2/molecule. [Repro-duced with permission from ref. 39 (Copyright 1987, AmericanPhysical Society).]

-j

z

CD

zLU

:Da

Lu

U_ 2950 2850

Dr FREQUENCY (cm-)

FIG. 4. SFG spectra at different interfaces in the (p-out, p-vis,p-infrared) polarization combination. ---, Silica-hexadecane interface;*, silica-OTS-CCl4 interfaces; and A, silica-OTS-hexadecane inter-face. [Reproduced with permission from ref. 40 (Copyright ElsevierScience).]

was also taken for the hexadecane/quartz interface withoutthe OTS monolayer. As shown in Fig. 4, no prominent spectralfeatures can be easily identified, indicating that the spectrumdue to bulk hexadecane must be very weak.The chain conformation of a surfactant monolayer at an

interface depends on the chain length, the chain density, andthe interfacial environment (P. Miranda, V. Pflumio, H. Saijo,and Y.R.S., unpublished). For a fully packed chain density, thechain-chain interaction effectively keeps the chains in theall-trans configuration, which is not likely to be affected by theopposing medium. This is the case of the OTS monolayer onquartz discussed above. The spectrum of OTS shown in Fig. 4remains essentially unchanged whether the chains are exposedto CC14, or hexadecane, or water, or air (40). The same is nottrue for surfactant monolayers with lesser chain densities. Thereduced chain-chain interaction and the spacing betweenchains now allow trans-gauche defects to develop. For a givenchain density, the chain conformation may depend on the fluidmedium the chains are facing. From what we have observed inthe cases of dioctadecyldimethyl ammonium chloride(DOAC), n,n-dimethyl-n-octadecyl-3-amino propyltrimethox-ysilylchloride, and other low-chain-density (>40 A2/molecule)surfactant monolayers adsorbed at liquid/solid interfaces, werealize that the conformational changes can be drastic(Miranda et al., unpublished).We consider here DOAC monolayers as an example. The

chain density of a full DOAC monolayer adsorbed on quartzis 3 x 1014 chains/cm2 (35 A2/chain), about two times less thanthat of a full OTS monolayer. Fig. 5 shows SFG vibrationalspectra in the CH stretch region of the adsorbed DOACmonolayer immersed in various liquids. In nonpolar liquid

.-

Cd'aCD

(0cn

2800 2900

Frequency ( cm 1 )3000

FIG. 5. SFG spectra with the (s, s, p) polarization combination ofa DOAC monolayer at various quartz/liquid interfaces. 0, Quartz/CC14 interface; *, quartz/(CCl4 plus 10% CDCl3) interface; figchr;V,quartz/(CCl4 plus 60% CDCl3) interface; v, quartz/CD30D inter-face; E, quartz/D20 interface. (Mirandra et al., unpublished.)

6 0

0

0~ ~~~1

,o~~04 0I

#0 0

I00 0I/ 0% 00 0

Proc. Natl. Acad. Sci. USA 93 (1996)

j

II

Dow

nloa

ded

by g

uest

on

June

25,

202

0


CC14, the spectrum is not very different from that of a DOACmonolayer exposed in air, and is characterized by the presenceof CH2 (at 2840 and 2920 cm-1) and CH3 (at 2875 and 2940cm-') peaks of nearly equal strengths. This indicates theexistence of appreciable amount of defects on the chains. InCCl4-CDCl3 liquid mixtures, as the fraction of the polarcomponent increases, the spectrum decreases in its overallintensity. Finally, in hydrogen-bonding (or associative) liquidssuch as water and methanol, the spectrum completely disappears.These results suggest that the number of defects on each chainmust have increased significantly as the polarity or hydrogen-bonding character of the liquid increases. Presumably in a lessfavorable solvent, the chains like to fold into a more compactconformation to reduce their surface areas exposed to thesolvent. Unfortunately, theoretical calculation that can give anobserved SFG spectrum a quantitative evaluation of the corre-sponding chain conformation is not yet available.Immersion of the adsorbed DOAC monolayer in alkane

liquids could also have the chain conformation changed dras-tically (Miranda et al., unpublished). In a study using alkane ofvarious chain lengths, we have found that the amount of defectson the DOAC chains is greatly reduced when the alkane chainlength is close to that of DOAC. Fig. 6 provides an example.The spectrum of the DOAC monolayer in hexadecane showsthat all the CH2 peaks are now clearly suppressed, indicatingconversion to a nearly all-trans conformation for the DOACchains. The chain-chain interaction between DOAC andhexadecane must be responsible for the change. The sameinteraction may not be sufficient to eliminate the chain defectsif the surfactant chain length is too short or the chain densitytoo low (Miranda et al., unpublished).

Liquid Interfaces. For science and technology, liquid inter-faces are at least as important as solid surfaces. Their prop-erties directly influence many aspects of our life. Yet becausefew techniques are applicable to liquid interfaces, their exper-imental studies have been rather limited. In particular, on pureliquid interfaces, few spectroscopic investigations have everbeen reported. Conventional spectroscopic techniques do nothave the needed surface specificity and therefore the observedspectra are generally overwhelmed by the bulk contribution. Inthis respect, SFG is currently the only effective spectroscopictool applicable to liquid interfaces (apart from studies ofadsorbates at interfaces). We consider here a few examples ofhow SFG vibrational spectroscopy can be used to deduceuseful information about liquid interfacial structures.Water is the most important liquid on earth (41). Water

interfaces play key roles in many important problems ofscience and technology, namely, wetting and drying, electro-chemistry, environmental chemistry, corrosion, membrane

24

1!

0-

0

U._U)

1'

2800 2900

Frequency (cm 1)3000

FIG. 6. SFG spectrum with the (s, s, p) polarization combination

of a DOAC monolayer at the quartz/hexadecane interface. Thepolarization combination is (s, s, p). (Mirandra et al., unpublished.)

formation, protein hydration, and so on. Despite numeroustheoretical works on the subject, experimental investigationshave been largely limited to the use of surface tension, surfacepotential, and ellipsometry measurements. Comparison be-tween theory and experiment has failed to provide definitiveinformation about the water interfacial structures. Now sur-face vibrational spectroscopy permits a change of the situation.

Depicted in Fig. 7 are the SFG spectra of the air/waterinterface in the OH stretch region (11). They are among thefirst vibrational spectra ever obtained for a neat liquid inter-face. The one with the (s, p, s) polarization combination isweak and shows hardly any discernible features. The (s, s, p)spectrum, on the other hand, exhibits three prominent peaksat 3680, 3400, and 3200 cm-. They can be attributed to OHstretches associated with dangling OH bonds, OH with its Hconnected to a neighboring molecule in a local disorderedstructure, and OH connected to neighbors in an orderedice-like structure, respectively. The 3680-cm-1 peak is partic-ularly significant. Its presence indicates that the spectrum mustcome from the surface monolayer of water since no danglingOH bonds should exist in the sublayers. Furthermore, toaccount for the dangling OH bonds, at least part of the surfacewater molecules must be oriented with one H pointing out ofthe liquid. Quantitative calibration of the spectrum allowed usto find that about 25% of the water molecules in the surfacemonolayer must have contributed one dangling bond permolecule. This is the same as what one would find on abulk-terminated crystalline ice surface. The result thereforesuggests that the surface water structure at the air/waterinterface is ice-like (43). However, unlike ice, the hydrogen-bonding network of surface water is highly disordered asevidenced by the presence of the strong 3400-cm-1 peak.Physically, the ice-like structure is the consequence of atendency for the surface layer to have the maximum numberof hydrogen bonds. In a recent molecular dynamics calculation,Benjamin (44) was able to reproduce the spectrum of Fig. 7 forthe air/water interface fairly well.Also shown in Fig. 7 is a spectrum of an air/water interface

with a fully packed long-chain alcohol monolayer adsorbed atthe interface (11). The 3680-cm-1 peak is now completelysuppressed as expected since the dangling OH bonds must beeliminated by the adsorption of alcohols. However, the 3400-cm-1 peak also seems to have disappeared, making the spec-trum resemble that of ice (shown as the dashed curve in Fig.7). This suggests that the alcohol monolayer, is capable of

1.0 [

.-

Xo 0.5cA

nnA3000 3200 3400 3600 3800

Wavenumber (cm 1)XBL 92114"S

FIG. 7. SFG spectrum (dotted line), with the (s, s, p) polarizationcombination, of water surface covered by a stearyl alcohol monolayercompressed to 10 mN/m. For comparison, the SFG spectrum of purewater surface (dashed line) and the IR absorption spectrum of ice(solid line) (from ref. 42) are also shown. [Reproduced with permis-sion from ref. 11 (Copyright 1993, American Physical Society).]

0~~~~

5 ~~~~~~or-5 IC

6?~~~* 'P t

I ~ ~ ~ ~ I


Dow

nloa

ded

by g

uest

on

June

25,

202

0


inducing a surface water structure more ordered like ice. Thatthis is indeed possible has been pointed out by Gavish et al.(45). They realize that the lattice structure of the alcoholmonolayer matches fairly well with that of hexagonal ice,rendering ice nucleation by epitaxial induction under thealcohol monolayer possible. The result here is an example ofhow SFG vibrational spectroscopy can be used to probe crystalnucleation at a surface in a solution. Crystal growths in manybiological systems are believed to be initiated this way (46).

Hydrophobicity and hydrophilicity of a surface are of greatimportance in chemistry and biology, but they have never beencharacterized at the molecular level. Now surface vibrationalspectroscopy allows us to do so (13). We found that thedangling OH peak of water at 3680 cm-' is a signature ofhydrophobicity (or nonwetting). Its disappearance at a water/hydrophilic interface signifies termination of the dangling OHbonds at the interface by hydrophilic interaction. On the otherhand, its presence at a water/hydrophobic interface indicatesthat the dangling OH bonds remain free against the hydro-phobic surface. Fig. 8 displays the SFG vibrational spectra forthree water/hydrophobic interfaces: water/air, water/OTS/coated quartz substrate, and water/hexane liquid. In all cases,the presence of the 3680 cm-' peak is clear, and its strengthindicates that -25% of water molecules in the surface mono-layer have contributed one dangling OH bond per molecule.Again this suggests that the interfacial water structure isice-like. However, the three spectra in the bonded OH region(3000-3600 cm-') are quite different. For the water/hydrophobic solid interface, the spectrum closely resembles thatof ice (shown in Fig. 8D) with the 3400-cm-1 peak very muchsuppressed, but for the water/air and water/oil interfaces, the3400 cm-' peak appears strong. Clearly, the ice-like interfacialwater structure is much more ordered like ice in the water/hydrophobic solid case. This can be understood ifwe realize thatagainst a solid wall, the surface water layer is no longer so flexibleas to allow much disordering in its hydrogen bonding network.The experimental result described here agrees well with theprediction of Lee et al. (47). For the water/air and water/oilinterfaces, the intrinsic flexibility of a fluid surface naturallycauses more disordering of the bonding network.The water/quartz interface is hydrophilic. As expected, the

3680-cm-1 peak is absent in the vibrational spectrum (12).Usually a clean quartz surface is terminated by hydrogen. In

1.0

0.8

0.6

0.4

~ 0.2

0.01.0

, 0.8C,)

0.6

0.4

0.2

0.0

_ (A) I

AA.

Xl~~~III.. ..

_ (C)

_tJk .

1 0.

(B)

06 N

0.4 1-04~~

0 20

2.5

2.0

1.5

1.0

0.5

0.03000 3500 4000 3000 3500 4000

contact with water, the surface can remain neutral or benegatively charged by deprotonation depending on the pHvalue of water. The vibrational spectrum of water at theinterface varies accordingly, as shown in Fig. 9 (12). Both highand low pH values yield a spectrum resembling that of ice,indicating that the hydrogen-bonding network of the interfa-cial water layer is well ordered. Intermediate pH values lead toa spectrum representing a more disordered bonding network.This can be understood as follows. Water molecules in theinterfacial layer are bonded into a more or less ice-likestructure, but they also experience two oppositely orientingforces created by the surface: a hydrogen-bonding force thatwould orient the water molecules to have their oxygen bondedto hydrogen on the surface, and a surface field from surfacecharges that would orient the water molecules with theirpermanent dipoles facing the surface (or their oxygen pointingaway from the surface). The former dominates at low pHvalues, while the latter dominates at high pH values. In bothcases, the surface-originated orienting force helps to establisha more ordered interfacial water structure. For intermediatepH values, the two forces are of comparable magnitude butopposite in direction. Their competition in aligning the watermolecules would cause disordering in the water interfacialstructure. We note that the above picture predicts oppositeaverage orientations of surface water molecules in the low andhigh pH cases. This was actually observed as a 180° phase shiftin the SFG output between the two cases (12). The results hereshow that the surface water structure next to an electrode canbe highly ordered. This finding should be important in oursearch for a better understanding of electrochemistry. SFGspectroscopy is also an effective probe for in-situ studies ofelectrochemical reactions at surfaces (14-16). It has alreadybeen used in several cases to identify molecular species ap-pearing and disappearing at the electrode surface in anelectrochemical process (14-16).SFG spectra for other neat liquid interfaces have also been

measured (48-51). They all provide useful information aboutthe liquid interfacial structures. For example, n-alkane(CnH2n+2) liquids with n 2 16 exhibit a surface monolayer

15 0.6Q '

1.2 f (a) 05 r(b6

* O.4~~0*4-

0.90.o-

50. 3000.6 * ~~~ - F.

0.3 0.jiu 0.0 0.0'

1.5 6

<1.2 ~~(c) 5- (d)

0.9 -

co 3 :.0.6-

0.3 1 1

0.0 0cu 10 ....2.5

/)8 1.0

CT.~~~~~~~~.

0.02500 3000 3500 4000 2500 3000 3500 4000

Wavenumber (cm ) Wavenumber (cm-) Wavenumber (cm-')

FIG. 8. SFG spectra from the quartz/OTS/water interface (A), theair/water interface (B), the hexane/water interface (C), and thequartz/ice interface (D). The sum frequency (SF) output, visibleinput, and IR input were s-, s-, and p-polarized, respectively. [Repro-duced with permission from ref. 13 (Copyright 1994, AmericanAssociation for the Advancement of Science).]

FIG. 9. SFG spectra from the quartz/water interfaces with differ-ent pH values in the bulk water: a, pH 1.5; b, pH 3.8; c, pH 5.6; d, pH8.0; and e, pH 12.3. (f) SFG spectrum from the quartz/ice interface.The lines are guides for the eyes; a typical error bar for the measure-ment is shown in b. [Reproduced with permission from ref. 12(Copyright 1994, American Physical Society).]

Wavenumber (-m-')


Dow

nloa

ded

by g

uest

on

June

25,

202

0


freezing transition at a temperature several degrees above thebulk freezing temperature (52-54). The surface vibrationalspectra of n-alkane show clearly that the monolayer freezingtransition is characterized by a conformational change ofsurface molecules from a configuration of vertically orientedchains with significant defects to a well-ordered nearly all-transconfiguration (51). For some liquid mixtures, namely, water-ethanol and water-CH3CN, SFG spectra also suggest theoccurrence of a phase transition in the surface structure whenthe bulk composition is varied (55, 56). In general, one wouldhope that through surface vibrational spectroscopy, enoughinformation could be obtained to permit deduction of ageneral understanding of liquid surfaces and interfaces. Cur-rently the difficulty often lies in our inability to properlyidentify the observed spectral features. Much theoretical andexperimental work is needed before we can unambiguouslyrelate a surface spectrum to a surface structure.

Surface Reactions and Catalysis. In situ probing of adsor-bates and surface reactions is another area we have foundunique applications for SFG spectroscopy. There already existmany techniques for such studies in ultrahigh vacuum (UHV)(57), but few of them are applicable to samples in a realatmosphere. IR spectroscopy, for example, can be employed insome cases, but its surface sensitivity and specificity are oftennot sufficient. Furthermore, none of the techniques has a timeresolution in the picosecond or subpicosecond range.SFG spectroscopic studies of adsorbates on metals and

semiconductors have been carried out by a number of groups(5-9, 14-16, 19-28). Here we consider adsorptions of H andCH3 on diamond C(111) studied in our laboratory (58-61).For the understanding of diamond film growth by chemicalvapor deposition (62-64), a clear understanding of how H andCH3 adsorb and how CH3 converts to CH on the diamondsurfaces are necessary. In situ surface vibrational spectra couldprovide the needed information. Indeed high-resolution elec-tron energy loss spectroscopy has been employed for thispurpose (65-67), but the reported spectra all show insufficientspectral resolution that could distinguish CH from CH2 andCH3. IR spectroscopy with a multiple reflection geometrycould be used to yield clear, highly resolved spectra (68).Unfortunately, this requires a diamond sample of a size that isprohibitively expensive. Thus SFG vibrational spectroscopybecomes a unique tool for such studies.

Fig. 10a shows the SFG spectra of CH stretching andbending modes ofH on a well-ordered C(111)-(1 x 1) surface(58, 59). Similar to H/Si(111), a single sharp peak appears inboth the CH stretch and bend regions. This result together withthe polarization dependence of the spectrum indicates that Hsits nearly vertically on top of C, in agreement with thetheoretical prediction (69). As in the case of H/Si(111), otherproperties of the CH stretch mode of H/C(111), such asanharmonicity and vibrational lifetime, could be measuredand compared with theories (58, 59). The observation of thesharp CH stretch and bend peaks is actually quite importantsince it asserts that allH must have been adsorbed at the stable,well-ordered C(l 1 1)-(1 x 1) top sites with an Sp3 bonding.They can therefore be used as signature of a well-ordered,H-covered, C(111)-(1 x 1) surface. Such a surface, we found,can be reliably created by dosing H on C(111) at 850°C, and isbelieved to be the functional surface in the chemical vapordeposition (CVD) diamond growth along [111]. If the surfaceis contaminated or disordered with H appearing at metastablesites, the SFG spectrum would exhibit weaker and broaderpeaks with subsidiary features around them. One metastablesite that H would first adsorb at shows up as a CH stretch peakblue-shifted by 30 cm-1 from the stable one (58, 59, 70) (seeFig. 10b). It can be converted to the stable CH stretch byincreasing H coverage or annealing of the substrate.On SFG spectroscopic studies of adsorption of CH3 and

conversion of CH3 to CH on C(111), an example is depicted in

a1.0

° 0.8._

o 0.60

0 0.4

r._X' 0.2

0.n

b

0-1

c0

-..

cn

1310 1320 1330 1340 2830 2840 2850 2860Frequency (cmI)

2750 2800 2850 2900

Wavenumber (cm-1 )FIG. 10. (a) SFG spectra (s, s, p) of CH bending and stretching

modes for H/C(111) at 300 K (0) and 700 K (0). [Reproduced withpermission from ref. 60 (Copyright 1995, American Physical Society).](b) SFG spectra of H/C(111) from a freshly formed (1 x 1) surfacewith different H coverages. (After refs. 58 and 59.)

Fig. 11 (61). In the experiment, the methyl radicals wereproduced by thermal decomposition of di-per-butyl peroxide[(CH3)3(OOC(CH3)3] through a hot filament at '800°C. Theiradsorption on C(111) at room temperature is evidenced by thepresence of three resonant modes in the spectrum of Fig. 11,spectrum b: the CH3 symmetric stretch, antisymmetric stretch,and stretch-bend Fermi resonance at 2895, 2975, and 2935cm-1, respectively. (The unusual lineshapes result from inter-ferences between various terms in Eq. 2 as discussed in theIntroduction.) To see that the adsorbed CH3 can be convertedto CH, we show in Fig. 11, spectra b-g, how as the substratetemperature increases, the CH3 peaks diminish while the CHstretch peaks grow. Note that here two CH stretch peaks showup in the spectra. The high-frequency one corresponds to Hadsorbed at the metastable site discussed above. It disappearsby conversion into the low-frequency mode corresponding toH adsorbed at the normal well-ordered site when the substrateis annealed at high temperatures. The spectra of Fig. 11 showthat conversion of CH3 to CH on C(111) begins at a temper-ature less than 350°C, which is many times lower than thetemperature required for decomposition of CH3 in the gasphase. Obviously the observed process is surface-mediated.This suggests that CVD diamond growth via adsorption andconversion of CH3 is possible.We should remark, however, that all the above experiments

were carried out under high vacuum, and therefore the pro-cesses observed may not be the same as those occurred in a realCVD atmosphere. The same difficulty may exist in studies ofsurface catalytic reactions. In the past decades, a great deal ofinformation on various surface catalytic reactions have beendeduced from investigation of catalytic systems in UHV (57).One must question whether they can actually be used tounderstand surface catalysis in a real atmosphere. The lack ofsuitable techniques has so far prevented in situ studies in the

a

a aa a a

a

v11 /


Dow

nloa

ded

by g

uest

on

June

25,

202

0


25

fl~~~~~~~~~~~~~~~f20

UE(e)

15

(d)

.m

103LL 10

u)

(c)

(b)

(a)

o . . . ., .. .....I...2750 2800 2850 2900 2950 3000 3050

Frequency (cm-1)

FIG. 11. SFG spectra exhibiting the effects of surface annealingafter dosing of di-tert- butylperoxide (DTBP) (through an o800°C hotfilament) on C(111) at room temperature. Spectra: a, nearly-baresurface; b, as-dosed surface; c-g, dosed surface annealed at 350°C,500°C, 620°C, 740°C, and 800°C, respectively. Spectra c-g have beenincrementally shifted upwards by 5 units for clarity. [Reproduced withpermission of ref. 61 (Copyright American Physical Society).]

latter case, but now with SFG spectroscopy, they have becomepossible. Discussed below as an example is an experiment onhydrogenation of ethylene (C2H4) to ethane (C2H6) on Pt(111)carried out jointly with G. Somorjai's group (17).

Ethylene hydrogenation on Pt is an important catalyticreaction. The widely accepted model used to describe thereaction was first proposed by Horiuti and Polanyi in 1934 (71).According to the model (Fig. 12a), ethylene first chemisorbson Pt by forming two or bonds with Pt while H2 adsorbs anddissociates into two H atoms on Pt. The di-of-bonded ethylenethen reacts with H on Pt and is stepwise hydrogenated throughan ethyl (C2H5) intermediate to ethane. To check whether themodel is correct, we need to monitor in situ the surface speciesduring the reaction in the real atmosphere. This can be donewith SFG vibrational spectroscopy. Our experiment was per-

Clea di-a bonded ehyl Clen

Othylene

(a)

x-boadedethylene

(b)

FIG. 12. (a) Horiuti-Polanyi mechanism for ethylene hydrogena-tion on platinum. (b) New mechanism for ethylene hydrogenation on

platinum.

formed with a Pt(111) sample in a batch reactor. The gas witha fixed composition of ethylene, hydrogen, and helium was

circulated constantly in a closed loop that contained a gaschromatograph for measuring the enthane production. SFGspectroscopy monitored simultaneously the appearance ofvarious species on the Pt(111) surface.

Fig. 13 describes the SFG surface vibrational spectra of tworelated cases with essentially the same ethane production rate.In the first case (Fig. 13a), the atmosphere was composed of35 torr ethylene, 100 torr H2, and 625 torr He; the reactionstarted with a clean Pt(111) surface at 295°K. In the secondcase (Fig. 13b), the conditions were the same except that thereaction started with Pt(111) fully covered by ethylidyne. Inboth cases, the ethane production rate was 11 ethane mol-ecules per surface Pt atom per second. The three peaks at 2875,2910, and 3000 cm-' in the spectra of Fig. 13 have beenidentified as signatures for ethylidyne (C2H3), di-o--bondedethylene, and i-bonded ethylene (ethylene bonded weakly toPt through its X orbital), respectively. Ethylidyne resulted fromdehydrogenation of di-cr-bonded ethylene, but as alreadyshown by others (72), it is only a spectator in the ethylenehydrogenation process. In our batch reactor, the variousspecies appearing on the Pt(111) surface were all in dynamicequilibrium with species both on the surface and in theatmosphere. The spectra of Fig. 13 show that in both cases thePt surface was covered mostly by ethylidyne at the 3-foldhollow sites (0.15 monolayer in one case and 0.20 monolayerin the other, with 0.25 monolayer being the saturation cover-

age) and partly by di-o- and I-bonded ethylene. The amountof w-bonded ethylene remained unchanged in the two cases,but that of di-c-bonded ethylene changed by a factor of 4. Sincethe ethane production rates in the two cases were the same, theresult clearly indicates that the di-or-bonded ethylene cannot bethe surface species responsible for the catalytic reaction of

a

b

i, I00

000

'. 0000

di-a bode

0 00000o00000 0°°

l

2850 2900 2950 3000 3050

Frequency (in cm")

2850 2900 2950 3000 3050

Frequency (in cmr')

FIG. 13. (a) SFG spectrum of the Pt(111) surface during ethylenehydrogen with 100 Torr H2, 35 Torr C2H4, and 615 Torr He at 295 K.(b) The SFG spectrum under the same conditions as a, but on a surfacethat was precovered in UHV with 0.25 monolayer of ethylidyne.[Reproduced with permission from ref. 17 (Copyright 1996, AmericanChemical Society).]

0

00q0

co

di-ae boXeded,_

0 0000 0 °°oo°°000e,,.000 oo


Dow

nloa

ded

by g

uest

on

June

25,

202

0


ethylene hydrogenation into ethane. The w-bonded ethylene isthen more likely the responsible one. The correspondingrelevant steps for the hydrogenation process are described inFig. 12b. We notice- that the Tr-bonded ethylene is not theprimary adsorbed species on Pt(111) and is only weakly bondedto the surface. This is presumably why ethylene hydrogenation isnot sensitive to the Pt surface structure (72-74).The above finding suggests that the more weakly bonded

surface species may actually play a dominant role in a surfacecatalytic reaction. If this is true in general then UHV studiesof catalysis would yield very different results from thoseobtained in real atmosphere, because the weakly bondedspecies may not even appear on the surface in UHV. Catalyticreactions can also occur at liquid/solid interfaces. Beingcapable of probing buried interfaces, SFG spectroscopy shouldalso be ideally suited for investigation of such reactions.

CONCLUSIONI have discussed here a number of selected examples toillustrate the unique applications of SFG as a surface spec-troscopic tool. There are certainly many others worth discuss-ing. As a probe that can be applied to all interfaces accessibleby light, SFG is capable of finding applications in manyareas-for example, studies of interfacial biological systems,reactions at liquid/solid interfaces, corrosion, etching, andchemical vapor deposition processes, surface dynamics atliquid interfaces, and others. All of these are problems of greatsignificance but their investigations have been very limited sofar because of lack of suitable techniques. SFG surface spec-troscopy is likely to change the scene. Currently, the IR tuningrange of a table-top SFG setup is limited to '10 m on thelong-wavelength side, and could perhaps be extended to -20,um. With a free electron laser, however, the spectral range canbe easily extended to cover the entire IR region. Surfacevibrational spectroscopy will then become even more powerfuland versatile as a surface analytical tool.

This work was supported by the Director, Office of Energy Research,Office of Basic Energy Sciences, Materials Sciences Division, of the U.S.Department of Energy under Contract DE-AC03-76SF00098.1. Shen, Y. R. (1994) Surf Sci. 299/300, 551-562.2. Shen, Y. R. (1984) The Principles of Nonlinear Optics (Wiley, New York),

pp. 13-41; 479-504.3. Shen, Y. R. (1994) in Frontiers in Laser Spectroscopy, Proceedings of the

International School of Physics "Enrico Fermi," Course CXX, eds. Hansch,T. & Inguscio, M. (North Holland, Amsterdam), pp. 139-165.

4. Bloembergen, N. & Pershan, P. S. (1962) Phys. Rev. 128, 606-622.5. Heinz, T. F., Himpsel, F. J., Palange, E. & Burstein, E. (1989) Phys. Rev.

Lett. 63, 644-647.6. Daum, W., Krause, H. J., Reichel, U. & Ibach, H. (1993) Phys. Rev. Lett.

71, 1234-1237.7. Yeganeh, M. S., Qi, J. Yodh, A. G. & Tamargo, M. C. (1992) Phys. Rev.

Lett. 68, 3761-3764.8. Yeganeh, M. S., Qi, J., Yodh, A. G. & Tamargo, M. C. (1992) Phys. Rev.

Lett. 69, 3579-3582.9. Qi, J., Angerer, W., Yeganeh, M. S., Yodh, A. G. & Theis, W. M. (1995)

Phys. Rev. Lett. 75, 3174-3177.10. Superfine, R., Huang, J. Y. & Shen, Y. R. (1991) Phys. Rev. Lett. 66,

1066-1069.11. Du, Q., Superfine, R., Freysz, E. & Shen, Y. R. (1993) Phys. Rev. Lett. 70,

2313-2316.12. Du, Q., Freysz, E. & Shen, Y. R. (1994) Phys. Rev. Lett. 72, 238-241.13. Du, Q., Freysz, E. & Shen, Y. R. (1994) Science 264, 826-828.14. Guyot-Sionnest, P., Tadjeddine, A. & Liebsch, A. (1990) Phys. Rev. Lett. 64,

1678-1681.15. Guyot-Sionnest, P. & Tadjeddine, A. (1990) Chem. Phys. Lett. 172,341-345.16. Tadjeddine, A. & Guyot-Sionnest, P. (1991) Electrochim. Acta 36, 1849-1854.17. Cremer, P. S., Su, X., Shen, Y..R. & Somorjai, G. A. (1996) J. Am. Chem.

Soc. 118, 2942-2949.18. Boyd, G. T., Shen, Y. R. & Hansch, T. W. (1986) Opt. Lett. 11, 97-99.19. Harris, A. L., Rothberg, L., Dubois, L. H., Levinos, N. J. & Dhar, L. (1990)

Phys. Rev. Lett. 64, 2086-2089.20. Harris, A. L., Rothberg, L., Dhar, L., Levinos, N. J. & Dubois, L. H. (1991)

J. Chem. Phys. 94, 2438-2448.21. Morin, M., Levinos, N.J. & Harris, A. L. (1992) J. Chem. Phys. 96,

3950-3956.22. Guyot-Sionnest, P. (1991) Phys. Rev. Let. 67, 2323-2326.23. Guyot-Sionnest, P. (1991) Phys. Rev. Lett. 66, 1489-1492.24. Heinz, T. F. (1991) in Nonlinear Surface Electromagnetic Phenomena, eds.

Ponath, H. E. & Stegeman, G. I. (North Holland, Amsterdam), pp. 353-416.

25. Richmond, G. L., Robinson, J. M. & Shannon, V. L. (1988) Prof Surf Sci.28, 1-70.

26. McGilp, J. F. (1989) J. Phys. Condens. Matter 1, SB85-SB92.27. Eisenthal, K. B. (1992) Annu. Rev. Phys. Chem. 43, 627-661.28. Corn, R. M. & Higgins, D. A. (1994) Chem. Rev. 94, 107-125.29. Zhang, J. Y., Huang, J. Y. & Shen, Y. R. (1995) in Laser Science and

Technology, eds. Letokhov, V. S., Shank, C. V., Shen, Y. R. & Walther, H.(Harwood, Lausanne, Switzerland), Vol. 19.

30. Peremans, A., Tadjeddine, A., Suhren, M., Prazeres, R., Glotin, F.,Jaroszyunski, D. & Ortega, J. M. (1993)1. Electron Spectros. Relat. Phenom.64/65, 391-393.

31. Peremans, A., Tadjeddine, A., Guyot-Sionnest, P., Prazeres, R., Glotin, F.,Jaroszynski, D., Berset, J. M. & Ortega, J.-M. (1994) Nucl. Instrum.MethodsA 341, 146-151.

32. Chen, C. K., de Castro, A. R. B. & Shen, Y. R. (1981) Phys. Rev. Lett. 46,145-148.

33. Chen, C. K., Heinz, T. F., Ricard, D. & Shen, Y. R. (1981) Phys. Rev. Lett.46, 1010-1012.

34. Adamson, A. W. (1990) Physical Chemistry of Surfaces (Wiley, New York),5th Ed.

35. Rosen, M. J. (1989) Surfaces and Interfacial Phenomena (Wiley, New York),2nd Ed.

36. Sperline, R. P., Song, Y. & Freiser, H. (1992) Langmuir 8, 2183-2191.37. Rothberg, L., Higashi, G. S., Allara, D. L. & Garoff, S. (1987) Chem. Phys.

Lett. 133, 67-72.38. Dierker, S. B., Murray, C. A., LeGrange, J. D. & Schlotter, N. E. (1987)

Chem. Phys. Lett. 137, 453-457.39. Guyot-Sionnest, P., Hunt, J. H. & Shen, Y. R. (1987) Phys. Rev. Lett. 59,

1597-1600.40. Guyot-Sionnest, P., Superfine, R., Hunt, J. H. & Shen, Y. R. (1988) Chem.

Phys. Lett. 144, 1-5.41. Franks, F., ed. (1972) Water, A Comprehensive Treatise: The Physics and

Chemistry of Water (Plenum, New York).42. Whallhey, E. & Bertie, J. E. (1967) J. Chem. Phys. 46, 1264.43. Pople, J. A. (1951) Proc. R. Soc. London A 205, 163-178.44. Benjamin, I. (1994) Phys. Rev. Leu. 73, 2083-2086.45. Gavish, M., Popovitz-Biro, R., Lahav, M. & Leisorowitz, L. (1990) Science

250, 973-975.46. Mann, S., Archibald, D. D., Didymus, J. M., Douglas, T., Heywood, B. R.,

Meldrum, F. C. & Reeves, N. (1993) Science 261, 1286-1292.47. Lee, C. Y., McCammon, J. A. & Rossky, P. J. (1984) J. Chem. Phys. 80,

4448-4455.48. Stanners, C. D., Du, Q., Chin, R. P., Cremer, P., Somorjai, G. A. & Shen,

Y. R. (1995) Chem. Phys. Lett. 232, 407-413.49. Hatch, S. R., Polizzotti, R. S., Dougal, S. & Rabinowitz, P. (1992) Chem.

Phys. Lett. 196, 97-102.50. Messmer, M. C., Conboy, J. C. & Richmond, G. L. (1995)J. Am. Chem. Soc.

117, 8039-8040.51. Sefler, G. A., Du, Q., Miranda, P. B. & Shen, Y. R. (1995) Chem. Phys. Lett.

235, 347-354.52. Wu, X. Z., Sirota, E. B., Sinha, S. K., Ocko, B. M. & Deutsch, M. (1993)

Phys. Rev. Lett. 70, 958-961.53. Wu, X. Z., Ocko, B. M., Sirota, E. B., Sinha, S. K. & Deutsch, M. (1993)

Physica A 200, 751-758.54. Wu, X. Z., Ocko, B. M., Sirota, E. B., Sinha, S. K, Deutsch, M., Cao, B. H.

& Kim, M. W. (1993) Science 261, 1018-1021.55. Zhang, D., Gutow, J. H., Eisenthal, K. B. & Heinz, T. F. (1993) J. Chem.

Phys. 98, 5099-5101.56. Wolfrum, K., Graener, H. & Laubereau, A. (1993) Chem. Phys. Lett. 213,

41-46.57. Somorjai, G. A. (1993) Introduction to Surface Chemistry and Catalysis

(Wiley, New York), pp. 19-28.58. Chin, R. P., Huang, J. Y., Shen, Y. R., Chuang, T. J., Seki, H. & Buck., M.

(1992) Phys. Rev. Rapid Commun. 45, 1522-1524.59. Chin, R. P., Blase, X., Shen, Y. R. & Louie, S. (1995) Eur. Phys. Lett. 30,

399-404.60. Chin, R. P., Huang, J. Y., Shen, Y. R., Chuang, T. J. & Seki, H. (1995) Phys.

Rev. B Condens. Matter 52, 5985-5995.61. Chin, R. P., Huang, J. Y., Shen, Y. R., Chuang, T. J. & Seki, H., Phys. Rev.

B Condens. Matter, in press.62. Davis, R. F., ed. (1993) Diamond Films and Coatings: Development, Prop-

erties, and Applications (Noyes, Park Ridge, NJ).63. Spear, K. E. & Dismukes, J. P. eds. (1993) Synthetic Diamond: Emerging

CVD Science and Technology (Wiley, New York).64. Pan, L. ed., (1994) Diamond: Electronic Properties and Applications (Klu-

war, Norwell, MA).65. Waclawski, B. J., Pierce, D. T., Swanson, N. & Celotta, R. J. (1982) J. Vac.

Sci. Technol. 21, 368-370.66. Lee, S. T. & Apai, G. (1993) Phys. Rev. B Condens. Matter 48, 2684-2693.67. Aizawa, T., Ando, T., Kamo, M. & Sato, Y. (1993) Phys. Rev. B Condens.\t Matter 48, 18348-18351.68.>hang, H. C., Lin, J. C., Wu, J. Y. & Chen, K. H. (1995) J. Phys. Chem. 99,

11081-11088.69. Zhu, X. & Louie, S. G. (1992) Phys. Rev. B Condens. Matter 45, 3940-3943.70. Zhigilei, L. V., Srivastava, D. & Garrison, B. J. (1996) Phys. Rev. B Condens.

Matter, in press.71. Horiuti, I. & Polanyi, M. (1934) Trans. Faraday Soc. 30, 1164-1178.72. Davis, S. M., Zaera, F., Gordon, B. E. & Somorjai, G. A. (1985) J. Catal.

(Dalian) 92, 240-246.73. Beebe, T. & Yates, J. (1986) J. Am. Chem. Soc. 108, 663-671.74. Rekoske, J. E., Cortright, R. D., Goddard, S. A., Sharma, S. B. & Dumesic,

J. A. (1992) J. Phys. Chem. 96, 1880-1888.


Dow

nloa

ded

by g

uest

on

June

25,

202

0

a applications nonlinear optical spectroscopyvibrational spectroscopy is capable of providing most...

Documents