Phase Reference in Phase-Sensitive Sum-Frequency Vibrational Spectroscopy

Transcription

1 Phase Reference in Phase-Sensitive Sum-Frequency Vibrational Spectroscopy Shumei Sun 1, Rongda Liang 1, Xiaofan Xu 1, Heyuan Zhu, Y. Ron Shen 1, and Chuanshan Tian 1, 1 Department of Physics, State Key Laboratory of Surface Physics, and Key Laboratory of Micro- and Nano-Photonic Structures (MOE), Fudan University, Shanghai 00, China Department of Optical Science and Engineering, Fudan University, Shanghai 00, China Collaborative Innovation Center of Advanced Microstructures, Fudan University, Shanghai, 00, China Department of Physics, University of California, Berkeley, CA 0, United States Phase-sensitive sum-frequency vibrational spectroscopy (PS-SFVS) has been established as a powerful technique for surface characterization, but for it to generate a reliable spectrum, accurate phase measurement with a well-defined phase reference is most important. Incorrect phase measurement can lead to significant distortion of a spectrum, as recently seen in the case for the air/water interface. In this work, we show theoretically and experimentally that a transparent, highly nonlinear crystal, such as quartz and barium borate, can be a good phase reference if the surface is clean and unstrained and the crystal is properly oriented to yield a strong SF output. In such cases, the reflected SF signal is dominated by the bulk electric dipole contribution and its phase is either +0 o or -0 o. On the other hand, materials with inversion symmetry, such as water, fused quartz, CaF etc., are not good phase references due to the quadrupole contribution and phase dispersion at the interface. Using a proper phase reference in PS-SFVS, we have found the most reliable OH stretching spectrum for the air/water interface. The positive band at low frequencies in the imaginary component of the spectrum, which has garnered much interest and been interpreted by many to be due to strongly hydrogen-bonded water species, is no longer present. A weak positive feature however still exists. Its magnitude approximately equals to that of air/d O away from resonances, suggesting that this positive feature is unrelated to surface resonance of water. I. INTRODUCTION Sum-frequency vibrational spectroscopy (SFVS) has been established as a versatile 1

2 analytical tool for study of surfaces and interfaces including those of neat liquids and solids. 1- More recently, phase-sensitive (PS) SFVS has been proven to be more powerful as it can produce a vibrational spectrum of the imaginary part of the effective surface nonlinear susceptibility, Im, that directly characterizes the surface resonances. - In the experiment, Im is obtained from the measured and the corresponding phase. Error in deducing Im resulting from inaccuracy of is usually much larger when is closer to zero. The phase of a wave generally is measured against a reference wave with a known phase. In the case of PS-SFVS, a non-centrosymmetric transparent crystal with non-resonant second-order nonlinearity dominated by the bulk should generate a reflected SF signal with a phase of 0 o and is supposedly an ideal reference. For SF vibrational spectra in the CH and OH stretching range, quartz is often the preferred choice. - In a recent publication, Tahara and coworkers, however, reported observation of SF reflection from quartz with a phase different from 0 o 1, 1 in the aforementioned spectral range. They suspected that the deviation originated from contaminants on the quartz surface. Using, instead of quartz, the air/d O interface as the reference, and assuming that the phase of the reflected SF wave from it is o, they obtained an,,,, 1, 1 that, unlike those reported in the literature, exhibited Im spectrum for the air/h O interface no weak positive band at low frequencies (<00 cm -1 ). This result is important because the presence of the positive band has generated much theoretical interest in recent years. 1- Therefore, we must be assured that the results of Tahara and coworkers are correct, i.e., whether the quartz crystal is indeed a poor phase reference and the air/d O a good reference. More generally, we would like to learn how to find an appropriate phase reference for PS-SFVS. In this paper, we show that the Z-cut quartz is actually a good phase reference if properly oriented, and so are the other transparent, highly nonlinear crystals. If, however, the crystal is so oriented that the bulk contribution to second-harmonic/sum-frequency generation (SHG/SFG) is strongly suppressed, the surface contribution from the crystal to SHG/SFG may become non-negligible and lead to an output phase of SHG/SFG different from 0 o. This is most likely the reason why the strength of the low-frequency positive band in the reported spectra of Im for the air/water interface varies in the literature. With the SF signal from Z-cut quartz

3 tuned to maximum, the measured phase of the SF output from the air/water interface was found to be ~ o in the low-frequency region, leading to a rather weak positive low-frequency feature. Thus, the conclusion of Tahara and coworkers that the positive band does not exist appears partially correct. However, their assertion that the phase of non-resonant SFG from the air/d O interface was o is incorrect. The reason is that even off resonance, has an imaginary component arising from the wave propagation phase effect in the interfacial layer as we shall explain. This is generally true for all centro-symmetric transparent media such as H O, D O, fused silica, Al O and CaF. We describe in the following sections the underlying theory for the phase measurement, the experimental arrangement and procedure, and details of the experimental results. We present at the end the obtained with the proper phase reference. II. UNDERLYING THEORY Im spectrum for the air/water interface Our focus here is on the reflected SFG and its phase from a transparent medium (far away from resonances). To make later discussion easier, we briefly review here the underlying theory of SFG from an interface at z=0 between air and a semi-infinite medium., The reflected SF field E (,0) at z=0 from the medium (z>0) originates from nonlinear polarization P B (, z) induced by incoming fields E1( 1, z) E1( 1, z)exp( ik1 r) and E(, z) E (, z)exp( ikr) in the medium, and can be written as i 1 1 ckz E (,0) : E (,0) E (,0) (1) The effective surface nonlinear susceptibility, Seff,, is what to be measured in SFVS and is generally a complex quantity. Including electric-dipole (ED) and electric-quadrupole (EQ) contributions, it has the expression

4 S, eff ijk 0 i B, ijk j 1 k ( ) [ f (, z) ( z) f (, z) f (, z) + fi(, z) q,( zi) jk( z) f j( 1, z) fk(, z) z fi(, z) q1, i( zj) k( z) f j( 1, z) fk(, z) z ikz z + fi(, z) q, ij( zk) ( z) f j( 1, z) fk(, z)] e dz z Here, kz k1z kz kz is the mismatch of the input and output wave-vectors along z. f (, z) E (, z)/ E (,0) describes the normalized field variation near the interface, which i i i II equals to 1 for i=x or y, but equals to the ratio of the optical dielectric constants, (, z) / ( ), for i=z. The bulk nonlinear susceptibility of the medium, Bijk,, is given by () where the subindices d and q refer to ED and EQ contribution associated with the field E ( ), respectively. Far away from resonances,,,and f (, z) d, ijk i are all real quantities. We now divide the semi-infinite medium into two parts: one is the interfacial layer in which the medium response coefficients and the field amplitudes vary appreciably with z and the other is the bulk region in which both the response coefficients and the field amplitudes are uniformly constant, independent of z. Accordingly, by splitting the integral in Eq. into with 0 + being thickness of the surface layer, we find ( ) (1ik 0 )/ ik S, eff ijk S, ijk B, ijk z z 0 S, ijk 0 i B, ijk j 1 k [ f (, z) ( z) f (, z) f (, z) + fi(, z) q,( zi) jk( z) f j( 1, z) fk(, z) z fi(, z) q 1, i( zj) k( z) f j( 1, z) fk(, z) z ikz z + fi(, z) q, ij( zk) ( z) f j( 1, z) fk(, z)] e dz z ikz z + [ ( z) ( z)] e 0 B, ijk IQ, ijk S 0 dz and dz 0 0 (),

5 Note that the phase factor i kzz in Eq. () was neglected in early literature because it was believed to be small for the very small thickness of the interfacial layer,, but we will show in the following that this may not be appropriate. We first consider the case of non-centrosymmetric crystals in which the EQ contribution is negligible compared to that of the ED, i.e., and in Eq. () can be simplified to () Here, <...> S denotes average over the interfacial region. The imaginary term d,ijk ik z z S 0 is neglected in Eq. () because obviously < d,ijk ik z z> S 0 d,ijk / ik z. In the case that d,ijk S d,ijk, i.e. ED in the interfacial region is different from that in the bulk, where both terms are real, the measured is complex even far off resonance. This 1 1 may happen if the surface structure of a nonlinear crystal is partially disordered due to mechanical polishing which leads to smaller d, ijk for the interfacial layer. Assuming 1 ( d,ijk S d,ijk ) ~ 1, then the phase of d,ijk is with =k z 0 / derived from Eq. (). For 0 + ranging from 1- nm and k z ~ 1/ (0nm), the phase from such a surface would be 0 o (1 o ~ o ). In searching for a well-defined phase reference, it is important to be sure that the chosen nonlinear crystal has good surface quality. On the other hand, for a medium with inversion symmetry, such as water, fused silica, etc, the ED contribution to second-order nonlinearity only comes from the interface. The EQ contribution may not necessarily be negligible unless the surface is strongly polar-ordered. Eq. () becomes ()

6 Far away from resonances, BQ is imaginary, but and are real. We can write d IQ ( S,eff with Re( S,eff Im( S,eff ) ijk Re( S,eff ) ijk ( d,ijk ) ijk i( BQ,ijk ) ijk i Im( S,eff IQ,ijk ) ijk ) S 0 BQ,ijk ik z S BQ,ijk )0 ( d,ijk IQ,ijk )k z z S 0 () The phase of ( ) is given by S, eff ijk tan 1 { i( BQ,ijk S BQ,ijk )0 ( d,ijk IQ,ijk )k z z S 0 } () IQ,ijk ) S 0 BQ,ijk / ik z ( d,ijk It is seen that deviates from 0 o or o and can even be significant if the terms in the denominator of Eq. () is of the same order of magnitude, but different signs. III. EXPERIMENTAL ARRANGEMENT We used collinear beam geometry with picosecond (ps) input pulses for our PS-SFVS measurement. The experimental setup is sketched in Fig. 1; details on the system have been described elsewhere., Briefly, the visible pulse ( 1 ) at nm and the tunable IR pulse ( ) from a ps Nd:YAG/OPA/DFG system were combined collinearly. They then passed through a 0-m y-cut quartz acting as a SF local oscillator and a fused quartz slab as a phase plate. Finally, the beams were incident on the sample at angle = o and the SF signal generated in reflection interfered with the SF wave from the local oscillator. An interferogram was obtained by tuning the phase plate and recorded by a PMT. The phase of the SF signal from the sample with respect to that from the y-cut quartz could be deduced from the interferogram. The same geometry was used for phase-sensitive second harmonic generation (SHG) with input at nm, but it was much simpler in arrangement because there existed only one input beam. The beam polarizations used for SFVS were SSP (denoting S-, S-, and P-polarized SF output, visible input and IR input, respectively) and for SHG were P for both the input and output beams.

7 FIG. 1. Schematic optical layout of a collinear PS-SFVS setup. We carried out phase measurement on reflected SHG and SFVS from a large number of samples including non-centrosymmetric crystals [Z-cut quartz, Z-cut -barium borate (-BBO), and bismuth borate (BiB O, i.e. BiBO) with surface normal oriented at =. o,= 0 o ], centrosymmetric solids (fused silica, CaF, and Al O ), and liquids (H O and D O). Our goal was to find proper phase references for PS-SFVS. To minimize surface contamination, all solid samples were cleaned before measurement by sonication in acetone for 0 minutes followed by baking at o C for half an hour. The glass cell for liquid samples was cleaned by soaking in sulfuric acid with NOCHROMIX (Godax laboratories Inc.) for a few hours, then rinsed with ultra-pure water (1. Mcm) and baked to dry before use. IV. RESULT AND DISCUSSION A. Phase Measurements on BiBO, BBO, and Z-cut Quartz in Air We first conducted phase measurements on non-centrosymmetric transparent crystals. As mentioned earlier, we expected the phase of reflected SHG/SFG from such a crystal to be 0 o if the signal is completely dominated by bulk ED contribution. This is likely the case for BBO and BiBO that are known to have high bulk nonlinearity. If the expected phase is indeed found on these crystals, then they can be confidently taken as the phase reference to probe other samples. To simplify the problem, we started with phase measurement on reflected SHG from BBO and BiBO using input frequency at nm. The incident plane was chosen to coincide with the Y-Z plane of the crystals. The BiBO had the surface normal of the polished surface oriented at =. o,= 0 o with respect to the crystalline Z and X axes, respectively. The BBO surface was Z-cut. The non-resonant reflected SH output fields from bulk BiBO and BBO came from bulk

8 nonlinear susceptibility elements,, and d, YYY d, YZZ d, ZYZ d, ZZY dyyy,, respectively, all of which are real. The measured SHG outputs from the two crystals were ~0 times larger than that from a Z-cut quartz. The phase of the fields should be 0 o if the surface contribution to SHG is negligible. (See Eq.(); +0 o or -0 o depends on the sign of d,ijk.) In that case, the relative phase between SHG from surfaces of different crystals is expected to be either 0 o or o. The measured SH phases from the Z-cut BBO and the two inverted surfaces of BiBO are listed in Table I. Note that the measured phase of a sample, M, actually contains two parts: M =+ 0 with being the phase of for the sample and 0 an additional constant phase arising from wave propagation from the y-cut quartz reference to the sample (See Fig.1). In our case, 0. o. It is seen that within our experimental accuracy of about 0. o, the phase is indeed 0 o for the surfaces of BBO and BiBO and -0 o for the inverted surface of BiBO. Thus, we can trust that both BBO and BiBO with high non-resonant SHG/SFG output can be used as a good phase reference with = 0 o or -0 o. TABLE I. Measured phases of SHG from BiBO, BBO and Z-cut quartz M = + 0 BiBO BiBO ( inverted face ) BBO 01. o ±0. o. o ±0. o 0. o 0. o #1 Z-cut quartz 0.±0. o # Z-cut quartz 0.±0. o We then checked whether the Z-cut quartz could also be a good phase reference for SHG/SFG, although its nonlinearity is much lower than BBO and BiBO. Two Z-cut quartz samples, labeled as #1 and # Z-cut quartz, were measured with their azimuthal orientations set to generate maximum reflected SHG. The results are also presented in Table I. The phase of #1 Z-cut quartz agrees well with those of BBO and BiBO, but that of # Z-cut quartz deviates by about o. This result indicates that one must be careful in selecting Z-cut quartz as phase

9 reference. The deviation is likely due to structural distortion of the surface introduced in the surface polishing process. By testing on a number of different Z-cut quartz crystals, we came to realize that even the opposite surfaces of a crystal supplied by a manufacturer could be different because they are often polished differently. As discussed in the underlying theory section, if the surface layer is structurally different from the bulk, the phase of ( ) can deviate from by a fraction of k z 0 /. A larger distorted surface layer thickness (0 + ) leads to a larger phase deviation. To be sure that transparent, highly nonlinear, crystal could be taken as a good phase reference for SFG, we repeated the phase measurement on reflected SFG from BiBO using = 000 cm -1 and 1 = nm, and again found = 0 o within 0. o. We should emphasize that to make the surface contribution negligible in SHG/SFG, the non-centrosymmetric crystal should be oriented to maximize the bulk ED contribution. This is particularly important in the case of quartz because its nonlinearity is relatively weak in comparison with BiBO and BBO. When the surface contribution becomes appreciable with respect to the bulk contribution, the phase of SHG/SFG will deviate significantly from 0 o. We show in Fig. the measured phase and intensity of SFG from the #1 Z-cut quartz sample with = 000 cm -1 and 1 = nm as a function of the azimuthal angle,, defined as the angle between the X-axis of the crystal and the incident plane. The -fold symmetry of the Z-cut quartz about its Z-axis results in ( ) cos( ) whose absolute value has maxima at =n/ (n=integers) and zero at S, eff ijk =/+n/. As expected, deviates more and more from 0 o as approaches the latter value. However, for within n//, the phase deviation is less than o. ijk

10 1 1 1 FIG.. Measured phase and intensity of SFG from #1 Z-cut quartz versus the azimuthal angle of the quartz with the IR input set at 000 cm -1. The solid lines are theoretical fit. B. Phase Measurements on H O, D O, Fused Silica, CaF and Al O in Air Suspecting that the phase of quartz might deviate from 0 o, Yamaguchi et al. proposed the use of the air/d O interface as a phase reference for PS-SFVS in the OH stretching region. 1 They believed that the phase of ( ) S, eff ijk for reflected non-resonant SFG from surfaces of all centrosymmetric media in air should be o. We now used the phase-calibrated #1 Z-cut quartz with set at 0 o as the phase reference to measure the phase of ( ) ( ) for S, eff SSP S, eff reflected SFG, with the IR input fixed at 000 cm -1, from the surfaces of H O, D O, fused quartz, CaF and Al O in air. In all cases, the phase was found to deviate from o. As depicted in Table II, it has a value in the range of 1 o to 1 o, which is in good contrast with the statement that the phase of the nonresonant background cannot be arbitrary but has to be 0 or. In particular, the air/d O interface has a phase of o ± o, significantly different from o 1 assumed in Ref [1] and Ref [1]. We see from Eqs. ()-() that off resonance Im( ) S, eff for 1 such surfaces may indeed be non-negligible compared to Re( ) S, eff, and the phase of ( ) S, eff could be appreciably different from 0 and. If such a phase difference is ignored when adopting the surface of a centrosymmetric medium as the phase reference in PS-SFVS, the resulting phase-sensitive spectrum of a sample will certainly be distorted. Specifically, we shall

11 discuss in the following the effect on the OH stretch spectrum of the air/water interface. We note, however, that the OH bending spectrum could be similarly and perhaps even more seriously affected because the bending resonances are significantly weaker. -0 TABLE II. Off resonance phases of SFG at 000 cm -1 from different materials Materials CaF Fused silica Al O H O D O Phase 1 o ± o 1 o ±. o 1 o 1. o 1 o ±. o o ± o C. Surface versus Bulk Contribution to Off-Resonant SFG from the Air/Water Interface. The phase of the air/h O or air/d O interface at = 000 cm -1 was found to be ~ o. For such a phase value, Eq. () for with k z ~ 1/ (0 nm)for reflected SFG and 0 + ~1 nm shows that the surface contribution ( ) 0 and the bulk contribution d, IQ, S i k BQ, / z 1 are roughly of the same order of magnitude, but opposite in sign. Our measurement of the reflected SF intensity yielded ( S,eff ) ( d, IQ, ) S 0 BQ, / ik z 0. 1 m /V. Thus, we expect 1 ( ) 0 and d, IQ, S / i k are both around BQ, z 1 m /V. From Eq. (), we 1 notice that since k z ~ k, the bulk contribution / i k should be of the same order of BQ, z 1 magnitude as or q,( zy) zy ( ), which could be measured by off-resonant q, yz ( zy) q,( zy) zy 1 1 transmitted SFG. Indeed, we found at = 000 cm -1, ( q,yz( zy) D. Spectra of the Air/H O Interface in the OH Stretching Region q,( zy)zy ) ~ 1 m /V.

12 FIG.. (a) Comparison of the measured phase spectra of ( IR ) and (b) sin( IR ) from different groups. (c) and (d) Comparison of our spectra with Nihonyanagi s by shifting the latter downward by o. The insets in (a) and (c) are the enlarged view of ( IR ) below 0 cm -1. Using #1 Z-cut quartz as the phase reference, we re-measured the spectrum of ( ) in S, eff the OH stretching region for the air/h O interface. Because of the importance of phase measurement in deducing the spectra of Re( ) and S, eff Im( ) S, eff, we first compare our phase spectrum with those of Yamaguchi et al. (the one with air/d O as reference), 1 Nihonyanagi et al. (with air/d O as reference), 1 and Tian et al. (with Z-cut quartz as reference) 1 in Fig. (a). (They were obtained from = -1 tan [Im / Re ] with Re and Im extracted from data in the corresponding papers.) In order to show the similarity and difference 1

13 more clearly, we also give the spectra of sinwhich is directly related to Im( ), in Fig. (b). Over the range between 00 and 00 cm -1, phase spectra from all groups are essentially the same, i.e., no significant difference that could appreciably affect the Im( ) spectra. Our overall spectrum agrees very well with that of Tian et al., except in a narrow range between and 00 cm -1 where the latter data have larger fluctuations, but differs with the two others, more obviously on the low and high frequency sides of the spectrum. The spectrum of Yamaguchi et al. has poorer quality, so we focus on comparison with that of Nihonyanagi et al. The zero value of sin below 0 cm -1 and above 00 cm -1 in the latter spectrum resulted from = o measured by taking the air/d O interface as the phase reference. We now know that in those ranges, should be o instead. Comparison of our spectrum with Nihonyanagi s in Figs. (c) and (d) by shifting the latter downward by o indeed shows much better agreement between the two. We then compare the Im( ) spectra obtained by the various groups in Fig.. All spectra have the same general profile between 00 and 00 cm -1. Our spectrum agrees quantitatively with Tian s in the entire spectral range, but differs appreciably with Nihonyagani s and Yamaguchi s (Fig. (a)). Normalization of all spectra to the area under the dangling OH peak reduces the disagreement significantly (Fig. (b)). Further correction of the spectrum of Nihonyanagi et al. by taking into account the o phase shift (Fig. (c)) brings their spectrum and our spectrum almost in total agreement except that the dangling OH mode of the former is broader and lower in peak strength because of the poorer spectral resolution with the fs broadband SFVS scheme used by Nihonyanagi et al.. 1

14 FIG.. Comparison of Im( ) spectra before removal of the Fresnel Factor from different S, eff groups. (a) Im( ) spectra as reported. (b) S, eff Im( ) spectra normalized to the overall S, eff strength of the free OH mode. (c) Same as (b), but with the Im( ) S, eff spectrum of Nihonyanagi et al. corrected by a phase downshift of o. The unit in (a) is - m /V 1

15 We note that the spectra of Re( ) and Im( ) in Fig. are directly deduced from measurements with contribution of Fresnel coefficients not yet removed. Removal of the Fresnel coefficients should be made in order to attain the intrinsic spectra of the interface. In Fig., we show two different models to calculate the Fresnel coefficients: one assumes the dielectric constant,, of the interfacial layer along the surface normal to be the same as the bulk ( ) dielectric constant of water, =, and the other assumes '. It is seen that the Fresnel coefficients calculated by using different affect the spectral profile of quantitatively in the frequency range of -00 cm -1. Im( ) only We now address the question whether the positive feature below 00 cm -1 in the Im( ) spectrum for the air/h O interface exists or not. Our spectrum shows that it does although it is very weak, while the spectrum of Nihonyanagi et al. shows that it does not. This is because Nihonyanagi et al. assumed that the phase of reflected SFG from the air/d O interface taken as the phase reference is o in that spectral range. We notice that even away from resonance above 0 cm -1, Im( ) is finite and positive because o. We also show 1 the measured Im( ) and Re( ) for the air/d O interface in Figs. (b) and (c) 1 respectively: in the entire spectral range displayed, the value of Im( ) for air/d O is positive and essentially constant with a value same as that of air/h O below 0 cm -1 and above 00 cm -1. Thus, the positive feature below 00 cm -1 in the Im( ) spectrum of air/h O is not likely to be associated with surface resonances, but arises from phase dispersion and EQ contribution in the interfacial layer as discussed earlier. In any case, the positive feature would be too weak to be significant even if it were associated with a surface resonance. The effect of strong H-bonding or many-body interaction among interfacial water molecules suggested by MD simulations to explain the previously reported positive band may not be important

16 FIG.. (a) Fresnel factor versus input IR frequency for the air/water interface calculated with two different as indicated. Spectra of (b) Im( ) S, eff and red) and air/d O (black) interfaces after removal of the Fresnel factor. The solid curves in (b) and (c) are guide to the eye and the unit is -1 m /V. 1 and (c) Re( ) for the air/h O (blue

17 V. CONCLUSION We have shown that reflected non-resonant SHG/SFG from properly oriented highly nonlinear crystals is dominated by bulk contribution, and has a phase of 0 o as expected. Such crystals can be adopted as good phase reference in phase-sensitive SF spectroscopy. The Z-cut crystalline quartz oriented to yield maximum SHG/SFG output falls into this category. On the other hand, opposite to what was believed, 1, 1 transparent centrosymmetric materials cannot be taken as good phase references. Because bulk contribution is non-negligible compared to surface contribution, reflected SHG/SFG from such materials has a phase deviating from o, and the deviation depends on the material. We present in this paper the Im( ) spectrum of the air/h O interface taken with the proper phase reference. A positive feature below cm -1 still appears in the spectrum, but it is much weaker than those reported earlier.,, 1, 1,, Its magnitude approximately equals to that of air/d O in the non-resonant frequency region, suggesting that this positive feature is unrelated to surface resonance and hence the interfacial structure of water. ACKNOWLEDGEMENT CST acknowledges support by the NSFC (No.0, No.011, and No. 0), NCET (No. ). YRS acknowledges support from the Director, Office of Science, Office of Basic Energy Sciences, Materials Sciences and Engineering Division, of the U.S. Department of Energy under Contract No. DE-AC0-SF000. REFERENCES 1 Y. R. Shen. Nature, 1 (1). K. Eisenthal. Chem. Rev., 1 (1). G. Richmond. Chem. Rev., (00). W.-T. Liu and Y. R. Shen. Phys.Rev. Lett. 1, 011 (00). C. Tian and Y. Shen. Surf. Sci. Rep., (01). N. Ji.; V. Ostroverkhov.; C. S. Tian and Y. R. Shen. Phys.Rev. Lett. 0, 0 (00). C. Tian and Y. Shen. Proc. Natl. Acad. Sci., (00). X. K. Chen.; W. Hua.; Z. S. Huang and H. C. Allen. J. Am. Chem. Soc. 1, (0). N. Ji.; V. Ostroverkhov.; C. Y. Chen and Y. R. Shen. J. Am. Chem. Soc. 1, 0 (00). S. Nihonyanagi.; S. Yamaguchi and T. Tahara. J. Chem. Phys., 00 (00). C. S. Hsieh.; M. Okuno.; J. Hunger.; E. H. G. Backus.; Y. Nagata and M. Bonn. Angew. Chem. Int. Edit., 1 (01). 1 S. Nihonyanagi.; R. Kusaka.; K. Inoue.; A. Adhikari.; S. Yamaguchi and T. Tahara. J. Chem. Phys. 1, (01). 1

18 S. Yamaguchi. J. Chem. Phys. 1, 00 (01). 1 C. S. Tian and Y. R. Shen. J. Am. Chem. Soc., 0 (00). 1 D. Hu and K. C. Chou. J. Am. Chem. Soc. 1, 1 (01). 1 T. Ishiyama and A. Morita. J. Chem. Phys., 1 (00). 1 T. Ishiyama and A. Morita. J. Phys. Chem. C., 1 (00). 1 P. A. Pieniazek.; C. J. Tainter and J. L. Skinner. J. Chem. Phys. 1, 001 (0). 1 S. Nihonyanagi.; T. Ishiyama.; T. K. Lee.; S. Yamaguchi.; M. Bonn.; A. Morita and T. Tahara. J. Am. Chem. Soc. 1, 1 (0). 0 P. A. Pieniazek.; C. J. Tainter and J. L. Skinner. J. Am. Chem. Soc. 1, 0 (0). 1 T. Ishiyama.; H. Takahashi and A. Morita. Phys. Rev. B., 00 (01). T. Ishiyama.; T. Imamura and A. Morita. Chem. Rev., (01). T. Ohto.; K. Usui.; T. Hasegawa.; M. Bonn and Y. Nagata. J. Chem. Phys. 1, (01). G. R. Medders and F. Paesani. J. Am. Chem. Soc. 1, 1 (01). Y. R. Shen. J. Phys. Chem. C., (01). S. Sun.; C. Tian and Y. R. Shen. Proc. Natl. Acad. Sci., (01). S. Yamaguchi and T. Tahara. J. Phys. Chem. C., 11 (01). A. D. Quast.; A. D. Curtis.; B. A. Horn.; S. R. Goates and J. E. Patterson. Anal. Chem., 1 (01). M. Vinaykin and A. V. Benderskii. J. Phys. Chem. Lett., (01). 0 Y. Nagata.; C. S. Hsieh.; T. Hasegawa.; J. Voll.; E. H. Backus and M. Bonn. J. Phys. Chem. Lett., 1 (01). 1 C. S. Tian.; S. J. Byrnes.; H. L. Han and Y. R. Shen. J. Phys. Chem. Lett., 1 (0). X. Zhuang.; P. B. Miranda.; D. Kim and Y. R. Shen. Phys. Rev. B., 1 (1). C. S. Tian.; N. Ji.; G. A. Waychunas and Y. R. Shen. J. Am. Chem. Soc., (00). C. S. Tian and Y. R. Shen. Chem. Phys. Lett. 0, 1 (00). 1

19

20 p h a s e (d e g re e ) a z im u th a l a n g le ψ (d e g re e ) in te n s ity (a.u.)

21

22

23