269
Surface Science 182 (1987) 269-286 North-Holland. Amsterdam
CHARGE-TRANSFER BAND AND SERS MECHANISM FOR THE PYRIDINE-Ag SYSTEM Haruka YAMADA, Department Received
Hisao NAGATA,
Kazuhiro TOBA and Yoshiko NAKAO
of Chemistry, Kwansei Gakuin University, Nishinomiya 18 September
1986; accepted
for publication
25 October
662, Japan 1986
Transmission spectra in the ultraviolet and visible region as well as SERS spectra were measured for chemisorbed pyridine on vacuum evaporated (island) Ag films and on annealed (heated) Ag films. Strong and clear SERS spectra were observed with an enhancement factor of - lo6 and were definitely assigned to the N-bonded pyridine. On adsorption of pyridine a new peak was detected at - 600 nm as the charge-transfer (CT) band for the pyridine-Ag interaction, which could neither be ascribed to acy impurity nor to aggregated Ag particles. The surface morphologies of the Ag films observed by a SEM (highest resolution 1.S nm) did not show any appreciable changes on adsorption of pyridine, when the Ag substrates ‘were properly prepared. The molar extinction coefficient of the CT band was lo3 -lo4 z!’molt ’ cm- I. On the annealed Ag films the CT band, which was well isolated from the Ag band, appeared on adsorption of pyridine and disappeared on desorption of pyridine, reversibly. The stronger the CT band, the stronger the SERS spectrum. The excitation profile of the Raman 1010 cm -t band overlapped well with the CT band. The number of adsorbed pyridine molecules was estimated, from the r-r* band of pyridine at 250 nm, to be (6-15)X1O15 molecules/cm* corresponding to a coverage factor of 20-60. The contributions of the electromagnetic (EM) effect and the CT effect to the apparent SERS enhancement factor were estimated to be approximately lo-lo2 and 103, respectively, The reason why the CT band had not been detected as an optical absorption band was discussed. The CT band observed is compared with those reported by EELS or reflection measurements.
1. Introduction Surface enhanced Raman scattering (SERS) can be understood as an overlapping effect of multiple mechanisms [l-4]. One of the main mechanisms is the electromagnetic (EM) effect, especially caused by the excitation of a surface plasmon of a rough metal surface or collective electron resonance on a metal colloid. The other important mechanism is known as the chemical effect, or charge-transfer (CT) effect which is essentially the same as the resonance Raman effect for the adsorbate-adsorbent interaction [5-91. Most work on the SERS mechanism was done for the pyridine-Ag system and the enhancement factor was usually determined from the comparison of the Raman intensities between pyridine adsorbed on Ag substrate and liquid pyridine. Strong SERS always showed a spectrum which differed from that of 0039-6028/87/$03.50 0 Elsevier Science Publishers B.V. (North-Holland Physics Publishing Division)
270 H. Yamuda et ul. / Charge-transfer
band und SERS
mechanism for the p_widine-Ag
system
liquid or condensed pyridine and was ascribed to the chemisorbed species and not to the physisorbed one [l-6,10]. Recently we have carefully studied the Raman spectra of dyes adsorbed on metal films [ll], in which strong and stable resonance Raman spectra of adsorbed dyes were obtained. The maximum Raman intensities were always observed when the absorption maxima of the adsorbed dyes coincided with the extinction maxima of the Ag films. The excitation profile of the Raman band was almost coincident with that of the resonance Raman spectrum of the dye and the maximum intensity was not so different from that expected for the resonance Raman effect of the dye itself [12]. Based on these results, the conclusion was that the EM mechanism should make a rather small contribution to the SERS of adsorbed dyes. We have also observed the very strong SERS spectrum of cyclohexene adsorbed on Ag sols [13], which differed from that of cyclohexene-Ag+ (CT complex) or liquid cyclohexene. On the same Ag sols, however, no enhanced Raman spectrum was detected for adsorbed H,O, which was definitely expected from the EM mechanism. The excitation profile of the Raman band of the adsorbed cyclohexene did not overlap with the extinction band of the Ag sols, and also differed from that expected only by the EM mechanism. All of these facts provide evidence for the chemical mechanism of SERS, in addition to the reported phenomena in the literature [ll-201. The CT mechanism, however, has been supported mostly by indirect evidence and the CT levels were detected only for a few cases by electron energy loss spectroscopy (EELS) [21] or optical reflectivity measurement on the Ag electrode [22]. Several attempts to observe the CT band [23] as optical transmission spectrum have so far been made without success. The present work is an attempt to observe the CT band for pyridine adsorbed on Ag films under suitable conditions. Then an attempt is made to estimate the contribution of each effect to the total (apparent) enhancement factor, something which is earnestly desired by the SERS researchers.
2. Experimental 2.1. Ag substrates Silver thin films (so-called island films) were prepared by vacuum evaporation of Ag wires (99.99% purity, Nissin EM Co.) on a tungsten basket on one side of an optically flat quartz or glass plate (35 X 9 X 1 mm) at 10P5-10Ph Torr at room temperature by using a Hitachi HUS-5GB or JEOL JEE-4X evaporator. The silver as well as the tungsten basket had been outgassed before the films were prepared. By adjusting current, evaporation speed and
I% Yamada et al. / Charge-transfer band and SERS mechanism for the pyndine-Ag
system 211
distance between the basket and the sample plate which was placed right above the basket, various kinds of Ag films were obtained. The mass thicknesses of the films were estimated to be l-6 nm from the wavelengths at the extinction maxima by measuring the transmission spectra of the films, as described earlier [11,24] and were also monitored with a quartz oscillator (Sloan Co.) using a CRTM (ULVAC) gauge. After the island Ag film, having maximum extinction at 500-550 nm (3-4 nm in thickness), was heated at 150-200°C for 1 h in 50 Torr hydrogen gas, the heated Ag film showed a rather sharp extinction band at 400-420 nm and very low absorbances in the longer wavelength region. We used it as an annealed Ag film, which consisted of more spherical particles of about the same size [25-271. 2.2. Adsorption
of pyridine
on the Ag island film
Spectroscopic grade pyridine was obtained from Tokyo Chem. Co. and was used after degassing at liquid nitrogen temperature in a vacuum system with an oil diffusion pump (10e6 Torr). The vacuum system consisted of a quartz or glass (1 cm) cell, whose three sides (right, left and bottom) were transparent and optically flat and of which the top had a stopcock through an O-ring joint. Greaseless cocks and joints, MKS Baratron gauges and an ionization gauge were used to avoid any contamination with grease and mercury (amalgam). The Ag film, kept in the cell at 10-5 Torr, was exposed to the saturated vapor of pyridine for 20-30 min at room temperature. Then the sample was pumped out at 10m5 Torr to remove excess and physisorbed pyridine, so that only chemisorbed pyridine molecules (monomolecularly adsorbed) were left on the Ag film. 2.3 Ultraviolet and visible spectra Absorption (extinction) spectra of the samples in the vacuum cells were obtained by measuring the transmittances with a Shimadzu UV-360 spectrophotometer. The base lines were determined by measuring the spectra of the quartz or glass plates in the same cell. For the UV region the samples on quartz plates in the quartz cells were used. 2.4. Raman
spectra
Raman spectra of the samples were measured with a Jasco R-300 spectrophotometer (DC amp.) connected with a NEC GLG-3200 Ar+ laser or a CR-590 dye laser combined with a CR-6 Ar+ laser (514.5 nm). The sample in the vacuum cell was irradiated from the bottom as shown in fig. 1, where the Ag film was irradiated through a quartz (or glass) plate with an incident angle
272 H. Yumada et al. / Charge-transfer
hand and SERS
mechanism for the pyridine-Ag
system
or quartz
I c
laser
Fig. 1. Irradiation
method
for Raman
measurements
of 75” at the surface of the plate. Also the sample was irradiated directly at an incident angle of 60” with s- or p-polarization. The optical arrangement was the same as that described earlier [6,11]. Interference filters (Nippon Shinku Optical Co. A&B), a polarizer and ih and ax plates were also used. The laser powers used were measured accurately by using a JLP-03 power-meter for less than 50 mW, in close vicinity of the sample. The spectral slit widths used were as follows: 2.4-4.8 cm-’ for 514.5, 488.0 and 457.9 nm excitations, and 1.6-3.3 cm-’ for 630.0, 623.8, 615.8, 607.9, 600.0, 593.1, 586.7, 578.9 and 569.0 nm excitations. For the estimation of Raman intensities, spectral slitwidths used, the sensitivity curve for the photomultiplier (Hamamatsu TV R-374) and the laser powers used were taken into account. 2.5. Scanning
electron microscope
(SEM)
In order to learn the surface morphologies of the Ag films, SEM images were observed for the evaporated and annealed Ag films, before and after the adsorption of pyridine, using a JEOL JSM-880 apparatus. The JSM-880 is the highest resolution SEM (1.5 nm), and lower voltages (3-5 kV) were used to avoid charging at tilting angles of 40” -60”.
H Yumada et al. / Charge-transfer
band and SERS
mechanism for the pyridine-Ag
system
213
3. Results and discussion 3.1. Pyridine adsorbed on the Ag island film (soon after the evaporation) The Ag evaporated film prepared at 10e6 Torr showed an extinction maximum at 450-550 nm and was called a Ag island film (rough surface) [ll], whose mass thickness was estimated to be 2-5 nm. The extinction band is known to be due to the Ag particles, excitation of the surface plasmon polariton (SPP) [27]. After the Ag film was exposed to pyridine vapor at room temperature for 30 min, excess and physisorbed pyridine were pumped out at lo-’ Torr and only chemisorbed species, monomolecularly adsorbed, were left on the Ag surface, which are well known for adsorbed pyridine on porous metal oxides [6,10,12]. When a homogeneous island Ag film was prepared, a rather sharp extinction band was observed with low absorbances at the longer wavelength tail, as shown in fig. 2a. After pyridine was adsorbed on it, spectrum b was obtained, where the Ag peak was slightly decreased and blue-shifted by the coating effect [11,28] and a new shoulder appeared at - 600 nm. In fig. 2, the UV spectrum is shown together with the visible spectrum, where the 250 nm band assigned to the +rr* transition of pyridine is observed for the adsorbed species. The scanning electron microscope (SEM) image observed for the homogeneous Ag film is given in fig. 3A, where the Ag particles are about 25 nm in diameter (rather small) and well isolated from each other. When pyridine was
0
1,
1
200
Fig. 2. Absorption
L
400
spectra
of (a) an evaporated (homogeneous, pyridine adsorbed on (a).
,I-
____L
600
80(
island) Ag film (on quartz)
and (b)
274 Il. Yumadu et al. / Charge-trunsfrr
hund und SERS
mechanism
for thep~~ridine-Ag .ystenl
R
Yamada et al. / Charge-transfer
band and SERS
mechanism for the pvridine-Ag
system
215
adsorbed on that Ag film, no appreciable change in morphology was observed as in fig. 3B. On the Ag colloid the adsorption of pyridine usually gave rise to aggregated sol particles, which showed an absorption maximum in the visible region [31]. For the Ag island films, however, the particles might be well isolated, so that aggregated particles were hardly formed on adsorption of pyridine. Although the SEM images depended on observing the position in the sample and also changed from sample to sample, no appreciable change in morphology could be observed by the adsorption of pyridine, when the Ag film was properly prepared. The Raman spectrum of the same sample showed very strong and clear SERS as shown in fig. 4a, which is definitely ascribed to the chemisorbed pyridine, i.e. N-bonded pyridine [6,10], showing the totally symmetric ring breathing vibration at 1010 cm-’ shifted from 993 cm-’ of free (liquid) pyridine, the CH stretching vibration at 3070 cm- ’ shifted from 3063 cm-l and the Ag-N (adsorption bond) stretching band at 230 cm-’ [29]. The SERS spectrum observed with s- or p-polarization at 514.5 nm is in agreement with those of pyridine adsorbed on Ag films [30] and also on a thicker Ag film (smooth surface) [ll]. If the spectrum in fig. 4a is compared with that on the smooth Ag surface, it is found that the former is much stronger and clearer than the latter. The reasons why SERS was stronger than on the flat surface is
Fig. 4. (a) SERS
spectrum
of pyridine chemisorbed on the evaporated spectrum of liquid pyridine.
Ag film, (b) Raman
276 H. Yumada et al. / Charge-transfer
hand and SERS
mechanism for the pyridine-Ag
system
of course due to the rough surface, and also due to the irradiation method. By the evanescent wave as in fig. 1, the SERS intensity for the sample (- 5 nm Ag thick) was about twice as strong as that observed by the direct irradiation. The e~anc~ment factor for the present SERS spectrum in fig. 4a is estimated as - lo6 if the number of pyridine molecules was assumed to be the same as that on the smooth surface (see section 3.3). The excitation profile of the SERS band was measured for the strongest band at 1010 cm-’ of chemisorbed pyridine as sh,own in fig. 5. The maximum is found at - 600 nm and the profile is well overlapping with the new shoulder band at - 600 nm, but it is shifted from the Ag band. For the Raman intensity measurements we used a special attachment which held the sample as well as a standard sample, simultaneously, around a rotating axis. Alternative measurements of the Raman intensities for the sample and the standard were made reproducibly by replacing the sample with the standard. The estimations of the Raman intensities were carefully made as described earlier 161. If the - 600 nm band for pyridine-Ag is caused from some impurity of the materials and SERS is based on the EM mechanism, SERS of that impurity should be detected with the same enhancement factor. Since we observed only the typical SERS of N-bonded pyridine, the - 600 nm band should not be assigned to any impurity. Since the pyridine does not absorb in the visible region [lo], the - 600 nm band is probably due to a new band produced by the pyridine-Ag interaction. We have measured many samples on the various Ag films and confirmed the appearance of the new band at - 600 nm. The new band was clear on the Ag film whose extinction band was rather sharp with low absorbances at the longer wavelength tail.
Fig. 5. Excitation
profile of the Raman
1010 cm _ 1 band for the pyridine-Ag
island film
II. Yamada et al. / Charge-transfer band and SERS mechanism for the p_vridine-Ag system 211
1
00 Fig. 6. Absorption
400
spectra
500
600
of (a) an evaporated
700
60C
Ag film (on glass), and (b) pyridine
adsorbed
on
(a).
The transmission spectra of some examples are shown in figs. 6 and 7. Curve b in fig. 6, after pyridine adsorption, might consist of two extinction maxima. The band maximum of the Ag film is probably located at - 450 nm
--
I
0.3
*\
’
-
\ \
/
,’ I I
\ \
\
a \ ,
\ \
,
O400
Fig. 7. Absorption
spectra
I
500
of (a) an evaporated
600
700
Ag film (on glass), and (b) pyridine (a).
600
adsorbed
on
278 H. Yumudu et al. / Charge-transfer
band and SERS
mechanism for the p.vridine-Ag
system
and the other at - 600 nm. The band of the Ag film showed a decrease in maximum absorbance and was blue-shifted on adsorption of pyridine, as the coating effect [11,28]. The - 600 nm band, overlapping with the strong Ag band, was a new band which was produced by the pyridine-Ag interaction. In fig. 7, the new band at - 600 nm is clearly observed with the Ag peak which is more shifted than that in fig. 6. When the samples involved larger or aggregated particles and were inhomogeneous Ag films, high absorbances at the longer wavelengths were always observed. The curves in fig. 6 are probably overlapping with the bands caused by the larger particles. Since the appearance of the - 600 nm band was common to all the cases, the - 600 nm band might be reasonably ascribed to the pyridine-Ag interaction band, such as a CT band, apart from aggregated particles or impurities.
3.2. Pyridine ~~~rbed
on the unnerved Ag ~~~
After the Ag island film (- 4 nm) had been prepared and was heated at 150-200°C for 1 h in hydrogen gas, a Ag film showed the absorption band at 410 nm with a narrowed width as shown in fig. Sa which is called an annealed Ag film. Because the annealed Ag film had low absorbances at wavelengths longer than 500 nm, the new band caused by the pyridine-Ag interaction is expected to be easily detected. According to the literature 125-271, the anneal-
400
500
600
700
800
2 / nm Fig. 8. Absorption spectra of (a) an annealed (heated) Ag film (on glass), (b) pyidine (a) and (c) heated (desorbed) (b).
adsorbed on
II. Yamada et al. / Charge-transfer band and SERS mechanism for the pyridine-Ag s_ystem 219
ing process causes large irregularly shaped islands to break up and form a more dense array of smaller islands exhibiting a higher degree of rotational symmetry. The present results by the SEM observations certainly confirm it, as indicated in figs. 3A and 3C. When the annealed Ag film was exposed to the pyridine vapor, and then was pumped out at 10P5 Torr to remove excess and physisorbed pyridine, the sample showed transmission spectrum b in fig. 8, where a new peak at - 600 nm was clearly observed for chemisorbed pyridine on the Ag film, as expected. After the sample (b) was pumped at 150°C to remove the adsorbed species for - 15 min, transmission curve c was observed, where the - 600 nm band disappeared; curve c was similar to curve a. When the pyridine vapor was adsorbed on the sample (c) again, curve b was observed again. Thus the 600 nm band reversibly appeared on the chemisorption of pyridine and disappeared on the desorption of pyridine. We repeated the experiments on the various Ag anneale,d films and the new peaks were always observed for the samples which gave the SERS spectra. The new band at - 600 nm did not show any shift for several hours after the pyridine was adsorbed. The SERS spectrum of the sample was always observed as in fig. 4, which was ascribed to pure W-bonded pyridine (chemisorbed). The coincidence of the excitation profile with the - 600 nm band indicates that the - 600 nm band is caused by the transition of the pyridine-Ag interaction and the resonance Raman effect takes place with the CT level of the pyridine-Ag complex. Thus the results on the annealed Ag films confirm that the new band is produced by the pyridine-Ag interaction and is the CT band. The stronger the CT band, the stronger the SERS spectrum. The intensity of the CT band, however, was smaller than that on the non-heated Ag film. For the pyridine-Ag colloid system, absorption curves similar to curve b were observed [13,31], when pyridine molecules were a.dsorbed on the Ag colloids. However, the long wavelength bands observed for the colloid systems were gradually shifted to the longer wavelength side after the pyridine molecules had been adsorbed on the colloids, which was mainly caused by the aggregated Ag clusters as longitudinal plasmon excitation [31]. Because the - 600 nm band did not shift on the present Ag film, it was different from the band caused by the aggregated Ag clusters. The SEM images observed after the adsorption of pyridine in fig. 3D almost coincided with that in fig. 3C, so that the - 600 nm band was not assigned to aggregated clusters. Although the SEM image D is poor due to the occurrence of severe charging, no appreciable change in morphology is found. When water or ethylalcohol was adsorbed on the same Ag film from the vapor in a similar manner, a little blue-shifted Ag peak ‘was observed with a slight decrease in maximum absorbance, but no new peak in the longer wavelength region was found, supporting that the new band at - 600 nm
observed interaction
for the band.
3.3. Numbers
pyridine-Ag
film
can
be
ascribed
to the
pyridine-Ag
of adsorbed molecules
The pyridine molecule has a weak n-r * band and an intense T--~T* band in the ultraviolet region. The n--71 * band appears at 275-295 nm with a molar extinction coefficient t = 100, while the r-r’* band has a strong peak at about 250 nm with e = 2000-3000 8 malll cm-’ (2000 in a cyclohexene solution and 3000 in an aqueous solution) [32]. If we measure the r--r* band of the pyridine adsorbed on the Ag film in the ultraviolet region, we can estimate approximate numbers of adsorbed molecules on the Ag films. It is known that the intense m-r* band is hardIy affected by complexation and the band intensity can be assumed to be essentially the same as that of the free molecule /33]. The absorption spectra were measured in the region 200-800 nm for the samples. The absorption spectrum in the UV region is given in fig. 2 for the pyridine was non-heated Ag film, where the p--71 * band of chemisorbed observed at 250 nm, on adsorption of pyridine. The measurements were repeated on unannealed as well as annealed Ag films. The optical density for the r-r* band at 250 nm was obtained as 0.02-0.05, depending on each sample, which was almost comparable with that of the CT peak. A little larger optical density was observed on the unannealed Ag film than on the annealed one. When we assume e = 2000-3000 for the r-r* band, numbers of adsorbed molecules can be estimated to be (6-15) X 1015/cm2. These are 24-60 for on the smooth times larger than the value of 2.5 X lOI molecules/cm’ (flat) surface, where the molecular area of pyridine was assumed to be 0.4 nm2 for the monomolecular adsorption. Even though the extinction coefficient of the 250 nm band for adsorbed pyridine might be changed from that of pyridine, the order of magnitude for the number of the adsorbed molecules is still correct. This is, indeed, a large number and we have to take an at least 10 times larger value for the coverage on the rough Ag surface than that on the smooth surface. Few researchers on the SERS mechanism have measured the number of adsorbed molecules on the Ag rough surfaces including the electrodes, until the recent observation of porosity at UHV was reported {34]. Since our SEM results showed that any appreciable changes in particle size were not observed on adsorption of pyridine, when the Ag films were properly prepared, it is deduced that the chemisorbed pyridine species must be penetrated inside the Ag particles as those on porous oxides or must be sitting on atomically rough surfaces with large surface area. This seems consistent with the roughness factor of 20 observed on the Ag cold island film [34] and also with the idea of cavities [40].
H. Yamada et al. / Charge-transfer band and SERS mechanism for the pyndine-Ag
system 281
This might be one of the reasons why SERS spectrum was so weak on the smooth surface compared with that on the rough surface. 3.4. Extinction
coefficient
of the CT band
By using the numbers of adsorbed pyridine molecules obtained from the V-VT* band measurements, the molar extinction coefficient of the CT band for the adsorbed pyridine can also be estimated to be lo3 - lo4 / mol-’ cm- ‘. Since the CT band at - 600 nm is comparable in peak optical density with the ~-VT* band at 250 nm, as shown in fig. 2, it is also found straightforwardly that the extinction coefficient of the CT peak is of about the same order as that of the 7~--7~*band. 3.5. Contribution
of each effect to the SERS
total enhancement
Let us now estimate the contribution of each effect to the total SERS - lo6 for the enhancement factor of the enhancement factor. We estimated SERS spectrum shown in fig. 4 with 514.5 nm excitation on the unannealed Ag film. As was mentioned above, the number of adsorbed molecules on the Ag island film was found to be 20-60 times larger than that on the flat surface, that is, the coverage factor should be taken into consideration as at least one order. The SERS intensities were certainly stronger on the unannealed Ag films than on the annealed ones. The intensity difference between them can be caused by the electromagnetic (EM) effect, namely the excitation of surface plasmon polariton (SPP), which is found to be lo-lo*, because the Ag (- 500 nm) band must be in resonance with the incident 514.5 nm light, while the - 410 nm band was shifted from the resonance condition. If we assume 20 for the coverage factor and 50 for the SPP effect, then we can estimate the CT contribution as - 103, which is comparable with the enhancement factor on smooth Ag or Ni films [6]. Although these values, of course, vary with each individual sample, and with each individual surface, the result indicates a rough idea of the contribution of each effect to the total enhancement. Then the contributions are summarized as follows: 20.-60: coverage factor, lo--100: SPP (EM) factor, - 103: CT factor. The present result seems to support Efrima’s view of the SERS mechanism [39]. Although the SPP effect is rather small in this case, on the practical surfaces, it may become much larger sometimes, for instance, for a sample which has a uniformly rough surface with the irradiation at the plasmon angle.
282 H. Ycrmnda et ul. / Charge-transfer
3.6. Comparison
hand and SERS
with the other experimental
mechunrsm
for the p.yridlne-Ag .qxtenz
results
The observed CT band at - 600 nm corresponds to an energy of 2.1 eV which shows good agreement with the CT bands observed by electron energy loss spectroscopy (EELS), 1.9 eV for pyridineeAg(ll1) and 2.5 eV for pyridine-Ag evaporated film [21]. In spite of a considerable effort to detect the CT band optically, there has not been found any report except one which showed the reflectivity change measured on the Ag electrode at - 750 nm [22]. From the present study it becomes clear that there is some experimental difficulty to detect the CT band in the visible absorption spectrum. The reason why the CT absorption band was not detected can be understood as follows: (i) On the rough Ag surface the strong extinction band ascribed to the excitation of the Ag surface plasmon was so intense that it obscured the CT band which was broad and rather weak. (ii) For the Ag colloidal system, the strong extinction bands caused by the Ag particles, whose intensities and the peak positions varied with the degree of aggregation of the particles, overlapped closely with the CT band. Since the CT band caused by the monomolecularly adsorbed species was rather broad and much weaker than the bands of the Ag particles, it is also difficult to detect the CT band separably from the Ag bands. For the UHV experiments pyridine molecules can be adsorbed on flat or rough Ag surfaces only at low temperatures. The adsorbed pyridine deposited at low temperatures is usually (not always) physisorbed (or condensed) and not chemisorbed. Even when the sample is heated up, the adsorbed species (pyridine) remain mostly as physisorbed or desorb rather than change into chemisorbed, because the temperature of the Ag substrate, of high thermal conductivity, rises faster than that of the physisorbed pyridine. It is, therefore, hard to obtain a large sample which consists of mostly chemisorbed pyridine molecules at UHV for optical measurement, and especially difficult on the flat surface where the number of adsorption sites is very small. Thus, the CT absorption band ascribed to the chemisorbed pyridine is difficult to measure at UHV. On the other hand, SFRS could rather likely be observed for the samples prepared at UHV, because a spot of the sample was used for the Raman measurement and the laser irradiation might provide some of the energy required for the conversion of physisorbed to chemisorbed pyridine without heating the Ag substrate too much. The small number of the chemisorbed species might be enough to give SERS, because the enhancement factor for SERS is very large. A wider area of the sample is usually used for the optical measurement than for the Raman measurement. This might be the reason why SERS was detected at low temperatures at UHV, but not the absorption band of the CT transition, even on rough surfaces. It has recently been confirmed by Penning ionization spectroscopy (PIS) that the PIS band of physisorbed pyridine at 80 K on the cold Ag substrate did not show any shift
IL Yamada et al. / Charge-transfer band and SERS mechanism for the pyridine-Ag system 283
corresponding to the conversion of the physisorbed to chemisorbed species, when the sample was heated or annealed [35]. By the way, the EELS results were observed for the samples deposited at higher temperatures at 140 K [21]. For the UHV experiments so far reported [15,23,36,37] the chemisorbed pyridine was identified only by the Raman - 1000 cm-’ band of adsorbed pyridine, the totally symmetric ring breathing vibration, whose shift from that of the isolated molecule was smaller than that of the N-bonded pyridine, in the present work or already reported as L-pyridine [lo]. The (rather broad) Raman peaks reported at UHV as chemisorbed species [23,36], therefore, must be overlapping bands for mixtures of a small part of chemisorbed species with mostly condensed (physisorbed) species, because the enhancement factor for the chemisorbed species is much larger than that for the condensed species [23,36]. If the CT band is assigned to the N-bonded pyridine, as reported by Demuth and coworkers [21], the Ag-N stretching vibration should be observed at - 230 cm-’ as in the present work. None of these researchers, however, reported the Ag-N stretching band at UHV. Therefore, it is uncertain whether the - 1000 cm-’ bands reported are really to be ascribed to the N-bonded pyridine having the resonating CT band. The intensity of the Ag-N band will also give the amount of N-bonded pyridine molecules. If we look at the excitation profiles of SERS of pyridine and CO adsorbed on the same island Ag film (deposited at 15 K and annealed at 70 K) reported by Seki [23], maximum Raman intensities are found with 500 nm excitation for both pyridine and CO. He observed the totally symmetric ring breathing vibration of pyridine at 1003 cm-l, which was shifted from 993 cm-’ of liquid pyridine but not so much as the 1010 cm-’ band for the chemisorbed (N-bonded) pyridine. The 500 nm maxima of the excitation profiles reported by Seki are probably interpreted as the resonance effect with the SPP of the Ag substrates. The 1003 cm-’ band might be ascribed to a mixture of mostly condensed species with a little of chemisorbed species. On the cold Ag film which was covered on the island film, he observed the strongest Raman band at 1005 cm-‘, a little more shifted than that on the Ag island film. On the cold Ag film more chemisorbed species might exist than on the Ag island film, because the cold Ag film is supposed to have higher conductivity than the Ag island film (cf. SEM images reported by Seki). The excitation profile for the 1005 cm-’ band had a much stronger maximum with the 600 nm excitation than for the 1003 cm-’ band, which must be due to the resonance effect with the CT band of the chemisorbed species, coinciding with the CT absorption band in the present work. The transmission measurements by Seki [23] with a high sensitivity apparatus did not show any CT peak clearly. He observed a peak at 440 nm which might be due to the Ag film, slightly blue-shifted from that of the Ag substrate itself by a coating effect. Thus the absorption curve caused by the Ag substrate might be altered on adsorption from that of the original Ag substrate. In order
284 H. Yamada et al. / Charge-transfer
hand and SERS
mechanrsm
for the pyridrne-Ag
system
to ascertain the true absorption curve for the pyridine-Ag interaction, the apparently observed curve should be corrected for the background. This makes it difficult to find the true CT band for the sample which had only a small amount of N-bonded species at low temperature. If the new band observed at - 600 nm is caused by the effect of the “image” theory [l], a kind of EM effect, the enhanced Raman spectrum should be observed for physisorbed pyridine as well as for the N-bonded pyridine. Because the strong SERS was always observed for the N-bonded species and not for the physisorbed species, the possibility of the image interpretation can be eliminated. Mullins and Campion [37] have reported unenhanced Raman scattering from pyridine physically adsorbed on the smooth surfaces of the Ag single crystal and also from chemically adsorbed species on the Ag(540) surface. The charge transfer mechanism can be applicable only when the incident light is matching with the CT excitation due to the adsorbent-adsorbate interaction. So the chernisorbed species do not always give SERS. Mullins and Campion’s result was only on the Ag(540) surface, where no CT band has been reported so far, which did not give any evidence against the CT mechanism. Also their result has been referred to as inadequate sometimes [38], where the EM effect is emphasized for the main SERS mechanism.
4. Conclusion
(1) The transmission spectra in the ultraviolet and visible region were measured for pyridine chemisorbed on vacuum evaporated (island) Ag films and annealed (heated) Ag films. On adsorption of pyridine a new peak was (CT) band produced at - 600 nm, which was ascribed to the charge-transfer of the pyridine-Ag interaction, neither to impurity nor to aggregated Ag particles. The molar extinction coefficient was found to be 103-lo4 e mall’ cm-.‘. (2) On the annealed Ag films the CT band was well isolated from the Ag band, which appeared on adsorption of pyridine and disappeared on desorption of pyridine, reversibly. (3) The SEM (highest resolution) images were observed for the surface morphologies of the Ag films. They did not show any appreciable changes on adsorption of pyridine, when the Ag substrates were properly prepared, Ag indicating that the - 600 nm band was not caused by the aggregated particles. (4) The strong and clear SERS was observed with an enhancement factor of - lo6 and was definitely assigned to the N-bonded pyridine, with the clear bands at 1010 cm-’ for ring breathing and at 230 cm-’ for Ag-N stretching.
H. Yamada et al. / Charge-transfer band and SERS mechanism for the pyridine-Ag
system 285
(5) The excitation profile of the Raman 1010 cm-’ band overlapped well with the CT band, but not with the Ag band. (6) The stronger the CT band, the stronger the SERS spectrum. (7) The number of adsorbed pyridine molecules was estimated from the V-T* band observed for the adsorbed pyridine at 250 nm, to be (6-15) X 1015 molecules/cm*, which was 20-60 times larger than the value for monomolecular adsorption on the flat surface. (8) Based on the SERS intensity difference between on the unannealed and annealed Ag films, the contribution of the electromagnetic (EM) effect to the SERS enhancement was estimated to be lo-lo*. Then, the contribution of the CT effect must be - lo3 to the apparent enhancement. (9) The reason why the CT band had not been detected as optical absorption band was discussed.
Acknowledgements We express our thanks to Professor S. Takayama of Kwansei Gakuin University for preparing the Ag films and also to Mr. K. Ogura of JEOL for the SEM observations. We also acknowledge valuable discussions with Professor E. Burstein of University of Pennsylvania, Professor S. Efrima of Ben Gut-ion University of the Negev, Professor S. Hayashi of Kobe University, Professor H. Ueba of Toyama University, Professors S. Ushioda and W. Suetaka of Tohoku University and Dr. H. Seki of IBM.
References [l] R.K. Chang and T.E. Furtak, Eds., Surface Enhanced Raman Scattering (Plenum, New York, 1982). [2] H. Ueba and H. Yamada, Eds., Proc. Intern. Symp. on Spectroscopic Studies of Adsorbates on Solid Surfaces [Surface Sci. 158 (1985)]. [3] A. Otto, in: Light Scattering in Solids, Vol. 4, Eds. M. Cardona and G. Gtintherodt (Springer, Berlin, 1984). [4] C.R. Brundle and H. Morawitz, Eds., Vibrations at Surfaces (Elsevier, New York, 1983). [5] H. Ueba, S. Ichimura and H. Yamada, Surface Sci. 119 (1982) 433: H. Ueba, Surface Sci. 129 (1983) L267; 131 (1983) 347. [6] H. Yamada and Y. Yamamoto, Surface Sci. 134 (1983) 71. [7] T.E. Furtak and S.H. Macomber, Chem. Phys. Letters 95 (1983) 328. [8] F.J. Adrian, J. Chem. Phys. 77 (19i2) 5302. [9] J.R. Lombardi, R.L. Birke, T. Lu and J. Xu, J. Chem. Phys. 84 (1986) 4174. [lo] H. Yamada and Y. Yamamoto, J. Chem. Sot. Faraday Trans. I, 75 (1979) 1215. [II] H. Yamada, H. Nagata and K. Kishibe, J. Phys. Chem. 90 (1986) 818. [12] H. Yamada, Appl. Spectrosc. Rev. 17 (1981) 227. [13] H. Yamada, H. Nagata and H. Teranishi, J. Phys. Chem. 90 (1986) 2384. [14] B. Pettinger and A. Gerolymatou, Ber. Bunsenges. Phys. Chem. 88 (1984) 359; Surface Sci. 156 (1985) 859.
286 H. Yumrrda et (II. / Churge-trunsfer
hand and SER.5 mechmism
for the p.yridzne-Ag
s_vstem
[15] A. Otto, 3. Billmann, J. Eickmans, U. Erttirk and C. Pettenkofer, Surface Sci. 138 (1984) 319; C. Pettenkofer, J. Eickmans, U. Erttirk and A. Otto, Surface Sci. 151 (1985) 9: A. Otto, K.H. Frank and B. Reihl, Surface Sci. 162 (1985) 891. [16] H. Seki, J. Vacuum Sci. Technol. 20 (1982) 584; J. Chem. Phys. 76 (1982) 4412. [17] H. Yamada, N. Tani and Y. Yamamoto, J. Electron Spectrosc. Related Phenomena 30 (1983) 13; H. Yamada, Y. Yamamoto and N. Tani, Chem. Phys. Letters 86 (1982) 397. [18] H. Sate, M. Kawasaki, K. Kasatani. E. Kawai and H. Sasaki, Chem. Phys. Letters 123 (1986) 355. [19] M. Takahashi, M. Goto and M. Ito, Chem. Phys. Letters 121 (1985) 458. [20] M. Kobayashi and M. Imai, Surface Sci. 158 (1985) 275. [21] Ph. Avouris and J.E. Demuth, J. Chem. Phys. 75 (1981) 4783; J.E. Demuth and P.N. Sanda, Phys. Rev. Letters 47 (1981) 57; N.J. Dinardo, Ph. Avoutis and J.E. Demuth, J. Chem. Phys. 81 (1984) 2169. [22] B. Pettinger, U. Went@ and D.M. Kolb, Ber. Bunsenges. Phys. Chem. 82 (1978) 1326. [23] H. Seki, J. Electroanal. Chem. 150 (1983) 425; J. Electron Spectrosc. Related Phenomena 30 (1983) 287. [24] Y. Nakao and H. Yamada, Surface Sci. 176 (19X6) 578. [25] S. Hayashi. Japan. J. Appl. Phys. 23 (1984) 665. [26] C.F. Eagen, Appl. Opt. 20 (1981) 3035. [27] S.W. Kennerly, J.W. Little, R.J. Warmack and T.L. Ferrell. Phys. Rev. B29 (1984) 2926. [28] M. Kamei and H. Yamada, unpublished work. [29] J.R. Lombardi, E.A.S. Knight and L. Birke, Chem. Phys. Letters 79 (1981) 214; B.H. Loo, J. Electroanal. Chem. 131 (1982) 381. [30] E. Burstein, C.Y. Chen and S. Lundquist, in: Light Scattering in Solids, Eds. J.L. Birman, H.Z. Cummins and K.K. Rebane (Plenum, New York, 1979) p. 479. [31] J.A. Creighton, C.G. Blatchford and M.G. Albrecht, J. Chem. Sot. Faraday Trans. II, 75 (1979) 790. [32] UV-Atlas of Organic Compounds, Vol. 2 (Verlag Chemie, Weinheim/Butterworths, London, 1967) GS,‘-G5,‘2. [33] J. Yarwood, Ed., Spectroscopy and Structure of Molecular Complexes (Plenum, New York, 1973); R.P. Lang, J. Am. Chem. Sot. 84 (1962) 1185; A.F. Grand and M. Tamres, J. Phys. Chem. 74 (1970) 208. [34] J. Eickmans and A. Otto, Surface Sci. 171 (1986) 415. [35] T. Suzuki, T. Kondow and K. Kuchitsu, private communication. [36] M. Udagawa, Chih-Cong Chou, J.C. Hemminger and S. Ushioda, Phys. Rev. B23 (1981) 6843: J. Giergicl, S. Ushioda and J. Hemminger, Phys. Rev. B33 (1986) 5657. [37] D.R. Mullins and A. Campion. Chem. Phys. Letters 110 (1984) 565; Surface Sci. 158 (1985) 263. [38] M. Moskovits, Rev. Mod. Phys. 57 (1985) 783. [39] S. Efrima, in: Modern Aspects of Electrochemistry, Vol. 16. Eds. B.E. Conway, R.E. White and J.O’M. Bockris (Plenum, New York, 1985) p. 253. [40] H. Seki and T.J. Chuang, Chem. Phys. Letters 100 (1983) 393; H. S&i. in: Proc. 10th Intern. Conf. on Raman Spectroscopy, Eds. W.L. Peticolas and B. Hudson (Univ. of Oregon Print, Eugene, 1986) p. 5514.