8844
J. Phys. Chem. 1991, 95, 8844-8849
Adswp#on and Swface Reaction of Actklhre in Sllver Sol: Surface-Enhanced Raman Spectroscopic M Y Seong Tae Ob,Kwan Kim,* and Myung So0 Kim* Department of Chemistry and Research Institute of Molecular Sciences, Seoul National University, Seoul 151-742, Korea (Received: March 18, 1991; In Final Form: June 3, 1991) Surface-enhanced Raman scattering (SERS) of acridine in silver sol has been investigated. The presence of halide ions was a prerequisite for the observation of SERS. Different SER spectra were obtained, in the presence of C1-, depending on the bulk pH. At neutral pH, acridine was found to adsorb on the silver surface via its nitrogen lone pair electrons, while in the acidic medium it was found to adsorb as the acridinium-chloride ion pair through the chlorine atom. The charge-transfer effect appeared to play important roles in the latter case along with the electromagnetic effect. On the other hand, surface photoreaction readily occurred in a highly basic medium. This could be understood in terms of the competitive adsorption of chloride and hydroxide ions on silver. Nonetheless, the acidity of acridinium ion seemed to be far greater on the surface than in the bulk aqueous phase.
Introduction Vibrational spectroscopy is a useful method for the investigation of molecular adsorption on surfaces.' In particular, surfaceenhanced Raman scattering (SERS), which is especially sensitive to the first layer of adsorbates on the surfaces of metals such as Ag, Au, and Cu, is finding increasing application to the study of surface-adsorbate interaction? Also, interpretation of the identity, structure, orientation, and conformation of an adsorbate is available from SERS studies.' The effect of solution pH on the SERS behavior of adsorbate has been investigated by numerous investigatomCB The fraction of protonated species on the surface was observed to change with solution pH, as in the bulk phase.+l This led to a change in either the adsorption mechanism&' or the orientation of adsorbate.* The SERS arch prototype, pyridine, is a particularly favorable example. Pyridine can bond easily to a metal either as a r-bonded species or as a u-bonded species through the lone pair electrons of the ring n i t r ~ g e n . ~However, to obtain the SER spectrum of its conjugate acid, pyridinium, the presence of a halide ion is necessary. Pyridinium is believed not to be directly adsorbed on the metal surface but rather to be located in the diffuse double layer and associated with specifically adsorbed halide to form an ion pair."l The catalytic activity of the silver surface can be varied significantly as the surface is exposed to the halide.t"t2 In the absence of coadsorbed chloride ions, trans-cinnamonitrile is, for example, converted to rrans-cinnamic acid by a surface reaction." On the other hand, chloride and hydroxide ions are known to competitively adsorb on the silver ~ u r f a c e . ' ~ As ~ ' ~the solution pH increases, the surface sites become more favorable to hydroxide ions rather than chloride ions. The catalytic activity of the metal surface will thus be substantially affected by the change in solution pH.I2 In the present work, a rather detailed SERS investigation has been carried out for acridine adsorbed on a silver sol surface. The major purpose of the present investigation is to study the effect of solution pH and the influence of surface CI- on the SER spectra of acridine. Like pyridine, acridine is a heterocyclic compound belonging to the point group C,. The acidity of acridinium (pK, = 5.60)is also very similar to that of pyridinium (pK, = 5.2). Hence, the SER spectral pattern of acridine is discussed in conjunction with that of pyridine, which remains the most studied molecule in the field. Experimental Section Acridine was purified by vacuum sublimation and then by recrystallization in an ethanol-water solution. Acridinium chloride was prepared as reported by Brigodiot and Lebas.I4 The details To whom all correspondence should be addressed.
of silver sol preparation have been reported d s e ~ h e r e . ' ~All the chemicals used were reagent grade, and triply distilled water was used for the preparation of the solutions. A small amount (10 ML)of 0.1 M acridine in ethanol was mixed with 2 mL of silver sol solution. The sol solution changed in color from yellow to either purple or green depending on the presence of halide ions (Cl-, Br-). Poly(vinylpyrro1idone) (PVP, M W 40000) was then added to the solution as a stabilizer. The pH of resulting solution was adjusted by using H$04 or NaOH. The concentrations of acridine, halide, and PVP were 5 X 10' M,1 X M, and 0.02% respectively, in the final solution. The instrumental details for Raman measurement have also been reported previously.3c Either a spinning cell spun at 3000 rpm or glass capillary cell was used as the sampling d e v i ~ e . ' ~ Results and Discussion Acridine (2,3,5,6-dibenzopyridine)is hardly soluble in water, but its solubility increases in acidic medium, forming its conjugate acid, acridinium ion. Figure l a shows the ordinary Raman (OR) spectrum of 0.2 M aqueous acridinium chloride solution a t pH 1.7. Since the pKAvalue of acridinium ion is 5.60,16 the spectrum in Figure l a can be attributed almost entirely to the acridinium ion. Even though acridine is soluble in aprotic solvents such as CC14, the quality of its OR spectrum is somewhat unsatisfactory ( 1 ) (a) Willis, R. F., Ed. Vibrational Spectroscopy of Adsorbates; Springer-Verlag: Berlin, 1980. (b) Yates, J. T., Madey, T. E., Eds. VibrationcdSpectroscopy of Motecules on Swfaces; Plenum; New Yotk, 1981. (2)Chang, R. K., Furtak, T. E., Eds. Swfie-Enhanced Raman Scattering; Plenum: New York, 1982. (3) (a) Creighton, J. A. Sur$ Sei. 1983, 124,209.(b) Moskovits, M.;Suh, J. S.J. Phys. Chem. 1984,88,5526.(c) Joo, T.H.; Kim, K.; Kim, M.S. J . Phys. Chem. 1986,W, 5816. (d) Moskovits, M.;Suh, J. S. 1.Phys. Chem. 1988,92,6321.(e) Joo, T. H.; Yim, Y. H.; Kim, K.; Kim, M. S. J . Phys. Chem. 1989,93,1422. ( f ) Yim, Y.H.; Kim, K.; Kim. M. S. J. Phys. Chem.
1990,94,2552. (4) Birke, R. L.; Bcmard. I.; Sanchez, L. A.; Lombardi, J. R. J. Elect r ~ ~Chem. l . 1983,150,447. ( 5 ) Rogers, D. J.; Luck, S.D.; Irish, D. E.; Guzonas, D. A.; Atkinson, G. F. J . Electroanal. Chem. 1984, 167, 231. (6)Sun, S.C.; Bernard, I.; Birke, R. L.; Lombardi, J. R. J . Electroanal. Chem. 1985,196,359. (7) Anderson, M.R.; Evans, D. H. J . Am. Chem. SOC.1988,110,6612. (8) Takahashi, M.;Furukawa, H.; Fujita, M.; Ito, M. J. Phys. Chem. 1981, 91, 5940. (9) (a) Demuth, J. E.; Christmann, K.; Sanda, P. N. Chem. Phys. Lett. 1980,76, 201. (b) Fleischmann, M.;Graves, P. R.; Robinson, J. J . Electroanal. Chem. 1985, 182, 73. (10)Brandt, E.S.J . Electroanal. Chem. 1983, I50,97. ( 1 1 ) Chun, H. A.; Kim, M. S.; Kim, K. J . Mol. Struet. 1990,221,127. (12)Park, H.;Kim, M. S.;Kim, K. To be published in Chem. Phys. (13)Salaita, G. N.;Lu, F.; Laguren-Davidson, L.; Hubbard, A. T. J . ElectroanaI. Chem. 1987,229, 1. (14) Brigodiot, M.; Lebas, J. M. J. Chim. Phys. Phys. Chim. Biol. 1972, 69,964. (15) Joo. T. H.; Kim, K.; Kim, M. S. Chem. Phys. Lett. 1984, 112.65. (16)Schulman, S.G.; Capomacchia, A. C. J . Am. Chem. Soe. 1973,95. 2763.
0022-365419112095-8844$02.50/0 0 1991 American Chemical Society
The Journal of Physical Chemistry, Vol. 95, No. 22, 1991 8845
Reaction of Acridine in Silver Sol
-I
8
P
000
ImO
lJ00
"2m
3IC
RAMAN SHIFT, cm"
Figure 1. Ordinary Raman spectra of (a) 0.2 M aqueous acridinium chloride solution at pH 1.7 and (b) acridine in the solid state, respectively. Excitation with a 514.5-nm line of an argon ion laser.
due to photoreaction induced by an argon ion laser used for Raman excitation. Nonetheless, the spectral pattern is very similar to that of acridine in the solid state as shown in Figure lb. Peaks appearing in the O R spectra of acridine and acridinium ion are summarized in Table I together with the symmetry assignments of Brigodiot and Lebas.14 The SER spectra of acridine have been obtained in the presence of 1 x M C1-. Parts a 4 of Figure 2 show the SER spectra obtained at pH 2.4,4.7, 8.8, and 12.2, respectively. The positions of the peaks in the SER spectra are listed in Table I also. The SER spectrum taken at pH 4.7 has the composite character of the SER spectra obtained a t pH 2.4 and 8.8. The SER spectrum a t pH 12.2 can be hardly correlated with the O R spectrum of either acridinium or acridine species. For instance, the peaks at 371, 541, 557, 1350, 1385, and 1599 cm-' in Figure 2d cannot find their counterparts not only in the O R spectra but also in the infrared spectra of acridine and acridinium. In the absence of chloride in the silver sol solution, a SER spectrum resembling Figure 2d was obtained even at pH 8.8. In both cases, the intensities of the above bands are greatly suppressed when the sampling cell was spun at 3000 rpm, indicating that the bands originate from certain surface photoreaction products. The origin and identity of the species responsible for the above additional peaks are uncertain. As will be discussed later, however, chloride ion adsorbed on the surface seems to reduce or prevent the formation of the unknown photoproduct. Based on the above description, it is interesting to find that there are two different SER spectra depending on the bulk pH that can be correlated with the OR spectra of acridine or acridinium ion. Variations in the SER spectra with changes in the bulk pH were reported previously. These were usually attributed either to a change in orientation of adsorbates with respect to the metal surfaces or to a change in chemical nature of the adsorbates.M When compared to the cases reported in the literature! the spectral variation observed in the present case seems to be too dramatic to be attributed to simple reorientation of adsorbed acridine. Hence, it is thought that the above variation is mainly due to the change in the chemical nature of adsorbed acridine. The reason for such an argument will become clearer through detailed spectral analysis. At pH 8.8, acridine exists as the unprotonated form in aqueous solution. Considering the chemical nature of this molecule, it seems likely that the chemical species adsorbed on the silver particle surface under this condition is the neutral acridine. In
Figure 2. SER spectra of 5 X IO-' M acridine in silver sol solution containing 1 X IOm3 M KCI at (a) pH 2.4, (b) pH 4.7, (c) pH 8.8, and (d) pH 12.2, respectively. pH's being adjusted in the final step for the same batch of sol solution. Excitation with a 514.5-nm line of an argon ion laser. (*) Art fluorescence line.
fact, the bands in the SER spectrum shown in Figure 2c correlate satisfactorily with the corresponding bands in the O R spectrum in Figure lb, even though some of the bands exhibited substantial peak shifts upon surface adsorption. Interestingly, the bands whose peak positions are markedly affected by surface adsorption are the ring deformation and C-H bending vibrations such as the bands at 517,744, and 1163 cm-' in the O R spectrum (Figure 1 b). The fact that the same bands are affected by protonation (Figure la) indicates that adsorption of acridine on the silver surface occurs via its base center, namely the nitrogen atom. The argument presented so far supports the fact that the species responsible for the SER spectrum in Figure 2c is the neutral acridine molecule adsorbed on the silver surface. At pH 2.4, acridine will exist as acridinium ion in the bulk aqueous phase. However, this does not necessarily mean that the adsorbed species under this condition is also the acridinium ion. In fact, several cases have been reported in which pK,, values of acids are modified by surface adsorption.6J' The SER spectrum taken a t p H 2.4 can be satisfactorily correlated with the O R spectrum of acridinium chloride solution, however. This can be seen in Table I. The ring stretching bands a t 1270, 1277, and 1556 cm-' in the O R spectrum of acridine are the most affected by protonation, appearing at 1280, 1294, and 1588 cm-l in the OR spectrum of acridinium ion. The corresponding bands appear at 1268, 1279, and 1565 cm-l, respectively, in the SER spectrum at pH 8.8, while the same bands appear at 1277, 1291, and 1584 cm-l in the SER spectrum at pH 2.4. Such spectral correlations form the basis for the identification of the chemical species adsorbed at pH 8.8 and 2.4 with acridine and acridinium ion, respectively. Nevertheless, the adsorbed species responsible for Figure 2a may not be acridinium ion itself. A positively charged cation is not likely to adsorb efficiently on the surface since silver (17) Anderson, M.R.;Evans, D.H.In Electrochemical Surface Science; Soriga, M. P., Ed.; American Chemical Society: Washington, DC, 1988.
8846 The Journal of Physical Chemistry, Vol. 95, No. 22, 1991
Oh et al.
TABLE I: Spectral Data for Acridine
ORS' acridine solid
saltb
SERS
Ad
ape
401 417
9
238 401 (31)r
8
523
11
525 (9)
IO
618
9
615 (8)
9
744
9
758 (24)
9
I010
9
1015 (IO) 1176 (IO) 1 I88
IO
1280 (17) 1294
CCI2
Au
DH 8.8
Av
PH 2.4 217 243
9
401 (26)
239 240 400 416 474 517 580 613 658h 744 958 1009 1 I63 I I76 I209 I270 1277 1315 1401 1415 I463 I478 1514 I556 1573 1628 3058
1161 1174 1 I90 1267 1275 1314 1401 1417 I464 1480
405 (24) 420 476 526 (4) 625 (4) 663 (2) 759 (23)
9
1 I68 (14)
9
1268 (21) 1279 (8) 131 1 1405 (100) 1425 1461 (7) 1481 1519 1565 (36)
7
1408 (100)
11
9
1490 (15)
12
1558 1579
9
11
1622
9
1588 (IO) 1568 1604 I628 1654
Au
svmmetrvl
8
523
13
615 (23) 660 (3) 756 (IO)
7 8 9
IO
1171 (19)
8
8 7
1277 (15) 1291
9 8
IO
1401 (100)
IO
1462 1485
IO
1584 (35)
1619
1621
3049
3963 1785 1799 1888 1979 2016 2194 2470 2677 2799 2857 2979
IO
IO
615 + 1171 = 1786 401 + 1401 = 1802 401 + 1485 = 1886 401 + 1584 = 1985 615 + 1401 = 2016 615 + 1584 = 2199 1171 + 1401 = 2572 1277 1401 = 2678 1401 X 2 = 2802 1277 + 1584 = 2861 1401 + 1584 = 2985
+
#Ordinary Raman spectrum. * Acridinium chloride. cTaken in CCI4 solution. dFull width at half-maximum (FWHM).eTaken in aqueous solution at pH 1.7. /Reference 14. gValues in parentheses are normalized Raman intensities. *IR peak.
sol particles would function as a Lewis acid. Considering that Cl- is needed to obtain the SER spectrum of acridinium ion without photoreaction, the species responsible for Figure 2a may be acridinium ion paired with surface chloride. A similar mechanism has been reported for other positively charged species that are not likely to bind directly to a metal For example, the presence of specifically adsorbed anions such as halides has been known to a prer uisite for the observation of a SER spectrum of pyridinium ion3 In this case, it was supposed that pyridinium did not adsorb directly on the metal surface but rather was in the diffuse double layer, being associated with adsorbed chloride to form an ion pair. Similarly, we propose that the acridinium ion forms an ion pair with the chloride and that the two are adsorbed together on the silver surface, resulting in the SER spectrum shown in Figure 2a. When the acridine and the acridinium-hloride ion pair coexist on the surface, the SER spectrum would appear as a superposition of the spectra of two different species. As mentioned above, the peaks appearing in the SER spectrum at pH 4.7 can be attributed to either adsorbed acridine and acridinium-chlorideion pair. For instance, the peak at 1584 cm-' in Figure 2b c a n be attributed to the adsorbed ion pair, while the 1565-cm-' peak can be attributed to the adsorbed acridine molecule. Some bands in Figure (18) (a) Tadayyoni, M.A.; Faquharson, S.;Weaver. M.J. J . Ckm. Phys. 1% 80,1363. (b) Faquharson, S.;Lay. P.A.; Weaver, M.J. Spectrochim. Acta 1984, IOA, 907.
2b seem to be slightly broader than the corresponding bands in Figure 2a or 2c. This may occur due to overlapping of bands belonging to the same type of vibration but originating from two different chemical species. Hence, the spectrum in Figure 2b can be interpreted in terms of the coexistence of two species adsorbed on silver. We have mentioned already that the SER spectrum of acridine obtained in the absence of C1- or in a highly basic medium with Cl- is due to unidentified photoreaction product(s). In addition, such spectra were observed to have broad backgrounds around 1200-1 600 cm-'. This is supposed to arise from carbon overlayers on silver formed through photodecomposition of acridine or an unknown reaction product. The exact cause of photoreaction and background growth with a pH increase is not certain. However, suppression of the surface photoreaction by chloride ion may be explained in terms of the competitive adsorption of chloride and hydroxide ions on silver. The effect of solution pH on SERS of halide species has been studied previously.'2 The higher the pH of the silver sol solution, the more surface sites are occupied by hydroxide ions rather than chloride ions. The competitive adsorption of OH- and C1- on silver surfaces was reported to occur at around pH lO.I3 Accordingly, the decrease in the intensity of the Ag-Cl stretching vibration at higher pH could be attributed mainly to the decrease in the surface coverage of chloride on silver.'* Moreover, in the SER spectrum of pyridine-chloride at basic pH, a broad background appeared distinctly around 1200-1600 cm-l. In contrast to the SER spectra of chloride, the
The Journal of Physical Chemistry, Vol. 95, No. 22, 1991 8847
Reaction of Acridine in Silver Sol SER spectra of bromide hardly changed with raising the solution pH. Even a t pH 12, the intensity of the Ag-Br stretching vibration was nearly the same as that at pH 5 . This was explained on the basis of the stronger adsorption strength of Br- on silver compared to that of CI-.12 Namely, due to the strong adsorbing power of bromide, hydroxide ion could not compete effectively with bromide for adsorption on silver surface, while it could dominate chloride at basic condition. Brandt'O studied the electrochemical reaction of hydrogen peroxide at a silver electrode and found that the catalytic activity of the silver surface was reduced significantly when the surface had been exposed to halide adsorbates. Farquharson et aI.l9 used the SERS technique to investigate the kinetics of various electrochemical reactions. It was suggested that surface sites displaying efficient Raman scattering might also provide centers of catalytic activity. Based on the above information, the enhanced photoreaction of acridine at high pH (>IO) is very likely to be caused by depopulation of surface chloride. The catalytic sites would become more available in such a condition for the adsorption and subsequent reaction of acridine to unknown product and carbon. Surface oxygen is known to possess a very strong Bronsted basicity and nucleophilicity.? The species is also known to create Lewis acid sites that may be catalytically active. There are many reports suggesting a dramatic effect of surface oxygen atoms on the carbon-carbon bond activation or decomposition of adsorbate.21*22It can thus be supposed that the surface oxygen species induces the photoreaction of acridine. Once hydroxide ions adsorb on the silver sol surface, various oxygen species, including surface oxygen itself, may be formed on the surface. These active species may play some role in the photoformation of an unknown reaction product in the present case. Acridine may bind to the surface either via its ring K system or via nitrogen lone pair electrons. For aromatic molecules, it has generally been known that the frequencies of ring stretching vibrations decrease or red shift by more than 10 cm-' and their bandwidths increase substantially when the molecules adsorb on the metal surface via their A systems.*) Such a red shift usually arises due to the weakening of the bonds in the ring system caused by backdonation of the metal d election to the antibonding A* orbitals of the ring system.23 Additional vibrational relaxation caused by interaction between the metal surface and the ring is responsible for the increase of the bandwidth. In the case of acridine, the ring stretching modes in the SER spectrum in Figure 2c exhibit a blue shift by 4-10 cm-I from the corresponding positions in the OR spectrum in Figure 1b. On the other hand, the bandwidths were hardly affected by surface adsorption. This indicates that the surface-ring n-orbital overlap does not occur significantly when acridine adsorbs on the silver surface. Hence, it is likely that acridine is adsorbed on silver via its nitrogen lone pair electrons. It can be seen from the OR spectral data that the frequencies of most of the vibrations of acridine increase as the molecule becomes protonated at the nitrogen atom. If the acridine molecule adsorbs on silver via its nitrogen atom, a similar increase in vibrational frequencies would be expected. This is in general agreement with what has been observed in the SER spectrum shown in Figure 2c. For the acridinium-chloride ion pair, nitrogen lone pair electrons are no longer available for surface coordination. Hence, the adduct should adsorb on silver via the chlorine atom and/or via its ring 7r system. It is seen that the ring stretching modes in Figure 2a have been more or less red shifted from the corresponding peaks in Figure la. However, the magnitude of the red shift has been 7 cm-' at the largest. Moreover, the bandwidths were hardly affected by the surface adsorption. Hence, it is likely that the acridinium-chloride ion pair does not interact with the surface via its ring A system. If the ion pair adsorbs on silver via the ~
(19) Farquharson, S.;Milner, D.; Tadayyoni, M. A.; Weaver, M. J. J . Electroanal. Chem. 1984, 178, 143. (20) Madix, R. J . Science 1986, 233, 1159. (21) Sault, A. G.; Madix, R. J. Sutf Sri. 1986, 172, 598. (22) Capote, A. J.; Madix, R. J. Surf. Sei. 1989, 214, 276. (23) Gao. P.; Weaver, M. J. J . Phys. Chem. 1985.89, 5040.
chlorine atom, the frequency of the Ag-CI stretching vibration will become different from that not involved in ion pair formation. Indeed, a noticeable change in the v(Ag-CI) band is seen in Figure 2a when compared to the same band in Figure 2c. According to previous investigation^,^^ the A g x l stretching vibration is known to appear at around 240 cm-I. The peak at 239 cm-l in Figure 2c may thus be assigned to this vibration for the CI-not involved in ion pair formation. There is a possibility, however, that this band may be due to the vibrational mode of acridine because a weak band is present at around 240 cm-' in the OR spectra (Figure l a and Ib). To check such a possibility, SER spectra were obtained over a wide pH range again, this time CI- being substituted by Br-. Even though other spectral features remained almost the same regardless of the halide species added, the band at 240 cm-' was absent in the SER spectra obtained in the presence of B i . Accordingly, the band around 240 cm-I in Figure 2 can be safely assigned to the Ag-C1 stretching mode. Before going further, it may be informative to notice that neither the photoreaction nor the surface carbon formation occurs when the SER spectrum of acridine is recorded in the presence of Breven under a highly basic condition. The SER spectrum thus obtained at pH 12.2 was nearly identical with that obtained in the presence of CI-at pH 8.8 (Figure 2c). Recalling that the adsorption strength of Br- on silver is far greater than that of CIand that of OH- cannot compete with Br- for adsorption on the silver surface,12 the above observation supports the earlier conclusion that the halide ions adsorbed on silver reduce the photoreaction of adsorbed acridine. The similarity between the SER spectrum obtained in the presence of CI- at pH 8.8 and that with Br- at pH 12.2 also supports that the species responsible for Figure 2c is the adsorbed acridine itself. As can be seen from Figure 2, the v(Ag-Cl) band becomes more asymmetric as the sol pH decreases. At pH 2.4, a shoulder peak at 217 cm-I can be seen distinctly. Although the 243-cm-I peak in Figure 2a can be readily attributed to the Ag-CI stretching vibration not affected by acridine adsorption, the origin of the band at 217 cm-* is a matter of conjecture. The asymmetrical feature of the band a t 240 cm-I has been reported previously. Pettinger et aLZ4identified two peaks a t 216 and 234 cm-' by taking the difference of SER spectra for CI- obtained at different electrode potentials. These were attributed to the different number of coordinated ligands in the adatom-halide-water surface complex, AgX,,(H,O),,,. Rogers et alesreported very little effect of pyridine on the shape of the v(Ag-Cl) band. Broadening of this band toward the lower frequency side from the peak center was attributed to a distribution of sites to which chloride ions were chemisorbed. On the other hand, Sun et a1.6 reported the 235-cm-l peak to become distorted in the presence of pyridine or pyridinium. It was suggested that interaction between pyridinium and chloride lowered the frequency of the Ag-CI stretching vibration, resulting in the observed distortion. A similar observation could be made in the present case also. As the acidity of sol solution was raised, the 217-cm-' peak became more distinct when acridinium existed together with CI- ions. Figure 3 shows the 100-300- and 15001600-~m-~ regions of SER spectra obtained at various pH's. As the sol pH was lowered, the intensity of the 1584-cm-' band increased while that of the 1565-cm-' band decreased concurrently. It is to be recalled that these bands have been attributed to adsorbed acridinium and acridine, respectively. As the latter band becomes more distinct, the bimodal character of the v(Ag-CI) band diminishes gradually. On the other hand, the v(Ag-Cl) band shape and bandwidth were observed to be comparatively less sensitive to the sol pH variation in the absence of acridine. The present observation suggests that acridinium in the interface near the silver surface affects at least in part the Ag-CI stretching vibration by forming an ion pair with chloride. It has been concluded that acridine adsorbs on the silver surface via its nitrogen lone pair electrons, while acridinium adsorbs as an ion pair via the chlorine atom. Orientation of the adsorbed 85.(24) 2746.Pettinger, B.; Philpott, M. R.; Gordan, M. J., 11. J . Phys. Chem. 1981,
8848 The Journal of Physical Chemistry, Vol. 95, No. 22, 1991
Oh et al.
z ‘5
-1
I
71
= 61. a s
q
I
I
h
ob
’
I
’
i
I
’
6
.
b
’
Ib
PH
Figure 4. pH dependence of SERS intensity for the most intense band at around 1400 cm-I. Excitation with 488 ( 0 ) and 514.5 nm (0). Concentrations of acridine and KCI in the sol solution were 5 X lo‘ and 1 X IO-’ M, respectively.
A pH6.1
Figure 3. SER spectra of 5 X IO-‘ M acridine in silver sol solution containing 1 X IO” M KCI at various pH’s in the 100-300- and 15001600-cm-1 regions. pH’s being adjusted in the final step for the same batch of sol solution. Excitation with (a) 488- and (b) 514.5-nm lines, respectively.
species cannot be determined conclusively, however, because the surface selection rule for SERS has not been established unequivocally. Nevertheless, according to the electromagnetic surface selection r ~ l e , ~a ~vibrational . ~ , ~ ~ mode with its normal mode component perpendicular to the metal surface is likely to become more enhanced than the parallel one. In particular, the C-H stretching vibrations were reported to be relatively unambiguous probes for adsorbate orientation.’*26 For a molecule belonging to the C, point group such as acridine, the A2 vibrational mode should appear distinctly in the SER spectrum if the molecule assumes a face-on stance with respect to the metal surface, while the B2 modes become more enhanced for edge-on adsorption. In the charge-transfer mechanism, only those vibrations that are totally symmetric with respect to the symmetry elements of the “moleculesurface composite system” are enhanced.27 Vibrations satisfying such a requirement are totally symmetric vibrations of the free adsorbate and modes in which equivalent atoms move in phase perpendicular to the surface. In the SER spectra of both acridine and acridinium, the totally symmetric AI vibrational modes were observed to be substantially enhanced upon surface adsorption. Although the B2-type modes can be seen in the SER spectra, the A2 vibrational bands are hardly detectable. More importantly, the C-H stretching bands appear distinctly in all the SER spectra. All these observations suggest that both species, acridine and acridinium, should assume perpendicular or at least tilted stances with respect to the silver surface. The observation that the bandwidths are hardly affected by surface adsorption also supports that flat orientation of these species with respect to the surface is not likely. Nevertheless, the tilt angle of the acridine ring with respect to the silver surface cannot be determined at the moment. As mentioned above, pyridinium has been reported to be associated with chloride to form an ion pair on the silver surface.&’ The model suggests that pyridinium should be separated from the (25) (a) Hallmark, V. M.;Campion, A. J . Chem. Phys. 1986,84,2933. (b) Hallmark, V. M.;Campion, A. J . Chem. Phys. 1986,84, 2942. (26) Gao. X.; Davies, J. P.;Weaver, M.J. J. Chem. Phys. 1990,94,6858. (27) (a) Adrian, F. J. J . Chem. Pbys. 1982, 77, 5302. (b) Lombardi, J. R.; Birke, R. L.; Lu, T.; Xu, J. J . Chem. Phys. 1986, 84, 4174.
surface by Cl-. This may dictate that the SER enhancement of pyridinium is more or less different from that of pyridine, which adsorbs on the surface directly via its nitrogen lone pair electrons. Rogers et aLs reported a 5- to IO-fold decrease in intensity of the ring breathing modes at pH 1.9 compared to the case at pH 8.5. Since electromagnetic field enhancement was not likely to diminish appreciably at 0.34 nm (chloride ion diameter) away from the metal surface,28 they attributed the intensity decrease to the decrease in chemical contribution to the Raman scattering enhancement.’ A similar phenomenon may be expected to occur in the present system. In contrast with the case of the pyridinium-pyridine system, however, the SERS bands of acridinium at pH 2-4 are observed, on the average, to be 3 times more intense than those of acridine at pH 6-9. Such an intensity variation may be attributed to a change in the state of sol aggregation caused by the pH variation. To further investigate the causes for the difference in the surface enhancement between acridine and acridinium, SER spectra were obtained at two different excitation wavelengths, namely, 514.5 and 488.0 nm of argon ion laser. It was assumed initially that the relative SERS enhancement would not be affected much in the two cases if a difference in the degree of sol aggregation is the main reason for intensity change because the excitation wavelengths were not much different. However, as shown in Figure 3, the relative intensity of bands at 1565 and 1584 cm-’ was observed to depend strongly on the excitation wavelength even at a given pH. This may imply that a change in the state of sol aggregation is not a priori cause for a stronger surface enhancement of acridinium. To obtain further insight, the intensity variation of the strong SERS band at around 1400 cm-I has been examined. It is Seen from Figure 4 that the SERS intensity of acridinium at 488-nm excitation is around 2 times stronger than at 514.5-nm excitation. On the other hand, SERS of acridine at pH 8.8 is rather insensitive to the excitation wavelength. With 488.0-nm excitation, the SERS intensity of acridinium at pH 2.4 is thus 5-6 times greater than that of acridine at pH 8.8. It has been well-known that, in the electrochemical experiments, the electrode potential has a marked effect on the SERS intend ~ The . potential ~ ~ at which the SERS intensity reaches a maximum for a given adsorbate is reported to shift with the change in the excitation wavelength. This can be interpreted in terms of a resonance Raman process associated with new electronic transitions, metal to adsorbate or adsorbate to metal, which become possible when a molecule is placed at the metal surface. Such a mechanism clearly accounts for the differing enhancement for different molecules since only certain molecules will have donor (28) (a) Ncviere, M.;Reineish, R. Phys. Rea. B 1982, 26, 5403. (b) Murray, C. A.; Allara, D. L. J . Chem. Phys. 1982, 76, 1290. (29) (a) Furtak, T. E.;Macomber, S.H.Chem. Phys. Le??.1963,95,328. (b) Furtak, T. E.; Roy, D.Phys. Reu. Le??.1983, 17, 1301. (c) Lombardi, J. R.; Birke, R. L.; Sanchez, L.A.; Bernard, 1.; Sun, S.C. Chem. Phys. Le??. 1984, 104, 240.
The Journal of Physical Chemistry, Vol. 95, No. 22, 1991 8849
Reaction of Acridine in Silver Sol
or acceptor levels of the correct energy to participate in the charge transfer to or from the surface. We have already inferred that the substantial enhancement of the totally symmetric vibrational modes might be caused by the charge-transfer mechanism. On the basis of the above information, the excitation wavelength dependence of the SERS signal in Figure 3 can be attributed to the involvement of the charge-transfer process. As described below, this is supported further by a more closer examination on the SER spectral features. A particularly noteworthy feature in the SER spectrum of acridinium is the appearance of numerous bands assignable to either combination or overtone of the fundamental vibrations. Within the experimental uncertainty, the calculated frequencies were in good accord with the observed ones. For instance, the 2799-cm-' peak in Figure 2a can be assigned to the overtone of the band at 1401 cm-I. The detailed data are collectively listed in Table 1 also. Few observations of overtone and combination bands30 have been made in SERS except for adsorbed dye, which shows surface-enhanced resonance Raman scattering (SERRS). Since the aqueous solution of acridinium chloride does not exhibit any electronic absorption in the excitation wavelength region,3' SERRS does not seem to be a likely mechanism for the observation of overtone and combination bands in this case. As an alternative mechanism, overtone and combination bands can be observed distinctly when the charge-transfer effect contributes substantially to the surface enhancement of Raman The observation that the SERS relative intensities of acridinium show an excitation wavelength dependence, while those of acridine do not, can also be understood based on the charge-transfer mechanism. Hence, it is thought that not only the usual electromagnetic effect but also the charge-transfer effect play important roles in the SERS of acridinium. We have already concluded that the 217-cm-' peak is related at least in part to the acridinium-chloride ion pair. In Figure 3b, the 21 7-cm-' peak appears as a shoulder peak, being much weaker than the 243-cm-l peak. This is in marked contrast with that observed in Figure 3a. That is, the 217-cm-' peak becomes more distinct, in relative sense with 488-nm excitation. Assuming that the acridinium-chloride ion pair experiences a charge-transfer effect and that the 21 7-cm-' peak has relevance to the ion pair, the above observation can be easily understood. One may expect that interaction of acridine or acridinium with the silver surface or with the specifically adsorbed chloride would change the acid dissociation constant of acridinium. In this regard, the SERS band intensities of acridinium and acridine have been analyzed in terms of the sol pH variation. For the following surface equilibrium,
Y
0
O
"I 3
.-
.05
Y
PH
Figure 5. Surface concentration ratio of conjugate acid-base pairs vs pH of sol. Excitation with (0)488- and ( 0 )514.5-nm lines, respectively. The dotted line corresponds to the concentration ratio vs pH in the aqueous solution.
Hence, an effective acid dissociation constant at the surface may be determined once a method to measure the ratio of surface
concentrations of an acid and its conjugate base can be devised. In the present analysis, the SERS band intensities of the bands a t 1565 and 1584 cm-l have been taken as relative measures of acridine and acridinium ion concentrations. The major problem in such a determination of concentration ratio arises from the fact that the surface enhancement factor containing contribution from the electromagnetic and charge-transfer effects can be different for acridine and acridinium ion. To account for such a difference, the intensity of the strongest SERS band at around 1400 cm-l was measured as a function of bulk pH. The intensity of this band at pH 2-4 was 3.2 and 5.4 times larger than that at pH 6-9 with 5 14.5- and 488-nm excitation, respectively. These values were used to estimate the concentration ratios from the observed intensity ratios. Such a treatment necessitates an assumption that the total number of acridines and acridiniums adsorbed on the surface at pH 6-9 and pH 2-4, respectively, are equivalent, which can be erroneous by as much as an order of magnitude. The common logarithm of the concentration ratio thus evaluated is plotted as a function of bulk pH in Figure 5. Two log-log plots obtained for 488- and 5 14.5-nm excitations are nearly indistinguishable, indicating that the charge-transfer contribution to the surface enhancement has been adequately accounted for, in relative sense. The PKA' value determined from the x intercept is 3.7 compared to the corresponding value of 5.60 in the bulk aqueous solution. This means that the acid dissociation constant of acridinium has increased by 100 on the surface compared to that in the bulk aqueous phase. For such data to be of quantitative value, it is of course needed to determine the total numbers of acridine and acridinium ion at their respective optimum pH values. One of the intriguing features in Figure 5 is the fact that the slopes of the log-log plots are different from -1, which is the expected value in the bulk-phase equilibrium. This may arise from the failure of our assumption of equal number of surface sites accessible to both chemical species. Alternatively, it may possibly be due to the buffering effect of the silver surface toward the pH variation in the bulk aqueous solution.
(30) (a) Pettinger, B. G e m . Phys. Lett. 1981, 78, 404. (b) Takahashi, M.;Goto, M.; Ito, M. Chem. Phys. Left. 1985, 121, 458. (31) Kasama, K.; Kikuchi, K.; Yamamoto, S.; Uji-ie, K.; Nishida, Y.; Kokubun. H. J . Phys. Chem. 1981.85, 1291.
Acknowledgment. This work was supported by the Ministry of Education, Republic of Korea, and the Korea Science and Engineering Foundation. Registry No. Ag, 7440-22-4; CI, 16887-00-6; acridine, 260-94-6.
AH+(s) = A(s)
+ H+(b)
where s and b indicate surface and bulk species, respectively, the acid dissociation constant, KA', can be expressed as PH = PKA' - log ([AH+(s)l/ [A(s)l)