J. Phys. Chem. 1992, 96, 4593-4598
4593
Photon Emission at the Metal/Acetonitrile Solution Interface: Effects of Redox Species and Electrode Metal Kei Murakoshi and Kobei Uosaki* Physical Chemistry Laboratory, Department of Chemistry, Faculty of Science, Hokkaido University, Sapporo 060, Japan (Received: November 13, 1991; I n Final Form: February 12, 1992)
To investigate the details of the photon emission at metal/electrolyte interfaces induced by an electron-transfer reaction, the effects of electrode metals (Au, Pt, and Pd) and redox species (benzonitrile, trans-stilbene, and benzophenone) on emission properties were studied. The high-energy threshold of the spectrum, Eth,changes linearly with the electrode potential, U , and is almost equal to -e(V - U,)where V is the redox potential of the chemical species. The peak position of the spectrum, Ep,also depends though less sensitively on U,and V.Although E* is solely determined by U,and V ,Ephas metal dependence. Emission efficiency, a, increases, saturates, and then decreases as U,becomes more positive. The potential dependence of also depends on the choice of metal. Based on these results, several p i b l e mechanisms for the photon emission are considered, and it is concluded that the major contribution to the photon emission is the charge-transfer reaction inverse photoemission process involving bulk band states.
Introduction An electron-transfer reaction is one of the most important elementary steps of many processes at solid/electrolyte interfaces and is a subject of numerous investigations. Most studies are phenomenological and macroscopic. Recently, various in-situ techniques have been introduced to obtain moIecular (IR) and atomic (STM/AFM) structures of the interface. Information of the electronic states of the interfaces is also very important, since they certainly play key roles in electron-transfer reactions. Although high-energy electron and photon scattering techniques such as ultraviolet photoemission spectroscopy (UPS)’ and inverse photoemission spectroscopy (IPS)* can be used to elucidate the electronic states of a solid surface under ultrahigh-vacuum (UHV) conditions, they are not suitable as in-situ techniques in electrochemical systems. Recently, McIntyre and Sass reported that charge-transfer reactions induce photon emission at electrochemical system^.^" They considered that photons are emitted from a metal electrode as a result of radiative relaxation of injected energetic electrons to empty electronic states in a metal, i.e., the charge-transfer reaction inverse photoemission (CTRIP) process (Figure 1). It is expected that the spectrum of emitted photon reflects the electronic structure of the metal. Although several reports after their observation attempted to determine the electronic states participating to the radiation process,’-1o the interpretation of the spectrum is not satisfactory. One of the difficulties of the analysis arises from the complex scattering process of the injected electrons. Light emission from a metal induced by injected electrons with excess energies of a few electronvolts has been observed also at metal-insulator-metal (MIM) tunneling junction (TJ)” and STMl23l3systems. A totally different model is proposed for the light emission at these systems. In both MIM-TJ and STM (1) Pendry, J. B. Surf. Sci. 1976, 57, 679. (2) Smith, N . V.: Woodruff. D. P. Pro,, Surf Sci. 1986, 21, 295. (3) McIntyre, R.;Sass, J. K. Phys. Rei. Leti 1986, 56, 651. ( 4 ) McIntyre, R.; Sass, J. K. J. Electroanal. Chem. 1985, 196, 199. (5) McIntyre, R.;Roe,D. K.; Sass, J. K.; Storck, W. J. Electroanal. Chem. 1987, 228, 293. (6) McIntyre, R.; Roc,D. K.; Sass, J. K.; Gerischer, H. Electrochemical Sutface Science; Soriaga, M. P., Ed.;ACS Symposium Series 378; American Chemical Society: Washington, DC, 1988; pp 233-244. (7) Ouyang, J.; Bard, A. J. J . Phys. Chem. 1987, 91,4058. ( 8 ) Ouyang, J.; Bard, A. J. J . Phys. Chem. 1988, 92, 5201. (9) Uosaki, K.; Murakoshi, K.; Kita, H. Chem. Lett. 1990, 1159. (10) Uosaki, K.; Murakoshi, K.; Kita, H. J . Phys. Chem. 1991, 95, 779. ( 1 1 ) Lambe. J.: McCarthv. S.L. Phvs. Rev. Lett. 1976. 37, 923. (12) Coomb$, J. H.; Gimzbwski, J. K:; Reihl, B.; Sass, J. K.; Schlitter, R. R. J . Microsc. 1988, 152, 325. (13) Gimzewski, J. K.; Sass, J. K.; Schlitter, R.R.;Schott, J. Europhys. Lett. 1989, 8, 435.
systems, it is suggested that the electron generates visible light in the tunneling process via a surface plasmon excitation-decay process. Thus, it seems to be necessary to evaluate the contribution of this process to the light emission process in an electrochemical system to analyze the emitted photon spectrum. Another difficulty is caused by the complexity of the energy distribution of electrons which is determined by the energy distribution of chemical species in solution and the energy dependence of the electron-transfer process at the electrochemical interface3” and the scattering processes within the metal. For a full understanding of the spectrum, it is necessary to discuss each step quantitatively. In the present work, we report the photon emission at metal (Au, Pt, and Pd) electrodes in acetonitrile solution containing a redox species (benzophenone, trans-stilbene, or benzophenone). The CTRIP and the surface plasmon excitation-decay processes are considered as possible routes for the photon emission process, and it is concluded that the CTRIP process involving bulk band states of the metal contributes most. Experimental Section Materials. Acetonitrile (spectral grade, Dojin Chemical Laboratories) and tetrabutylammonium tetrafluoroborate ((C4H9)4NBF4;reagent grade, Aldrich Chemicals) were used as a solvent and a supporting electrolyte, respectively. The redox species were benzophenone, trans-stilbene, and benzonitrile (reagent grade, Wako Pure Chemicals). All chemicals were purified and dehydrated by the usual techniques.14 The electrolyte solutions used here contain 50 mM redox species and 0.2 M (C4H914NBF4. Metal electrodes (Au, Pt, and Pd) were prepared on clean glass by vacuum deposition using a pure metal wire (99.99%)as a source in 10”-Torr vacuum. The temperature of the glass substrate during Au and Pd deposition was controlled by a hot plate controller (Chino Co., DB-01-3) at 300 and 450 OC, respectively. Pt was deposited onto a glass substrate at room temperature. The deposition rate and the thickness of the film were measured with a quartz crystal thickness monitor (ULVAC, CRTM- 1000). Deposition rate (0.1 nm/s for Au, 0.01 nm/s for Pt and Pd) was controlled by changing the current passed through tungsten wire, around which metal wire was wound. Surface smoothness of the metal films was examined by a scanning tunneling microscope (STM; Digital Instruments, NanoScope I). Apparatus. A spectroelectrochemical cell made of Teflon with a UV-quartz (Fujihara Factory) optical window was used.1° A metal working electrode (area 0.28 cm2) and a platinized Pt mesh (14) Perrin, D. D.; Armarego, W. L. F.; Perrin, D. R. Purification of Laboratory Chemicals, 2nd ed.; Pergamon: New York, 198 1 .
0022-365419212096-4593%03.00/0 0 1992 American Chemical Societv
4594 The Journal of Physical Chemistry, Vol. 96,No. 11, 1 9'92
Murakoshi and Uosaki
Figure 1. Schematic energy diagram of the photon emission process during the oxidation of the high-energy organic cation radical at the metal electrode.
counter electrode were held firmly in the cell so that they were in parallel positions. A Ag/O.Ol M AgN03 electrode was used as a reference electrode. All potentials reported herein are referred to this reference electrode. Electrolyte solutions were prepared in a 10-3-Torr vacuum line and transferred through a Teflon tube to the cell by positive N, (99.999%) pressure. Cyclic voltammetric and potential step experiments were performed with a potentiostat (Nikko Keisoku, NPGS-301s) and a programmable function generator (Toho Technical Research, FG-02). Current transients were recorded by a wave memory ( N F Circuit, WM-811). Light emitted from the working electrode was focused by a convex lens into an imaging spectrograph (Jobin Yvon, CP-200). Spectra were obtained by a multichannel detector (Hamamatsu Photonics, IMD-C3330) equipped with an image intensifier. Charge coupled device (CCD) elements of the detector were cooled to -25 OC. This system covers a photon energy range of 1.4-3.5 eV, limited by the optical properties of the CCD elements. The resolution of the optical system was ca. 4 nm. Spectrum Measurements. In emission measurements, the electrode potential was stepped first to a negative potential, Vi, where the anion radical of the chemical species in solution is formed and then to various positive potentials, U , where the anion radical is oxidized. The pulse width at Vi was 200 ms. When the emission was weak, the pulse width a t Vi was extended to 1 s to generate more radical anion and stronger emission. The exposure time for spectrum measurement at the CCD system was synchronized to the anodic pulse duration followed by the reduction pulse at Vi. The signal from the CCD system was fed into a personal computer (NEC Co., PC-9801/RA51) via a DMA interface and averaged 16 times a t each potential. The spectrum response of the detector system was calibrated by using a tungsten-halogen lamp. As reported previously, the emitted photon spectrum of this system is extremely weak and has a broad shape with the full width at half-maximum (fwhm) of ca.0.4-1.0 eV."O Noise reduction of the spectrum was carried out by a Fourier transformation/inversion procedure with a quadratic filter function.I5 The fwhm of the spectrum becomes broader by ca. 0.05 eV by this noise reduction process. All the measurements were carried out at room temperature. ReSultS
All systems investigated in this study showed reversible redox behavior. No electrochemical reaction other than the redox of the species was observed in the voltammogram within the potential region investigated. Redox potentials of each species, UO,were determined as the average of the cathodic and anodic peak potentials and are -2.15, -2.6, and -2.7 V for benzophenone, trans-stilbene, and benzonitrile, respectively. During the cathodic pulse at vi, a reduction current flowed due to the generation of anion radical but no emission was detected. When the potential was stepped to Uf, an anodic current flowed due to the oxidation of anion radical and light was observed. If (15) Aubarel, E. E.; Myland, J. C.; Oldham. K. B.; Zoski, C. G.J . Elecrroanal. Chem. 1985, 184, 239.
2.0
3.0
Photon Energy / eV
Figure 2. Emission spectra of gold electrode. The potential was stepped from Ui= -2.6 V to various electrode potentials. Ufwas measured in acetonitrile solution containing 50 mM benzonitrile and 0.2 M (C4"4.
Figure 3. Emission spectra of gold electrode in acetonitrile solution containing 50 mM (a) benzonitrile (CP = -2.7 V), (b) tram-stilbene (CP = -2.6 V), and (c) benzophenone (CP = -2.15 V). Uf= 0.4 V.
vi was set too negative, the decomposition of the chemical species contained in the electrolyte solution took place, leading to a noticeable decrease of the emission intensity and the current during the repeat on the double-potential steps. Thus, Vi was usually set to a potential 0.1 V more positive than UO,unless otherwise stated. The emission spectra of Au obtained at various Uf in a solution containing benmnitrile are shown in Figure 2. These spectra show that the high-energy threshold of each spectrum, Eth, and the peak energy of each spectrum, Ep,increase as Uf becomes more positive. The increment of Ethis larger than that of E,. The emission spectra obtained in acetonitrile solutions of trans-stilbene and benzophenone show similar potential dependences. The E t h and Epof these spectra depend also on UO. Figure 3 shows the emission spectra of Au obtained in benzonitrile, trans-stilbene, and benzophenone solutions at a given Uf (=0.4 V). The E t h and Epshift to high-energy direction as UO becomes more negative. In all cases, Eth seems to be equal to - e ( V - Uf) which represents the energy difference between the energy of the highest occupied state in solution and the Fermi level of the electrode (cf. Figure l), Le., the highest energy of injected electrons with respect to the Fermi level, Einj(Eth = -e(UO - U,) = Einj).This relation is further confirmed by the comparison of the spectra obtained in various solutions a t a given E,. Figure 4 shows the spectra of Au in benzonitrile, trans-stilbene, and benzophenone solutions obtained
The Journal of Physical Chemistry, Vol. 96, No. 11, 1992 4595
Photon Emission at Metal/Acetonitrile Interface
Photon Energy
1 eV
Figure 4. Emission spectra of gold electrode in acetonitrile solution containing 50 mM (-) benzonitrile (V = -2.7 V; Vf = 0.2 V), (--) trans-stilbene (VO = -2.6 V; V, = 0.3 V), and (- -) benzophenone (V = -2.15 V; CJf = 0.7 V).
-
-_
..__ ,.> ._/
2.0
VS.
1 1)
Ag/Ag'
Figure 7. Emission efficiency-potential relation: (0) Au, Pd. The conditions are as in Figure 2. 41
I
I
I
-0.5
3.0
/ eV
.-.-.-.-. I
0.5
0 Ut
Figure 5. Emission spectra of Au, Pt,and Pd electrodes. The conditions are as in Figure 2.
t
0.5
Ur 1 V
(m) Pt, (A)
\
--.> Pt
Photon Energy
0
-0 5
1V
1.o
vs. Ag/Ag+
~;l-;-;-;.~
Figure 8. Emission efficiency-potential relation of Pt electrode in acetonitrile solution containing 50 mM benzonitrile (VO = -2.7 V, I) or benzophenone (V = -2.15 V, 0).
31 Ur / V
VS.
Ag/Ag+
0 -3.0
-2.9 -2.8 -2.7 Potential / V vs. Ag/Ag+
-2.6
Figure 6. Potential dependence of the peak energy position, Ep,of the spectra at Au (0), Pt (I), and Pd (A) electrodes. The conditions are as in Figure 2.
Figure 9. Emission efficiency-negative potential limit, Vi, relation. Vf = 0.2 V. The conditions are as in Figure 2.
at a given E , (=2.9 eV) by varying Uf.Three spectra are superposed upon each other relatively well, and the differences of Eth and E , among these spectra are less than 0.2 eV. Photon emission was observed also at Pd and Pt electrodes, although the intensity was very weak at Pt and Pd compared with that at Au. The Eth for Pt and Pd are also given by -e(W - Uf) as is the case for Au. Figure 5 shows the spectra of emitted light at Pt, Pd, and Au obtained in benzonitrile solution at Uf = 0.6 V. The potential dependences of E, for the three metal electrodes are shown in Figure 6. When Uf was more negative than 0 V, the emission intensity at Pt and Pd was very weak and, therefore, it was very difficult to determine E,. Thus, only the values obtained at potentials more positive than 0 V are shown in Figure 6. The E, for Au shifts almost linearly ta high-energy value as Uf becomes more positive. The E, for Pd and Pt shift more steeply than that for Au and saturate in a region of relatively positive potential. The values of E , for Pd and Pt are larger than that for Au when Uf is more positive than 0.1 V. Figure 7 shows the emission efficiency, @, in benzonitrile solution as a function of Up The @ is defined as the integrated emission intensity divided by the anodic charge passed for a given duration. The @ increases drastically as U,becomes positive but saturates or even decreases at still more positive Uts.The potential dependence of @ is different from metal to metal. The 3 at Au is always larger than that at Pt and Pd. At Au electrodes saturation is reached around 0.6 V. Experiments at more positive U,
were not possible on Au due to instability of the electrode. At Pd electrodes 9 h m e s maximum around 0.6-0.8 V and declines slightly at more positive potentials. The 9 at Pt shows a maximum at 0.5 V and then decreases as Uf becomes more positive. Since 0 at 0.5 V was recovered after the experiment at 0.9 V, the decrease of 9 in a positive potential region should not be due to an irreversible surface change nor to formation of an organic polymeric compound. Figure 8 shows the potential dependence of 0 at Pt obtained in benzonitrile and benzophenone solutions. It seems that the two curves coincide if the log *Uf relation in benzophenone solution is shifted negatively by ca. 0.5 V, which is almost equal to the difference of V of these two redox couples. The Q dependence of the emission properties was studied at Au electrodes in benzonitrile solution by keeping Uf constant at 0.2 V. Vi was varied from -2.6 ( X P ) to -2.9 V (