Interparticle Electron Transfer in Metal ... - ACS Publications

Sep 25, 1997 - Benjamin H. Meekins and Prashant V. Kamat ... Matthew R. Jones , Kyle D. Osberg , Robert J. Macfarlane , Mark R. Langille , and Chad A...
0 downloads 0 Views 181KB Size
© Copyright 1997 by the American Chemical Society

VOLUME 101, NUMBER 39, SEPTEMBER 25, 1997

LETTERS Interparticle Electron Transfer in Metal/Semiconductor Composites. Picosecond Dynamics of CdS-Capped Gold Nanoclusters Prashant V. Kamat* and Bhairavi Shanghavi† Notre Dame Radiation Laboratory, UniVersity of Notre Dame, Notre Dame, Indiana 46556 ReceiVed: March 14, 1997; In Final Form: August 6, 1997X

Composite nanoclusters of gold and CdS with core/shell geometry have been synthesized using a chemical precipitation method in aqueous media. These nanoclusters are photoactive and exhibit transient bleaching in the 400-600 nm region when subjected to 355 nm laser pulse excitation. Capping of gold colloids with ultrasmall CdS nanoclusters (particle diameter ∼4 nm) significantly alters the picosecond dynamics of the gold core. The bleaching of the surface plasmon absorption of the gold core is achieved by exciting the CdS shell. The major fraction of the interparticle electron transfer between CdS and Au nanoclusters is completed within the laser pulse duration of 18 ps.

Introduction Nanometer-sized semiconductors and metals have drawn considerable attention in recent years because of their sizedependent optical properties.1 While conduction and valence bands of semiconductors are separated by a well-defined bandgap, metal nanoclusters have close lying bands and electrons move quite freely. The surface plasmon absorption band of a metal cluster is dependent on both the cluster size and chemical surroundings.2-4 Of particular interest are the optical properties of gold nanoclusters that are dependent on the laser excitation. For example, the surface plasmon absorption band is bleached in the subpicosecond time scale when excited with a UV or visible laser pulse. The electron dynamics of Au and Ag nanoclusters have been investigated recently by picosecond transient grating spectroscopy5 and femtosecond transient absorption spectroscopy.6,7 Some similarities between metal and semiconductor nanoclusters can be drawn with respect to their physical properties.8 * Address correspondence to this author. E-mail: [email protected] or http://www.nd.edu/∼pkamat. † Co-op student from Department of Chemistry, University of Waterloo. X Abstract published in AdVance ACS Abstracts, September 15, 1997.

S1089-5647(97)00946-2 CCC: $14.00

They are optically transparent and act as dipoles. Significant efforts have been made in our laboratory and elsewhere to investigate the photophysical and photochemical behavior of single and multicomponent semiconductor nanoclusters.9,10 Composite semiconductor nanoclusters of different materials are especially of interest in light energy conversion systems since they promote photoinduced charge separation and enhance the efficiency of photocatalytic reactions. Photoinduced deposition of noble metals such as Pt or Au on semiconductor nanoclusters has often been employed to enhance their photocatalytic activity.11,12 Recently efforts have been made to bind semiconductor nanoclusters to metal surfaces using a self-assembled monolayer approach13 or to synthesize multilayered metal nanoclusters.14 In our continued efforts to explore the optical properties of composite nanoclusters, we have now elucidated the kinetics and mechanism of the interparticle electron transfer between semiconductor and metal nanocrystallites (Scheme 1). We report here for the first time the optical properties of Au/CdS composite nanoclusters prepared in aqueous medium. © 1997 American Chemical Society

7676 J. Phys. Chem. B, Vol. 101, No. 39, 1997 SCHEME 1: Photoinduced Charge Separation in a Semiconductor-Metal Composite Nanocluster

Experimental Section Au colloids were prepared by the conventional reduction method of reducing HAuCl4 (0.12 mM) in water with sodium citrate at near the boiling temperature.15 The concentration of sodium citrate employed for the reduction was kept low (0.42 mM) to avoid the presence of excess citric acid in gold suspension The ruby-colored solution was cooled and then surface-modified with a thio compound to facilitate binding with CdS nanoclusters. A known amount of 2-mercaptonicotinic acid (MNA) solution in ethanol was mixed with the suspension of Au colloids, and the surface complexation was monitored from the absorption changes in the surface plasmon band. As the surface modifier complexed with the gold colloids the plasmon band broadened with the tail absorption extending into the red region (700 nm). Since the MNA concentration was sufficiently low, we did not encounter any aggregation effects. To check the effects of citric acid, Au colloidal suspension was dialyzed using standard cellulose dialysis tubing (Spectrapor membrane from Scientific products; MW cutoff 12000-14000). These surface-modified Au colloids were further capped with CdS nanocrystallites by adding a known amount of Cd2+ solution and then exposing to H2S under controlled conditions. The colloidal suspension turned brownish as the Au colloids were capped with a layer of CdS nanoclusters. These colloids will be referred to as Au/CdS. The synthetic strategy of Au/ CdS colloids is illustrated in Scheme 2. A separate batch of CdS colloids stabilized with MNA was also prepared by exposing aqueous solution containing 0.25 mM Cd2+ and 3.9 µM MNA to H2S. All colloidal samples were prepared fresh and used within 2-3 h after their preparation. For transmission electron microscopic examination of colloids, a drop of the colloid sample was applied to a carboncoated copper grid. Particle sizes were determined from the photographs taken at a magnification of 150000× using a Hitachi H600 transmission electron microscope. The pictures were further magnified by photographic enlargement. Absorption spectra were recorded with Milton Roy 3000 Diode Array spectrophotometer. Picosecond laser flash photolysis experiments were performed with 355 nm laser pulses from a modelocked, Q-switched Continuum YG-501 DP Nd:YAG laser system (output 2-3 mJ/pulse, pulse width ∼ 18ps). The white continuum picosecond probe pulse was generated by passing the fundamental output through a D2O/H2O solution. The details of the experimental setup and its operation are described elsewhere.16,17 Results and Discussion Gold colloids prepared by the citrate reduction method produce nearly spherical particles with a diameter ranging from 10 to 30 nm. Various spectroscopy techniques have been employed in the past to characterize such gold nanoclusters.18 The transmission electron micrographs of gold and Au/CdS composite nanoclusters synthesized in this study are shown in

Letters parts A and B of Figure 1 respectively. The gold particles in both these samples are fairly monodisperse with a particle diameter of ∼20 nm. Dialysis to remove excess citric acid or capping with MNA of gold colloids did not induce any noticeable changes in the particle size. The CdS nanoclusters prepared by precipitation of Cd2+ with H2S in CdS/Au composite are smaller than the Au nanoclusters. The particle diameter of the CdS nanoclusters is around 4 nm. Thus, the CdS shell that surrounds the core of Au nanoclusters is essentially composed of ultrasmall CdS nanoclusters. A similar observation has also been made earlier by Alivisatos and co-workers19 in their study with CdS/HgS composite semiconductor nanoclusters. Only the particles, which establish interparticle interaction, should cause the changes in the photophysical behavior of the composites. Though not essential, the use of surface modifier, MNA, was beneficial in achieving the colloidal stability and keeping the two particles in close proximity. It is interesting to note that most of the Au nanoclusters interact quite effectively with CdS nanoclusters. However, some inhomogeneity was evident from TEM analysis as can be seen from the presence of few isolated CdS nanoclusters. By increasing the concentration of gold particles it was possible to maximize the interaction between the two particles. This was evident from the increased quenching of CdS emission with increasing concentration of gold core. The absorption spectrum of the Au/CdS colloidal suspension in water is compared with that of the individual Au and CdS colloidal suspensions in Figure 2. CdS colloids exhibit an absorption edge around 520 nm corresponding to the bandgap of 2 eV. The 2-mercaptonicotinic acid modified and unmodified Au colloids exhibit an absorption maximum around 530 nm corresponding to the surface plasmon band. Because of the complexation with the thio group of MNA, the plasmon band is slightly decreased and red-shifted by about 5 nm. As shown earlier, the surface plasmon absorption band of metal nanoclusters is very sensitive to the adsorbed ions. For example, chemisorbed I-, SH-, and C6H5S- ions results in damping of the surface plasmon band of colloidal silver particles.20,21 Alternatively, one can also observe bleaching of the surface plasmon band by depositing electrons from radiolytically produced radicals.22 A more detailed discussion on the damping effects caused by surrounding material can be found elsewhere.2,3,23 The inset in Figure 2 shows an enhanced view of the effect of MNA on the plasmon band. The exclusion of citric acid from colloidal Au suspension by dialysis or surface complexation of Au colloids with MNA had relatively small influence on the plasmon absorption band. The Au/CdS colloids exhibit strong absorption in the UV and visible region, with absorption extending up to 700 nm. The absorption spectra recorded in Figure 2 did not indicate any aggregation of thio-capped gold particles in the present experiments. The absorption of the Au/ CdS composite system has a plasmon absorption band with absorption maximum at 545 nm. The sharp increase in the absorption at wavelengths below 500 nm confirms the formation of CdS nanoclusters. It is interesting to note that the absorption of Au/CdS in the visible region is greatly increased and the surface plasmon absorption band is red-shifted by about 15 nm compared to that of Au colloids. This suggests that the absorption properties of Au/CdS are not the result of a simple addition of the spectra of two nanoclusters, but indeed, the CdS shell influences the surface plasmon absorption of the gold core. Such an influence of a semiconductor shell on the electronic properties of the core material has been recently reported for a variety of semiconductor systems.10 Semiconductor composites

Letters

J. Phys. Chem. B, Vol. 101, No. 39, 1997 7677

Figure 1. Transmission electron micrographs (TEM) of Au nanoclusters (left) and Au/CdS (right) composite nanoclusters. (An enlarged view of the CdS-capped gold nanocluster is shown in the inset.) The pictures were originally recorded with a magnification of 150 000 and were further magnified photographically

SCHEME 2

with electronic properties that differ from simple superposition of the electronic properties of two semiconductor particles have been observed for the CdS/HgS system.24,25 In the case of CdS@PbS colloids, the excited energy gap and photoluminescence energy were shown to be dependent on the core/shell ratio of the composite.26 Figure 3 shows the emission spectra of CdS colloids at different Au core concentrations. Separate precipitations of CdS were carried out in acetonitrile solutions containing different amounts of gold colloids. The amount of CdS deposited on the gold core in these experiments was kept constant. We observed a quenching of CdS emission as we increased the concentration of the gold core. More than 70% emission was quenched when we increased the gold core concentration to 13.7 µM. The emission quenching is indicative of the occurrence of electron transfer from excited CdS into Au core. Since the conduction band energy of CdS is around -1.0 V vs NHE and the Fermi level of gold is at +0.5 V vs NHE, we expect a favorable energetics for such an electron transfer between the two nanoclusters. Similar emission quenching with electro transfer has also been observed in other composite systems such as CdS/TiO2, CdS/ZnO, and CdS/SnO2 nanoclusters. The photophysical properties of Au/CdS nanoclusters were further probed by exciting the colloidal suspension in a picosecond laser flash photolysis apparatus. The transient absorption spectra were recorded at different delay times to compare the transient behavior of Au/CdS nanoclusters with that of CdS and Au nanoclusters. The difference absorption spectra recorded immediately after the laser pulse excitation of CdS, Au and Au/CdS colloids are shown in Figure 4. The difference absorption spectrum of gold colloids shows an intense bleaching of the surface plasmon band at 520 nm. The plasmon

band of metal particles as explained on the basis of Mie theory involves dipolar oscillations of the free electrons in the conduction band that occupy energy states immediately above the Fermi level.2,27 Once these electrons are excited by a laser pulse, they do not oscillate at the same frequency as that of the unexcited electrons, thus resulting in the decrease of the plasmon absorption band. These aspects have been addressed in recent spectroscopic investigations.5,7 The recovery of the plasmon band at the picosecond time scale is mainly due to electronphonon and phonon-phonon relaxations. The CdS colloids, on the other hand, show bleaching close to the absorption edge with a maximum at 480 nm. This bleaching of CdS absorption arises as a result of a shift in the band edge as charge separation occurs in these particles. The dynamics of this bleaching and its recovery has been the topic of many recent investigations.25,28-35 The broad absorption at wavelengths >500 nm is attributed to the hole trapping that further leads the chemical changes at the CdS surface to produce - 17 . This absorption is prominantly seen in the spectra Ssurf recorded at longer time scales (not shown in the figure), as it evolves with a rate constant of 5 × 108 s-1. The transient absorption spectrum recorded following 355 nm laser pulse excitation of Au/CdS colloids shows two distinct transient bleaching maxima at 480 and 545 nm which correspond to the CdS shell and Au core, respectively. The difference absorption spectrum (spectrum c in Figure 4) observed in this set of experiments is different than the simple addition of spectra a and b in Figure 4. It is unlikely that the transient absorption changes of the Au core are due to its direct excitation. Since the absorbance of CdS at 355 nm is nearly twice that of Au colloids, we expect CdS to absorb most of the excitation laser pulse in the composite system. Yet, the net bleaching of the CdS at 480 nm in the Au/CdS system is smaller than that of the CdS nanoclusters alone. On the other hand, the surface plasmon absorption of the composite system is slightly more than that observed for the gold nanocluster. While the surface plasmon bleaching in pristine gold colloids is seen with direct laser pulse excitation, the indirect process of electron

7678 J. Phys. Chem. B, Vol. 101, No. 39, 1997

Letters

Figure 2. Absorption spectra of (a) 0.12 mM colloidal Au stabilized with citric acid; (b) 0.12 mM Au colloids after surface modification with MNA; (c) 0.25 mM colloidal CdS; (d) colloidal Au/CdS composite containing the same amount of Au and CdS as in (b) and (c). Inset shows a magnified view of the plasmon absorption band of gold colloids and the changes observed as a result of (a, a′) dialysis and (b, b′) addition of MNA. Solutions in (b)-(d) contained 3.9 µM 2-mercaptonicotinic acid as surface modifier, which was added before precipitation of CdS.

Figure 3. Emission spectra of 1 mM CdS in acetonitrile at various Au core concentrations: (a) 0 M; (b) 0.83; (c) 6.7; (d) 13.3; (e) 53.3 µM. Appropriate amount of concentrated gold colloids was added to 1 mM Cd2+ solution before its exposure to H2S. Excitation wavelength was 400 nm.

Figure 4. Transient absorption spectra recorded immediately after 355nm laser pulse (pulse width 18 ps) excitation of colloidal suspension (degassed with N2): (a) CdS, (b) Au, and (c) Au/CdS composite. The concentrations were the same as those in Figure 2.

injection mainly causes the bleaching at 545 nm in the Au/CdS composite. The former is due to heating of the electronic gas, and the latter is due to the shift in Fermi level of the gold core. The situation of electron injection in the Au/CdS core is similar to the one observed during the reaction of noble metal colloids with radiolytically generated radicals.9 The photogenerated electrons in CdS nanoclusters are quickly transferred to the Au core, causing the surface plasmon absorption to bleach. Similar observations have been indirectly made in earlier studies. For example, the noble metals (e.g., Pt, Au) deposited on semiconductor particles have been shown to improve the efficiency of photocatalytic reactions.36 These noble metal clusters act as an electron sink and thus promote charge separation in metal-deposited semiconductor particles. The example discussed here demonstrates for the first time the picosecond dynamics of an interparticle electron transfer between semiconductor and metal nanoclusters. We also followed the recovery of the bleaching of the absorption bands corresponding to CdS and Au/CdS nanoclusters at 480 nm (Figure 5A) and Au and Au/CdS nanoclusters

in the 525-540 nm region (Figure 5B). The bleaching recovery of the CdS alone is quite fast (τ ) 76 ps). A small positive is also observed at absorption corresponding to the Ssurf 17 wavelengths greater than 500 nm. The recovery of the bleaching at 480 nm for the Au/CdS composite was slower (τ ) 390 ps). The fraction of the bleaching that fails to recover essentially represents the loss of CdS in the irreversible degradation process. The bleaching recovery of the pristine Au colloids has been shown to consist of a fast (τ ) 2.5 ps) and a slow process (τ > 50 ps).7 These fast and slow recoveries are attributed to the relaxation of “hot” electrons via electron-phonon coupling and phonon-phonon relaxation of the lattice, respectively. Since our excitation laser pulse has a half-width of 18 ps, we couldn’t resolve the fast component of the recovery. The slower component of the recovery (trace a in Figure 5B) had a lifetime of 170 ps. However, a different picture emerged from the bleaching recovery of the Au/CdS composite at 540 nm (trace b in Figure 5B). During the initial 50-60 ps time period, the bleaching did not recover; instead, it grew by about 10%. Although it is difficult to resolve this growth on our experi-

Letters

J. Phys. Chem. B, Vol. 101, No. 39, 1997 7679 spectroscopy methods. Further experiments are underway to improve the efficiency of photoinduced charge separation of composite semiconductor nanoclusters and to utilize them in photocatalytic reactions. Acknowledgment. We thank Mr. Chouhaid Nasr for his help in kinetic analysis. The work described herein was supported by the Office of the Basic Energy Sciences of the U.S. Department of Energy. This is Contribution No. 3990 from the Notre Dame Radiation Laboratory. References and Notes

Figure 5. (A) Bleaching recovery as monitored from the transient absorption at 480 nm (ex. 355 nm): (a) CdS and (b) Au/CdS composite. (B) Bleaching recovery as monitored from the transient absorption at maximum bleaching (ex. 355 nm): (a) Au (525 nm) and (b) Au/CdS composite (540 nm) nanoclusters. The experimental conditions were the same as those in Figure 4.

mental time scale, it indicates the maximum time limit with which the photogenerated electrons from the CdS shell are transferred into the Au core. The major fraction of the electron transfer seems to be completed within the laser pulse duration of 18 ps. Since the bleaching of plasmon absorption in the Au/ CdS composite is mainly caused by the indirect excitation of CdS, we do not expect to see any fast recovery that otherwise would have resulted from the relaxation of electrons within the Au core. We observe only a slow component of the recovery that possibly arises from the back electron transfer to the in the CdS cluster. On a longer time scale most of the Ssurf bleaching at 540 nm recovered with a lifetime of 400 ps, which is slower than the one observed with the colloidal Au system alone. Conclusions Our preliminary transient absorption results presented here highlight the possibility of developing metal-semiconductor composite nanoclusters with properties that are different than those of the individual components of the composites. The electron transfers from CdS to Au core result in the bleaching of surface plasmon band and occur within the time frame of our laser pulse (18 ps). The slower recovery of the transient bleaching suggests improved charge separation in the Au/CdS composite system. The results presented here show the feasibility of monitoring interparticle electron transfer in metal/ semiconductor composite system using transient absorption

(1) Henglein, A. Ber. Bunsen-Ges. Phys. Chem. 1995, 99, 903. (2) Kreibig, U.; Vollmer, M. Optical Properties of Metal Clusters; Springer: Berlin, 1995. (3) Kreibig, U.; Gartz, M.; Hilger, A.; Hovel, H. In Fine Particles Science and Technology; Pelizzatti, E., Ed.; Kulwer Academic Publishers: Boston, MA, 1996; p 499. (4) Mulvaney, P. Langmuir 1997, 12, 7788. (5) Heilweil, E. J.; Hochstrasser, R. M. J. Chem. Phys. 1985, 82, 179. (6) Roberti, T. W.; Smith, B. A.; Zhang, J. Z. J. Chem. Phys. 1995, 102, 3860. (7) Ahmadi, T. S.; Logunov, S. L.; El-Sayed, M. A. J. Phys. Chem. 1996, 100, 8053. (8) Mulvaney, P. In Semiconductor Nanoclusters-Physical, Chemical and Catalytic Aspects; Kamat, P. V., Meisel, D., Eds.; Elsevier Science: Amsterdam, 1997; p 99. (9) Henglein, A. J. Phys. Chem. 1993, 97, 5457. (10) Kamat, P. V. In Semiconductor Nanoclusters-Physical, Chemical and Catalytic Aspects; Kamat, P. V., Meisel, D., Eds.; Elsevier Science: Amsterdam, 1997; p 237. (11) Kraeutler, B.; Bard, A. J. J. Am. Chem. Soc. 1978, 100, 4317. (12) Borgarello, E.; Harris, R.; Serpone, N. NouV. J. Chim. 1985, 9, 743. (13) Colvin, V. L.; Goldstein, A. N.; Alivisatos, A. P. J. Am. Chem. Soc. 1992, 114, 5221. (14) Mulvaney, P.; Giersig, M.; Henglein, A. J. Phys. Chem. 1993, 97, 7061. (15) Turkevich, J.; Stevenson, P. L.; Hillier, J. Discuss. Faraday Soc. 1951, 11, 55. (16) Ebbesen, T. W. ReV. Sci. Instrum. 1988, 59, 1307. (17) Kamat, P. V.; Ebbesen, T. W.; Dimitrijevic, N. M.; Nozik, A. J. Chem. Phys. Lett. 1989, 157, 384. (18) Rao, C. N. R.; Vijayakrishnan, V.; Aiyer, H. N.; Kulkarni, G. U.; Subbanna, G. N. J. Phys. Chem. 1993, 97, 11157. (19) Alivisatos, P. J. Phys. Chem. 1996, 100, 13226. (20) Linnert, T.; Mulvaney, P.; Henglein, A. J. Phys. Chem. 1993, 97, 679. (21) Gutierrez, M.; Henglein, A. J. Phys. Chem. 1993, 97, 11368. (22) Henglein, A.; Mulvaney, P.; Linnert, T. Faraday Discuss. 1991, 92, 31. (23) Persson, B. N. J. Phys. ReV. B 1989, 39, 8220. (24) Haesselbarth, A.; Eychmueller, A.; Eichberger, R.; Giersig, M.; Mews, A.; Weller, H. J. Phys. Chem. 1993, 97, 5333. (25) Eychmueller, A.; Vobmeyer, T.; Mews, A.; Weller, H. J. Lumin. 1994, 58, 223. (26) Zhou, H. S.; Sasahara, H.; Honma, I.; Komiyama, H.; Haus, J. W. Chem. Mater. 1994, 6, 1534. (27) Mie, G. Ann. Phys. B 1908, 25, 377. (28) Henglein, A.; Kumar, A.; Janata, E.; Weller, H. Chem. Phys. Lett. 1986, 132, 133. (29) Dimitrijevic, N. M.; Kamat, P. V. J. Phys. Chem. 1987, 91, 2096. (30) Kamat, P. V.; Dimitrijevic, N. M.; Nozik, A. J. J. Phys. Chem. 1989, 93, 2873. (31) Wang, Y.; Suna, A.; McHugh, J.; Hilinski, E. F.; Lucas, P. A.; Johnson, R. D. J. Chem. Phys. 1990, 92, 6927. (32) Brus, L. Appl. Phys. 1991, A53, 465. (33) Kamalov, V. F.; Little, R.; Logunov, S. L.; El-Sayed, M. A. J. Phys. Chem. 1996, 100, 6381. (34) Klimov, V. I.; Bolivar, P. H.; Kurz, H. Phys. ReV. B 1996, 53, 1453. (35) Klimov, V. I.; Bolivar, P. H.; Kurz, H.; Karavinskii, V. A. Superlattices Microstruct. 1996, 20, 395. (36) Kalyanasundaram, K.; Graetzel, M.; Pelizzetti, E. Coord. Chem. ReV. 1986, 69, 57.