2008, 112, 12074–12076 Published on Web 07/17/2008
Ultrafast Electron Transfer Dynamics in CdSe/CdTe Donor-Acceptor Nanorods Chad J. Dooley, Stoichko D. Dimitrov, and Torsten Fiebig* Merkert Chemistry Center, Boston College, 2609 Beacon Street, Chestnut Hill, Massachusetts 02467 ReceiVed: May 7, 2008; ReVised Manuscript ReceiVed: June 27, 2008
Photoinduced charge transfer dynamics in type-II CdSe/CdTe donor-acceptor nanorods were studied in realtime using femtosecond broadband pump-probe spectroscopy. Following photoexcitation of CdTe (λex ) 620 nm), spectral bleaching from the state-filling signals of the band-edge optical transitions was measured using a femtosecond white light pulse (λprobe ) 350-750 nm). Excitation of the lowest energy CdTe transition shows an ultrafast change in the bleaching signals of both the CdTe and CdSe lowest energy 1S transitions as well as the characteristic carrier transfer band. Our results indicate that interfacial interconduction band electron transfer CdTe to CdSe occurs on the 500 fs time scale in these heteromaterials. Over the past 15 years, reports of semiconductor nanoparticle (NP)-based devices, both all-inorganic1 and organic/inorganic hybrids,2,3 have emerged as scientists have continued to search for new materials to replace silicon-based technologies for solar energy conversion. Second generation heteromaterial-nanostructures with two or more components within a single quantum confined NP are potentially well suited for photonics and photovoltaic applications as carrier donor and acceptor band structures can easily be tuned by varying particle geometries.4 Here we present real-time measurements on the dynamics of photoinduced electron transfer from CdTe to CdSe in colloidal heteromaterial nanorods (hNR) on the femtosecond time scale. Charge carrier dynamics in both spherical and nonspherical NPs have been studied extensively over the past decade using ultrafast pump-probe spectroscopies.5–12,14 Following optical excitation, carrier cooling occurs through multiple processes ranging from subps timescales to times longer than nanoseconds.9 In order to achieve high charge separation quantum efficiencies, carrier transfer (CT) needs to compete with the ultrafast relaxation dynamics. Saykally et al. have studied CdSe/ CdTe nanotetrapods using femtosecond spectroscopy and have observed carrier cooling dynamics for interband, band-edge and intraband transitions, however, they were not able to identify dynamics that could be attributed to CT.5 Studying a type-I CdS/ HgS/CdS quantum dot-quantum well structure, El-Sayed and co-workers have observed electron and hole transfer from CdS to HgS occurring with ∼1.5 and 0.4 ps time constants, respectively.7 Similarly, using a tunable pump pulse and broadband (350-750 nm) white-light probe pulse, we have studied the carrier dynamics in CdSe/CdTe hNRs with welldefined band structures and separable dynamics specific to electron transfer. While type-II core/shell NPs have been synthesized and shown to exhibit photoinduced CT,13–15 these materials are presumably not suitable for photovoltaic applications as one of the carriers is always confined to the particle core. Heteromaterial nanorods can be regarded as a logical extension of electron donor-acceptor substituted molecules. Scholes and co-workers * To whom correspondence should be addressed. E-mail:[email protected]
10.1021/jp804040r CCC: $40.75
Figure 1. Representative steady-state absorption and emission spectra of CdSe/CdTe hNR used in these experiments. The inset identifies the 3 absorption bands of interest corresponding to the lowest energy CdSe, CdTe and CT optical transitions (a-c respectively).
recently reported the single-pot synthesis of type-II CdSe/CdTe hNRs using classical colloidal chemistry and have identified characteristic optical absorption and emission bands for the CT state.16,17 By describing the photoluminescence decay dynamics with a stochastic kinetic model, they showed that trap-states play an important role in carrier recombination and concluded that electron transfer from the conduction band of CdTe to that of CdSe occurs on the 200 ps time scale.18 In this letter, we present experimental data that indicate that the electron transfer occurs on the time scale of several hundred femtoseconds. Adopting a similar method from the procedure in ref 16, we have grown CdTe on CdSe seed particles to give CdSe/CdTe hNRs. The materials reported here are on average 3.5 nm in diameter and 10-15 nm in length with CdSe seeds of the same diameter and 6-8 nm in length as characterized by TEM (see Supporting Information) and exhibit three important absorption bands. The steady state absorption and emission spectra of a representative batch of hNR in toluene are shown in Figure 1. The first two bands (a and b) correspond to the lowest energy 1S transitions of CdSe and CdTe (here referred to as CdSe1S, 570 nm and CdTe1S, 620 nm, respectively). The third feature (c) is a broad 2008 American Chemical Society
J. Phys. Chem. C, Vol. 112, No. 32, 2008 12075
Figure 2. Energy level diagrams of the two excited states that are populated by the pump pulse. The arrows illustrate the possible statefilling signals that can be observed as bleaching bands in the femtosecond broadband pump probe spectra.
peak at >670 nm corresponding to the CT band where the electron and hole are spatially separated. This band is not present in mixed solutions of CdTe and CdSe NRs. Comparison with the absorption spectrum of the seed material identifies the CdSe 1S transition in the hNR spectrum and suggests that the CdSe NRs do not grow during the CdTe deposition stage. Optical pumping at 620 nm leads to the formation of two distinctly different excited states, i.e., CdTe1S-CdSe (where one electron has been promoted from the valence band edge of CdTe to the conduction band edge of CdTe) and CdTe+-CdSe(where one electron has been promoted from the valence band of CdTe to the conduction band of CdSe). Figure 2 illustrates these states and identifies three optical transitions that are observable as spectral bleaching (after photoexcitation) due to the difference in electron occupation numbers in the respective conduction bands.19 These three state-filling signals are the CdTe/CdSe CT transition (∼680 nm), and the lowest energy transitions (1S) of CdTe (620 nm) and CdSe (550 nm). Hence, by monitoring the temporal evolution of the three bleaching signals one can probe the time-dependent populations of the (optically prepared) CdTe1S-CdSe and CdTe+-CdSe- states. To simplify the observed dynamics and to extract the contributions from electron transfer, a pump wavelength slightly to the red of the absorption maximum of CdSe has been chosen to effectively eliminate the population of the CdTe-CdSe1S state and thus to ensure that dynamic population of the CT state may solely originate from the CdTe1S-CdSe precursor state. In the present experiment, we split a 775 nm, 100 fs pulse from a regenerative amplified Ti:sapphire laser operating at 1 kHz to generate a white light continuum (350-750 nm) probe pulse and a home-built tunable (480-700 nm) pump pulse. The tunable pump pulse was generated using a dual noncollinear optical parametric amplifier (NOPA) as detailed earlier.20 The pump pulse was tuned to ∼620 nm, i.e., just to the red-side of the 1S transition of CdTe. In order to minimize the effects of multiexciton generation on cooling dynamics and avoid Augertype processes, a low-intensity pump pulse was used such that the average number of excitons per particle was less than 0.05. This value was calculated by Nex ) σj, where Nex is the average number of excitons per particle, j is the flux of the excitation pulse, and σ is the absorption cross-section at the excitation wavelength which can be calculated from the extinction coefficient. The extinction coefficient was determined based on the CdTe1S absorbance energy.21 Figure 3 displays the temporal evolution of the hNR pump probe spectrum which consists of three separate bleaching bands with maxima at 570, 620, and ∼680 nm. As pointed out above, these bands are the state-filling signals that represent the populations of the CdTe1S-CdSe and CdTe+-CdSe- states. As a control, we examined a mixture of CdTe and CdSe nanorods with similar dimensions (d ) 3.5 nm, l ) 5-10 nm) in a 1:1 mol ratio which yielded spectra dominated by bleaching of the
Figure 3. Pump probe spectra of (a) CdSe/CdTe hNR and (b) mixed CdSe and CdTe nanorods at selected time points following photoexcitation of CdTe. Excitation of CdTe in hNRs leads to a strong statefilling band from electron injection into the 1S(e) level of CdSe (570 and 680 nm). In mixed colloids of similar sized CdSe and CdTe, no state-filling from CdSe is observed.
Figure 4. Kinetics of CdSe, CdTe and CT state-filling signals during (a) the first 2 ps and (b) 100 ps following photoexcitation show the electron transfer from CdTe to CdSe occurs on the 500 fs time scale. Differences in decay kinetics on the 10-100 ps timescales originate from partial spectral overlap of the CdTe and CT spectral bands.
CdTe1S transition, however, without the CT band or the statefilling from CdSe1S (Figure 3b). Plotting the normalized change in absorption at the maxima of the CdSe1S, CdTe1S, and CT bands as a function of time reveals an ultrafast decay of the CdTe1S-CdSe state population (as identified by the CdTe1S band), accompanied by a corresponding increase of the CdSe1S and CT state-filling bands (which are assigned to the CdTe+-CdSe- state population, Figure 4). It is noteworthy that the CdSe1S and CT bleaching bands exhibit nearly identical buildup dynamics during the first 2 ps after excitation. Following a sub-100 fs buildup, a second rise with a time constant of ∼500 fs is observed. Similarly, a fit of the CdTe1S-CdSe population decay at 620 nm in Figure 3a reveals a time constant of ∼600 fs, a value that is within the error bar of the CT risetime. The similarity in the rise times of the charge separated CdTe+-CdSe- state on the one hand and the decay time of the electron donor CdTe1S-CdSe state on the other is consistent with the injection of an electron from the CdTe conduction band into the conduction band of CdSe.
12076 J. Phys. Chem. C, Vol. 112, No. 32, 2008 The 500 fs buildup only accounts for about 20% of the total signal amplitude indicating that most of the electron population in the CdSe conduction band level arrives on the time scale of the pump pulse. This quasi-instantaneous buildup can occur by either excitation of a CdSe valence band electron or via the direct optical electron transfer of an electron from the valence band of CdTe to the conduction band of CdSe without forming the CdTe1S-CdSe intermediate state. Since our pump wavelength (620 nm) does not overlap with the CdSe1S transition, one must conclude that direct optical CT takes place upon photoexcitation.22 Entirely consistent with this conclusion is the presence of a (weak) CT absorption band in the steady-state UV/vis spectrum (see Figure 1) as well as the state-filling bleaching band. Furthermore, since the dynamics of the CT and CdSe1S bands are coupled, the optical transition at >680 nm must involve at least one electron or hole from the CdSe particle and cannot be associated with an optically allowed transition to a shallow trap-state within the CdTe bandgap. If the broad, red-shifted emission observed stemmed from a trap-state then it would involve states confined within the CdTe band structure and not contribute to the CdSe signal. Following similar buildup dynamics, the CdSe1S band exhibits a 280 ps decay while the CT band decays with a 200 ps time constant (see Figure 4b). We note that if the dynamics of the CT band were monitored between 690 and 700 nm (instead of 685 nm) this discrepancy in the time constants was less pronounced. On the other hand, monitoring the absorption dynamics at wavelengths shorter than 685 nm resulted in a larger discrepancy in the two time constants. On the basis of this observation, we tentatively attribute the difference in dynamics to spectral inhomogeneities in the state-filling bleaching signal due to spectral overlap with the CdTe1S band. Thus, at wavelengths closer to the CdTe1S band, the spectrum is more affected by the decay of the CdTe1S population. Furthermore, as detailed previously,18 the various dynamical processes involved in carrier relaxation are sufficiently complicated that some deviations from our simple model are not surprising. With an injection time of 500 fs, the interfacial electron transfer event in CdSe/CdTe hNR occurs 3 orders of magnitude faster than previously assumed18 and on a time scale that is competitive to the ultrafast relaxation dynamics observed in quantum dots.9 The observed time scale is similar to that of electron-transfer in model dye-sensitized semiconductor systems,23 hole-transfer across the CdS/HgS interface, and about three times faster than electron transfer at the CdS/HgS interface.7 The similarity with the observed kinetics in the CdS/ HgS system can be explained qualitatively by considering the energy overlap of the respective conduction and valence bands as done in ref 7. At the CdS/HgS interface, both valence band states are composed of atomic orbitals from S2- giving them more closely matching carrier effective masses24 and, correspondingly, favorable condition for resonant CT. In the typeII CdSe/CdTe hNR band structure, both conduction band levels are composed of orbital contributions from Cd2+. By an extension of this analysis, we would expect the hole-transfer process from CdSe1s to CdTe to be slightly slower as a result of less favorable overlap between valence bands states of Te2and Se2-. This transfer event, however, cannot be isolated from state-filling bleaching signals alone as the hole-transfer dynamics, activated by pumping at or above the CdSe bandgap energy,
Letters would be mixed in with the electron transfer mechanism and a number of other cooling processes occurring on a similar time scale. In conclusion, we have examined the electron transfer dynamics in type-II CdSe/CdTe hNRs using ultrafast broadband pump-probe spectroscopy to simultaneously monitor and correlate the state-filling dynamics of the CdSe1S, CdTe1S, and CT bands on the subpicoseconds to nanoseconds time scale. By selective photoexcitation of CdTe, we have found that photoinduced electron transfer from CdTe to CdSe occurs with a time constant of 500 fs. We have argued that this ultrafast electron transfer, comparable to the hole transfer in CdS/HgS, is the result of favorable resonance conditions of the corresponding conduction band states. Acknowledgment. T.F. is grateful for financial support from the ACS Petroleum Research Fund. Supporting Information Available: This material is available free of charge via the Internet at http://pubs.acs.org. References and Notes (1) Gur, I.; Fromer, N. A.; Geier, M. L.; Alivisatos, A. P. Science 2005, 310, 462–465. (2) Huynh, W. U.; Dittmer, J. J.; Alivisatos, A. P. Science 2002, 295, 2425–2427. (3) Gunes, S.; Fritz, K. P.; Neugebauer, H.; Sariciftci, N. S.; Kumar, S.; Scholes, G. D. Sol. Energy Mater. Sol. Cells 2007, 91, 420–423. (4) Brus, L. E. J. Chem. Phys. 1983, 79, 5566–5571. (5) Peng, P.; Milliron, D.; Hughes, S.; Johnson, J.; Alivisatos, A.; Saykally, R. Nano Lett. 2005, 5, 1809–1813. (6) Chen, C.; Cheng, C.; Yu, J.; Pu, S.; Cheng, Y.; Chou, P.; Chou, Y.; Chiu, H. J. Phys. Chem. B 2004, 108, 10687–10691. (7) Braun, M.; Link, S.; Burda, C.; El-Sayed, M. Phys. ReV. B 2002, 66, 205312. (8) Ellingson, R.; Beard, M.; Johnson, J.; Yu, P.; Micic, O.; Nozik, A.; Shabaev, A.; Efros, A. Nano Lett. 2005, 5, 865–871. (9) Klimov, V. I.; McBranch, D. W. Phys. ReV. Lett. 1998, 80, 4028– 4031. (10) Yu, P.; Nedeljkovic, J.; Ahrenkiel, P.; Ellingson, R.; Nozik, A. Nano Lett. 2004, 4, 1089–1092. (11) Mohamed, M.; Burda, C.; El-Sayed, M. Nano Lett. 2001, 1, 589– 593. (12) Xu, S.; Mikhailovsky, A. A.; Hollingsworth, J. A.; Klimov, V. I. Phys. ReV. B 2002, 65, 045319. (13) Kim, S.; Fisher, B.; Eisler, H.; Bawendi, M. J. Am. Chem. Soc. 2003, 125, 11466–11467. (14) Chou, P.; Chen, C.; Cheng, C.; Pu, S.; Wu, K.; Cheng, Y.; Lai, C.; Chou, Y.; Chiu, H. ChemPhysChem. 2006, 7, 222–228. (15) Halpert, J.; Porter, V.; Zimmer, J.; Bawendi, M. J. Am. Chem. Soc. 2006, 128, 12590–12591. (16) Kumar, S.; Jones, M.; Lo, S.; Scholes, G. Small 2007, 3, 1633– 1639. (17) Scholes, G.; Jones, M.; Kumar, S. J. Phys. Chem. C 2007, 111, 13777–13785. (18) Jones, M.; Kumar, S.; Lo, S.; Scholes, G. J. Phys. Chem. C 2008, 112, 5423–5431. (19) Klimov, V. J. Phys. Chem. B 2000, 104, 6112–6123. (20) Raytchev, M.; Pandurski, E.; Buchvarov, I.; Modrakowski, C.; Fiebig, T. J. Phys. Chem. A 2003, 107, 4592–4600. (21) Yu, W.; Qu, L.; Guo, W.; Peng, X. Chem. Mater. 2003, 15, 2854– 2860. (22) Note that following the formation of the CT state, hole-transfer to the CdSe valence band would yield a CdSe1s state but this transition is endothermic in a type-II structure and thus energetically less favorable. While electron-hole coupled Auger-like relaxation dynamics might be able to stimulate such a transfer, the low pump energy and intensity used should exclude such complications due to multiexciton generation suggesting again the signal originates from an electron injection event. (23) Anderson, N. A.; Lian, T. Annu. ReV. Phys. Chem. 2005, 56, 491– 519. (24) Jasko´lski, W.; Bryant, G. W. Phys. ReV. B 1998, 57, R4237.