Hot-electron transfer in quantum-dot heterojunction films

Thermalization losses limit the photon-to-power conversion of solar cells at the high-energy side of the solar spectrum, as electrons quickly lose their energy relaxing to the band edge. Hot-electron transfer could reduce these losses. Here, we demonstrate fast and efficient hot-electron transfer between lead selenide and cadmium selenide quantum dots assembled in a quantum-dot heterojunction solid. In this system, the energy structure of the absorber material and of the electron extracting material can be easily tuned via a variation of quantum-dot size, allowing us to tailor the energetics of the transfer process for device applications. The efficiency of the transfer process increases with excitation energy as a result of the more favorable competition between hot-electron transfer and electron cooling. The experimental picture is supported by time-domain density functional theory calculations, showing that electron density is transferred from lead selenide to cadmium selenide quantum dots on the sub-picosecond timescale.


1.
P.2 line 50 has text discussing exciton cooling processes. The authors discuss the phonon bottleneck and experimental investigations therof. But this entire discussion is done at the level of 1998 investigations, largely by Klimov and Guyot-Sionnest. There has been much advance in our understanding based upon state-resolved exciton dynamics which gives a far more detailed understanding than those prior works. In particular, the phonon bottleneck and that discussion was merely based upon too simple a theory. A more comprehensive theory deals with several relaxation channels that are now precisely measured and corroborate more comprehensive theory: Phys. Rev. Lett., 98, 177403 (2007); Phys. Rev. B., 78, 245311 (2007); J. Phys. Chem. C, 115, 22809 (2011).

2.
The following paragraphs on p2 and p3 are well done, well conceived, and nicely illustrate the idea here. This is a great idea with great execution.

3.
The text on p3 line 115 continues the discussion of how to analyze the transient absorption (TA) signals. Again, this could be better done. There is Ground State Bleach, Stimulated Emission, Excited State Absorption. And in the case of photoinduced absorptions these arise from excitation in to biexcitons and not a trapped carrier Stark effect. The latter was merely old language which is not up to date: J. Phys. Chem. C, 115, 22809 (2011);J. Phys. Chem. Lett, 3, 1182(2012.

4.
The authors, like others cited here discuss the process of hot electron transfer. What is missing is a more detailed discussion in light of the very well established physical chemistry of electron transfer in molecular systems. I suggest the author familiarize themselves with some of the classic literature on this topic so as to elevate the discussion. Here is one of many excellent such articles: Contemporary Issues in Electron Transfer Research, J. Phys. Chem., 1996, 100 (31), pp 13148-13168.

5.
Once one is familiar with the salient processes of relaxation vs transfer, one can better discuss the phenomena. Nothing here is specific to nanocrystals, and even less so to the process of charge transfer from nanocrystals. These are well studied phenomena. But this specific work does a great job of showing a classical effect in a new system with a great motivation. With that, the general idea is the competition kinetics between hot exciton cooling and hot exciton transfer. The transfer of charge could be an electron or hole. But the ideas are general. Cooling takes place on a ~ 1 ps timescale so perhaps transfer takes place on a ~10 ps timescale to give some yield of 0.1. These are merely examples, but the idea remains. Indeed this idea has been explored in detail for hot exciton trapping to the surface of the NC. Surface trapping is not the same as charge transfer that the authors do here, but the ideas are the same. The following papers all give detailed discussion and measurement of hot exciton surface trapping: J. Chem. Phys., 129, 084701 (2008);Nano Lett. 10, 3062 (2010);J. Chem. Phys., 134, 094706 (2011). In addition to TA measurements of the very similar process, one can have steady state measurements of the quantum yield spectrum for PL: J. Phys. Chem. C, 118, 7730 (2014). The point is that there is a large body of work that is highly relevant and supports the results here. It is important to note that the idea of hot carrier electron transfer is not something new that was discovered in this report. But that this report shows this classical effect in a compelling way.

6.
The authors use fluence to report on excitation conditions. But it would be useful to estimate the mean exciton occupancy <N> in the standard methods.

7.
The data in Fig2e is spectacular. Very compelling and very rich. P5 line 180 discusses these data. The main features are the CdSe 1S bleaching buildup and recovery, and also the photoinduced absorptions. As noted above, the state-resolved exciton dynamics in CdSe model systems enable a much more detailed discussion of the signals and what processes give rise to them. For example, how fast is the buildup compared to CdSe 1S pump? It should be IRF limited there, but with a buildup time given by the hot electron transfer time. The depletion is the back electron transfer (note this mirrors the discussion of photoinduced ET in the review by Barbara et al). with b-ET, one has buildup of charges on PbSe and these give the photoinduced absorptions shown here. Not a Stark effect. Absorption into biexcitons. The detailed works on multiexcitons explain the signals in detail: J. Phys. Chem. C, 115, 22809 (2011);J. Phys. Chem. Lett, 3, 1182(2012J. Chem. Phys., 134, 094706 (2011). The worked up data in Fig3 here reproduces the earlier works.

8.
The data on effiency is very nice too. But I would like to see data with CdSe 1S pump. The idea is to see the buildup time. Is there some energy dependence to the buildup time?

9.
Can the authors suggest what goes into the efficiency? One might think the efficiency is just given by competition kinetics of cooling vs transfer. And both will have an energy dependence. But this is what happens for charge trapping at the surface. The transfer of charge to the other NC may be something different.
Reviewer #2 (Remarks to the Author): In this manuscript, the author observed bleaching of the 1S transition in CdSe QDs through transient absorption spectroscopy even with sub-bandgap excitation when assembled with PbSe QDs to form type-I band alignment. It was attributed to hot-electron transfer between the two QD components in the HJ film. The hot-electron transfer efficiency was quantified by integrating the differential absorbance at the 1S transition of CdSe QDs and shown to have excitation energy dependence. By fitting the change of transfer efficiency with time, the author was also able to extract the electron transfer rate, energy loss rate, etc. during the process. All the experimental observations were consistent with the result obtained from the DFT modeling.
I think this manuscript is clearly written with conclusions significant enough for publishing with minor revisions: 1.
How monodisperse are the CdSe and PbSe QDs? Based on the UV-Vis absorption spectra, the exciton peak of PbSe QDs is noticeably broader. How does the charge/energy transfer within different-sized PbSe/CdSe QDs compete with the hot-electron transfer across the two QD species?

2.
How many electrons are generated in PbSe QDs and transferred to CdSe QDs under the experimental fluences? The increase of electrons in CdSe QDs from hot-electron transfer can facilitate Auger recombination, which can be another cause of the fast decay of the bleach at CdSe 1S energy besides back-transfer.

3.
In the middle of page 5, the author tries to verify the back-transfer by exciting CdSe and counting bleach at PbSe 1S transition, which I doubt its sufficiency because it can be excitons and/or energy besides just electron transfer from excited CdSe to PbSe QDs.

4.
Line 271, the equation for bleach cross-section is miswritten.
Addressing the concerns of reviewer 1:

Reviewer comments: This paper reports on the migration of electrons from one quantum dot (QD) to another.
The specific focus is on the transfer of hot electrons, meaning electron transfer prior to exciton cooling. The process itself is of basic science interest. And the process may be useful in solar cell and related applications. In short, the topic and context is highly suitable for this Journal. The quality of the science is excellent. The data and the writing are great, and a pleasure to read. The control experiments are convincing. My only issue is that the basic spectroscopy and exciton dynamics deserve to be more rigorously discussed. Details below.
Reply: We thank the reviewer for this favorable evaluation. As discussed below we have used his suggestions to strengthen the discussion on the spectroscopy and exciton dynamic.  (2007); J. Phys. Chem. C, 115, 22809 (2011).
Reply: We fully acknowledge the complexity of the phonon bottleneck effect, and note that we have studied this in great detail recently for PbSe QDs in solution (ACS Nano 10, p695 and ACS Nano 11, p6286). We have expressed this more strongly in the revised introduction. We mentioned already in line 59 that experimental evidence on the presence of phonon bottlenecks remain scattered, and that high carrier relaxation rates are found for quantum dots. Following his remark, we expanded the comment on the phonon mentioning that to properly exploit this process one needs to take care of the possible surface related relaxation channels. We have included the first and third citation suggested by the reviewer in the manuscript.
2. The following paragraphs on p2 and p3 are well done, well conceived, and nicely illustrate the idea here. This is a great idea with great execution.
Reply: We thank the reviewer for this supportive comment!  C, 115, 22809 (2011);J. Phys. Chem. Lett, 3, 1182(2012. Reply: We acknowledge the reviewer's remarks on the nomenclature of the origin of the induced absorption features observed in TA, stressing the role of biexcitonic effects over a nomenclature suggesting a purely electrostatic picture. Thus we changed the reference to a "Coulomb shift" in the text, changing them into a "biexciton shift", where appropriate (e.g. line 149 and 220 of the revised manuscript), including the references suggested by the reviewer. At the same time we remark that spectral shifts are not always due to biexciton effects and some of the shifts we observe at longer delay times are induced by trapped charges. Here, we argue, the term Coulomb shift is more appropriate. Reply: We fully agree that HET is not specific to QDs and admit that this possibility was not mentioned explicitly in the manuscript. In the revised manuscrtipt we included a general statement about the occurrence of HET (see line 64 of revised manuscript), while expanding our literature citations on the reports of HET towards localized surface states, including the first and third reference suggested by the reviewer.

The authors use fluence to report on excitation conditions. But it would be useful to estimate the mean exciton occupancy <N> in the standard methods.
Reply: Following the reviewer's suggestion we included for each reported TA measurement the estimation of the average exciton population in the PbSe QD component upon photoexcitation.  detail: J. Phys. Chem. C, 115, 22809 (2011);J. Phys. Chem. Lett, 3, 1182(2012J. Chem. Phys., 134, 094706 (2011). The worked up data in Fig3 here reproduces the earlier works.

The depletion is the back electron transfer (note this mirrors the discussion of photoinduced ET in the review by Barbara et al). with b-ET, one has buildup of charges on PbSe and these give the photoinduced absorptions shown here. Not a Stark effect. Absorption into biexcitons. The detailed works on multiexcitons explain the signals in
Following the suggestion from the reviewer we added in the Supporting Information a section displaying rise times of the CdSe bleach signal for different excitation energies, both below and above the threshold for direct excitation of CdSe QDs. Our data show a strong reduction of the rise time for direct CdSe excitation compared to PbSe excitation followed by HET, indicating that, in contrast to the IRF limited rise of the bleach signal upon CdSe 1S pump, injecting carriers in CdSe via HET requires additional time. The rise time of the HET CdSe bleach signal is determined both by the transfer rate and by cooling within the manifold of CdSe states, the latter preventing facile extraction of transfer rates from the rise time. We can exclude that the photoinduced absorptions arising after 100 ps from photoexcitation at the bandgap energy of CdSe QDs arise from charges located at the conduction band edge of PbSe QDs. In fact the derivative-like feature appears in a time-window in which the PbSe band-edge bleach decays, showing little decay well after the PbSe QD 1S absorption is fully recovered. Thus we interpreted this behavior in terms of electrons trapped at the interface between PbSe and CdSe QDs, and not as biexciton generation. With the information provided by the measurement we cannot establish whether the electrons are trapped at the CdSe or PbSe QD surface.

The data on effiency is very nice too. But I would like to see data with CdSe 1S pump. The idea is to see the buildup time. Is there some energy dependence to the buildup time?
Reply: For direct CdSe 1S pump the signal generated due to direct excitation largely overpowers the signal due to HET from PbSe QDs. In this case the rise time of the signal is simply the IRF, as addressed in the reply to point 7. The immage shown below shows the version of Figure 4a and 4b including the data for direct CdSe excitation as well.
Efficiencies of initial electron injection in CdSe QDs are below unity even for direct CdSe excitation, due to the concurrent absorption from PbSe QDs. Reply: Indeed the efficiency trend can be explained in terms of a direct competition between electron cooling and electron transfer. Increasing the excitation energy increases the time-window for HET to take place, thus increasing the effective cooling time necessary for the electron to cross the threshold where transfer is no longer possible. At the same time higher electron energy leads both to a higher degree of electron delocalization, as indicated by the TDDFT results in the text, and to a higher number of final CdSe QD states, both effects increasing the rate of electron transfer. This picture is sufficient to justify the trend reported experimentally.

Can the authors suggest what goes into
Addressing the concerns of reviewer 2: Reply: The observed decay of the CdSe QD bleach is independent from photon fluence. The fact that the fast decay remains for excitation conditions generating less than 0.1 excitons per PbSe QD, combined with the ~5% yield of the transfer process, suggest that the situation involving two transferred electrons in a CdSe QD is negligible. Furthermore holes are energetically restricted from transferring to CdSe QDs by the valence band offset between the two materials, which should decrease significantly the impact of Auger recombination on the bleach induced by the transferred electrons.
3. In the middle of page 5, the author tries to verify the back-transfer by exciting CdSe and counting bleach at PbSe 1S transition, which I doubt its sufficiency because it can be excitons and/or energy besides just electron transfer from excited CdSe to PbSe QDs.
We thank the reviewer for this comment. In quantum dot samples typical FRET lifetime span the 100ps to 1ns range (see Appl. Phys. Lett. 84, 2904(2004; doi: 10.1063/1.1702136). Transfer of electrons out of the CdSe QD conduction band edge happens on a ~1ps to 10ps timescale, suggesting the behavior is associated with charge transfer rather than excitonic transfer. Furthermore, the constant bleach for absorbed photon across the whole excitation range, showed in Figure S7, suggests that all of the photoexcited electrons and holes reach PbSe quantum dots within the cooling time.