Two-step photon up-conversion solar cells

Reducing the transmission loss for below-gap photons is a straightforward way to break the limit of the energy-conversion efficiency of solar cells (SCs). The up-conversion of below-gap photons is very promising for generating additional photocurrent. Here we propose a two-step photon up-conversion SC with a hetero-interface comprising different bandgaps of Al0.3Ga0.7As and GaAs. The below-gap photons for Al0.3Ga0.7As excite GaAs and generate electrons at the hetero-interface. The accumulated electrons at the hetero-interface are pumped upwards into the Al0.3Ga0.7As barrier by below-gap photons for GaAs. Efficient two-step photon up-conversion is achieved by introducing InAs quantum dots at the hetero-interface. We observe not only a dramatic increase in the additional photocurrent, which exceeds the reported values by approximately two orders of magnitude, but also an increase in the photovoltage. These results suggest that the two-step photon up-conversion SC has a high potential for implementation in the next-generation high-efficiency SCs.

4. Related to comment 2, a proof of concept of the proposed device would be to measure Voc higher than the bandgap of the low-bandgap material (GaAs,1.4 eV). In this sense, the measured 0.7 V (Figure 4a), with an increase of 1 mV due to the second photon flux, is not sufficient. In addition, it would be unexpected that the measured Voc did not increase following an increase in Jsc.
5. It is said that 1300 nm photons (0.95 eV) could not trigger interband transitions in the InAs quantum dots, yet there is no experimental support for such an important statement (for example, some kind of spectroscopy of the quantum dots).
6. Concerning the photocurrent measurements shown in Figure 2. Samples were illuminated first with chopped monochromatic light, secondly adding a continuous-wave 1300 nm light source. The power density of the continuous-wave light source was equivalent to almost 4 suns and samples were not cooled during the measurements. These conditions cannot exclude thermal effects altering the results.
In conclusion, the submitted manuscript fails to meet some of the publication criteria of Nature Communications, in particular the following two: -"The data is technically sound" -"The paper provides strong evidence for its conclusions".
For all these reasons I suggest it is not accepted for publication in Nature Communications.
We deeply appreciate valuable comments from the reviewers. All the issues pointed out by the reviewers have been addressed as follows one-by-one in detail. According to the reviewer's comments, we updated discussions, experimental results, and figures.

RESPONSE:
'Transmission loss' is also called 'transparency loss' or 'below E g loss', which is distinguished from 'recombination loss' [Ref. 2]. Transmission loss is caused by below-gap photons passing through a solar cell (SC). Below-band gap photons with energy smaller than the band gap of SC are not absorbed and do not contribute to create carriers. On the other hand, recombination loss is a loss caused by recombination of photo-created carriers, where photons absorbed in SC play an important role. We updated texts in page 2, lines 7-8.
(#1-2) COMMENT: lines 16-17 ---dramatic increase in the photocurrent, which exceeded the reported value by approximately two orders of magnitude, but also an increase in the photovoltage. 'Did the photocurrent or the change in the photocurrent exceed the reported value?'

RESPONSE:
The additional photocurrent (the change in the photocurrent) caused by TPU exceeded the reported value by approximately two orders of magnitude. We updated texts in page 1, line 16. What is 'detailed valance framework'. Give more details of the model, its assumptions and approximations. I suggest proofreading it.

RESPONSE:
We owe the reviewer our thanks. The spell of "valance" was incorrect. It is, of course, "balance". The reviewer's suggestion may be on this fault. Our calculation in Fig. 5 is based on a framework called "detailed balance" which was originally proposed by Profs. W. Shockley and H. J. Queisser in Ref. 1. As described in Ref. 1, the detailed balance framework is a framework considering a steady state between carrier generation and recombination at the optimum operation point of SC. This model has been widely used to calculate an ideal, maximum conversion efficiency of SC. Here, we ignore nonradiative processes in SC for predicting the ideal limit of the conversion efficiency. The total photon emission flux, N, with the energy range between and is calculated by using the generalised Planck equation incorporating the effect of chemical potential, : where is temperature, is the Planck's constant, is the light velocity, k b is the Boltzmann constant. By using Eq. (1), generation rates of in wide-gap semiconductor (WGS), in narrow-gap semiconductor (NGS), and for TPU can be expressed by (see Fig. R1): and where is the solar concentration factor, 2.16 10 is the solid angle of the sun, 6,000 K is the temperature of sun, and are bandgap energies of WGS and NGS, respectively, is the conduction band offset between WGS and NGS, 300 is the temperature of the SC. Relation of , , , and valence band offset, , is given by: Each recombination rate is given by: and where and are the quasi-Fermi level separation in WGS and NGS, respectively and is the quasi-Fermi level separation due to TPU (see Fig. R1). Here, we take into , , , , ∞, , , , , , , , , , , account electrons accumulated at the hetero-interface and TPU occurring in the conduction band.
Similar thing happens when holes are accumulated at the hetero-interface and TPU occurs in the valence band. In this case, in Eqs. (7) and (8) is replaced by . According to these relations, the total current, J, generated in TPU-SC is obtained by: where q is electronic charge. In TPU-SC, the following current matching condition of TPU must be satisfied: The output voltage of TPU-SC is given by Finally, the total electrical power generated in TPU-SC is calculated as a product of VJ and, hence, the expected conversion efficiency can be estimated by VJ divided by the total incident photon energy. Detailed theoretical model and calculated results for the conversion efficiency of TPU-SC will be presented elsewhere.
In this paper, we added several texts describing the fundamental framework of the detailed  μ WGS and μ WGS are the quasi-Fermi level separation in WGS and NGS, respectively and μ up is the quasi-Fermi level separation due to TPU. ΔE c and ΔE v are the CB and VB discontinuity, respectively. G WGS and G NGS carrier-generation rates in WGS and NGS, respectively, and G UP is the carrier-generation rate due to up-conversion. R WGS , R NGS and R UP is the carrier-recombination rates in WGS, NGS, and at the hetero-interface, respectively. , . (#2-1) COMMENT: Even though the obtained enhancement in external quantum efficiency of the investigated device is undoubtedly large, I think that a more detailed discussion on this effect is missing. To be clearer, the improvement is claimed in the abstract and in the manuscript as the best "ever reported" but for supporting this claim, only reference 16, from the same authors and using a similar dot in well structure, is cited. For a journal like Nature Communications I would expect a more extensive discussion on such a claim, with respect to a larger suite of literature results.
Relevant and more updated works proposing different approaches and analysing the competition between equilibrium and non-equilibrium charge-transfer processes must be cited in this respect:

RESPONSE:
We agree with your opinion, and thank you very much for presenting the papers. In the revised manuscript, we added the following references in page 2, line 18 to page 3, line 1.  (#2-2) COMMENT: In the introduction, it is stated that the "key factor" at the base of the increase in TPU efficiency is "the spatial potential fluctuations at the heterointerface". It seems to me that this effect is inferred to the specific potential barrier, not to any fluctuations. This point should be clarified.

RESPONSE:
Electrons excited in GaAs are accumulated at the Al 0.3 Ga 0.7 As / GaAs hetero-interface and up-converted into Al 0.3 Ga 0.7 As by the second photoexcitation and the internal electric field. TPU itself occurs at the hetero-interface of Al 0.3 Ga 0.7 As / GaAs without InAs QDs. In that sense, the TPU effect might be due to the specific potential barrier. It is noted that the interface containing InAs QDs improves the TPU efficiency as shown in Figs All along the text, electrons are said to be accumulated at the hetero-interface AlGaAs/GaAs/InAs/GaAs. Which is actually the working interface? Data about energy barrier height and thermal escape evaluation should be provided.

RESPONSE:
This strongly relates to the last comment #2-2. The interface structure used is complicate. As we mentioned above, TPU itself occurs at the hetero-interface of Al 0.3 Ga 0.7 As / GaAs. InAs QDs play a role enhancing the TPU efficiency. According to these results, we believe that the working interface is Al 0.3 Ga 0.7 As / GaAs.
Based on a well-known band discontinuity of the Al 0.3 Ga 0.7 As / GaAs hetero-interface, the conduction band offset can be estimated to be 220 meV which corresponds to the barrier height for electron [Ref. 30]. Excited electrons in GaAs can be accumulated at the Al 0.3 Ga 0.7 As / GaAs hetero-interface with the large potential barriers, though a small number of electrons are thermally pumped out, which reduces the output voltage. Here, we carefully measured the temperature dependence of the current-voltage characteristics for TPU-SC with InAs QDs. Figure S1 shows the temperature dependence of the current density when irradiated by the 780-nm LD. The current density increases with increasing the temperature. The inset of Fig. S1 indicates the applied bias voltage dependence of the estimated thermal activation energy E A . E A monotonically decreases with increasing the electric field because of lowering the effective barrier height at the hetero-interface. E A shows the maximum of 221 ± 3 meV at 0.02 V as shown in Fig. S1. Conversely, applying higher positive bias voltage weakens the internal electric field significantly and makes flatter the band. As the forward current increases even at the same bias condition with increasing the temperature, the detected photocurrent decreases rapidly with flatten the band. Thereby, E A decreases and finally becomes negative with increasing the bias voltage. The maximum E A excellently coincides with the estimated conduction-band discontinuity between Al 0.3 Ga 0.7 As and GaAs [Ref. 32].
We added texts describing the thermal escape property in page 6, line 19 to page 7, lines 2. In addition, we added Section 1 and Fig. S1 in Supplementary Information. (#2-4) COMMENT: Another important issue is the comparison with a reference cell without InAs quantum dots which is only mentioned, when discussing EQE variation with IR light (figure 2). All the spectra shown in the manuscript must also show the behaviour of this reference cell for direct  Fig. 2d,EQE is obviously generated even in TPU-SC with the hetero-interface of Al 0.3 Ga 0.7 As / GaAs without InAs QDs. Therefore, the working interface is Al 0.3 Ga 0.7 As / GaAs. Comparison between the EQE spectra suggests that the hetero-interface containing InAs QDs improves the TPU efficiency. As we mentioned for the comment #2-2, three-dimensionally confined QD relaxes the optical selection rule, and, therefore, electrons at the Al 0.3 Ga 0.7 As / GaAs hetero-interface are easily pumped into the conduction band of Al 0.3 Ga 0.7 As by the excitation light irradiating the two-dimensional plane containing QDs perpendicularly. We updated Fig. 2 and relevant texts in page 8, lines 6 to page 9 lines 2.

Wavelength (nm)
the EQE variation with the IR illumination in the wavelength below the band gap of Al 0.3 Ga 0.7 As was obvious for all the devices we tested, our discussion regarding TPU is reliable.
We updated texts in page 7, lines 17 to page 8, lines 3.

Figure S3 | EQE and ΔEQE for a different TPU-SC with InAs QD.
(#2-6) COMMENT: I have some concerns about the theoretical prediction of the conversion efficiency for the proposed cell. The band profiling is extremely simplified with a two level system, E(AlGaAs) and E2. The introduction of InAs QDs actually complicates this profile, by adding QD confined states as well as the WL. These features are neglected by the authors.

RESPONSE:
As we mentioned above, TPU itself occurs at the Al 0.3 Ga 0.7 As / GaAs hetero-interface, and InAs QDs play a role enhancing the TPU efficiency. In our theoretical estimation of the conversion efficiency, we neglected QD states enhancing the TPU efficiency because we simply assumed a perfect TPU at the hetero-interface. Our theoretical prediction of the conversion efficiency for the proposed TPU-SC is based on several ideal assumptions such as complete optical absorption, TPU, and carrier collection efficiency. Besides, we ignored any nonradiative process. We added texts in page 15, lines 5-8. Finally, authors declare that the device used in the study has the purpose only to demonstrate this improvement in TPU, so it is not the optimal device which, as they claim, should exhibit thickness optimization, doping, window layer and A/R coating. However, what are the performances of this "un-optimized" device (FF, efficiency, Voc and Jsc) under standard AM1.5 illumination?

RESPONSE:
The device used in this study was designed and fabricated to demonstrate the TPU phenomena.  is not discussed. This discussion is crucial for understanding the high Voc of the device (higher than the bandgap of the low-bandgap material).

RESPONSE:
We added an illustration of TPU-SC at the operating condition in Fig. 1b  3. The drift-driven accumulation of electrons at the interface is presented as the enabling factor for long-lived electrons, therefore enhancing two-step photon absorption. At high voltages, close to Voc, the electric field is strongly reduced (flatter bands); therefore, the cause for strong two-step absorption is reduced. This issue is not discussed. At high voltages, close to V oc , the electric field, yes, is reduced, and the band diagram becomes flatter. Therefore, the cause for TPU reduces. The electric field at the operating point exhibiting the maximum output power is also not so strong. We need to perform detailed simulations of the band profile at the operating point in order to maintain a moderate internal electric field even at the operating point by controlling the doping profile near the hetero-interface. We added texts discussing this point in page 18, lines 7-13. Experimental: (#3-4) COMMENT: 4. Related to comment 2, a proof of concept of the proposed device would be to measure Voc higher than the bandgap of the low-bandgap material (GaAs, 1.4 eV). In this sense, the measured 0.7 V (Figure 4a), with an increase of 1 mV due to the second photon flux, is not sufficient. In addition, it would be unexpected that the measured Voc did not increase following an increase in Jsc.

RESPONSE:
We agree with your opinion. The increase of 1 mV is not sufficient as compared with the band gap difference between Al 0.3 Ga 0.7 As and GaAs. As given in Eq. (4), V oc is an increase against V oc,single which is the open-circuit voltage measured at the single-color excitation without the 1,300-nm LD illumination. The increase in V oc includes effect of the voltage boost effect at the hetero-interface, which follows an increase in the extra photocurrent J sc created by the additional 1,300-nm LD illumination. It is difficult to extract the contribution of the voltage boost effect at the hetero-interface from fitting the curve of V oc in Fig. 4b. However, we have clearly demonstrated a difference between the TPU effect caused by the optical process and the thermal population effect. To confirm the contribution of 1,300-nm LD illumination to V oc , we carefully measured V oc as a function of J sc controlled by the 1,300-nm LD illumination or temperature. The results are summarized in Fig. 4d. The blue circles indicate V oc recorded by changing J sc controlled by temperature. Here, the 1,300-nm LD does not shine the device. With increasing the temperature, J sc increases because of increasing thermal carrier population, and, resultantly, V oc reduces. This is a well-known phenomenon. As the band gap change in this temperature variation is approximately 4.5 meV which is given by 5 × 10 -4 eV/K of the temperature dependence of the band gap of GaAs, the observed change in V oc was almost caused by the thermal carrier population effect. Conversely, when the 1300-nm LD with the excitation power density of 300 mW/cm 2 illuminates the device at 290 K, V oc slightly increases, despite increasing J sc similarly. The clear difference between the thermal effect and the TPU by the second photon flux proves the concept of the proposed TPU-SC. We added several texts discussing the voltage boost in page 12, lines 16 to page 13, lines 13 and the following figure as Fig. 4d.

RESPONSE:
We added a photoluminescence (PL) spectrum measured at 300 K. The wavelength and power density of the excitation laser were 660 nm and 80 mW/cm 2 , respectively. The PL peak appeared at 1,180 nm corresponds to the fundamental state of the QD transition, which is shorter than that of 1,300 nm of the second-excitation laser. Thus, 1,300-nm photons (0.95 eV) could not trigger the interband transition in InAs QDs. We added the PL result in Supplementary Section 2.

RESPONSE:
We completely agree with you. We need to consider contribution of both TPU and thermal population to the device operation. The maximum power density of the 1,300-nm LD used in this study was equivalent to almost 17 suns. Here, we took into account Air Mass 1.5G solar spectrum. TPU requires photon absorption in the wavelength region between 1,180 nm (the fundamental state of InAs QD) and 5,640 nm (CB discontinuity). Thereby, the estimated photon flux consumed by TPU is approximately 1.4 × 10 17 photons/cm 2 at 1 sun. As the 1,300-nm LD with the power density of 360 mW/cm 2 corresponds to 2.4 × 10 18 photons/cm 2 , the equivalent solar concentration in this study becomes 17 suns. As we mentioned for the comment #3-4, we successfully distinguished the difference between the thermal effect and the TPU caused by the second photon flux. We updated the relevant discussion in page 12, lines 16 to page 13, lines 13, and page 18, line 1. Dear Authors, Your detailed responses to the the reviewers' comments are highly appreciated. The clarity of the resubmitted manuscript has been greatly improved, and several essential references have been included and discussed. Therefore, I recommend that this revised version of the article is accepted for publication in Nature Communications.
Reviewer #2 (Remarks to the Author): I appreciate the significant effort faced by the authors to improve the manuscript quality. However, there are still some points which lead me to evaluate the paper as not technologically sound and strongly supported by data as required for Nature Communications. First of all, I think that a clear band profile with all the energy levels available within the structure must be provided. This would require a more careful spectroscopic data analysis with respect to carrier dynamics. For example EQE or PL as a function of temperature should be analysed by Arrhenius plots to clearly assess inter level processes within the cell structure. Only after a clear picture depicted on available energy states for confined carriers authors could discuss about the AlGaAs/GaAs interface role. Based on the structure scheme in figure 6 (which, in any case, should be presented earlier in the text)QDs are embedded within GaAs and this is ignored in the discussion. Moreover, I do not understand the difference at high energy between the EQE curve of the proposed cell and the reference one(without InAs quantum dots). It seems that in the short wavelength region without IR additional illumination, these cells behave differently. Considering the revisions presented by the authors and involving fundamental understanding of the structure effect on device operation, I am sorry I cannot support publication on Nature Communications.

Reviewer #3 (Remarks to the Author):
In this new version the authors have tackled satisfactorily most of the points I raised on the original manuscript.
The increase in voltage is still low taking into account the high IR illumination density (17 suns), but it is valid as proof of the potential of the proposed device now that thermal effects have been discarded.
I do not have further comments on the manuscript and, therefore, my opinion is that it is ready for publication.
I encourage the authors to investigate carefully the impact of positive bias on the TPU effect.
Thank you very much for reviewing our manuscript. We deeply appreciate your valuable, constructive comments and warm encouragement. All the issues pointed out by the reviewers have been addressed as follows one-by-one in detail. In particular, according to the comments of reviewer #2, we conducted PL and EQE experiments pointed out by reviewer #2. All the updated outcomes strongly support the TPU effects at the hetero-interface.
Reviewer #1: COMMENT: Dear Authors, Your detailed responses to the reviewers' comments are highly appreciated. The clarity of the resubmitted manuscript has been greatly improved, and several essential references have been included and discussed. Therefore, I recommend that this revised version of the article is accepted for publication in Nature Communications.

RESPONSE:
We deeply appreciate your careful review and are very happy that you find the potential interest of our manuscript.
Reviewer #2: (#2-1) COMMENT: I appreciate the significant effort faced by the authors to improve the manuscript quality.
However, there are still some points which lead me to evaluate the paper as not technologically sound and strongly supported by data as required for Nature Communications.

RESPONSE:
We deeply appreciate your constructive comments. We conducted PL and EQE experiments pointed out. We added new spectroscopic data and analysis with respect to carrier dynamics, in particular, carrier excitation process, as follows.
(#2-2) COMMENT: First of all, I think that a clear band profile with all the energy levels available within the structure must be provided. This would require a more careful spectroscopic data analysis with respect to carrier dynamics. For example EQE or PL as a function of temperature should be analysed by Arrhenius plots to clearly assess inter level processes within the cell structure. Only after a clear picture depicted on available energy states for confined carriers authors could discuss about the AlGaAs/GaAs interface role. Based on the structure scheme in figure 6 (which, in any case, should be presented earlier in the text) QDs are embedded within GaAs and this is ignored in the discussion.

RESPONSE:
Thank you very much for your constructive and useful comments regarding the band profile influencing on carrier dynamics at the hetero-interface. We conducted PL and EQE measurements as a function of temperature.
Figures S1a and S1b show PL spectra for TPU-SC with InAs QDs measured at various temperatures and the temperature dependence of the integrated PL intensity, respectively. The  The temperature dependence of the PL intensity reflects the change in the recombining carrier density in QDs. That change is caused by the thermal carrier escape from the confined state. The thermally populated carriers will increase the current. We carefully measured the temperature dependence of EQE. Figure S2 shows the EQE spectra for TPU-SC with InAs QDs measured at various temperatures and the temperature dependences of the current density obtained at typical excitation wavelengths.  fitting. E A is the estimated thermal activation energy. Figure Figure S2c shows the results. We recorded the current at the bias of 0.02 V. Here, the excitation light was produced by a supercontinuum laser, passed through a 270-mm single monochromator. The monochromatic excitation-laser line width was 9.6 nm. The EQE line width of the wetting layer state is approximately 15 nm and the temperature drift of the wetting layer state is approximately 2.9 nm, so that we fixed the excitation wavelength in this experiment. The evaluated thermal activation energy was 254 ± 5 meV.
These experimental results of PL and EQE as a function of temperature clarify thermal carrier population processes occurring at the hetero-interface. Photo-excited electrons are thermally populated from the GaAs edge to the Al 0.3 Ga 0.7 As one, from the InAs wetting layer state to the Al 0.3 Ga 0.7 As edge, and from the QD GS to the GaAs edge. We did not confirm an obvious change caused by thermal population of holes, suggesting photo-excited holes reach the p-GaAs contact without captured at the hetero-interface. The following Fig. S3 summarizes these results we obtained. This clear picture clarifies available energy states for confined carriers at the hetero-interface of TPU-SC with InAs QDs. As shown in Fig. 2, ΔEQE at the InAs QD GS of 1,186 nm is very weak, suggesting that optical absorption in the single InAs QD layer with the in-plane QD density of ~1.0 × 10 10 cm -3 is not enough to contribute to the change in the current generation at the QD GS, whereas QDs obviously improves the TPU efficiency as shown in Fig. 2b. It is noted that ΔEQE is generated even in TPU-SC with the hetero-interface of Al 0.3 Ga 0.7 As / GaAs without InAs QDs. Thus, electrons at the hetero-interface obey the selection rule modified by QDs and are efficiently pumped into the conduction band of Al 0.3 Ga 0.7 As by the 1,300-nm LD illumination.
In our theoretical estimation of the conversion efficiency, we simply assumed a perfect TPU at the hetero-interface in our model.
We combined Supplementary Sections 1 and 2 for barrier height estimation at the hetero-interface of TPU-SC with InAs QDs. We included all the new experimental outcomes in the updated Supplementary Section 1 and updated discussion in page 7, lines 8-14, page 8, lines 5-8, page 8, line 17 to page 9, line 8, and the figure caption of Fig. 1. Furthermore, as you pointed out, we added sentences briefly describing the sample structure earlier in the text in page 4, lines 5-7.
(#2-3) COMMENT: Moreover, I do not understand the difference at high energy between the EQE curve of the proposed cell and the reference one (without InAs quantum dots). It seems that in the short wavelength region without IR additional illumination, these cells behave differently.

RESPONSE:
As we explained in the last response regarding this comment, we confirmed that cell-to-cell variability in the trend of ΔEQE for decreasing wavelength (lower than 680 nm) exists in our  Reviewer #3: COMMENT: In this new version the authors have tackled satisfactorily most of the points I raised on the original manuscript.
The increase in voltage is still low taking into account the high IR illumination density (17 suns), but it is valid as proof of the potential of the proposed device now that thermal effects have been discarded.
I do not have further comments on the manuscript and, therefore, my opinion is that it is ready for publication.
I encourage the authors to investigate carefully the impact of positive bias on the TPU effect.

Wavelength (nm)
Reviewers' comments: Reviewer #2 (Remarks to the Author): The authors have made an heavy job in improving data analysis. There are still some overselling points, especially the claims in the abstract related to the increase in the quasi-Fermi gap and the generation of the substantial additional photocurrent in the TPU-SC, resulting in a high conversion efficiency for intermediate-band SCs. Such an increment is actually not reported experimentally but only proposed as a potential effect of a further optimization of this design. I think this is the main limitation for this manuscript to be published in such a relevant journal as Nature Communications. Therefore, a re-focusing of the abstract in this direction is at least required. A minor concern deals with the usage of sentences such as in line 111 " electrons are thermally populated", which has been used also elsewhere in the text and which must be corrected.
Thank you very much indeed for reviewing our manuscript. We deeply appreciate your valuable, constructive comments. All the issues pointed out by the reviewers have been addressed as follows one-by-one in detail.
Reviewer #2: COMMENT: There are still some overselling points, especially the claims in the abstract related to the increase in the quasi-Fermi gap and the generation of the substantial additional photocurrent in the TPU-SC, resulting in a high conversion efficiency for intermediate-band SCs. Such an increment is actually not reported experimentally but only proposed as a potential effect of a further optimization of this design. I think this is the main limitation for this manuscript to be published in such a relevant journal as Nature Communications. Therefore, a re-focusing of the abstract in this direction is at least required.

RESPONSE:
We agree with your opinion. We re-focused the abstract. In this study, we observed a dramatic increase in the additional photocurrent that we have never seen before, which exceeds the reported values by approximately two orders of magnitude. Conversely, for the quasi-Fermi gap we observed, the increase of 1 mV is not sufficient as compared with the band gap difference between Al 0.3 Ga 0.7 As and GaAs. As our paper do not report an increment in the conversion efficiency, the last sentence of the abstract is not appropriate. We updated the last part of the abstract according to the Discussion section. We believe that the generation of the substantial additional photocurrent in the TPU-SC is promising for a potential effect of the TPU-SC.

COMMENT:
A minor concern deals with the usage of sentences such as in line 111 " electrons are thermally populated", which has been used also elsewhere in the text and which must be corrected.

RESPONSE:
Thank you very much for your careful review. We corrected relevant sentences to "electrons are thermally excited". Thank you very much again.

1
We deeply appreciate your careful review and warm encouragement. We would like to tackle investigation of impact of positive bias on the TPU effect in our future work and open a new field of SC based on the TPU effect.