Cite this Article
Kabir I, Mahmood S A. Analysis of highly efficient perovskite solar cells with inorganic hole transport material. Chinese Physics B, 2019, 28(12): 128801  
Permissions
Analysis of highly efficient perovskite solar cells with inorganic hole transport material
Kabir I, Mahmood S A †
Department of Electrical and Electronics Engineering, Bangladesh University of Engineering and Technology, Dhaka, Bangladesh

 

† Corresponding author. E-mail: asifmahmood@eee.buet.ac.bd

Abstract

Organo-halide perovskites in planar heterojunction architecture have shown considerable promise as efficient light harvesters in solar cells. We carry out a numerical modeling of a planar lead based perovskite solar cell (PSC) with Cu2ZnSnS4 (CZTS) as the hole transporting material (HTM) using the one-dimensional solar cell capacitance simulator (SCAPS-1D). The effects of numerous parameters such as defect density, thickness, and doping density of the absorber layer on the device performance are investigated. The doping densities and electron affinities of the electron transporting material (ETM) and the HTM are also varied to optimize the PSC performance. It has been observed that a thinner absorber layer of ∼220 nm with a defect density of 1014 cm−3 compared to the reference structure improves the device performance. When doping density of the absorber layer increases beyond 2 × 1016 cm−3, the power conversion efficiency (PCE) reduces due to enhanced recombination rate. The defect density at the absorber/ETM interface reduces the PCE as well. Considering a series resistance of 5 Ω · cm2 and all the optimum parameters of absorber, ETM and HTM layers simultaneously, the overall PCE of the device increases significantly. In comparison with the reference structure, the PCE of the optimized device has been increased from 12.76% to 22.7%, and hence the optimized CZTS based PSC is highly efficient.

PACS: 88.40.H-;88.40.hj;88.40.fc
Keyword:CH3NH3PbI3;Cu2ZnSnS4 (CZTS);SCAPS-1D;absorption coefficient
1. Introduction

The rate of energy consumption increases to cope with economic development. However the source of nonrenewable energy decreases gradually. Among other renewable energy sources, solar cells have drawn a great deal of attention from researchers. Over the past decade, many studies have been carried out on organo-halide perovskite solar cells (PSCs).[1–4] In comparison with traditional solar cells, they have simpler processing techniques, higher power conversion efficiency (PCE), and lower cost.[5] In 2009, Miyasaka et al. first introduced the PSCs using MAPbI3 (MA = CH3NH3) and MAPbBr3 as light harvesters although their PCEs are low (3%–4%).[6] Recently, the PCEs of lead based PSCs have reached ∼22.1%.[1,7] The PCEs have reached above 25% when PSC is used in silicon based tandem solar cells.[8] The absorption coefficient of MAPbX3 is very high (103–104 cm−1) along with high charge carrier mobility (2–66 cm2 · V−1 · s−1).[7] The tunable band gap (1.55–2.50 eV) property of this material can be utilized to make colorful solar designs.[9] In addition, MAPbX3 based thin films have charge carrier life time of ∼270 ns and carrier diffusion length of 100–1900 nm.[7]

The planar PSC structure, where a thin perovskite layer is sandwiched between layers of a hole transporting material (HTM) and an electron-transporting material (ETM), is preferred due to simple fabrication technique.[10] According to the order of preparation, the planar PSC structure can be either n–i–p or p–i–n. Organic compounds, such as spiro-MeOTAD, PEDOT:PSS, PTAA and P3HT are the commonly used HTMs in the case of high-efficient PSCs.[2,11–13] However, most of the organic based HTMs are unstable and costly. The reduction in PCE of spiro-MeOTAD based PSC is more than 50% after 15 days under moisture and heat.[14] The complex and time consuming synthetic process make the spiro-MeOTAD based PSCs costly as well.[2] PEDOT:PSS is chemically unstable due to its acidic and hygroscopic nature.[15] The contraction of PEDOT conductive grains leads to the thermal instability of PEDOT:PSS films.[15] The PCE of PEDOT:PSS based PSC reduces by 40% within 10 days.[16] The highest PCE achieved by PTAA based PSC is 21.2% and the PCE degradation is ∼5% in 20 days.[11,17] The PCE of pristine P3HT based PSC is only 14.3% and degraded by 20% within 29 days.[18]

Inorganic HTMs are preferable as they are cheaper, more stable, and simple to synthesize. Recently, Cu based HTMs such as CuI, CuSCN, Cu2O, CuInGaSSe, and Cu2ZnSnS4 (CZTS) draw attention of researchers.[19–23] The reported PCE of CuI based PSC is 17.6% and the PCE decreases by 8% within 90 days.[22] Although its stability is comparable to PTAA based PSC, the PCE is lower than the organic counterpart. Its mesoporous structure increases the process complexity and hence industrially less preferable.[24] The PCE of CuSCN based inverted planar PSC is 15.6% and degraded by 20% within 2 days.[19] A PCE of 13.35% is achieved by Cu2O based inverted planar PSC and sustained above 90% of initial value for 70 days.[20] Another mesostructured PSC with CuInGaSSe as HTM achieves a PCE of 9.15% and sustained 90% of initial value for 5 days. Kesterite materials, e.g., CZTS, are an efficient light harvester for thin film solar cells due to their high absorption coefficient (∼104 cm−1), optimal band gap (1.4–1.5 eV) for solar energy conversion, high hole mobility (6–35 cm2·V−1 · s−1), and earth abundant constituents.[25] However, the highest PCE of CZTSSe based thin film solar cell is 12.6%.[21] Wu et al. have shown that Cu2ZnSnS4 (CZTS) can be used as a nontoxic and low cost HTM in planar lead based PSC.[25] It increases the light absorbance of the device and hence increases the short circuit current density of the PSC. Surface recombination can also be reduced by using CZTS layer as an HTM in PSCs.[25] Zuo et al. have employed CZTS nanoparticle as an HTM in lead based planar PSC and achieved a PCE of 13.41%.[26] They have also demonstrated an improved crystallinity of the perovskite film. In another work it has been exhibited that CZTS based PSC sustains more than 98% of its initial PCE for 3 days.[21] For ETM, TiO2 is the most widely used materials both in planar and mesostructured PSC.[27] Currently the PCE of planar PSCs is lower than the mesostructured devices.[28] Apart from experimental works, a theoretical investigation is necessary to improve the performance of the planar PSC. Many researchers have used one dimensional solar cell capacitance simulator (SCAPS-1D) to simulate the PSC and other thin film solar cells.[29]

Recently, Kabir et al. have analyzed the MAPbI3 based PSC with CZTS as the HTM, and optimized a few parameters such as absorber layer thickness and doping density of different layers to improve the PCE of the device.[30] In this paper, the defect density, doping concentration, and thickness of the absorber layer are varied to optimize the device performance. The effects of doping concentration and the electron affinity of the HTM and the ETM on the device performance have been investigated. The hole mobility of the HTM and the defect density at the ETM/absorber and absorber/HTM interfaces are also varied to improve the PCE of the device. This work will be very helpful to design a highly efficient lead based planar PSC with CZTS as the HTM.

2. Simulation Method and Device Model

Many researchers used the SCAPS-1D solar cell simulator to analyze the device performances such as quantum efficiency (QE), energy band diagram, J–V curve, fill factor (FF), power conversion efficiency (PCE), short circuit current density Jsc and open circuit voltage Voc. Figure 1 shows the reference n–i–p planar PSC structure where the intrinsic absorber layer (MAPbI3) is placed between the p-type HTM (CZTS) and the n-type ETM (TiO2). The back metal contact is gold (Au) and the fluorine doped tin oxide (FTO) is a transparent conducting oxide. The values of shunt and series resistances of the reference PSC are 920.8 Ω · cm2 and 9.8 Ω · cm2, respectively.[25] For simulation, two nano-scale (∼ 10 nm) hypothetical interface layers (IL1 and IL2) are added between the absorber and the ETM/HTM to consider the charge carrier recombination.[31]

Fig. 1. Structure of an n–i–p planar PSC.

Table 1 shows the various material parameters of different layers considered in the simulation. The effective density states of valance band and conduction band are considered to be 1.8 × 1019 cm−3 and 2.2 × 1018 cm−3, respectively.[32] At the back contact, the surface recombination velocities are considered to be 1 × 105 cm/s for electrons and 1 × 107 cm/s for holes.[32] At the front contact, the surface recombination velocities are assumed to be 1 × 107 cm/s for electrons and 1 × 105 cm/s for holes. The defect type is neutral with gaussian distribution and the characteristic energy is assumed to be 0.1 eV.[32] The value of capture cross section for both holes and electrons is taken as 1 × 10−15 cm2. The work functions of front and back contacts are considered to be 4.41 eV and 5.10 eV, respectively.[32] The values of acceptor density, donor density and defect density of various layers are considered to validate the theoretical results with the experimental data.[25] All the simulations are performed under 1000 W/m2 illumination, 1.5 G air mass and 300 K operating temperature.

Figure 2 shows the absorption coefficients (α) of different materials which are taken from Website.[38] These values are taken to avoid the empirical α calculated by the simulator. It is evident from Fig. 2 that MAPbI3 has high α at wavelengths from 325 nm to 750 nm and hence absorbs most of the incident light at the visible spectrum. The peak value of α is at wavelength of 380 nm. Figure 2 also shows that CZTS has high α at wavelengths from 325 nm to 1250 nm. The absorption coefficient of CZTS is nearly constant at wavelengths from 325 nm to 1000 nm and then falls sharply at wavelengths from 1000 nm to 1320 nm. It is evident from Fig. 2 that MAPbI3 has higher α than CZTS up to 600 nm. Therefore CZTS contributes to the light absorption of the PSC from 600 nm to 1100 nm, which increases Jsc of the device. It is also evident from Fig. 2 that α of FTO decreases in the wavelength range from 325 nm to 560 nm since the extinction coefficient of FTO decreases up to wavelength 560 nm.[39]

Fig. 2. Absorption coefficients of different materials.
Table 1.

The material parameters of different layers considered in the simulation.

.
3. Results and discussion
3.1. Model validation

Figure 3 shows the energy band diagram of the reference PSC at equilibrium. It is evident from Fig. 3 that both the CZTS and MAPbI3 act as absorber layers due to their apposite band gaps. It is also evident from Fig. 3 that the band offsets are suitable for the movement of charge carriers to the metal contact escaping recombination. Considering the parameters mentioned in Table 1, the device performance is obtained as Voc of 1.06 V, Jsc of 20.49 mA · cm−2, FF of 58.73% and PCE of 12.76%. It has been found that the simulated results are very close to the experimental data of CZTS based PSC.[25]

Fig. 3. Simulated energy band diagram of the device.
3.2. Effect of the defect density of the absorber on the PSC performance

The defect density of the absorber layer has a significant impact on the device performance. Absorber with larger grain size has lower defect density, longer diffusion length, and higher charge carrier mobility.[40] Figure 4 shows the effect of defect density on the device performance. The other device parameters are the same as those listed in Table 1. It is evident from Fig. 4 that Voc, FF, Jsc and PCE do not change significantly up to defect density 1015 cm−3 and then decreases gradually. Low absorber defect density leads to longer diffusion length, which reduces the carrier loss due to recombination and hence the performance of the device improves. On the contrary, higher defect density causes shorter diffusion length and hence charge carrier recombination increases.[40,41] As a result, the device performance degrades significantly. Table 2 shows the relationship between the defect density and the diffusion length of the absorber layer. It is evident from Table 2 that the diffusion length of the charge carrier increases with the decrease in the defect density of the absorber layer. However, the electron diffusion length does not increase significantly by reducing defect density below 1014 cm−3. The results will be useful for assessing perovskite film quality of lead based PSC.

Fig. 4. Effect of defect density of absorber layer on the device performance.
Table 2.

Variations of the charge carrier diffusion length with the defect density of absorber layer.

.
3.3. Effect of the thickness of the absorber layer

The thickness of the absorber layer affects the device performance. Optimum absorber thickness ensures an efficient charge collection by balancing both absorption and recombination altogether in the device.[42] Figure 5 shows the device performance for various thicknesses of the absorber layer. The other device parameters are the same as those listed in Table 1. As the thickness of the absorber layer increases, Voc increases up to 380 nm, and then decreases marginally. The generation rate of the charge carrier increases due to higher absorption of light. For thinner absorber, when the generation rate is higher than the recombination rate, the Voc increases. Again for thicker absorber, when the recombination rate is higher than the generation rate, the Voc decreases. Figure 5 shows that Jsc also increases with the absorber layer thickness up to 350 nm and then decreases. More excess carrier generates with the increase in thickness due to enhanced light absorption, and hence Jsc increases. The Jsc decreases when the absorber thickness greater than the charge carrier diffusion length due to enhanced recombination. Figure 5 also shows that the FF decreases continuously with the absorber thickness. It is evident from Fig. 5 that the PCE is maximal at ∼ 220 nm thickness. As the absorber thickness increases more photon absorbs and hence, the PCE increases due to enhanced charge carrier generation rate. However for thicker absorber, the carrier transport is limited by the enhanced recombination which reduces the overall PCE. This happens when the thickness of the absorber layer is beyond the optimal value.[42,43]

Fig. 5. Effect of the thickness of absorber layer on the device performance.
3.4. Effect of electron affinities of ETM and HTM

The band offset of the PSC structure plays an important role in determining charge carrier recombination at the interfaces.[44] It also quantifies the Voc of the PSC. The band offsets can be tailored by changing the electron affinities of ETM and HTM. Figure 6 shows the effect of varying the electron affinity of the ETM on the device performance. The other device parameters are the same as those listed in Table 1. It is evident from Fig. 6 that Voc, Jsc, FF, and PCE increase with the electron affinity of the ETM and then decrease gradually. The PCE decreases as the electron affinity decreases below 3.80 eV. The PCE also decreases as the electron affinity increases beyond 3.90 eV. The PCE is optimized at 3.85 eV electron affinity of the ETM.

Fig. 6. Effect of electron affinity of ETM on the device performance.

Figure 7 shows the device performance by varying electron affinity of the HTM. The other device parameters are the same as those in Table 1. It is noticeable from Fig. 7 that the Voc decreases as the electron affinity decreases below 4 eV. The Voc also decreases as the electron affinity of the HTM increases beyond 4.10 eV. However, Jsc decreases significantly as electron affinity increases beyond 3.90 eV. The device performance is optimized at 3.90 eV electron affinity of the HTM. Considering the optimized electron affinities for both the HTM and the ETM, the device performance is found as Voc of 1.05 V, Jsc of 24.11 mA · cm−2, FF of 61.28% and PCE of 15.57%.

Fig. 7. Effect of electron affinity of the HTM on the device performance.
3.5. Effect of the defect density at the interface

The interface quality at the heterojunction affects the device performance. Specially, interface defect density at the absorber/ETM and absorber/HTM has a significant effect on the device performance.[31,32] Figure 8 shows the J–V characteristics of the PSC by varying defect density at the interface layer IL2. The other device parameters are the same as those in Table 1. It is evident from Fig. 8 that the defect density at IL2 does not affect the device performance significantly. Similar results have been demonstrated by another work.[32] Figure 9 shows the J–V characteristics of the PSC by varying defect density at the interface layer IL1. The other device parameters are the same as those in Table 1. It is noticeable from Fig. 9 that the J–V curve shifts downwards with the increase in defect density at IL1. As the defect density increases, more carriers are trapped at the interface and are not available for conduction. As a result, the device performance deteriorates.[32] The interface layer IL1 is closer to FTO than IL2 and hence, most of the charge carriers are generated near IL1. The presence of higher excess carrier near IL1 causes significant carrier loss due to trapping. As a result, defects at IL1 have more effects on the device performance than that at IL2. The similar effects for planar PSC structure have been observed by other researchers.[32] Therefore, a high quality absorber/ETM interface is essential for highly efficient lead based PSCs.

Fig. 8. J–V characteristics of PSC for various defect densities of IL2.
Fig. 9. J–V characteristics of PSC for various defect densities of IL1.
3.6. Effect of doping densities of absorber, ETM, and HTM

Doping plays a vital role to improve the performance of the halide based PSC. As doping density increases, charge carrier concentrations also increases. However, ionized impurity introduces scattering centers in the crystal structure which reduces the charge carrier mobility. Moreover, deep traps are generated due to high doping density.[45] Figure 10 shows the PCE of the PSC for various doping densities of the absorber, ETM, and HTM. Different ranges of doping density for three layers have been considered to show the significant effects. The other device parameters are the same as those in Table 1. Figure 10 manifests that the PCE of the PSC increases up to 2.0 × 1016 cm−3 doping density of the absorber layer and then decreases sharply. With the increase in doping density built in electric field enhances. As a result, the PCE of the device increases.[46] It is evident from Fig. 10 that the PCE increases for heavily doped ETM and HTM. The charge density increases with the increase in doping density, which increases the conductivity of the ETM/HTM and hence the PCE of the device increases. It has been found that the PCE is optimum considering a doping density of 2.0 × 1018 cm−3 and 1019 cm−3 for the ETM and the HTM, respectively.

Fig. 10. The PCE of the PSC for various doping densities of the absorber, ETM, and HTM.
3.7. Determination of the optimized PSC structure

Table 3 shows the optimum parameters of the device. The comparison of the device performance considering the various optimum parameters and the reference structure parameters are shown in Table 4. The series resistance has a significant effect on the device performance. Considering all the optimum parameters and a series resistance Rs of 9.8 Ω · cm2, the device performance is as Voc of 1.15 V, Jsc of 26.29 mA·cm−2, FF of 65.73%, and PCE of 19.83%. It has been reported that a lower Rs of ∼ 5 Ω · cm2 can be achieved in the planar lead based PSC.[43] For high performance thin film solar cells, Rs is less than 1 Ω · cm.[43] Considering all the optimum parameters and Rs of 5 Ω · cm2, the final optimum device performance is as Voc of 1.15 V, Jsc of 26.45 mA · cm−2, FF of 74.61%, and PCE of 22.7%. The simulated optimum device performance is similar to the experimental data of mesostructured double-layered halide architecture with costly organic HTM.[47] Figure 11 shows the J–V characteristic curves of the PSC considering the optimum parameters and the reference structure parameters. The performance of the PSC can be improved further by fabricating a highly uniform perovskite layer with large grain size and long carrier diffusion length.[43]

Fig. 11. J–V characteristics of the PSC. The dotted line and the solid line represent the simulated results considering the reference parameters and the optimum parameters, respectively.
Table 3.

Optimum parameters of the device.

.
Table 4.

Comparison of the performance between the reference structure parameters and the optimum parameters.

.

As mentioned above, CZTS has also been used as an active layer in thin film solar cells. The best reported PCE of CZTS based thin film solar cell is 11%.[48] Various synthesis techniques have been used to improve the PCE of CZTS based photovoltaic devices. The complex synthesis process forms secondary phases which degrades the CZTS film quality and hence limits the solar cell efficiency.[49]

4. Conclusion

A lead based PSC with CZTS as inorganic HTM has been analyzed by SCAPS-1D device simulator. Consistent material parameters are taken from the literature. The relative permittivity, electron affinity, band gap, and charge carrier mobility of the absorber layer are the same as those in the reference structure. For the ETM and the HTM, the thickness, relative permittivity, band gap, and defect density are also the same as those in the reference structure. The simulated result illustrates that the performance of the reference PSC can be improved taking an optimum absorber layer defect density of ∼ 5 × 1014 cm−3. The optimum thickness of absorber layer (∼ 220 nm) has been found, which increases the PCE. The doping in the absorber layer has also been optimized to increase the PCE from 12.76% to 15.69%. It has been observed that a doping density beyond 2 × 1016 cm−3 in the absorber layer reduces the PCE of the device due to enhanced recombination rate. Electron affinities of 3.9 eV and 3.85 eV for the HTM and the ETM, respectively are found to increase the PCE of the reference PSC from 12.76% to 15.57%. It has also been found that the defect density at the absorber/HTM interface has negligible effect on the device performance. The defect density at the absorber/ETM interface has significant effect on the device performance as most of the carrier generated near this interface. The overall PCE of the reference PSC increases from 12.76% to 22.7% by considering a series resistance of 5 Ω·cm2, and the optimum parameters of absorber, ETM and HTM layers simultaneously. The PCE of the optimized PSC has been increased significantly compared to the reference structure. Hence, the optimized PSC structure with CZTS as the HTM is highly efficient.

Reference
[1] Park N G Gratzel M Miyasaka T Zhu K Emery K 2016 Nat. Energy 1 16152
[2] Hawash Z Ono L K Qi Y 2018 Adv. Mater. Interfaces 5 1700623
[3] Kim H S Lee C R Im J H Lee K B Moehl T Marchioro A Moon S J Humphry-Baker R Yum J H Moser J E Grätzel M 2012 Sci. Rep. 2 591
[4] Liu D Kelly T L 2014 Nat. Photon. 8 133
[5] Jiang M Niu Q Tang X Zhang H Xu H Huang W Yao J Yan B Xia R 2019 Polymers 11 147
[6] Kojima A Teshima K Shirai Y Miyasaka T 2009 J. Am. Chem. Soc. 131 6050
[7] Fakharuddin A De Rossi F Watson T M Schmidt-Mende L Jose R 2016 APL Mater. 4 091505
[8] Werner J Niesen B Ballif C 2018 Adv. Mater. Interfaces 5 1700731
[9] Noh J H Im S H Heo J H Mandal T N Seok S I 2013 Nano Lett. 13 1764
[10] Baltakesmez A Biber M Tüzemena S 2018 J. Radiat. Res. App. Scis 11 124
[11] Kim Y Jung E H Kim G Kim D Kim B J Seo J 2018 J. Adv. Energy Mater. 8 1801668
[12] Nia N Y Matteocci F Cina L Carlo A D 2017 ChemSusChem. 10 3854
[13] Wang Y Hu Y Han D Yuan Q Caoc T Chen N Zhou D Cong H Feng L 2019 Org. Electron. 70 63
[14] Jena A K Numata Y Ikegami M Miyasaka T 2018 J. Mater. Chem. 6 2219
[15] Ava T Al-Mamun A Marsillac S Namkoong G 2019 Appl. Sci. 9 188
[16] Hu L Li M Yang K Xiong Z Yang B Wang M Tang X Zang Z Liu X Li B Xiao Z Lu S Gong H Ouyang J Sun K 2018 J. Mater. Chem. 6 16583
[17] Heo J H Han H J Lee M Song M Kim D H Im S H 2015 Energy Environ. Sci. 8 2922
[18] Zhou P Bu T Shi S Li L Zhang Y Ku Z Peng Y Zhong J Cheng Y B Huang F 2018 J. Mater. Chem. 6 5733
[19] Ye S Sun W Li Y Yan W Peng H Bian Z Lui Z Huang C 2015 Nano Lett. 15 3723
[20] Zuo C Ding L 2015 Small 11 5528
[21] Patel S B Patel A H Gohel J V 2018 Cryst. Eng. Comm. 20 7677
[22] Li X Yang J Jiang Q Chu W Zhang D Zhou Z Xin J 2017 ACS Appl. Mater. Interfaces 9 41354
[23] Xu L Deng L L Cao J Wang X Chen W Y Jiang Z 2017 Nanoscale Res. Lett. 12 159
[24] Yang S Fu W Zhang Z Chen H Li C Z 2017 J. Mater. Chem. 5 11462
[25] Wu Q Xue C Li Y Zhou P Liu W Zhu J Dai S Zhu C Yang S 2015 ACS Appl. Mater. Interfaces 7 28466
[26] Zuo Y Chen L Jiang W Liu B Zeng C Li M Shi X 2018 Mater. Tehno. 52 483
[27] Mahmood K Sarwar S Mehran M 2017 RSC Adv. 7 17044
[28] Wang R Mujahid M Duan Y Wang Z K Xue J Yang Y 2019 Adv. Funct. Mater. 1808843
[29] Burgelman M Decock K Khelifi S Abass A 2013 Thin Solid Films 535 296
[30] Kabir I Sadik F Mahmood S A 2018 10th International Conference on Electrical and Computer Engineering Dhaka, Bangladesh 20–22 December 2018 145
[31] Chouhan A S Jasti N P Avasthi S 2018 Mater. Lett. 221 150
[32] Tan K Lin P Wang G Liu Y Xu Z Lin Y 2016 Solid State Electron. 126 75
[33] Li H Yang Y Feng X Shen K Li H Li J Jiang Z Song F 2016 Nanomater Nanotechnol. 6 24
[34] Herz L M 2017 ACS Energy Lett. 2 1539
[35] Chihi A Boujmil M F Bessais B 2017 J. Electron. Mater. 46 5270
[36] Shin B Gunawan O Zhu Y Bojarczuk N A Chey S J Guha S 2013 Prog. Photovolt: Res. Appl. 21 72
[37] Wanda M D Ouédraogo S Tchoffo F Zougmoré F Ndjaka J M B 2016 Int. J. Photoenergy 1
[38]
[39] Sap J A Isabella O Jager K Zeman M 2011 Thin Solid Films 520 1096
[40] Ng A Ren Z Shen Q Cheung S H Gokkaya H C So S K Djurisic A B Wan Y Wu X Surya C 2016 ACS Appl. Mater. Interfaces 8 32805
[41] Zhou Y Long G 2017 J. Phys. Chem. 121 1455
[42] Baloch A A Hossain M I Tabet N Alharbi F H 2018 J. Phys. Chem. Lett. 9 426
[43] Tavakoli M M Gu L Gao Y Reckmeier C He J Rogach A L Yao Y Fan Z 2015 Sci. Rep. 5 14083
[44] Lim K G Ahn S Kim Y H Qi Y Lee T W 2016 Energy Environ. Sci. 9 932
[45] Thakur U Kisslinger R Shankar K 1996 Nano Mater. 7 95
[46] Jiang C S Yang M Zhou Y To B Nanayakkara S U Luther J M Zhou W Berry J J Van De Lagemaat J Padture N P Zhu K Al-Jassim M M 2015 Nat. Commun. 6 8397
[47] Jung E H Jeon N J Park E Y Moon C S Shin T J Yang T Y Noh J H Seo J 2019 Nature 567 511
[48] Ravindiran M Praveenkumar C 2018 Renew. Sustain. Energy Rev. 94 317
[49] Zhang X Fu E Wang Y Zhang C 2019 Nanomaterials 9 336