Efficiency of earth abundant and pure sulfide kesterite Cu-Zn-Sn-S (CZTS) solar cell has been stagnant around 9.4% for years, while its counterpart Cu-In-Ga-Se (CIGS) reports an efficiency of more than 22%. Low open circuit voltage (VOC) is the major challenging factor for low efficiency due to severe nonradiative interface recombinations. The existence of higher defect states at the conventional CZTS-CdS interface due to undesirable energy level alignment and lattice misfit promotes trap-assisted recombinations and results in low VOC. In this work, amorphous TiO2(Eg=3.8eV) is proposed as a promising substitute to the conventional and low bandgap CdS (Eg=2.4eV) layer. The surface and interface of the CZTS-TiO2 layer were investigated using X-ray photoelectron spectroscopy (XPS) and ultraviolet photoelectron spectroscopy (UPS). The result reveals favorable “spike”-like conformations at the CZTS-TiO2 interface with a conduction band offset value of 0.17 eV. The nanoscale probing of the interface by Kelvin probe force microscopy across CZTS-TiO2 layers shows a higher potential barrier for interface recombination at CZTS-TiO2 in contrast to the conventional CZTS-CdS interface. Finally, the fast decay response and lower persistent photoconductivity of photogenerated carriers for CZTS-TiO2 heterojunction based photodetectors further validate our results. The energy level alignment and nanoscale interface studies signify TiO2 as a promising alternate buffer layer for earth abundant CZTS solar cells.

Among various photovoltaic technologies, Cu2ZnSnS4 (CZTS) based solar cells have attracted worldwide attention for scalable thin film solar cell development. Several of its key features, such as suitable bandgap at ∼1.4–1.5 eV, earth abundant nontoxic elements, and high absorption coefficient of ∼104 cm−1, make it an attractive and potential choice for sustainable photovoltaic devices.1–3 However, the current record efficiency of 9.4% in CZTS solar cells is still not proportionate to commercially available technologies such as Cu-In-Ga-Se (CIGS) and CdTe. The lower performance of CZTS solar cells is mainly attributed to a significant voltage loss due to increased interface recombination.4–6 Fundamentally, it is known that the conduction band offset (CBO) and the low quality p-n junction at the CZTS-CdS (buffer) interface critically affect the open circuit voltage (VOC) of the device.7 Different conduction band offset (CBO) values have been reported for the CZTS-CdS heterojunction. Some studies claimed that CBO of CZTS-CdS is “spike”-like8 and others have reported it to be “cliff”-like.9,10 This disparity might be due to the variation in the interface behavior caused by different process conditions or treatments prior to CdS deposition. It was reported that “cliff” conformation acts as a barrier for the flow of injected electrons from the buffer to the absorber in the forward bias condition.4 These accumulated electrons at the interface recombine with the majority carriers (holes) in the p-type CZTS via interface states, thereby reducing VOC, while “spike”-like alignment within the CBO range (0–0.4 eV) is supposed to be favorable for severe interface recombination suppression.8 Moreover, from a device model of CZTS solar cells, Courel et al. have shown that the CZTS-CdS interface is prone to have high density of defect formation due to large lattice and thermal expansion mismatches between the CZTS absorber and the CdS buffer layer.7 Various efforts have also been made to improve p-n junction properties at the CZTS-CdS interface for better charge collection and reduced interface recombination.11–13 The charge carriers generated close to the space charge region (SCR) quickly get separated due to electric field, while charge carriers generated away from the junction recombine due to the short diffusion length. Thereby, a wider SCR region will assist in extracting charge carriers efficiently. Furthermore, the low bandgap of CdS (Eg=2.4eV) causes parasitic absorption and thus reduces short circuit current density (JSC) of the device.

To resolve aforementioned challenges with conventionally employed CdS, several alternate buffer layers have been reported, In2S3,10 Zn1−xSnxO,12 Zn1−xMgxO,14 Zn(O,S),10 and Zn1−xCdxS.15 The compound buffer layers have some disadvantages such as improper mixing and complicated reaction mechanism that further limits the scalable production. Even after proposing different approaches, there is a limit to reduce interface recombination and to improve power conversion efficiency (PCE) in CZTS solar cells.7–10 The PCE of CZTS solar cells is still far away from the level of capturing the commercial market. Therefore, a wide bandgap material with a suitable energy level alignment is required to substitute CdS in CZTS solar cells. Moreover, it is also crucial to closely analyze the chemical and electrical properties of the absorber-buffer interface.16 

This work investigates the interface energetics of the CZTS-TiO2 heterojunction that substantially proves amorphous TiO2 (Eg = 3.8 eV) as a potential substitute of CdS in CZTS solar cells. Amorphous TiO2 was deposited over CZTS, and the interface energetics were compared to the conventional CZTS-CdS interface. A detailed study of energy level alignment was carried out using X-ray photoelectron spectroscopy (XPS) and UV photoelectron spectroscopy (UPS). Our results show favorable “spike”-like band alignment between CZTS and TiO2 that can effectively reduce interface recombination. Furthermore, nanoscale probing of the CZTS-TiO2 interface was carried out using Kelvin probe force microscopy (KPFM) and compared with the conventional CZTS-CdS interface to understand the charge carrier dynamics at both the interfaces. The analysis of the surface potential (SP) variation at two interfaces exhibits that the p-n junction formed at the CZTS-TiO2 interface has a wider SCR and higher built-in potential, advocating excellent charge separation conditions. The energy level alignment and surface potential data from XPS and KPFM, respectively, were further correlated with the time dependent photoresponse of CZTS-CdS and CZTS-TiO2 heterojunction based photodetectors. The CZTS-TiO2 based photodetector has shown a fast decay response due to the effective separation of charge carriers that eventually suppresses the charge carrier recombinations at the interface which strongly approves the XPS and KPFM analysis. We have critically chosen the transient photoresponse at the absorber/buffer interface only, as the fabrication of solar cells would add other interfaces in the device. Therefore, to understand the charge separation mechanism, the transient photoresponse at the absorber/buffer interface is investigated. Our results illustrate that TiO2 has the potential to successfully substitute the toxic CdS buffer layer in earth abundant CZTS solar cells.

The CZTS thin film was prepared by the cosputtering of Cu/ZnS/SnS2 precursors on soda lime glass (SLG) via the magnetron sputtering technique. Prior to CZTS deposition, SLG was cleaned thoroughly using de-ionized (DI) water followed by ultrasonication in acetone and propanol for 30 min. The cleaned substrate was loaded in the sputtering chamber and a base pressure of 7.5 × 10−7 Torr was achieved. The cosputtering of three targets was carried out at 6 mTorr by providing the RF power to SnS2 and ZnS targets while a DC power to the Cu target at room temperature. Finally, a thickness of ∼1 μm of CZTS was deposited on the SLG substrate. The as-deposited CZTS film was then sulfurized in a quartz tube furnace for 3 h at 510 °C in 95% Ar and 5% H2S atmosphere. A 70 nm CdS layer was deposited over CZTS films using the traditional chemical bath deposition (CBD) method at 70 °C for 10 min. The aqueous solution of DI water (36 ml), aqueous ammonium hydroxide (8 ml), CdSO4 (10 mM), and thiourea (92.95 mM) was used for CdS deposition. After CdS deposition, samples were rinsed by DI water to remove less adhesive CdS particulates from the film surface. Amorphous TiO2 over CZTS films was deposited by RF magnetron sputtering with the RF power of 60 W for 30 min at room temperature. To investigate the CZTS-TiO2 and CZTS-CdS interface, half of the CZTS was covered and TiO2 and CdS layers were deposited on the other half.

The structural analysis of CZTS thin films was performed using X-ray diffraction (XRD) and Raman spectroscopy. Grazing incident X-ray diffraction (GI-XRD) was recorded from a PANalytical X-pert pro multipurpose diffractometer equipped with a high intensity Cu Kα energy source. Renishaw make Raman spectroscopy with excitation laser of wavelength 532 nm was used to identify the phase purity of CZTS thin films. Surface morphology and elemental distribution of CZTS samples were analyzed by scanning electron microscopy (SEM, JEOL 6610LV) coupled with an energy dispersive X-ray spectroscopy (EDX) system. A UV-Vis-NIR spectrophotometer (Perkin-Elmer, Lambda 950) was employed to measure the bandgap of CZTS, CdS, and TiO2 films.

To examine the band alignment at the buffer/CZTS interface, the valence band offset (VBO) and the conduction band offset (CBO) was determined by X-ray photoelectron spectroscopy (XPS) and Ultraviolet photoelectron spectroscopy (UPS). XPS measurements were carried out using an ESCALAB XI+ (ThermoFisher Scientific) under an ultrahigh vacuum analysis chamber. The X-ray source was monochromated Al kα (hν = 1486.68 eV) with a binding energy calibrated with the gold reference. In order to probe the buffer/CZTS interface and CZTS bulk, the buffer layers (CdS and TiO2) were sputtered with 1 keV Ar+ ions until we reach the CZTS layer from the top buffer layer. UPS was used to collect information about the valence band maxima (VBM) of different samples. UPS source was He I with an energy of 21.22 eV.

The nanoscale contact potential difference (CPD) of buffer/CZTS thin films is obtained by using noncontact KPFM with Agilent 5500 scanning probe microscopy. The KPFM system with Cr/Pt coating tip possessing a resonance frequency of 75 kHz with a force constant 3 N/m and a Multi75E-G budget sensor was used for CPD measurements. The separation between the tip and the sample was controlled by a first lock-in amplifier (L1A1) fed with the first resonance frequency of 67 kHz for topographic and phase imaging of the sample, whereas a second lock-in amplifier (L1A2) was fed by the second resonance frequency of 5 kHz for surface potential measurements. An electrical oscillation to the tip was provided by LIA2 at 5 kHz with a dc offset of −3 V and to attain a surface potential amplitude of 0.2 V, the drive percentage of LIA2 was approximately 15%. Surface potential with respect to z (SP vs z) spectroscopy was delicately measured to prevent any cross talk between topography and SP signal from the sample. The voltage bias was applied to the sample while performing the KPFM measurements. Finally, photodetectors were fabricated on the CZTS-CdS and CZTS-TiO2 heterojunction to correlate the XPS and KPFM analysis with photodetector performance. The transient photoresponses were determined using Keithley 6430 SMU with a 532 nm laser of power 5 mW/cm2. All the measurements were executed in a dark box to avoid any hindrance from the abandoned light from the surroundings.

The surface morphology of the sulfurized CZTS layer using SEM is shown in Figs. 1(a) and 1(b). The enlarged view in Fig. 1(b) shows a dense and crack-free top surface, which is composed of relatively small and closely packed grains. The composition ratios, Cu/(Zn + Sn) and Zn/Sn, of the prepared CZTS film are extracted from energy-dispersive X-ray spectroscopy are 0.93 and 1.10, respectively. Figure 1(c) shows the X-ray diffraction spectra (XRD) for CZTS films plotted against the standard CZTS (JCPDS 26-0575) data. XRD analysis affirms that the film is polycrystalline in nature with peaks corresponding to (101), (112), (200), (220), (312), and (224) planes of the Kesterite CZTS phase. It is well known that CZTS, Cu2SnS3, and ZnS have identical XRD patterns due to the similar crystal structure; therefore, just based on the XRD results, it is hard to determine the purity of the CZTS phase.17,18 In order to distinguish the CZTS phase from other secondary phases, Raman measurements were performed with an excitation wavelength of 532 nm laser and results are shown in Fig. 1(d). Raman spectra of CZTS sample show characteristics peaks at 288 cm−1, 337 cm−1, and 375 cm−1, which are in good agreement with the literature.19,20 No Raman peaks corresponding to Cu2SnS3 and ZnS were found. However, the little off stoichiometric composition in CZTS may result in secondary phases present in the absorber. Although, we have not found sharp peaks corresponding to different secondary phases during Raman measurements, as shown in Fig. 1(d). It is reported in the literature that ZnS generally segregates at the back interface.21,22 However, we have performed Raman with a 325 nm excitation laser to investigate the presence of ZnS as shown in Fig. 1(d), though, we have not found any significant Raman peak for it.23,24 It may be due to the strong absorption of 325 nm wavelength which leads to less penetration depth of the laser wavelength. The XRD pattern represented in Fig. 2(a) indicates that the deposited TiO2 film is amorphous in nature, while CdS has a polycrystalline structure. The polycrystalline CdS leads to a 7% lattice misfit at the CZTS-CdS interface.4 This lattice misfit may cause defects states within the bandgap and act as trap centers for charge carriers. It is also reported that lattice mismatching at the CZTS-CdS interface leads to lattice strain and damage to the interface structure,25 whereas an amorphous TiO2 can neutralize the effect of lattice misfit at the interface.4 The AFM images of CdS and TiO2 thin films are shown in Figs. 2(b) and 2(c), respectively. The measured root mean square (RMS) roughness for CdS and TiO2 thin films is 11 ± 1.5 nm and 0.9 ± 0.14 nm, respectively at the 1 μm scale. Thus, the surface roughness is low for the amorphous TiO2, while CBD deposited CdS offers higher surface roughness. Javey and his co-workers reported significant improvements in VOC for the CIGS solar cell using the amorphous TiO2 as an electron selective layer (ETL). However, nanocrystalline TiO2 films with higher roughness lower down the performance of the solar cell.26 The same group has applied the amorphous TiO2 with p-type InP and an enhanced open circuit voltage of 735 mV was achieved.27 Thus, an amorphous TiO2 buffer layer accompanied with low surface roughness obtained in this work may be beneficial for the absorber-buffer interface.

FIG. 1.

(a) SEM surface morphology of sulfurized CZTS thin films, (b) enlarged view of the top surface morphology, (c) X-ray diffraction spectra of CZTS films with JCPDS (26-0575) data, and (d) Raman spectra of sulfurized CZTS thin films.

FIG. 1.

(a) SEM surface morphology of sulfurized CZTS thin films, (b) enlarged view of the top surface morphology, (c) X-ray diffraction spectra of CZTS films with JCPDS (26-0575) data, and (d) Raman spectra of sulfurized CZTS thin films.

Close modal
FIG. 2.

(a) X-ray diffraction spectra of CdS and TiO2 films, (b) atomic force microscopy (AFM) images of deposited CdS thin films, and (c) TiO2 thin films on the bare SLG substrate.

FIG. 2.

(a) X-ray diffraction spectra of CdS and TiO2 films, (b) atomic force microscopy (AFM) images of deposited CdS thin films, and (c) TiO2 thin films on the bare SLG substrate.

Close modal

The depth profiling was employed using XPS in order to monitor the chemical nature of CZTS-TiO2 and CZTS-CdS interface regions, respectively. The XPS depth profiles corresponding to Cu, Sn, Zn, Ti, Cd, and S elements are shown in Fig. 3. The insets of the same figure show the schematic of the CZTS-CdS and CZTS-TiO2 interface structure. It was observed that Cd from CdS diffuses deeply inside the CZTS film, while there is negligible diffusion of Ti into CZTS. For the CZTS film with TiO2, the concentration of Cu, Zn, Sn, and S near the interface is quite homogeneous compared to the atomic composition for CZTS films with the CdS buffer. Since all samples were fabricated in a single run of deposition and have the same annealing (sulfurization) conditions, it can be expected that all CZTS samples have a similar atomic composition. Therefore, the changes observed in the concentration of elements for CZTS near the interface appear only after the buffer layer deposition. It was observed that Cd diffusion disturbs the atomic composition of elements not only at the interface but also deep inside the CZTS layer. The CZTS-CdS interface region, shown in Fig. 3(a), is depleted from Sn and Zn signals, establishing a Cu-rich and Zn-poor region near the interface. The Cu-rich layer near the interface is not desirable for device performance as with rise in the Cu content, hole carrier concentration (NC) increases.28 The higher value of NC deteriorates the photovoltaic performance, reduces minority carrier's (electrons in CZTS) lifetime, reduces mobility, and cut down the short-circuit current density (JSC) value of the devices.29 Moreover, it is reported that Cu rich and Zn poor conditions at interface may lead to high density of CuZn antisite defects, which are detrimental for heterojunction interface quality because of Fermi level pinning.30 Tajima et al. fabricated high-performance devices with improvements in photovoltaic parameters using the Cu-poor layer near the CdS interface and Cu-rich layer near the Mo interface.31 On the other hand, Cd diffusion deep in the CZTS replaces Zn rather than Cu because of isoelectronic configuration and comparatively much lower formation energy of CdZn antisites.32 The diffusion width of Cd inside CZTS is directly associated with the Zn depletion level toward the interface, as evident from Fig. 3(a). Therefore, from depth profiling, we can state that Zn site may be occupied by Cd. Cd replacing Zn would lead to the formation of CdZn antisite defects which are deep acceptor trap states.32 Our observation is completely in agreement with many reports that show Zn depletion by Cd insertion inside the CZTS film.32,33 However, there are reports that also show Zn diffusion inside the CdS buffer.34 It is reported that deep diffusion and higher Cd concentration inside CZTS deteriorate device performance,35 whereas Fig. 3(b) shows the presence of Cu, Zn, Sn, and S signals on the TiO2 buffer side and Ti diffusion up to a few nanometers (close to the interface) in the CZTS layer, clearly suggesting the intermixing of CZTS and TiO2 layers at the interface. Such a kind of intermixing at the interface is suggested to form a good quality heterojunction.26 No signs of Cu rich and Zn poor conditions at the CZTS-TiO2 interface unlike CZTS-CdS interface, rather all the elements are finely intermixed. The elemental interdiffusion improves the interface quality and believed to upgrade device performance by limiting charge carrier recombination at the interface.26 Therefore, XPS depth profiling revealed promising interface quality with the TiO2 buffer layer. It advocates that TiO2 over CZTS form a better interface than CdS over CZTS. This encouraged us to examine the band alignment and investigate the CBO at both the heterojunctions.

FIG. 3.

The XPS composition profile of CZTS with (a) CdS and (b) TiO2 buffer layer with etching time. The inset shows the schematic structure of the CZTS-CdS and CZTS-TiO2 heterojunction. The dashed (black) line represents the interface area.

FIG. 3.

The XPS composition profile of CZTS with (a) CdS and (b) TiO2 buffer layer with etching time. The inset shows the schematic structure of the CZTS-CdS and CZTS-TiO2 heterojunction. The dashed (black) line represents the interface area.

Close modal

To further investigate the chemical ambient from top of the buffer layer toward the bulk CZTS, XPS core level spectra on both the samples were collected after different Ar+ etching times. A careful investigation was conducted to examine any chemical and structural changes in the material after etching the surface with 1 keV Ar+ ions. No significant changes were observed in Sn 3p, Zn 2p, and Cu 2p core level B.E after XPS sputtering cycles, which is in complete agreement with the reported literature,15–36 while Ti 2p, Cd 3d, and S 2p line shapes show modifications due to variations in their chemical ambient that is elaborately explained further in the discussion. Figure 4 shows the core-level XPS spectra of the CZTS-TiO2 and CZTS-CdS interface as a function of distance from the top surface of the buffer to CZTS (bulk). Figures 4(a)–(d) and 4(f)–(i) exhibit core-level spectra of Sn 3d, Zn 2p, Cu 2p, and S 2p for CZTS films decorated with TiO2 and CdS layers, respectively. The binding energy peak of Sn 3d, Zn 2p, and Cu 2p core-levels is situated at 486.46 eV, 1021.75 eV, and 932.63 eV, with peak separation around 8.4 eV, 23.1 eV, and 19.8 eV, indicative of Sn(IV), Zn(II), and Cu(I), respectively.37,38 No peak found corresponding to Sn 3d and Zn 2p, as shown in Figs. 4(f) and 4(g), at the CZTS-CdS interface as this interface is depleted from Sn and Zn, discussed earlier. The analysis of core level states of S for the CZTS film with both buffer layers (TiO2 and CdS) shows doublets of S 2p3/2 and S 2p1/2 exists at 161.75 eV and 162.83 eV with a peak splitting of 1.1 eV, as shown in Figs. 4(d) and 4(i).38 In the case of CZTS-CdS, S 2p3/2 and S 2p1/2 has a peak shift of 0.3 eV toward higher binding energy when moving from CZTS toward the CdS layer, as shown in Fig. S1 of the supplementary material. The peak shift could be associated with the change in S-bonds within CZTS film and within the CdS layer. The sulfur is bonded to all elements, i.e., Cu, Zn and Sn in CZTS, while it is bonded to Cd only in CdS. However, S 2p doublet peaks were found to be consistent with an expected range of 160–164 eV for the sulfide phases. XPS analysis for elements Cu, Zn, Sn, and S, is in good agreement with the reported literature, thereby confirming the CZTS phase beneath the CdS and TiO2 buffer layers.37–39 The Ti 2p core level splits into 2p3/2 and 2p1/2 centered at 458.94 eV and 464.75 eV with peak separation 5.6 eV, indicating the Ti(IV) valence state and confirming the existence of the TiO2 phase, as shown in Fig. 4(e).40 The Cd 3d spectrum shown in Fig. 4(j) displays two peaks at 405.71 eV and 412.49 eV, with a binding energy difference of 6.7 eV.41 These peaks are characteristics of the Cd(II) valence state. There is a shift in the peak toward lower energy for Cd when XPS data were recorded from the top of CdS to the CZTS layer, as shown in Fig. S2 of the supplementary material. This peak shift may be attributed to Zn substitution by Cd, as discussed earlier. There is no existence of a satellite peak in the S 2p spectrum around 168 eV, recorded near the interface or inside the CdS layer, as shown in Fig. S3 of the supplementary material. This peak is assigned to the S–O bond in SO42− and SO32− species.41,42 This clearly indicates that deposited CdS was in pure sulfide form and no CdS-oxidized phase exists. As can be seen in Fig. 4(j), signals for Cd are detected inside the CZTS (bulk) due to deep Cd diffusion as discussed earlier. However, Ti peaks get recorded up to the CZTS-TiO2 interface, shown in Fig. 4(e). Moreover, weak intensity peaks for Cu, Zn, and Sn are observed on the TiO2 side, illustrating the interdiffusion of elements at the interface justified for the superior interface quality.

FIG. 4.

High resolution core-level XPS spectra of the CZTS-TiO2 interface (a) Sn 3d, (b) Zn 3p, (c) Cu 2p, (d) S 2p, (e) Ti 2p, and of the CZTS-CdS interface (f) Sn 3d, (g) Zn 2p, (h) Cu 2p, (i) S 2p, and (j) Cd 3d. XPS spectra are recorded in the direction from the top surface of the buffer layer to the CZTS (bulk). In (d) and (i), black lines are the original signals and other color lines are the fitting of experimental data.

FIG. 4.

High resolution core-level XPS spectra of the CZTS-TiO2 interface (a) Sn 3d, (b) Zn 3p, (c) Cu 2p, (d) S 2p, (e) Ti 2p, and of the CZTS-CdS interface (f) Sn 3d, (g) Zn 2p, (h) Cu 2p, (i) S 2p, and (j) Cd 3d. XPS spectra are recorded in the direction from the top surface of the buffer layer to the CZTS (bulk). In (d) and (i), black lines are the original signals and other color lines are the fitting of experimental data.

Close modal

In order to provide an insight into the energy level alignment at heterojunction interfaces, XPS core-level spectroscopy and UPS measurements were conducted. The CBO was determined by using VBO along with bandgap (Eg) values of different layers.

The VBO can be obtained by the following formula:36–43 

VBO=EVBbEVBa+Vbb,
(1)

where EVBb and EVBa represent the energy positions of the valence band edges of bulk buffer and bulk CZTS absorber, respectively, and Vbb is the band bending.36 The Vbb value was calculated by the following formula:

Vbb=(ECLaECLa(i))+(ECLb(i)ECLb).
(2)

Here, ECLa and ECLb are the core level energies of two selected elements in the bulk region of CZTS (a) and buffer (b) layer, respectively, while ECLa(i) and ECLb(i) represent the core level energies of the same corresponding elements measured at the interface. While calculating the Vbb values, we have assumed that the convention that binding energy beneath the Fermi level is negative. Consequently, VBO will be negative if the valence band edge of the buffer is lower than that of the CZTS.

Valence band maxima (VBM) positions of CZTS, TiO2, and CdS were measured by UPS and displayed in Fig. 5(a), predicted by the linear extrapolation approach.44 It can be seen from Fig. 5(a) that few energy states are occupied above VBM, which can be attributed to the presence of dangling bonds and other defect states present within the material.45 The presence of such defect states eventually leads to the tailing of the VBM edge. The VBM positions (relative to EF) determined by the linear fitting of the leading edge with uncertainties are found to be (−0.62 ± 0.03) eV, (−1.72 ± 0.02) eV, and (−3.0 ± 0.02) eV for CZTS, CdS, and TiO2, respectively. The VBM values calculated for CZTS, CdS, and TiO2 are consistent with the reported literature.9–46 To gain information about the band bending, the shift in the core level energies of different elements at the interface and bulk of CZTS and buffer layers was analyzed. The band bending at the CZTS-buffer interface was calculated from the core level energy shift of Cu, Zn, and Sn from CZTS, while Cd and Ti were chosen from CdS and TiO2, respectively. Figures S4 and S5 of the supplementary material show the core level energies of Cu (2p3), Zn (2p3), Sn (3d5), Cd (3d5), and Ti (2p3) with different etching times. The error on the band bending values is conservatively evaluated to be ±0.10 eV. The average band bending, Vbb, for CZTS-CdS and CZTS-TiO2 is estimated to be (0.08 ± 0.1) eV and (0.27 ± 0.1) eV, respectively. Finally, the VBO was calculated by using Eq. (1) and found to be (−1.02 ± 0.1) eV and (−2.1 ± 0.1) eV for CZTS-CdS and CZTS-TiO2 heterojunctions, respectively. The CBO for the interfaces was determined using VBO values through the following equation:

CBO=EgbEga+VBO.
(3)
FIG. 5.

(a) UPS spectra of CZTS, CdS, and TiO2 samples. The linear extrapolation (pink solid line) of leading UPS edges was used to determine VBM. The black dotted line marks the exact position of VBM on the x axis. (b) Schematic representation of the band alignment at the CZTS/TiO2 and CZTS/CdS interfaces. VBM, CBO, EF, and Eg are indicated (in electron volts).

FIG. 5.

(a) UPS spectra of CZTS, CdS, and TiO2 samples. The linear extrapolation (pink solid line) of leading UPS edges was used to determine VBM. The black dotted line marks the exact position of VBM on the x axis. (b) Schematic representation of the band alignment at the CZTS/TiO2 and CZTS/CdS interfaces. VBM, CBO, EF, and Eg are indicated (in electron volts).

Close modal

Here, Egb and Ega are the optical bandgap measured by UV-Vis spectroscopy for the pure buffer and the CZTS layer, respectively. Figure S6 of the supplementary material shows the Tauc plot of (αhυ)2 vs hυ, where α is the absorption coefficient and hυ is the incident photon energy. The bandgap was calculated by extrapolating the linear edge of the curve to meet the x axis (energy axis). The estimated bandgap values for CZTS, TiO2, and CdS are 1.54 eV, 3.8 eV, and 2.4 eV, respectively. Finally, the calculated CBO was found to be (−0.16 ± 0.15) eV at CZTS-CdS and (+0.17 ± 0.15) eV for the CZTS-TiO2 interface. A “spike”-like conformation is observed at the CZTS-TiO2 heterojunction interface, which lies within the optimal range of CBO, i.e., (0–0.4 eV) desired for high efficiency of CZTS solar cells.8 A schematic of the obtained energy level alignment at the CZTS-CdS and CZTS-TiO2 interface is summarized in Fig. 5(b). A “spike”-like conformation interface forbids interface recombination, specifically, recombination between electrons in the conduction band of buffer and holes in the valence band of CZTS. In contrast, “cliff”-like CBO significantly contributes to this type of recombination, particularly in the presence of interface defect states and thereby resulting in VOC deficit.8–10 Energy level alignment is one of the key areas for VOC enhancement of the device. The amorphous TiO2 providing a favorable CBO at the p-n junction may directly benefit the open circuit voltage of the device by alleviating the severe interface recombination.

Furthermore, nanoscale Kelvin probe force microscopy (KPFM) was carried out at CZTS-TiO2 and CZTS-CdS heterojunctions to probe the interface charge dynamics, which is critical for any high-performance devices. Figures 6(a) and 6(b) represent topography of the CZTS-TiO2 and CZTS-CdS interfaces, respectively. The surface potential distribution for these interfaces is shown in KPFM images [Figs. 6(c) and 6(d)]. In surface topography images, dark brown region shows CZTS and light brown part exhibits TiO2 in Fig. 6(a) and CdS in Fig. 6(b). Half side of the CZTS film was intentionally covered before depositing the buffer layer to investigate the surface potential across the interfaces, as shown in the inset of Figs. 6(a) and 6(b). A clear visible interface was formed when TiO2 and CdS were deposited over CZTS. Although the topography images do not display any noticeable change in the morphology at the interface region, a good contrast is observed for surface potential images, as shown in Figs. 6(c) and 6(d). The surface potential line scan, as shown in Figs. 6(e) and 6(f), displays the contact potential difference (CPD) between CZTS and TiO2, and CZTS and CdS interface. CPD acts as an energy barrier that has to be overcome for back carrier recombination at the interfaces. The CZTS-TiO2 interface exhibits a much higher CPD of ∼432 meV, while it is only ∼128 meV for the CZTS-CdS interface. Thus, the back carrier recombination can be significantly reduced when TiO2 will be used as a buffer layer in CZTS solar cells. The CPD is calculated by subtracting the average surface potential of CZTS from the average surface potential of TiO2 and CdS. The variation in the surface can be attributed to the inhomogeneous surface of the sample under investigation.

FIG. 6.

Topography of the (a) TiO2-CZTS interface and (b) CdS-CZTS interface. The inset shows the schematic of the CZTS-TiO2 and CZTS-CdS interfaces for KPFM measurement. Surface potential of the (c) TiO2-CZTS interface and (d) CdS-CZTS interface. Surface potential line profile across the (e) TiO2-CZTS interface and (f) the CdS-CZTS interface.

FIG. 6.

Topography of the (a) TiO2-CZTS interface and (b) CdS-CZTS interface. The inset shows the schematic of the CZTS-TiO2 and CZTS-CdS interfaces for KPFM measurement. Surface potential of the (c) TiO2-CZTS interface and (d) CdS-CZTS interface. Surface potential line profile across the (e) TiO2-CZTS interface and (f) the CdS-CZTS interface.

Close modal

The KPFM technique is widely used to investigate the junction properties for fundamental understanding to cast appropriate electron transport layer (ETL) and hole transport layer (HTL) in solar cells.47,48 Qiao et al. have used the KPFM technique to select suitable ETL and HTL in polymers and perovskite solar cells. The contact potential difference is measured at the acceptor-ETL interface and at the donor-HTL interface to adopt energetically favorable materials that facilitate charge transfer.48–51 Using this approach, Etgar et al. reported TiO2 as a favorable electron transport layer, while Al2O3 act as a scaffold in perovskite based solar cells.52 

To retrieve further information on charge dynamics, interface properties like space charge region width and electric field distribution were calculated from KPFM analysis. Electric field distribution was deduced by taking the first derivative of the contact potential difference (CPD), dCPD/dx, as shown in Fig. 7. The maximum electric field is located at the CZTS-TiO2 and CZTS-CdS interface, and the field drops down when we move away from the interface region. The maxima of electric field demonstrate the location of the p-n junction.47 The electric field of E ∼ 3.79 × 106 V/m was established at the CZTS-TiO2 heterojunction with the SCR width, WD ∼ 300 nm. While in the case of CZTS-CdS interface, the electric field was estimated to be E ∼ 1.59 × 106 V/m with the SCR width of ∼230 nm. The SCR width was calculated from the two points on the electric field curve which crosses the zero electric field line.52,53 The charge separation process at any heterojunction is assisted by electric field and space charge region. The higher electric field at heterojunctions will quickly sweep apart the photogenerated charge carriers and drift them toward collecting electrodes with lower possibility of recombination. Under weak electric field, Shockley-Read-Hall (SRH) recombinations will dominate via traps located within the forbidden gap.7 Thus, the existence of a higher electric field at the CZTS-TiO2 heterojunction will increase the charge separation efficiency compared to the CZTS-CdS interface. In the quasineutral region, the charge transport of minority charge carriers is carried out by diffusion, while the drift supervises the carrier transport within the space charge region. Therefore, the wide space charge region width at the CZTS-TiO2 interface will help in increasing the collection probability of the generated charge carriers. In comparison to the quasineutral region, the contribution of the space charge region to EQE is critically important, as most of the generated charge carriers within the space charge region get separated due to electric field.7 The concise space charge region at the CZTS-CdS junction will influence the collection probability and thus will reduce the light-generated current. The existence of the Cu-rich region near the CZTS-CdS junction leads to higher hole concentration, as discussed earlier, could be one of the reasons for the short space charge region as the width of the space charge region is inversely proportional to the carrier concentration.54Table I summarizes all the p-n junction parameters for CZTS-TiO2 and CZTS-CdS heterojunctions, calculated from the KPFM analysis. The value of the SCR width for the CZTS-CdS junction closely matches with the reported literature.11–55 The estimated built-in potential for both the interfaces was determined by calculating the area under the electric field curve from the maximum electric field up to the zero electric field line.53 The favorable p-n junction properties of the CZTS-TiO2 interface may facilitate excellent charge separation and suppress interface recombination that can directly assist in superior photovoltaic performance.

FIG. 7.

CPD profile and corresponding electric field, dCPD/dx, for the (a) TiO2-CZTS, and (b) CdS-CZTS interface. WD represents the SCR width at the p-n junction.

FIG. 7.

CPD profile and corresponding electric field, dCPD/dx, for the (a) TiO2-CZTS, and (b) CdS-CZTS interface. WD represents the SCR width at the p-n junction.

Close modal
TABLE I.

The junction parameters for CZTS-TiO2 and CZTS–CdS heterojunction interfaces.

InterfaceΔ CPD (meV) (at 10 distinct points)Max. electric field (106 V/m)Built-in potential (meV)SCR width (nm)
CZTS-TiO2 375.6–453.2 3.17 543.64 ∼300 
CZTS-CdS 100.9–136.3 1.59 230.84 ∼230 
InterfaceΔ CPD (meV) (at 10 distinct points)Max. electric field (106 V/m)Built-in potential (meV)SCR width (nm)
CZTS-TiO2 375.6–453.2 3.17 543.64 ∼300 
CZTS-CdS 100.9–136.3 1.59 230.84 ∼230 

Finally, photodetector measurements were performed at CZTS-CdS and CZTS-TiO2 heterojunctions to further examine the correlation of transient photoresponse with the energy level alignment and contact potential difference at the interfaces. The schematic diagrams of CZTS-TiO2 and CZTS-CdS based photodetector devices with Al contact under illumination are shown in Figs. 8(a) and 8(b), respectively. The time-dependent response of the device was taken by turning ON/OFF the 532 nm laser periodically under a constant −1.5 mV bias as shown in Figs. 8(c) and 8(d) to compare the decay time at two different heterojunctions (i.e., CZTS-TiO2 and CZTS-CdS). The photoconductive rise and the decay time constants of the CZTS-CdS and CZTS-TiO2 interfaces can be extracted from the transient response fit using the following equation:

I=I0+Aetτr+Betτd,
(4)

where I0 is the steady-state photocurrent, t is the time, A and B are constants, and τ is the relaxation time constant, τr and τd denote the rise and decay time constants, respectively. A repeatable photoresponse was observed with the number of cycles as can be seen in Figs. 8(c) and 8(d). The rise and decay time for CZTS-TiO2 photodetectors were found to be 35.4 ms and 54.6 ms, respectively. While CZTS-CdS photodetectors showed rise and decay time of 579.8 ms and 902.7 ms, respectively. The decay time for CZTS-CdS photodetectors was notably higher (∼16 times) than CZTS-TiO2, as shown in Figs. 8(e) and 8(f). The higher value of the decay time in the CZTS-CdS photodetector device indicates the slow decay of the generated photocurrent with time even after the removal of the light source, which manifests the persistent photoconductive effect.56 Photopersistent current (PPC) is attributed commonly due to the existence of defects, vacancies, and trap states in the semiconductor thin films and in the interface.56,57 The indication of comparative faster response time in the CZTS-TiO2 photodetector provides an insight into the minimization of trapping of charge carriers at this interface. Therefore, charge carriers at the CZTS-TiO2 interface get separated and transported efficiently as compared to the CZTS-CdS interface. It directly proves that the CZTS-TiO2 interface eventually results in the suppression of recombination, thereby improving the transient photoresponse. Moreover, the transient photoresponse of CZTS-TiO2 and CZTS-CdS heterojunction photodetectors evidently supports our arguments from XPS and KPFM analysis. It is also noticeable from Fig. 8(d) that the CZTS-CdS photodetector based device exhibits comparatively more dark current than the CZTS-TiO2 based photodetector.

FIG. 8.

Optoelectronic characterizations of as-deposited CZTS based photodetectors. (a) and (b) show the schematic of the CZTS-TiO2 and CZTS-CdS heterojunction photodetector, respectively, (c) and (d) time dependent photocurrent measurements of CZTS-TiO2 and CZTS-CdS, respectively, photodetector under illumination of a 532 nm wavelength laser, (e) and (f) represent the experimental fitting of CZTS-TiO2 and CZTS-CdS based photodetectors, respectively.

FIG. 8.

Optoelectronic characterizations of as-deposited CZTS based photodetectors. (a) and (b) show the schematic of the CZTS-TiO2 and CZTS-CdS heterojunction photodetector, respectively, (c) and (d) time dependent photocurrent measurements of CZTS-TiO2 and CZTS-CdS, respectively, photodetector under illumination of a 532 nm wavelength laser, (e) and (f) represent the experimental fitting of CZTS-TiO2 and CZTS-CdS based photodetectors, respectively.

Close modal

The higher value of dark current in the CZTS-CdS photodetector device may originate due to trap repopulation, which is detrimental for high quality heterojunctions.56 On the contrary, the shrinkage of dark current in CZTS-TiO2 photodetector devices indicates the reduction of defects and trap states, which is also suggested by the above discussed analysis.

Figure 9 shows the comparison of responsivity and decay time of CZTS-CdS and CZTS-TiO2 heterojunction photodetectors under a constant −1.5 mV bias voltage. The photoresponsivity is determined by the following equation:

R=IphIdarkPoptical.
(5)
FIG. 9.

Comparison of CZTS-TiO2 and CZTS-CdS heterojunction photodetector device performance parameters: responsivity and decay time.

FIG. 9.

Comparison of CZTS-TiO2 and CZTS-CdS heterojunction photodetector device performance parameters: responsivity and decay time.

Close modal

The photoresponsivity of about 56.6 μA W−1 and 30.3 μA W−1 was observed for TiO2 and CdS based CZTS photodetectors, respectively. The results of photodetector analysis confirm that the CZTS-TiO2 heterojunction provides a high-quality interface to encounter the requirements of high-performance optoelectronic devices.

Figure 10 summarizes the XPS, KPFM, and transient photoresponse for CZTS-TiO2 and CZTS-CdS heterojunctions. It can be seen that favorable “spike”-like conformation and maximum CPD along with faster photoresponse, together confirms that charge carrier separation and transportation will be efficient at the CZTS-TiO2 junction, whereas, CZTS-CdS interface provides “cliff”-like energy level alignment and concise SCR width that boost up the recombination of electrons from the buffer layer and holes in the CZTS valence band (VB), responsible for higher decay time in the photodetector response. Thus, TiO2 granting barrier for back carrier recombination in terms of appropriate energy level alignment measured from XPS, high contact potential difference measured from KPFM and fast decay for CZTS-TiO2 photodetectors provides an insight for the favorable electron transport layer in CZTS solar cells. Our investigation from XPS and KPFM analysis confirmed by photodetector measurements shows the potential of the TiO2 buffer layer to successfully substitute the conventional CdS layer.

FIG. 10.

CPD, SCR width, and decay time are represented by left (black) y axis, right (red) y axis, and right (blue) y axis, respectively, for CZTS-TiO2 and CZTS-CdS heterojunctions combined with the energy level alignment.

FIG. 10.

CPD, SCR width, and decay time are represented by left (black) y axis, right (red) y axis, and right (blue) y axis, respectively, for CZTS-TiO2 and CZTS-CdS heterojunctions combined with the energy level alignment.

Close modal

The open circuit voltage deficit, due to severe nonradiative interface recombinations, is one of the major bottlenecks in CZTS solar cells. In order to overcome this shortcoming, we have taken a step by employing TiO2 as a novel electron transport material. The energy level alignment at the CZTS-TiO2 and CZTS-CdS interface was determined by XPS and UPS. A favorable “spike”-like alignment with CBO 0.17 eV is observed at the CZTS-TiO2 interface alleviates the severe recombinations at the junctions. The p-n junction electrical parameters including electric field and SCR width were probed by the KPFM technique. Higher electric field and wider space charge region at the CZTS-TiO2 junction contribute to the efficient separation of charge carriers. It was further confirmed by the transient photoresponse of CZTS-CdS/TiO2 based photodetectors, where faster decay of generated photocurrent in the case of the CZTS-TiO2 heterojunction showing the suppression of recombination of charge carriers at the interface. By overcoming interface recombinations, the major challenge in CZTS solar cells, i.e., open circuit voltage deficit can be conquered, leading to a high performance of CZTS solar cells.

See Figs. S1 and S2 of the supplementary material for peak shifting data for S and Cd. Figure S3 shows the presence of the pure sulfide CdS phase. The core level band bending calculation for energy level alignment, CZTS-CdS and CZTS-TiO2, heterojunction is shown in Figs. S4 and S5. Tauc plots for different layers (i.e., CZTS, CdS and TiO2) are shown in Fig. S6.

We gratefully acknowledge the research grant from the Department of Science and Technology (DST) under Solar Energy Research Initiative (SERI), Government of India [(No. DST/TMC/SERI/FR/118)]. One of the authors (Nisika) sincerely acknowledges fellowship from the Ministry of Electronics & Information Technology (MeitY), Government of India. The U.S. authors acknowledge the financial support from the National Science Foundation (NSF) MRI program.

1.
K.-J.
Yang
,
S.
Kim
,
S.-Y.
Kim
,
K.
Ahn
,
D.-H.
Son
,
S.-H.
Kim
,
S.-J.
Lee
,
Y.-I.
Kim
,
S.-N.
Park
,
S.-J.
Sung
,
D.-H.
Kim
,
T.
Enkhbat
,
J.
Kim
,
C.-W.
Jeon
, and
J.-K.
Kang
,
Nat. Commun.
10
(
1
),
2959
(
2019
).
2.
M.
Kumar
,
A.
Dubey
,
N.
Adhikari
,
S.
Venkatesan
, and
Q.
Qiao
,
Energy Environ. Sci.
8
(
11
),
3134
3159
(
2015
).
3.
W.
Wang
,
M. T.
Winkler
,
O.
Gunawan
,
T.
Gokmen
,
T. K.
Todorov
,
Y.
Zhu
, and
D. B.
Mitzi
,
Adv. Energy Mater.
4
(
7
),
1301465
(
2014
).
4.
K.
Kaur
,
N.
Kumar
, and
M.
Kumar
,
J. Mater. Chem. A
5
(
7
),
3069
3090
(
2017
).
5.
Z. K.
Yuan
,
S.
Chen
,
H.
Xiang
,
X. G.
Gong
,
A.
Walsh
,
J. S.
Park
,
I.
Repins
, and
S. H.
Wei
,
Adv. Funct. Mater.
25
(
43
),
6733
6743
(
2015
).
6.
O.
Gunawan
,
T. K.
Todorov
, and
D. B.
Mitzi
,
Appl. Phys. Lett.
97
(
23
),
233506
(
2010
).
7.
M.
Courel
,
J.
Andrade-Arvizu
, and
O.
Vigil-Galán
,
Appl. Phys. Lett.
105
(
23
),
233501
(
2014
).
8.
R.
Haight
,
A.
Barkhouse
,
O.
Gunawan
,
B.
Shin
,
M.
Copel
,
M.
Hopstaken
, and
D. B.
Mitzi
,
Appl. Phys. Lett.
98
(
25
),
253502
(
2011
).
9.
M.
Bär
,
B.-A.
Schubert
,
B.
Marsen
,
R. G.
Wilks
,
S.
Pookpanratana
,
M.
Blum
,
S.
Krause
,
T.
Unold
,
W.
Yang
, and
L.
Weinhardt
,
Appl. Phys. Lett.
99
(
22
),
222105
(
2011
).
10.
C.
Yan
,
F.
Liu
,
N.
Song
,
B. K.
Ng
,
J. A.
Stride
,
A.
Tadich
, and
X.
Hao
,
Appl. Phys. Lett.
104
(
17
),
173901
(
2014
).
11.
J.
Li
,
H.
Wang
,
M.
Luo
,
J.
Tang
,
C.
Chen
,
W.
Liu
,
F.
Liu
,
Y.
Sun
,
J.
Han
, and
Y.
Zhang
,
Solids Energy Mater. Solids Cells
149
,
242
249
(
2016
).
12.
C.
Platzer-Björkman
,
C.
Frisk
,
J. K.
Larsen
,
T.
Ericson
,
S.-Y.
Li
,
J. J. S.
Scragg
,
J.
Keller
,
F.
Larsson
, and
T.
Törndahl
,
Appl. Phys. Lett.
107
(
24
),
243904
(
2015
).
13.
M.
Kauk-Kuusik
,
K.
Timmo
,
M.
Danilson
,
M.
Altosaar
,
M.
Grossberg
, and
K.
Ernits
,
Appl. Surf. Sci.
357
,
795
798
(
2015
).
14.
D.
Hironiwa
,
N.
Matsuo
,
N.
Sakai
,
T.
Katou
,
H.
Sugimoto
,
J.
Chantana
,
Z.
Tang
, and
T.
Minemoto
,
Jpn. J. Appl. Phys.
53
(
10
),
106502
(
2014
).
15.
K.
Sun
,
C.
Yan
,
F.
Liu
,
J.
Huang
,
F.
Zhou
,
J. A.
Stride
,
M.
Green
, and
X.
Hao
,
Adv. Energy Mater.
6
(
12
),
1600046
(
2016
).
16.
M.
Graetzel
,
R. A.
Janssen
,
D. B.
Mitzi
, and
E. H.
Sargent
,
Nature
488
(
7411
),
304
(
2012
).
17.
P. A.
Fernandes
,
P. M. P.
Salomé
, and
A. F.
Da Cunha
,
Thin Solid Films
517
(
7
),
2519
2523
(
2009
).
18.
L.-C.
Wang
and
Y.-C.
Lin
,
Appl. Surf. Sci.
354
,
14
19
(
2015
).
19.
A.
Alvarez Barragan
,
H.
Malekpour
,
S.
Exarhos
,
A. A.
Balandin
, and
L.
Mangolini
,
ACS Appl. Mater. Interfaces
8
(
35
),
22971
22976
(
2016
).
20.
C. W.
Hong
,
S. W.
Shin
,
K. V.
Gurav
,
S. A.
Vanalakar
,
S. J.
Yeo
,
H. S.
Yang
,
J. H.
Yun
, and
J. H.
Kim
,
Appl. Surf. Sci.
334
,
180
184
(
2015
).
21.
F.
Liu
,
J.
Huang
,
K.
Sun
,
C.
Yan
,
Y.
Shen
,
J.
Park
,
A.
Pu
,
F.
Zhou
,
X.
Liu
, and
J. A.
Stride
,
npg Asia Mater.
9
(
7
),
e401
(
2017
).
22.
J. J.
Scragg
,
T.
Kubart
,
J. T.
Wätjen
,
T.
Ericson
,
M. K.
Linnarsson
, and
C.
Platzer-Björkman
,
Chem. Mater.
25
(
15
),
3162
3171
(
2013
).
23.
J.
Tao
,
L.
Chen
,
H.
Cao
,
C.
Zhang
,
J.
Liu
,
Y.
Zhang
,
L.
Huang
,
J.
Jiang
,
P.
Yang
, and
J.
Chu
,
J. Mater. Chem. A
4
(
10
),
3798
3805
(
2016
).
24.
D. M.
Berg
,
M.
Arasimowicz
,
R.
Djemour
,
L.
Gütay
,
S.
Siebentritt
,
S.
Schorr
,
X.
Fontané
,
V.
Izquierdo-Roca
,
A.
Pérez-Rodriguez
, and
P. J.
Dale
,
Thin Solid Films
569
,
113
123
(
2014
).
25.
A.
Nagoya
,
R.
Asahi
, and
G.
Kresse
,
J. Phys. Condens. Matter
23
(
40
),
404203
(
2011
).
26.
W.
Hsu
,
C. M.
Sutter-Fella
,
M.
Hettick
,
L.
Cheng
,
S.
Chan
,
Y.
Chen
,
Y.
Zeng
,
M.
Zheng
,
H.-P.
Wang
, and
C.-C.
Chiang
,
Sci. Rep.
5
,
16028
(
2015
).
27.
X.
Yin
,
C.
Battaglia
,
Y.
Lin
,
K.
Chen
,
M.
Hettick
,
M.
Zheng
,
C.-Y.
Chen
,
D.
Kiriya
, and
A.
Javey
,
ACS Photonics
1
(
12
),
1245
1250
(
2014
).
28.
A.
Nagaoka
,
H.
Miyake
,
T.
Taniyama
,
K.
Kakimoto
, and
K.
Yoshino
,
Appl. Phys. Lett.
103
(
11
),
112107
(
2013
).
29.
H.
Sugimoto
,
C.
Liao
,
H.
Hiroi
,
N.
Sakai
and
T.
Kato
, paper presented at the 2013 IEEE 39th Photovoltaic Specialists Conference (PVSC).
30.
K.
Kaur
,
K.
Arora
,
B.
Behzad
,
Q.
Qiao
, and
M.
Kumar
,
Nanotechnology
30
(
6
),
065706
(
2018
).
31.
S.
Tajima
,
T.
Itoh
,
H.
Hazama
,
K.
Ohishi
, and
R.
Asahi
,
Appl. Phys. Express
8
(
8
),
082302
(
2015
).
32.
G.
Sai Gautam
,
T. P.
Senftle
, and
E. A.
Carter
,
Chem. Mater.
30
(
14
),
4543
4555
(
2018
).
33.
S.
Tajima
,
R.
Asahi
,
D.
Isheim
,
D. N.
Seidman
,
T.
Itoh
,
M.
Hasegawa
, and
K.
Ohishi
,
Appl. Phys. Lett.
105
(
9
),
093901
(
2014
).
34.
J.
Huang
,
C.
Yan
,
K.
Sun
,
F.
Liu
,
H.
Sun
,
A.
Pu
,
X.
Liu
,
M.
Green
, and
X.
Hao
,
Sol. Energy Mater. Sol. Cells
175
,
71
76
(
2018
).
35.
C.
Yan
,
J.
Huang
,
K.
Sun
,
S.
Johnston
,
Y.
Zhang
,
H.
Sun
,
A.
Pu
,
M.
He
,
F.
Liu
, and
K.
Eder
,
Nat. Energy
3
(
9
),
764
(
2018
).
36.
A.
Santoni
,
F.
Biccari
,
C.
Malerba
,
M.
Valentini
,
R.
Chierchia
, and
A.
Mittiga
,
J. Phys. D Appl. Phys.
46
(
17
),
175101
(
2013
).
37.
G. Y.
Ashebir
,
C.
Dong
,
Z.
Wan
,
J.
Qi
,
J.
Chen
,
Q.
Zhao
,
W.
Chen
, and
M.
Wang
,
J. Phys. Chem. Solids
129
,
204
208
(
2019
).
38.
R.
Pandiyan
,
Z. O.
Elhmaidi
,
Z.
Sekkat
,
M.
Abd-lefdil
, and
M. A.
El Khakani
,
Appl. Surf. Sci.
396
,
1562
1570
(
2017
).
39.
K.
Kaur
,
M.
Sood
,
N.
Kumar
,
H. H.
Nazari
,
G. S.
Gudavalli
,
T. P.
Dhakal
, and
M.
Kumar
,
Sol. Energy Mater. Sol. Cells
179
,
22
30
(
2018
).
40.
T.
Gan
,
Y.
Li
,
X.-Z.
Wang
,
X.-T.
Wang
, and
C.-W.
Wang
,
Appl. Surf. Sci.
408
,
60
67
(
2017
).
41.
J.
Yu
,
C.
Gong
,
Z.
Wu
,
Y.
Wu
,
W.
Xiao
,
Y.
Su
,
L.
Sun
, and
C.
Lin
,
J. Mater. Chem. A
3
(
44
),
22218
22226
(
2015
).
42.
A. V.
Okotrub
,
I. P.
Asanov
,
S. V.
Larionov
,
A. G.
Kudashov
,
T. G.
Leonova
, and
L. G.
Bulusheva
,
Phys. Chem. Chem. Phys.
12
(
36
),
10871
10875
(
2010
).
43.
A.
Klein
,
T.
Löher
,
Y.
Tomm
,
C.
Pettenkofer
, and
W.
Jaegermann
,
Appl. Phys. Lett.
70
(
10
),
1299
1301
(
1997
).
44.
E. A.
Kraut
,
R. W.
Grant
,
J. R.
Waldrop
, and
S. P.
Kowalczyk
,
Phys. Rev. Lett.
44
(
24
),
1620
(
1980
).
45.
H.
Yaghoubi
,
Z.
Li
,
Y.
Chen
,
H. T.
Ngo
,
V. R.
Bhethanabotla
,
B.
Joseph
,
S.
Ma
,
R.
Schlaf
, and
A.
Takshi
,
ACS Catal.
5
(
1
),
327
335
(
2014
).
46.
A.
Saha
,
A.
Moya
,
A.
Kahnt
,
D.
Iglesias
,
S.
Marchesan
,
R.
Wannemacher
,
M.
Prato
,
J. J.
Vilatela
, and
D. M.
Guldi
,
Nanoscale
9
(
23
),
7911
7921
(
2017
).
47.
C.-S.
Jiang
,
M.
Yang
,
Y.
Zhou
,
B.
To
,
S. U.
Nanayakkara
,
J. M.
Luther
,
W.
Zhou
,
J. J.
Berry
,
J.
Van De Lagemaat
, and
N. P.
Padture
,
Nat. Commun.
6
,
8397
(
2015
).
48.
N.
Adhikari
,
A.
Dubey
,
D.
Khatiwada
,
A. F.
Mitul
,
Q.
Wang
,
S.
Venkatesan
,
A.
Iefanova
,
J.
Zai
,
X.
Qian
, and
M.
Kumar
,
ACS Appl. Mater. Interfaces
7
(
48
),
26445
26454
(
2015
).
49.
M. K.
Siddiki
,
S.
Venkatesan
,
D.
Galipeau
, and
Q.
Qiao
,
ACS Appl. Mater. Interfaces
5
(
4
),
1279
1286
(
2013
).
50.
E. T.
Vickers
,
T. A.
Graham
,
A. H.
Chowdhury
,
B.
Bahrami
,
B. W.
Dreskin
,
S.
Lindley
,
S. B.
Naghadeh
,
Q.
Qiao
, and
J. Z.
Zhang
,
ACS Energy Lett.
3
(
12
),
2931
2939
(
2018
).
51.
N.
Adhikari
,
A.
Dubey
,
E. A.
Gaml
,
B.
Vaagensmith
,
K. M.
Reza
,
S. A. A.
Mabrouk
,
S.
Gu
,
J.
Zai
,
X.
Qian
, and
Q.
Qiao
,
Nanoscale
8
(
5
),
2693
2703
(
2016
).
52.
A.
Dymshits
,
A.
Henning
,
G.
Segev
,
Y.
Rosenwaks
, and
L.
Etgar
,
Sci. Rep.
5
,
8704
(
2015
).
53.
S. M.
Sze
,
Semiconductor Devices: Physics and Technology
(
John Wiley & Sons
,
2008
).
54.
D. A.
Neamen
,
Semiconductor Physics and Devices
(
McGraw-Hill
,
New York, NY
,
2012
).
55.
C.
Yan
,
K.
Sun
,
F.
Liu
,
J.
Huang
,
F.
Zhou
, and
X.
Hao
,
Sol. Energy Mater. Solid Cells
160
,
7
11
(
2017
).
56.
K.
Arora
,
N.
Goel
,
M.
Kumar
, and
M.
Kumar
,
ACS Photonics
5
(
6
),
2391
2401
(
2018
).
57.
S.
Lany
and
A.
Zunger
,
Phys. Rev. B
72
(
3
),
035215
(
2005
).

Supplementary Material