Solution-processed TiO2 as a hole blocking layer in PEDOT:PSS/n-Si heterojunction solar cells

The junction properties at the solution-processed titanium dioxide (TiO2)/n-type crystalline Si(n-Si) interface were studied for poly(3,4-ethylenedioxythiophene):poly(styrene sulfonate) (PEDOT:PSS)/ n-Si heterojunction solar cells by the steady-state photovoltaic performance and transient reverse recovery characterizations. The power conversion efficiency could be increased from 11.23% to 13.08% by adjusting the layer thickness of TiO2 together with increasing open-circuit voltage and suppressed dark saturation current density. These findings originate from the enhancement of the carrier collection efficiency at the n-Si/cathode interface. The transient reverse recovery characterization revealed that the surface recombination velocity Swas ∼375 cm/s for double TiO2 interlayer of ∼2 nm thickness. This value was almost the same as that determined by microwave photoconductance decay measurement. These findings suggest that solution-processed TiO2 has potential as a hole blocking layer for the crystalline Si photovoltaics.

The present study demonstrates the potential of solution-processed TiO 2 as an HBL to improve the photovoltaic performance of PEDOT:PSS/n-Si/TiO 2 double heterojunction solar cells. The junction properties of n-Si/TiO 2 cathode interfaces are also investigated in terms by transient reverse recovery T rr measurement to determine the effective surface recombination velocity S at n-Si/ TiO 2 interface.
2 Experimental procedure 2.1 Solution-processed TiO 2 and device fabrication Figure 1 shows the molecular structure of PEDOT:PSS and device structure of PEDOT:PSS/n-Si/TiO 2 double heterojunction solar cells. Both-side-polished 2 Â 2-cm 2 n-type (100) CZ c-Si wafers (1-5 V cm) with a thickness of 250 mm were used as the base substrate. Prior to the film deposition, the n-Si substrates were ultrasonically cleaned with acetone, isopropanol, and DI-water for 10 min each, followed by 5 wt.% HF aq treatment for 3 min to remove the native oxide. As a first step, a solution of PEDOT:PSS (prepared from Clevios R PH1000 by adding ethylene-glycol and capstone fluorosurfactant in a ratio of 93:7:0.16 wt.%) was spin-coated (SC) on top of the cleaned n-Si substrate, followed by thermal annealing at 140°C for 30 min to remove the residual solvent. Then Ag grid electrodes were screen printed at the top of the PEDOT:PSS. In a next step, a precursor solution of titanium tetraisopropoxide [Ti (OCH(CH 3 ) 2 ] 4 :TiP) diluted in isopropyl alcohol at three different concentrations of 0.5, 1, and 2 mg/ml was spincoated at 3000 rpm for 40 s on the rear side of the n-Si followed by thermal annealing at 140°C for 10 min to remove the residual solvent. The hydrolysis reaction described below was applied to synthesize titanium dioxide on the n-Si substrate as an HBL [34].
Two types of device structures were fabricated as shown in Figure 1b. One is a single layer of PEDOT:PSS (80 nm)/n-Si/TiO 2 double heterojunction solar cells of 1, 2, and 3 nm thickness TiO 2 , formed by adjusting the solution concentration on the top of the Ag grid electrode, to understand the thickness effect of TiO 2 on cathode interface. The other is the alternate coating of TiO 2 layers to suppress the junction area at the Ag/n-Si contact. This structure was fabricated by first forming a 1-nm-thick TiO 2 layer on the n-Si substrate, followed by a screen print of the Ag grid electrode. Then, another 2-nm-thick TiO 2 was spin coated on top of the Ag grid/TiO 2 /n-Si structure. Finally, the Al was evaporated in from the entire area of the rear side to form the cathode electrode.

Characterizations
The junction properties at the TiO 2 /n-Si interface were evaluated using atomic force microscopy (AFM), X-ray photoemission spectroscopy (XPS), effective minority carrier lifetime t eff , and the electroluminescence in solar cell under dark current injection in the forward bias condition.

XPS study
XPS measurements were performed for the Ti 4+ peak with a binding energy of 458.6 eV for 2p 3/2 and 464.7 eV for 2p 1/2 and the Si(2p) line region at 99.4 ± 0.3 eV using a monochromatized Al K a radiation of hn = 1486.6 eV [AXIS-Nova (Kratos Analytical)]. The formation of suboxides at the TiO 2 /n-Si interface was examined by deconvolution including metallic Si, Si + , Si 2+ and Si 4+ complexes in the 100-104 eV regions. The effect of Al metallization on the Al/TiO 2 /n-Si interface was examined by depositing Al a few nanometers thick by evaporation.

Carrier lifetime
The PEDOT:PSS and TiO 2 layers on n-Si n-type c-Si(1-5 V · cm) substrates were examined through a 2D map of minority carrier lifetime t eff measurements (SLT-1410A, KOBELCO). TiO 2 layers with different thicknesses were spin coated by adjusting the solution concentration and then thermally annealed at 140°C for 10 min prior to the lifetime measurement.

Characterization of solar cells
The current density-voltage (J-V) characteristics were measured in the dark and under exposure with simulated solar light of AM1.5G, 100 mW/cm 2 [Bunkoukeiki (CEP-25BX)]. The light exposure area was masked using a shadow mask to avoid the light leakage. The photovoltaic performance was studied using a 2 Â 2-cm 2 device under simulated of AM1.5 solar light exposure at 25°C. The short-circuit current density J sc , open-circuit voltage V oc , fill factor FF and power conversion efficiency PCE were determined from the photocurrent density-voltage (J-V) curves. The external quantum efficiency EQE was also measured with and without bias light exposure. The twodimensional (2D) map of EQE at 1000 nm was also characterized for devices with a 2 Â 2 cm 2 area using a Lasertec: MP Series.

Transient reverse recovery T rr measurement
T rr , unlike m-PCD, does not is need to use both sides of the symmetric TiO 2 coated samples to determine S. Hence, it can be used to determine the recombination velocity of the complete solar cell device structure. Figure 2 presents (a) the circuit diagram used for the T rr study and (b) the expected output current. Here, V ts is the transient bias source, R L (100 V) is the external load resistance, the blue dashed line area represents the simple equivalent circuit of the solar cell device, and R s and R sh are equivalent series and shunt resistance. The details of the T rr measurement are described in references [35,36]. First, a positive V ts higher than the built-in potential is applied to the circuit to achieve the steady forward current level I D and I sh . Then, a reverse bias is applied to the device under test and the time of recovery to a steady state was monitored by combining a programmable rectangle wave (WW2074 model of Tabor Electronics) of 1 KHz and the digital storage oscilloscope (DSO7054A model of Agilent Technologies) signal. The amount of stored charge inside the bulk can be calculated by: where, I is the maximum recovery current and t s is the storage time. Assuming that I si/HBL (t si/HBL ) and I si (t si ) are the transient currents (storage times) for devices with and without HBLs, then the storage charge ratio Q ratio can be determined by: If; Q ratio can be obtained from the diffusion coefficient D p and recombination velocity S as follows: Thus, S can be calculated by determining Q ratio without calculating the exact amount of excess hole density and the effect of bulk recombination. The t s value was also calculated by m-PCD using the following well-known equation to confirm the reliability of the S value [37]: where W is the thickness of the Si substrate and D is the minority carrier diffusion constant of n-Si.

Results and discussion
3.1 Solution-processed TiO 2 Figure 3a shows the AFM image and line profile of 2-nmthick TiO 2 spin coated from the precursor with 1 mg/ml concentration on an n-Si wafer. The RMS value was 0.215 nm in the 5 Â 5 mm 2 area, which value was almost same with that of ALD. In Figure 3b, the 2-dimensional map of t s is shown for 2-and 10-nm-thick TiO 2 . About 4 times higher average lifetime value was observed for the ∼2-nm-thick TiO 2 coated device compared to the bare-Si (∼7 ms), with slight non-uniformity, this non-uniformity may come from partial Si surface exposure to air due to the ultrathin TiO 2 layer. The ∼10 nm-thick-TiO 2 coated sample shows comparatively uniform and 5∼6 times higher lifetime value with respect to bare silicon. Although these lifetime values are much lower than the PEDOT:PSS value (∼230 ms) (Fig. 3c), which implies that the passivation level was worse compared to the PEDOT:PSS/n-Si anode interface. Thus, recombination properties of PEDOT: PSS/n-Si/TiO 2 structure is mostly dominated by cathode (Si/TiO 2 ) interface.  Table 1. J sc increased from 27.53 to 30 mA/cm 2 with increasing FF and V oc for TiO 2 thicknesses of 1 and 2 nm. This is due to the lowering of the work function of Al by inserting a TiO 2 layer as well as the enhanced hole blocking capability at the cathode interface. A large number of holes diffuse backward inside the bulk Si. As a result, the PCE increased from 11.23% for the pristine device to 13.08% for a TiO 2 HBL device on a plain substrate with a TiO 2 thickness of ∼2 nm. Figure 4b presents the EQE for PEDOT:PSS/n-Si devices with and without a 2-nm-thick TiO 2 HBL double layer. The inset shows the normalized EQE of the corresponding device. The EQE at the n-Si/cathode interface region corresponding to a wavelength of ∼1000 nm increased for the double-layer TiO 2 inserted device more than for the single-layer device. These findings originate from the reduction of carrier recombination at the Si/cathode interface. In addition, electroluminescence images at the far infrared region due to dark current injection from the cathode interface for the devises are compared (Fig. 4c). The emission image is more intense for   the device with a TiO 2 HBL than that without an insert HBL, suggesting the increased electron injection from the cathode by a TiO 2 HBL. Figure 5 shows the T rr current of a PEDOT:PSS/n-Si/Ag (Al) solar cell with different injection current (forward) levels. The T rr current increased with increasing forward current level together with an extended recovery time. This is because the number of diffused minority carriers (hole) pushed out from PEDOT:PSS to the bulk n-Si is higher for higher injection currents. Thus, the T rr study monitors the diffused (from PEDOT:PSS) minority carrier (hole) inside the bulk n-Si blocked at the cathode interface. Figure 6b shows the T rr current of PEDOT:PSS/n-Si heterojunction solar cells with 2-nm-thick single-and double-layer TiO 2 , as shown in Figure 1b. The hole storage time is ∼2 and 2.8 times longer for single-and double-layer devices, respectively, compared to the pristine device without a TiO 2 layer. The amount of stored charge is calculated by multiplying the corresponding t s with the maximum transient reverse current. An S of ∼750 cm/s is obtained for the single-layer TiO 2 inserted device, in which a 15.5% back area of the Si surface is in direct contact with metal (Ag). This value is in a good agreement with the S value measured by conventional m-PCD. An S value of ∼375 cm/s was obtained for the device with a coating alternating of TiO 2 layers (Fig. 1b). To understand the reliability of this value obtained from the T rr , a m-PCD measurement was performed using PEDOT:PSS and TiO 2 coated n-Si samples at both front and rear sides of c-Si substrate. The S of ∼700 cm/s and ∼60 cm/s were obtained   from both sides of the TiO 2 (2 nm) and PEDOT:PSS (80 nm) coated n-Si (1-5 V cm) substrates respectively, which suggest that the photovoltaic performance is largely determined by the cathode interface. However, these S values of the TiO 2 HBL devices are still higher than for PE-CVD SiN x or a-Si devices. This is because the thinner TiO 2 of ∼2 nm reacts with the underlying TiO 2 during the Al metallization. Figures 7a  and 7b show the XPS Ti(2p) core energy region of TiO 2 on n-Si before and after Al metallization. Compared to the spectrum of pristine TiO 2 , two additional peaks at 457.3 eV and 463 eV appeared, which originated from the Ti 3+ oxidation state. These findings suggest that TiO 2 react with the Al during the evaporation and forms a Ti-O-Al complex oxide, which degrades the hole-blocking ability and passivation quality of the TiO 2 layer.

Summary and conclusions
The junction properties at the solution-processed TiO 2 /n-Si interface were studied using PEDOT:PSS/n-Si heterojunction solar cells. A PCE of 13.08% was obtained for PEDOT:PSS/n-Si/TiO 2 double heterojunction solar cells by adjusting the TiO 2 layer thickness at the n-Si/Ag interface with increased J sc and V oc . These findings originate from the efficient carrier collection at the n-Si/ cathode interface, although surface recombination at the cathode interface dominate the photovoltaic performance. Trr provides the S value using the solar cell device structures with no need to examine both sides of TiO 2 coated c-Si.
This study was supported in part by the Ministry of Education, Culture, Sports, Science and Technology (MEXT) and the Takahashi Industrial and Economic Research Foundation.