Expanding application fields and improving power conversion efficiency (PCE) of solar cells have become inescapable issues to further promote the development of photovoltaic (PV) market. Semitransparent solar cells (ST-SCs) provide an attractive choice to address aforementioned problems, and thus receive overwhelming attention. For example, ST-SCs can be utilized not only as color-tunable building integrated PV (BIPV),^[1] wearable electronics,^[2] and PV vehicles^[3] to excite future applications, but also as sub-cell of multi-junction^[4] or bifacial PV devices^[5] to convert more solar energy into electricity, and therefore improve PCE. Organic–inorganic hybrid perovskite solar cells (PSCs) have emerged as a promising candidate for ST-SCs, because opaque PSCs have achieved PCE of 25.2%^[6] with unprecedented speed attributing to its direct band gap, high absorption coefficient, low exciton binding energy, and a long electron–hole diffusion length. Simple device architecture, facile fabrication process, and higher PCE make the planar heterojunction stand out from all PSCs device structures, especially the planar n–i–p device.^[7–9] Different from opaque solar cells, both charge collection efficiency and light transmittance of HTMs are vital to avoid influencing light absorption in perovskite active layer and realize high performance of ST-PSCs.^[10] Generally, 2,2’,7,7’-tetrakis (N,N-di-p-Methoxyphenylamine)-9, 9’-spirobifluorene (Spiro-OMeTAD)^[11] or poly (triarylamine) (PTAA)^[12] are used as HTMs in most efficient PSCs. Unfortunately, both of them are unstable and expensive, which will greatly affect their wide application in the future. Besides, the serious light absorption in short-wavelength range is also harmful to light harvesting of perovskite absorption layer. Therefore, it is urgent and essential to explore alternative HTMs.^[13]

Recently, copper thiocyanate (CuSCN) is paid much attention due to its excellent material stability, extremely low price, and compatible HOMO (highest occupied molecular orbited) level with perovskite. And it is noteworthy that the hole mobility of CuSCN of 0.01 cm² · V⁻¹ · s⁻¹–0.1 cm² · V⁻¹ · s⁻¹ is remarkably higher than that of common organic HTMs, such as 10⁻³ cm² · V⁻¹ · s⁻¹ of PTAA and 4 × 10⁻⁵ cm² · V⁻¹ · s⁻¹ of Spiro-OMeTAD.^[14] A report of the PSCs based opaque CuSCN with PCE over 20%^[15] confirmed that CuSCN can be employed as excellent HTM. In addition, the low parasitic absorption also allows it as potential candidate of HTMs for ST-PSCs. For instance, Jen and co-authors^[16] adopted CuSCN as HTM instead of PEDOT:PSS to prepare ST-PSCs to obtain PCE of 10% and average visible transmittance (AVT) of 25%. Fan and co-authors^[17] fabricated bifacial semitransparent n–i–p planar PSCs, which achieved a maximum efficiency of 12.47% and 8.74% by illumination from the front side and back side, respectively. Wang and co-authors^[18] used the CuSCN and ITO as the HTMs and back electrode in the PSCs, respectively. Compared with the Spiro-OMeTAD based ST-PSCs, the CuSCN-based ST-PSCs exhibited higher PCE of 14.2% and 13.3% by illumination from the front side and back side, respectively. Despite many efforts have been made in employing CuSCN as HTM, few attentions were focused on the poor electrical conductivity and small work function of CuSCN. Those problems will cause weak built-in field within limited film thickness, and at the same time cause energy level offset at the CuSCN/perovskite interface.^[19] Consequently, it can be deduced that there will be a large amount of hole accumulation and electron–hole recombination at the interface of CuSCN/perovskite due to the low holes transfer efficiency of intrinsic CuSCN.

In this work, F4TCNQ was doped into CuSCN to be used as the HTM in CsFAMA based n–i–p planar perovskite solar cells, which optimize the energy level alignment at the I/P interface in order to achieve better hole transport. Ultraviolet photoelectron spectroscopy (UPS) and x-ray photoelectron spectroscopy (XPS) indicated that, after F4TCNQ doping, the energy level positions of CuSCN HTMs changed, in which Fermi level (E_F) decreased from −4.82 eV to −4.90 eV and valence band maximum (VBM) −E_F decreased from 0.71 eV to 0.68 eV. Therefore, the average conversion efficiency of CsFAMA mixed cation PSCs was improved by about 11%, and a PCE of 16.6% for a champion cell. In addition, for bifacial semitransparent n–i–p planar PSC using CuSCN as HTM, maximum PCE of 14.8% and 12.5% were achieved by illuminated from the front side and the rear side, respectively.

The device schematic diagram of the undoped and doped CuSCN are illustrated in Fig. 1(a). At present, in order to enhance the characteristics of hole transport layers, lots of strategies were conducted,^[20–22] such as He and coauthors^[23] found that Cu doping NiO results in increased work function and better energy levels alignment with enhanced device performance; Liao and coauthors^[24] proved that the incorporation of copper salts achieved a p-type doping with efficient charge transfer complex, which resulted in improved film conductivity and hole mobility in spiro-OMeTAD:copper salts composite films. As a result, the PCE is largely improved from 14.82% to 18.02%. There are many methods to control the carrier concentrations in HTMs, and doping is one of the most effective methods.^[25,26] Strong electron acceptor of F4TCNQ can be easily ionized by attracting electrons.^[27] With the measurement of UV–vis absorption spectroscopy, the optical bandgap values of undoped CuSCN and doped CuSCN are determined to be 3.66 eV and 3.67 eV, respectively (see Fig. A1 in Appendix A). To explore the mechanism, electronic structures of CuSCN with and without F4TCNQ are probed with UPS.^[28] As shown in Fig A2(a), the valance band spectra show that after doping F4TCNQ, the difference between E_F and VBM of CuSCN narrows from 0.71 eV to 0.68 eV, indicating the hole concentration is enhanced. From the UPS measurement, the work functions (ϕ) of the undoped and doped CuSCN are calculated to be 4.82 eV and 4.90 eV, respectively (see Fig. A2(b)). The VBM of the doped CuSCN (5.58 eV) went deeper than undoped CuSCN (5.53 eV) as referred to vacuum level. From this result, we can draw the conclusion that the transferring of electrons from CuSCN to F4TCNQ was successful, which is consistent with some reports of F4TCNQ.^[29] The negative shifts of E_F and VBM effectively reduce the energy level offset between CuSCN and perovskite absorbers and enhance built-in field strength, and hence improve the device performance. From the XPS measurements, the Cu 2p core level peaks have unambiguous shifts about 0.2 eV after doping F4TCNQ (Fig A3), and the core level peaks of S 2p, C 1s, and N 1s also have unambiguous shifts about 0.1 eV–0.4 eV after doping F4TCNQ (Fig. A3) respectively, which further make us confirm that electron transfer from CuSCN to F4TCNQ happens.^[30]

Fig. 1.

	Figure Option ViewDownloadNew Window
	Fig. 1. (a) Schematic device architecture. (b) The energy band structure diagram of PSCs.

Fig. A1.

	Figure Option ViewDownloadNew Window
	Fig. A1. The relationship between (αhv)2 and energy of CuSCN and 3% F4TCNQ-doped CuSCN film. The band gap can be determined by tangent of the curve to the base line. It can be found that the band gaps of the CuSCN and 3% F4TCNQ-doped CuSCN are about 3.66 eV and 3.67 eV.

Fig. A2.

	Figure Option ViewDownloadNew Window
	Fig. A2. (a) Valance band spectra of the CuSCN and F4TCNQ-doped CuSCN. (b) UPS spectra of the CuSCN and F4TCNQ-doped CuSCN, measured under −21.22-eV bias. Band bending at CuSCN-perovskite interface of before (c) and after (d) doping F4TCNQ.

Fig. A3.

Figure Option ViewDownloadNew Window

Fig. A3. XPS spectra of CuSCN and F4TCNQ-doped CuSCN for the core level of Cu 2p (a), S 2p (b), C 1s (c), and N 1s (d); the core level of Cu 2p1/2 shifted ≈ 0.22 eV and Cu 2p3/2 shifted ≈ 0.27 eV after molecular doped; the core level of S 2p shifted ≈ 0.34 eV after molecular doped; the core level of C 1s shifted ≈ 0.15 eV after molecular doped; the core level of S 2p shifted ≈ 0.26 eV after molecular doped.

Based on the aforementioned problem, the band diagram of the CuSCN/perovskite in the PSCs is shown in Fig. 1(b) and the corresponding band bending at CuSCN-perovskite interface of before and after doping F4TCNQ is shown in Figs. A2(c) and A2(d), where the VBM and E_F of CsFAMA perovskite is set at 5.65 eV and at the middle of the band gap, respectively.^[23,31] It can be clearly seen that the energy level offset between CuSCN VBM and perovskite VBM for charge transfer is decreased from 0.12 eV to 0.07 eV. Owing to the simultaneous reduction of its work function, the band bending of CuSCN/perovskite interface was enlarged 0.08 eV, doped CuSCN/PVK interface present accelerated charge transfer and reduces holes accumulation, resulting in lower charge recombination and higher charge transport efficiency from perovskite.

Steady-state photoluminescence (PL) and time-resolved photoluminescence (TRPL) decay measurements are used to understand the hole extraction capability and recombination characteristics of the devices with different HTMs. As shown in Fig. 2(a), for the doped CuSCN, a stronger quenching effect is observed, indicating an enhanced hole extraction capability. The measurement of TRPL can detect the time intensity change of radiative processes. The TRPL decay can be ascribed to the perovskite bulk radiative or nonradiative recombination of electrons and holes, charge capture at the surface defects of perovskite surface, or the transfer of hole from perovskite to HTM. The measured TRPL curves are shown in Fig. 2(b). Fitting with exponential, results are shown as equation as follows:

Fig. 2.

	Figure Option ViewDownloadNew Window
	Fig. 2. Steady-state (a) and time-resolved (b) photoluminescence of glass/perovskite/CuSCN and glass/perovskite/F4TCNQ-doped CuSCN; SCLC of (c) ITO/NiOx/PVK/F4TCNQ-undoped CuSCN/Au, and (d) ITO/NiOx/PVK/F4TCNQ-doped CuSCN/Au.

Table A1.

Table A1.

Table A1. Summary of the carrier lifetime from fitting curves of the TRPL measurements. . Sample τ1/ns A1/% τ2/ns A2/% Glass/CsFAMA/undoped CuSCN 39.68 7.74 213.49 92.26 Glass/CsFAMA/doped CuSCN 25.46 3.63 328.81 96.37

Table A1. Summary of the carrier lifetime from fitting curves of the TRPL measurements. .

The dark current–voltage (J–V) characteristics of hole-only devices based on ITO/NiO_x/perovskite/undoped or doped CuSCN/Au were used to evaluate carrier recombination at the interface of CuSCN/Perovskite. The space-charge-limited currents (SCLC) curves is used to measure the trap state density state (n_trap). As shown in Figs. 2(c) and 2(d), the dark J–V curve can be divided into three parts. In the case where the applied voltage is higher than the voltage at the first kink point (n > 3), it can be concluded that all the trap states are filled with injected carriers, because the current exhibits a rapid non-linear growth state. The applied voltage at the kink point is the trap-filled limit voltage (V_TFL) defined by the trap state density^[34]

The photovoltaic performance of the CsFAMA perovskite PSCs with undoped and doped CuSCN HTMs is analyzed and is shown in Fig. 3. It can be seen that the cell based on doped CuSCN achieves a considerably higher short-circuit current density (J_sc), open-circuit voltage (V_oc), and fill factor (FF) and thus higher PCE. The results of the J–V curves of the best devices indicate that using F4TCNQ-doped CuSCN as HTMs can significantly improve device performance (Fig. 3(a)). Finally, the champion cell with PCE of 16.6% is obtained, while the undoped device achieves an optimal PCE of 13.5%. Table 1 summarizes all photovoltaic parameters. The EQE spectra are shown in Fig. 3(b). Devices using doped CuSCN as HTMs in the range from 300 nm to 800 nm have higher EQE values compared to reference devices, which indicates that doped CuSCN has stronger charge extraction and transporting capabilities. The photocurrents of the undoped and doped CuSCN based solar cells calculated from the results of EQE are 19.5 mA/cm² and 20.3 mA/cm², respectively.

Fig. 3.

Figure Option ViewDownloadNew Window

Fig. 3. (a) J–V curves from forward-biased (FB) to short-circuit (SC) of the optimal CsFAMA PSCs using CuSCN or F4TCNQ-doped CuSCN as HTMs under 1-sun illumination. The insert is the cross-sectional scanning electron microscope image of PSCs; (b) EQE and integrated Jsc spectra of the corresponding optimal devices. (c) Normalized PCE of PSC (undoped CuSCN and doped CuSCN) as a function of aging time at 25 °C in N2 atmosphere. (d) Statistics of PCE for CuSCN and F4TCNQ-doped CuSCN devices (25 cells are included).

Table 1.

Table 1.

Table 1. Summary of photovoltaic parameters of n–i–p planar PSCs based on doped and undoped CuSCN HTMs. . Devices scan direction Jsc/(mA/cm2) Voc/V FF/% PCE/% Rs/(Ω/cm2) Rsh/(kΩ/cm2) Undoped CuSCN reverse 20.1 1.06 61.3 13.1 8.6 2.3 forward 20.3 1.06 63.1 13.5 8.2 2.4 Doped CuSCN reverse 21.1 1.10 68.8 16.0 5.1 4.6 forward 21.5 1.10 70.5 16.6 4.5 4.9

Table 1. Summary of photovoltaic parameters of n–i–p planar PSCs based on doped and undoped CuSCN HTMs. .

The long-term stability of these PSCs can also be evaluated by the J–V characteristics monitoring of the devices stored in glove box (RH < 10%) at room temperature. As shown in Fig. 3(c), after 200 days, 88% of the initial PCE remains for the doped CuSCN, while only 76% remains for the undoped CuSCN. And we also conducted experiments about the steady-state PCEs for the devices with and without doping, which are 16.5% and 13.2%, respectively (Fig. A6). The 3% F4TCNQ is an optimal doping ratio (Fig. A4 and Table A2 in Appendix A), the statistics performances for 25 devices with 3% F4TCNQ doped or undoped devices are presented in Fig. 3(d) and Fig. A5. The averaged PCE for undoped devices is 13.1%, about 8.5% lower than that of the devices with F4TCNQ-doping (14.5%). From these results, the most important improvement is that V_oc increases from 1.06 V to 1.10 V (Fig. A7). In addition, after using F4TCNQ-doped CuSCN as HTMs, the series resistance (R_s) of device decreases, and the shunt resistance (R_sh) increases (Table 1). The increase of V_oc and FF means that the recombination at the CuSCN/perovskite interface can be reduced by doped CuSCN. The current–voltage curves measured under both reverse and forward scan directions were shown in Fig. A6, the parameters of the devices as shown in Table 1. And we also calculated the J–V hysteresis factor, which defined by (PCE_Reverse-PCE_Forward)/PCE_Reverse,^[36–38] the hysteresis factor of the champion PSCs based on doped and undoped CuSCN are 0.030 and 0.036, respectively. Both devices with negligible hysteresis (Fig. A7 and Table 1).

Fig. A4.

	Figure Option ViewDownloadNew Window
	Fig. A4. Forward scan J–V curves of the optimal CsFAMA PSCs with F4TCNQ-doped CuSCN HTM under 1-sun illumination.

Fig. A5.

	Figure Option ViewDownloadNew Window
	Fig. A5. Statistics of device parameters (Jsc, Voc, and FF) for PSCs based on CuSCN or F4TCNQ-doped CuSCN HTM (25 cells are included).

Fig. A6.

	Figure Option ViewDownloadNew Window
	Fig. A6. Steady photocurrent output at fixed bias of Vmpp and steady PCE of the corresponding optimal devices.

Fig. A7.

	Figure Option ViewDownloadNew Window
	Fig. A7. Reverse and forward J–V curves of the optimal CsFAMA PSCs using (a) CuSCN and (b) F4TCNQ-doped CuSCN as HTMs under 1-sun illumination.

Table A2.

Table A2.

Table A2. The device performance with changed F4TCNQ concentration. . HTM Averaged Jsc/(mA/cm2) Averaged FF/V Averaged FF/% PCE/% Average Max CuSCN +0% FTCNQ 20.25 ± 0.66 1.02 ± 0.04 63.2 ± 3.5 13.05 ± 0.85 13.58 CuSCN +1% F4TCNQ 20.52 ± 0.68 1.06 ± 0.04 64.5 ± 4.5 14.03 ± 0.96 15.88 CuSCN +3% F4TCNQ 20.86 ± 0.85 1.06 ± 0.04 65.5 ± 4.5 14.48 ± 1.11 16.61 CuSCN +5% F4TCNQ 20.44 ± 0.72 1.06 ± 0.04 63.2 ± 3.5 13.69 ± 1.02 14.57 CuSCN +7% F4TCNQ 19.85 ± 0.75 1.02 ± 0.04 60.3 ± 3.5 12.03 ± 0.88 12.66

Table A2. The device performance with changed F4TCNQ concentration. .

As depicted in Fig. 4(a), compared with the reference device, the dark current density of the device with doped CuSCN is about one order of magnitude lower. This shows that using doped CuSCN as HTM can effectively inhibit the leakage current density of perovskite. Therefore, photo-generated carriers are dissociated effectively in the photovoltaic process. In addition, according to Eq. (3), V_oc can be increased by reducing the leakage current density^[39]

Fig. 4.

	Figure Option ViewDownloadNew Window
	Fig. 4. (a) Dark J–V curves of PSCs with CuSCN or F4TCNQ-doped CuSCN; SEM [(b) and (c)] and AFM [(d) and (e)] topographic images of CuSCN and F4TCNQ-doped CuSCN on the CsFAMA perovskite films. The scale bar is 1 μm for each SEM image and the scan scale is 10 μm × 10 μm for AFM images. (f) Current–voltage curves of CuSCN and F4TCNQ-doped CuSCN thin films.

The vertical conductivity of the undoped and doped film are measured by J–V test system. Figure 4(f) shows that the conductivity of the doped CuSCN film is 2.51 × 10⁻⁴ S/cm, which is almost 2 times higher than the undoped CuSCN film (1.48 × 10⁻⁴ S/cm). The improvement of the conductivity of the HTM film will lead to the reduction of the R_s, thus improving the charge collection efficiency meanwhile improving the fill factor.^[40]

As shown in Fig. 5(a), the p-type CuSCN HTM shows excellent transparency throughout the entire UV–vis–NIR range. In addition, the F4TCNQ-doped CuSCN demonstrates almost equivalent transmittance compared with undoped sample. This high transparency ensures that the perovskite absorption layer will have enough light under illumination from HTL side, and thus potentially leads to higher photocurrent in n–i–p ST-PSCs. Accordingly, the wide bandgap CuSCN (E_g = 3.67 eV) may be used as an excellent transparent HTM, applied on the bifacial ST-PSCs. Figure 5(b) shows the structure of a bifacial semitransparent n–i–p planar PSCs: ITO-glass/SnO₂/perovskite/CuSCN/MoO_x (9 nm)/ITO(100 nm)/Au. The structure can harvest sun-light from both sides.^[41] To investigate the effect of F4TCNQ doping on charge collection in ST-PSCs, the J–V curves of ST-PSCs with illumination from CuSCN and ITO/glass sides are shown in Figs. 5(c) and 5(d), and Table 2 also lists the corresponding detailed parameters. The parameters of bifacial semitransparent device with doped CuSCN indicate a V_oc of 1.06 V, a J_SC of 20.4 mA/cm², a fill factor of 68% and a PCE of 14.8% by illuminating from ITO side, and a V_oc of 1.02 V, a J_SC of 19.2 mA/cm², an FF of 64% and a PCE of 12.5% by illuminating from doped CuSCN side. The corresponding EQE of PSCs is shown in Figs. 5(e) and 5(f). Compared with the undoped CuSCN device, the semi-transparent device with doped CuSCN exhibits higher PCE and external quantum efficiency values in the range from 300 nm to 800 nm.

Fig. 5.

	Figure Option ViewDownloadNew Window
	Fig. 5. (a) Transmittance spectra of a bare ITO/glass substrate, ITO-glass/undoped CuSCN substrate, and ITO-glass/doped CuSCN substrate, respectively; (b) ST-PSCs with two illumination directions. (c)–(f) J–V measurements and EQE spectra of the ST-PSCs with illumination incident from ITO/glass and transparent electrode side, respectively. The integrated Jsc for both sides are given in the legend.

Table 2.

Table 2.

Table 2. The J–V parameters for ST-PSCs with light incident from ITO/glass and transparent electrode side, respectively. . Sample Light incident direction Devices scan direction Jsc[EQE]/(mA/cm2) Jsc/(mA/cm2) Voc/V} FF/% PCE/% Undoped CuSCN incident from reverse 18.84 19.9 1.02 54.7 11.1 ITO/glass side forward 19.7 1.02 64.2 12.9 incident from reverse 17.82 18.9 0.98 54.1 10.1 transparent electrode side forward 18.7 0.98 60.1 11.1 Doped CuSCN Incident from Reverse 19.42 20.5 1.02 68.1 14.3 ITO/glass side forward 20.4 1.06 68.0 14.8 incident from reverse 18.24 19.4 1.02 58.9 11.6 transparent electrode side forward 19.2 1.02 63.6 12.5

Table 2. The J–V parameters for ST-PSCs with light incident from ITO/glass and transparent electrode side, respectively. .

In summary, a molecular doping strategy is used to tailor the electronic properties of inorganic CuSCN hole transporting layer. The doped CuSCN has a superior conductivity and well-matched energy level with perovskite, which is beneficial for hole transferring from perovskite to doped CuSCN and reducing the recombination at perovskite/CuSCN interface. As a result, the champion opaque cells with doped CuSCN exhibits PCE of 16.6%, J_sc of 21.5 mA/cm², V_oc of 1.10 V, FF of 70.5%. When employing as HTM in bifacial semitransparent n–i–p planar PSCs, the maximum efficiency of 14.8% and 12.5% are achieved by illuminating from the front side and rear side, respectively. This work can be used for reference to improve the efficiency of ST-PSCs based on CuSCN and will promote the development of BIPV.