Perovskite solar cells are considered to be promising photovoltaic devices thanks to their excellent properties, such as low cost^[1], high efficiency^[2], and solution processability^{[3, 4]}. To further boost the photo-conversion efficiency (PCE), both the hole transport layer (HTL)1)^{[5, 6]} and electron transport layer (ETL)^[7-10] are employed in perovskite solar cells (PSC). This interfacial engineering could effectively and sufficiently extract the photo-generated holes and electrons from the perovskite active layer and transport them to the corresponding electrodes; thus, PSCs have reached certified efficiencies of over 22.7%^[11]. To further improve the PCEs of PSCs, the interfacial engineering of the HTL for hole extraction, transport, and collection from the perovskite layer is an ongoing focus^[12–16]. One example is a device structure of ITO/PEDOT:PSS (poly(3,4ethylenedioxythiophene) polystyrene sulfonate)/perovskite/PC61BM/Ag, where PEDOT:PSS are used as the HTL^{[17, 18]}. However, due to electrical inhomogeneity that results from the electrical conductivity gradient from the near surface region to the bulk, and the large microstructure in PEDOT:PSS film morphology (order-of-magnitude variations), the ability of charge transfer within the PEDOT:PSS layer is weakened^[19]. Therefore, the electrical conductivity of the PEDOT:PSS layer is lower than that of common metal cathodes^[1]. In the PSCs, the poor conductivity of HTL leads to an unbalanced charge carrier transport and the accumulation of charge carriers^[20-23]. In addition, the poor electrical conductivity of the HTL layer causes a large series resistance, which results in energy loss during the process of charge carrier transport within the PEDOT:PSS HTL and diminishes the photocurrent and the PCEs of PSCs^{[5, 24]}. An effective hole transport layer with a good conductivity may further improve the photocurrent and FF of PSCs. However, the relative research on optimizing the conductivity of PEDOT:PSS HTL has rarely been reported^{[25, 26]}.

In our previous work^[27], it was found that doping MoO_x in PEDOT:PSS could improve the coverage of the perovskite layer and then obtain the better PCEs. However, the effect of doped MoO_x on electrical conductivity and interface polarization of HTL hs not received much attention. In this work, we further found that a trace amount of MoO_x not only led to a dramatical enhancement of electrical conductivity of the PEDOT:PSS HTL, which provided an efficient pathway for hole extraction, transport, and collection from the perovskite active layer to the ITO anode, but also resulted in an enhanced interface induced polarization between PEDTOT:PSS layer and perovskite. The short-circuit current density (J_sc) was significantly enhanced (from 17.75 to 19.64 mA/cm²), the open-circuit voltage (V_oc) had a slight increase (from 0.96 to 1.01 V), and the FF was significantly enhanced from 0.65 to 0.76, which all resulted in a dramatically enhanced PCE of 15.01%. Compared with the reference PSCs, the PCE had an ~36% enhancement. Moreover, the PSCs with the PEDOT:PSS-MoO_x HTL possessed high device reproducibility. We attributed these experimental results to the fact that the Mo⁵⁺ could enhance the interfacial polarization of PEDOT:PSS, which possesses a higher conductivity and a larger dielectric constant. The higher interfacial polarization between the perovskite and HTL could facilitate the separating of photo generated holes and electrons and thus suppress the recombination between the traps and free charges. Therefore, the obtained results give an effective strategy to build efficient hole transporting materials for improving the photovoltaic performance of PSCs.

First, 0.2 g of molybdenum powder (purchased from Aladdin) was dispersed in 20 mL ethanol and oscillated for 30 min in an ultra-sound bath. Then 0.7 mL hydrogen peroxide (H₂O₂) (30%) solution was added into the mixture. Ethanol was used to control the reaction rate between molybdenum powders and H₂O₂. After 20 h of reaction, the mix solution turned from gray to blue. Finally, the mix solution was vaporized in a dry box.

The detailed processing and characterization of planar perovskite devices have been reported in our previous work^[27]. The MoO₃ was acquired from Sigma-Aldrich (99.999%). It is well-known that the commercially available MoO₃ nanoparticles are slightly soluble in water^[28]. However, we found it easy to add into the PEDOT:PSS solution. The ability of MoO₃ as a carrier-selective material depends on its specific electronic properties^[29]. Since its energy bandgap (E_gap) lies within the O_2p- and metal d-bands, its conductivity and work function value depend on the occupancy of the d-states. The N-type MoO_x semiconductivity depends on the intrinsic oxygen vacancies in the atomic structure of MoO_3–x, that is, the fully stoichiometric configuration of MoO₃ is insulators and MoO₂ is metallic-like conductors^[30].

The conductivity of HTL films with 0–2 mg/mL MoO_x doped in PEDOT:PSS was tested using a four-point probe (Fig. 1(a)). Deposition and measurement of the films were performed in atmosphere under conditions. A significant increase in conductivity from 2.1 × 10⁻⁴ to 2.4 × 10⁻³ S/cm was observed. In addition, the conductivity of PEDOT:PSS-MoO_x was obviously better than that of the PEDOT:PSS-MoO₃. This indicates that the enhanced conductivity of HTL films is linear to MoO_x concentration. It is well-known that the number of O vacancies in MoO_x is larger than that of fully stoichiometric configuration of MoO₃. Therefore, the fact of enhancing conductivity of PEDOT:PSS-MoO_x is consistent to the previous report that the O vacancies determine the conductivity of MoO_x^[30]. To further prove the improvement of conductivity originating from the incorporation of MoO_x, the I–V curves of Ag/PEDOT:OSS-MoO₃/ITO and Ag/PEDOT:OSS-MoO_x/ITO were performed as shown in Fig. 1(b). The slope of Ag/MoO_x/ITO is higher than that of Ag/MoO_x/ITO, which indicates that the conductivity of PEDOT:OSS-MoO_x is higher than that of PEDOT:OSS-MoO₃.

To further prove the presence of O vacancies in MoO_x, the chemical composition of the MoO_x and MoO₃ was probed using X-ray photoelectron spectroscopy (XPS), as shown in Fig. 2. From Fig. 2(a) it can be seen that the major species Mo⁶⁺ was present in MoO₃. The Mo 3d peak was deconvoluted to obtain the contributions of Mo⁵⁺ (232.0 eV for Mo 3d5/2) and Mo⁶⁺ (232.7, 236.0 eV) and the major species present in MoO_x is the mixture of Mo⁵⁺ and Mo⁶⁺, as shown in Fig. 2(b). In addition, Mo⁵⁺ species was found to be 18.7 atomic rations (%). Mo⁵⁺ species is the origin of the metal-like electrical properties of MoO_x, which could low the series resistance of solar cell, and be beneficial to collect photo-generated charge carriers and finally enhance J_sc of solar cells.

The conventional planar architecture is adopted to further prove the effect of MoO_x on the performance of hybrid PSCs, which is constructed with indium tin oxide (ITO)/HTL/CH₃NH₃PbI_3–xCl_x/PC₆₁BM/Bphen/Ag (Fig. 3(a)). The PEDOT:PSS-MoO_x, PEDOT:PSS-MoO₃, and PEDOT:PSS were used to fabricate the HTLs. The performance of the devices with these three different HTLs is investigated under AM 1.5 G 100 mW/cm² illumination. According to our previous work^[27], 1 mol/L concentration of MoO_x and MoO₃ is doped into PEDOT:PSS as the holes transport layer. Fig. 3(b) shows the cross-sectional scanning electron microscopy (SEM) image. It can be seen that the perovskite layer completely covered the surface of the HTL layer and had an average thickness of 280 nm.

The current density–voltage (J–V) characteristics of PSCs using the different HTLs are shown in Fig. 4(a). The relevant performance parameters are also summed up in the inset of Fig. 4(a). Figure 4(b) shows the histogram of PSCs using PEDOT:PSS-MoO_x and PEDOT:PSPS-MoO₃. The device with HTL of PEDOT:PSS exhibited a PCE of 11.04%, with V_oc of 0.96 V, J_sc of 17.75 mA/cm², and FF of 0.65. The device with HTL of PEDOT:PSS-MoO₃ shows a significantly enhanced PCE of 12.75% with a V_oc of 1.01 V, J_sc of 18.22 mA/cm², and FF of 0.70. Compared to that of the PEDOT:PSS-MoO₃ device, the PCE of the PEDOT:PSS-MoO_x device furthermore improved from 12.75 to 15.01%.

By analyzing the data shown in the inset of Fig. 5, we can also see that the FF and J_sc values are improved and the series resistances (R_s) are reduced. The values of R_s of the PEDOT:PSS-based cells, PEDOT:PSS-MoO₃, and PEDOT:PSS-MoO_x are 18.65, 16.89, and 3.89 Ω·cm², respectively. The use of MoO₃ NPs additive induces oxidation doping of PEDOT:PSS, which promotes the networking between the conducting PEDOT units, and thereby greatly increases the electric conductivity of hybrid PEDOT:PSS-MoO₃ and PEDOT:PSS-MoO_x layer^[31].

It is supposed that the significant reduction of R_s ascribes to the good conductivity of PEDOT:PSS-MoO_x, and the high hole-conductivity could result in the high J_sc and the high FF due to the reduced R_s, which corresponds to the previous report by Nguyen et al.^[32]. Therefore, the cell photovoltaic performance is strongly affected by the conductivity of the HTLs.

The obvious increase of J_sc could be attributed to the incorporation of MoO_x into HTLs and the reduced R_s. Pristine PEDOT:PSS is a poor conductor, resulting in the low charge collection efficiency due to charge reorganization which dominates the charge transfer^[33], and causes the devices to suffer from the high R_s. By increasing the conductivity of the HTL with the addition of Mo⁵⁺, the ability of the hole transport is improved and the R_s is reduced, which allows improving the charge collection efficiency and increasing J_sc. Similar trends of J_sc improvements have been observed by using doped spiro-OMeTAD as HTLs^{[32, 34]}.

The addition of Mo⁵⁺ caused the conductivity of the HTL to increase, which resulted in a decrease of R_s, and the reduced R_s greatly affected FF, thus getting an increase in FF^[34]. The improved results may be due to the reduced recombination and optimization of the charge transport process caused by the increased conductivity of the HTL layer, which are consistent with previous reports^{[22, 35]}.

To further verify what caused the improvements on FF and J_sc, the atomic force microscopy (AFM) images and SEM images were carried out to research the surface properties of the HTLs and the perovskite layers. Fig. S1 displays AFM images of the PEDOT:PSS-MoO₃ and PEDOT:PSS-MoO_x, as well as SEM images of perovskite layers on the PEDOT:PSS-MoO₃ and PEDOT:PSS-MoO_x. The PEDOT:PSS-MoO₃ and PEDOT:PSS-MoO_x interface modification had a little effect on the morphology of perovskite films but it is not so obvious. The identical XRD patterns observed from the perovskite films indicated that perovskite crystallizations on PEDOT:PSS-MoO₃ and PEDOT:PSS-MoO_x were the same (Fig. S2). Thus, the difference of photovoltaic performance was solely caused by the different Mo valence states.

To understand the effect of Mo⁵⁺ on the resistance of the device, Impedance Spectroscopy (IS) measurements were carried out under dark conditions, which provided insights on the charge transport, recombination, and accumulation within the perovskite-based devices^{[36, 37]}. By doing so, the interface characteristics of PSCs can be further clarified by IS. Nyquist plots of the three types of devices measured in dark are shown in Fig. 6. The solid curves are the fitting results with a transmission line model. The measured frequency ranges from 10 to 10⁶ Hz with an AC amplitude of 10 mV at 0.8 V bias. These data were used to perform the fitting and to simulate the performance.

The single semicircle appears in Fig. 6, which is consistent with expectation. Moreover, the semi-arc gets smaller as the addition of Mo⁵⁺ into PEDOT:PSS. We attributed this result to the occupancy of the trap states at the perovskite/HTL interface, which is associated with the presence of a capacitance (C_p)^[38]. The equivalent circuit shown in the inset in Fig. 6 could further explain this behavior, where R_s is the series resistance and C_p is the capacitance of the device. These parameters include either contact resistance or the interface of the perovskite/HTL. Since C_p and R_s are strongly depended on the conductivity of the HTLs, the increasing conductivity of the HTLs caused the semi-arcs to become smaller. This signified that the C_p and R_s were decreased significantly, illuminating a lower density of carrier traps at the perovskite/HTL interface. The decreased carrier combination resulted in a smaller C_p^[39]. Therefore, the smaller C_p is demonstrated the presence of higher carrier density in the devices with high conductivity of HTLs.

We demonstrated the improved photovoltaic performance by using highly conductive HTLs, namely PEDOT:PSS-MoO_x. Therefore, we will discuss the possible mechanisms for improved performance from the molecular design perspective.

The open-circuit photovoltage decay (OCVD) measurements are to establish a photo-stationary state, and the device is illuminated during open circuit^{[37, 40]}. The light is then switched off and the decay in V_oc is detected as a function of time. The longer time scales of constant voltage maintaining were detected for the device of PEDOT:PSS-MoO_x HTL as shown in Fig. 7(a). Apparently, a remarkably slow decay of V_oc was exhibited for the cell of PEDOT:PSS-MoO_x HTL. The voltage remained at 0.8 V after 9 ms, and it had not completely decayed even at 0.1 s. In contrast, the open-circuit voltage of the cell with pristine PEDOT:PSS or PEDOT:PSS-MoO₃ HTL decayed more speedily, for it only takes 4.5 ms to be below 0.8 V. This observation provided an important insight showing that the highly conductive PEDOT:PSS-MoO_x HTL can maintain a persistent photovoltage at the longer time scales. The slow OCVD decay behavior is similar to the previous report by Baumann et al., in which the TiO₂ ETL was severally replaced by PC₇₀PCBM and MeOTAD HTL was replaced by PEDOT:PSS/polyTPD^[41]. Although the origin remains unclear, they speculated that these long-term transients are attributed to the interface presence of the perovskite/TiO₂, which might play an important role in maintaining the slow decaying^[37]. They attributed this change of the V_oc decay to the onset of a shunt pathway, which decayed the photovoltage more speedily. Our results show a clear change in the photovoltage decays over long periods of time with the highly conductive PEDOT:PSS-MoO_x, and this suggests that the good conductivity of the HTL could reduce the charges recombination which sustains photovoltages decay observed in perovskite solar cells^[42].

It is well known that the charge density in the device is closely related to V_oc with the following equation:

${{{V}}_{\rm{oc}}} = \frac{{kT}}{q}\ln\frac{{np}}{{{n_0}{p_0}}},$  (1)

which is supposed that the photovoltage is equal to the quasi Fermi level splitting^[43], where k represents the Boltzmann constant, q represents the elementary charge, T represents the temperature, p₀ and n₀ are the intrinsic hole and electron concentrations, and p and n are the total concentrations of electrons and holes, respectively. For the prestine device of PEDOT:PSS under the open circuit conditions, a reduction of concentrations of electrons and holes, which means the increase of the recombination of charges within the device, would result in the decrease of V_oc. For the PEDOT:PSS-MoO_x device, the increase of V_oc means the reducing recombination of charge carriers, and this result is consistent with the IS measurements results.

The dielectric constants ε_r of the perovskite device based on PEDOT:PSS-MoO₃ and PEDOT:PSS-MoO_x, are tested showed as Fig. 8(a), and the characteristic behavior of the frequency is the real part of the dielectric constant. In low frequency, a significant increase of ε_r of the perovskite device based on PEDOT:PSS-MoO_x is produced compared to that of PEDOT:PSS-MoO₃. At low frequencies—namely, below 10⁶ Hz—the dielectric response, ε_r, was assumed to be an ionic conduction, which largely stems from the accumulation of electrode interface charge. In the high-frequency zone, the change of ε_r value is mainly caused by the electronic conduction^[44].

The ε_r also affected the recombination of charges in the interface between the perovskite and the HTL. The holes were captured and ended up by the Coulombic interactions between the holes and the negatively charged trap states on the perovskite surface. The trap-assisted reorganization process takes place between the perovskite layer and HTL, as shown by Fig. 7(b). The critical distance (r_c) can be expressed as follows:

${r_{\rm c}} = \frac{{{q^2}}}{{4\pi {\varepsilon _0}{\varepsilon _r}KT}},$  (2)

where q, ε₀, ε_r, and T are the elementary charge, vacuum permittivity, the dielectric constant of HTL, and temperature, respectively^[45]. Under the aid of thermal energy (KT), the charges within the limits of r_c can escape the electrostatic force, and the holes have the greatest probability to be trapped and ended up in HTL within the critical distance^[46]. It is obvious that as the ε_r of PEDOT:PSS-MoO_x increases, the r_c value decreases, thus, the probability of trap-assisted recombination within the critical distance of r_c becomes low in the PSCs. This result indicates that the charge accumulation probability at the HTL/perovskite interface is decreased, which is subsequently related to the increase in V_oc and J_sc. In the device using PEDOT:PSS-MoO₃ as HTL, the r_c value was about 16.8 nm at 295 K, while it was significantly decreased to about 9.1 nm for the PSC based on PEDOT:PSS-MoO_x due to its higher ε_r. Therefore, the electrostatic force can be effectively screened and the trap-assisted recombination could be effectively suppressed in PSCs of PEDOT:PSS-MoO_x. This also explains the reason of the charge recombination effectively and improved performance of PEDOT:PSS-MoO_x PSCs.

To further distinguish the dielectric constant effect on the polarization performance, the J–V curves of the same device at the different scanning directions were performed in the dark as shown in Fig. 8(c), which could also explain the possible effect mechanism of interfacial polarization on charge recombination in PSCs. Interfacial polarization was tested with an external bias, as shown in Fig. 8(b). The applied external bias is ranged from 0 to 0.3 V, and J–V curves were obtained for both devices containing PEDOT:PSS-MoO_x and PEDOT:PSS-MoO₃.

Apart from the impact of MoO_x on conductivity of PEDOT:PSS, the current effect at the different directions of the applied DC field is also studied. The device with a positive electrode linked to a test anode had a similar electron transfer direction to PSC of PEDOT:PSS-MoO_x and PEDOT:PSS-MoO₃. In contrast, a device with a positive electrode linked to a test cathode had an electron transfer direction opposite the two devices. It can be seen from Fig. 8(b) that the measured current and the applied bias are positive correlation. After changing the direction of the applied field, it is found that the current of PSC based on PEDOT:PSS-MoO₃ is very similar to the current before the applied field is changed in the total applied field range. This result suggests that the charge transport in PSC based on PEDOT:PSS-MoO₃ makes no change. Once MoO₃ is changed to MoO_x (with the same electron transfer direction and applied DC power source), the interface induced polarization initiates the electron-hole dipole pairs to separate, then the current is increased. After the direction of the applied field was changed, the current was observed to be measured at room temperature and 0 V applied bias for the perovskite devices based on PEDOT:PSS-MoO_x. The current is lower than that with a positive field in the overall applied field range. This could be clarified to be the significant contribution of interface induced polarization between the perovskite and PEDOT:PSS-MoO_x. When both of the electron transfer directions from the applied DC power source and the interface effect are identical, interface induced polarization initiating the separation of electron-hole dipole pairs could contribute to the increasing current. Fig. 8(c) shows a schematic diagram of electric field distribution in perovskite device based on PEDOT:PSS-MoO_x. The polarization field (E_p) was consistent with the internal built field (E₀). Therefore, it facilitated the separation of photo generated holes and electrons. The accumulated photo-generated holes first at the perovskite/HTL interface are dragged into the PEDOT:PSS-MoO_x layer by E_p and then extracted at the anode, and this reduces the probability of non-radiative recombination by surface and bulk traps in the perovskite layer^[47].

We have revealed and highlighted the important effect of Mo⁵⁺ on the conductivity and polarization of HTLs using PSCs. The enhanced conductivity of the HTL by incorporating Mo⁵⁺ into PEDOT:PSS HTL significantly boosted the photocurrent and reduced charge recombination, which effectively improved device PCES due to the manipulation of the hole pathway by using the more electrically conductive MoO_x and the dielectric constant. We speculated that the higher dielectric constant can produce a polarization field and screen the reorganisation of the traps and the holes in the extraction layer, which could potentially reduce photogenerated charge recombination in the interface of the perovskite and the HTLs. This discovery affords an effective strategy for structuring efficient charge-transport materials to increase the overall performance of PSCs.

This work was supported by Colleges Universities in Henan Province Key Scientific Research Project Funding Scheme (No. 17A140020), Henan Province Nature Science Foundation Project (No. 182300410241) and Chinese Nature Science Foundation Committee (No. 61640406).