Charge Accumulation, Recombination, and Their Associated Time Scale in Efficient (GUA) x (MA)1-x PbI3-Based Perovskite Solar Cells.

Here, we study the influence of guanidinium (GUA) ions on the open-circuit voltage (V oc) in the (GUA) x (MA)1-x PbI3 based perovskite solar cells. We demonstrate that incorporation of GUA forms electronic and ionic accumulation regions at the interface of the electron transporting layer and perovskite absorber layer. Our electrochemical impedance spectroscopy results prove that the formed accumulation region is associated with the enhanced surface charge capacitance and photovoltage. Furthermore, we also demonstrate the influence of the GUA ions on the enhanced interfacial and bulk electronic properties due to more efficient charge transfer between the bulk and interfaces and the reduced electronic defect energy levels.

Introduction

Over the last decade, organometal halide perovskites have attracted significant attention as promising materials for next-generation thin film photovoltaic (PV) technology.[1,2] The appealing physicochemical properties of these compounds such as high carrier mobility and diffusivity,[3] low exciton binding energy,[4] tunable bandgap,[5] long charge-carrier diffusion length,[6] and tolerance to defects provide a broad application horizon from perovskite solar cells (PSCs) to light emitting diodes or lasers.[7−9] Despite the great achievements in the efficiency of the PSCs,[10−14] the optoelectronic properties of these devices are mostly influenced by the phenomena occurring at the interfaces between the perovskite layer and either the electron transporting layer (ETL) or the hole transporting layer (HTL).[14,15] A recent report on the surface recombination and collection efficiency in PSCs has demonstrated that the nonradiative recombination at the interface (ETL/perovskite) plays a crucial role in the Voc and charge transfer phenomenon.[16] The surface and bulk recombinations in the framework of Shockley Read Hall can occur in two fractions: (i) a doped semiconductor, in which the doped material undergoes rapid recombination due to the presence of excess carriers and (ii) an intrinsic semiconductor, in which the recombination is delayed, and this could be certainly beneficial for achieving higher photovoltage.[17] Small perturbation techniques, such as intensity-modulated photocurrent and photovoltage spectroscopy (IMPS and IMVS), and frequency domain measurements are able to efficiently extract the dynamic operating parameters of solar cells. These characterization tools have been widely employed for evaluation of different photovoltaic devices.[18,19] However, despite their great potential, only a few reports demonstrated the applications of these techniques in the PSCs.[20−23]

Recently, guanidinium (C(NH2)3+) has been regarded as an attractive candidate for fabrication of efficient and stable PSCs.[24−28] Remarkably, we demonstrated that the guanidinium (GUA) cation, despite its large size, can replace methylammonium (MA) at the A site of ABX3 perovskite and form a pure phase of GUAMA1–PbI3-type perovskite.[29] The resulting GUA-based PSCs show potential to increase the charge-carrier lifetime and Voc of the PSCs compared to the pristine MAPbI3 based devices. However, the origin of such improvements and the charge dynamic mechanism have not been widely investigated.

In this work, we investigate the role of GUA in charge carrier and recombination dynamics in fully working (GUA)(MA)1–PbI3-type PSCs. Our analysis using small perturbation techniques reveals that the incorporation of GUA causes electrode polarization at the interface, which could be responsible for the prominent hysteresis in GUA-based PSCs. The IMPS response permits the quantization of the charge transport phenomena, which affects the photogenerated charge collection. Furthermore, we demonstrate that the incorporation of GUA causes excess electronic charge accumulation at the interface, which is responsible for the observed enhancement in photovoltage. These results provide a pathway to achieve efficient solar cells along with an in-depth working mechanism.

Results and Discussion

To study the effect of GUA on the photovoltaic (PV) properties of the PSCs as also reported in the literature,[29] we fabricated a batch of solar cells with the following architecture: FTO/compact TiO2/mesoporous TiO2/perovskite/Spiro-OMeTAD/Au (more details can be found in the Experimental Section). Figure a shows the current density–voltage (J–V) characteristics of the best performing (GUA)(MA)1–PbI3 PSCs (where x = 0, 0.10, and 0.25) measured under reverse bias and Air Mass 1.5G (AM1.5G) (the statistic histogram of PV parameters are shown in Figure S1). We found that the replacement of 10 mol % MA with GUA increases the Voc and Jsc over 30 mV and 1.2 mA cm–2, respectively. As a result, the PCE of the prepared devices increases from 17.5 ± 0.5% for MAPbI3 to 18.0 ± 0.5% for (GUA)0.1(MA)0.9PbI3 PSCs. However, incorporation of 25 mol % GUA reduces the PCE to 16.0 ± 0.5% due to a decrease in the FF and Jsc. These data are consistent with our previous work,[29] where the introduction of a small amount of GUA into the perovskite composition increased the Voc and PCE. The MAPbI3 and (GUA)0.10(MA)0.90PbI3 PSCs with a PCE of 17.5 ± 0.5% were employed to study the role of GUA in the hysteresis effect and charge carrier and recombination dynamics.

Figure 1

(a) Current–voltage characteristics measured under 1 Sun illumination Air Mass 1.5G in the backward direction for MAPbI3, (GUA)0.10(MA)0.90PbI3, and (GUA)0.25(MA)0.75PbI3 devices. (b) Capacitance–frequency (C–f) measurements at zero bias under dark conditions for MAPbI3 and (GUA)0.10(MA)0.90PbI3 devices.

Hysteresis Effect

Figure S2 shows the hysteresis curves of the MAPbI3 (reference) and (GUA)0.10(MA)0.90PbI3 PSCs. We used the following formula to calculate hysteresis indices (HIs) of the PSCs[30]From the J–V result, the HI was calculated to be 18.5% for the reference device and 16.8% for the (GUA)0.10(MA)0.90PbI3 PSC. Despite the favorable properties of GUA cations such as zero dipole moment and hydrogen bonding capability, the hysteresis is more prominent in GUA-based devices. To rationalize the observed hysteresis feature, we measured the capacitance–frequency (C–f) responses at zero bias under dark conditions (Figure b). The distinct features in the low and high frequencies of the C–f response are clearly distinguishable.[30] A constant capacitance element in frequency >103 Hz for the reference and (GUA)0.10(MA)0.90PbI3 is associated with the dielectric response of the absorber material, while the capacitance in the low-frequency region has originated from the ionic characteristics.[31−34] It is clear that due to the presence of GUA cations in the perovskite, which have a larger size (278 pm) and weaker bonding ability than MA cations, GUA cations can pile up near the interfaces and distribute the local electric field, leading to a higher capacitance in the low-frequency range.[28] By assuming only the electrostatic interaction at room temperature, the space charge densities of 6.04 × 1017 and 1.1 × 1018 and were obtained for the reference and (GUA)0.10(MA)0.90PbI3 PSCs, respectively. The accumulation of excess GUA ions in the (GUA)0.10(MA)0.90PbI3 PSCs increases the capacitance in the low-frequency region and leads to the current hysteresis. Moreover, the ion movement under the applied bias amplifies the hysteresis by screening the internal electric field at the interface between TiO2 and perovskite.

Charge Transport and Time Responses

To elucidate the influence of GUA on the transport properties of the PSCs, the intensity-modulated photocurrent spectroscopy (IMPS) under constant light intensity was performed (Figure ). The IMPS response for both PSCs shows distinguishable characteristics in the low and high frequency regions and exhibits different time constants. We analyzed the frequency arcs according to the established transport mechanism for the TiO2/perovskite interface, TiO2 ETL, and the perovskite layer.[35,36] The slow response time in the low-frequency region is dominated by the transport process in the TiO2 or TiO2/perovskite interface, while the higher frequency of the arc from 1 to 10 Hz (faster transport) is ascribed to the perovskite layer. The low-frequency component (800 mHz to 1 Hz) of MAPbI3 perovskite corresponds to the time constant of 0.031–1 s. However, the low-frequency component of the (GUA)0.10(MA)0.90PbI3 device slightly shifts toward a higher frequency, which corresponds to the lower time constant, suggesting the modification in the interfacial electrical response. To have a better understanding of this phenomenon, we further examined the device with a higher amount of GUA (25%). As expected, the low-frequency component in the (GUA)0.25(MA)0.75PbI3 device shifts toward a higher frequency as compared to the (GUA)0.10(MA)0.90PbI3 device. In contrast, the high frequency responses (shorter time constant) centered at 7 Hz are almost similar in all of the investigated PSCs. The diffusion coefficient D = d2/2.35τ, where D is the diffusion parameter, d is the thickness of the perovskite absorber layer, and τ represents the time constant, are calculated to be 1.23 × 10–8 and 2.83 × 10–6 cm2 s–1 for the reference and (GUA)0.10(MA)0.90PbI3 PSCs, respectively. The charge transport at the TiO2/perovskite interface can also be visualized by the current transient measurements near the short circuit condition as shown in Figure S3. The shorter decay time in the (GUA)0.10(MA)0.90PbI3 device confirms the faster charge transport at the TiO2/perovskite interface as compared to the reference device. From the IMPS and transient responses, we conclude that the GUA cations improve the charge transport and diffusion coefficient.

Figure 2

Intensity-modulated photocurrent spectroscopy (IMPS) under constant light intensity for MAPbI3, (GUA)0.10(MA)0.90PbI3, and (GUA)0.25(MA)0.75PbI3 PSCs.

The pXRD patterns of (GUA)(MA)1–PbI3 films (where x = 0, 0.10, and 0.25) are shown in Figure S4, which are in good agreement with previous reports.[25,29]Figure S5a shows the UV–visible spectra of the perovskite films with different amounts of GUA cations. As seen, the perovskite film with 10% mol GUA has better absorption than the reference sample. Also, the photoluminescence (PL) emission of this sample is much stronger than the others (see Figure S5b). The observed better optical properties of (GUA)0.10(MA)0.90PbI3 perovskite together with the improved charge transfer properties could be the main reasons for the improved current density in these PSCs. Additionally, we study the morphology of the corresponding perovskite films using scanning electron microscopy (SEM) and atomic force microscopy (AFM) techniques, as shown in Figures S6 and S7. From the SEM images, the GUA-based perovskite films have different morphology with slightly lower grain sizes compared to pure MAPbI3. As mentioned in the literature,[25] the GUA-based perovskites have large crystal domains containing small grain sizes, which can affect the optoelectronic properties of the perovskite films. In fact, the AFM results show similar morphology and grain size to the SEM images. We found that the surface roughness of the (GUA)0.10(MA)0.90PbI3 perovskite film (21 ± 6 nm) is lower than that of the pure MAPbI3 sample (34 ± 9 nm). This result indicates that the GUA-based perovskite has a smoother contact than the reference sample with the HTL, suggesting better charge transfer at this interface with possibly lower recombination.[37,38]

To gain insight into the band diagram of the GUA-based perovskite, ultraviolet photoelectron spectroscopy (UPS) measurement was performed (Figure S8). The extracted data from the UPS curves are summarized in Table S2. Based on these values, we draw the band diagram of the perovskite films, as shown in Figure S8c. We found that by introduction of GUA cations into the perovskite composition, the energy levels of both valence and conduction bands become deeper and upon increasing the amount of GUA, the energy levels are further enhanced. We observed this enhancement for the Fermi level as well. Interestingly, these results indicate that addition of the GUA cations makes the perovskite film more p-type. This could help for a better charge transfer between the perovskite and TiO2 ETL, due to the presence of a stronger electrical field at the interface and also a closer conduction band of GUA-based perovskites to that of the TiO2 ETL.[38] However, if we look at the interface between the perovskite and spiro HTL, it is clear that pure MAPbI3 perovskite shows better charge transfer (hole) with respect to the spiro HTL, due to its closer valence band to that of the HTL. Notably, by increasing the amount of GUA to 25 mol %, the valence band of the perovskite film increases to 5.5 eV, which is much deeper than that of the spiro HTL, resulting in more difficult hole transfer and thus more interfacial recombination.

Charge Accumulation and Time Responses

The effect of GUA on the interfacial electronic charge accumulation properties were further studied by measuring the capacitance versus frequency response under illumination at different applied biases from knee voltages to the VOC of the PSCs. The C–f plot, shown in Figure , demonstrates that the capacitive element due to the surface charge accumulation (CS) dominates at low frequency (∼1 Hz) and the capacitance due to the dielectric response of the perovskite thin film (Cg) predominates at high frequency (>103 Hz). The low-frequency C–f plot (Figure b), assigned to the accumulation of ions and ionic space charge capacitance, exhibits values in the order of microfarad, which substantially increases to millifarad under illumination conditions, indicating that the interfacial capacitance is changed by the electronic phenomena (Figure ).[6,16,17] However, as the ionic capacitance under open-circuit conditions is limited by the Helmholtz capacitance, a higher value of CS is obtained for (GUA)0.10(MA)0.90PbI3 as compared to that of the MAPbI3 device. On the contrary, the Cg value remains unchanged under illumination for both PSCs and is attributed to the dielectric characteristics of the perovskite film. This mechanism shows that the electronic accumulation is kinetically controlled by GUA ionic species. The buildup of electronic accumulation capacitance at the TiO2/perovskite interface follows the exponential voltage dependence by Cs = εε0/√2LD exp(qV/2kBT), where is the Debye length and q and kBT are the elementary charge and thermal energy, respectively.[31,32,39]

Figure 3

Capacitance–frequency (C–f) measurements as a function of applied bias (the arrow indicates the increase in applied bias) under illumination for (a) MAPbI3 and (b) (GUA)0.10(MA)0.90PbI3 devices.

Recently, it was reported that a part of Voc in the PSCs has originated from the charge accumulation, while the other part of Voc has originated from band bending.[40,41] In the case of the (GUA)0.10(MA)0.90PbI3 PSC, ion accumulation causes the formation of an accumulation zone, where upon illumination and under open-circuit conditions, photogenerated charges get accommodated and forms the photovoltage.

To investigate the time response associated with the charge accumulation and dielectric phenomenon of the corresponding PSCs, the intensity-modulated photovoltage spectroscopy (IMVS)-derived time constant as a function of Voc was recorded for both PSCs and is shown in Figure a. One can easily calculate the time response by the inverse of frequency in the low and high frequency spectra of IMVS. The high frequency peak, assigned to the dielectric capacitance of the absorber layer, provides information about the carrier lifetime. The reference device exhibits a carrier lifetime of ∼4–5 μs, whereas the (GUA)0.10(MA)0.90PbI3 sample shows an enhanced charge-carrier lifetime of ∼6–7.5 μs, attributing to the lower charge recombination, which is in good agreement with the previous report.[24] In line with a similar effect observed by the other authors,[24,25] where the GUA enhanced the carrier lifetime and Voc of the PSCs,[24,25] the present study indicates that along with the enhanced carrier lifetime, the higher electronic accumulation at the interface also significantly contributes to the Voc.

Figure 4

(a) Recombination lifetime obtained from intensity-modulated photovoltage spectroscopy (IMVS) measurements at the open-circuit voltage for MAPbI3, (GUA)0.10(MA)0.90PbI3, and (GUA)0.25(MA)0.75PbI3 and (b) trap densities obtained from thermal admittance spectroscopy for MAPbI3 and (GUA)0.10(MA)0.90PbI3 devices.

To further understand the influence of GUA on the defect-state energy level of the perovskite films, the defect energy level and defect density of MAPbI3 and (GUA)0.10(MA)0.90PbI3 PSCs were investigated using thermal admittance spectroscopy (TAS), as shown in Figure b. TAS is a well-established technique to probe both shallow and deep defect energy levels with defect density on thin films.[42−44] The defect energy level is calculated by using the expression, ω0 = βT2 exp(−Ea/kBT), where Ea is the defect activation energy level and ω0 is the characteristic transition frequency related to the rate of carrier emission and capture in the defect states. The Ea of the reference device is calculated as 0.40 eV, while a significantly lower Ea of 0.22 eV was obtained for the (GUA)0.10(MA)0.90PbI3 PSC. The obtained defect energy levels are in accordance with the commonly quoted values in the literature and are potentially ascribed to iodine interstitials.[43,44] The trap densities corresponding to the obtained Ea can be calculated by using the expression of Nt = (Vbi/qW)(dC/dω)(ω/kBT), where Vbi is built in potential, W is the depletion width, dC/dω is the derivative of capacitance with respect to the angular frequency. By considering the Gaussian distribution, the defect states are centered at 0.40 eV for the reference and at 0.22 eV for the (GUA)0.10(MA)0.90PbI3 PSCs. Moreover, an integrated defect density of ∼1016 cm–3 was obtained for both PSCs.

Conclusions

In conclusion, we have investigated the role of GUA in the charge carrier and recombination dynamics and their associated time responses in the (GUA)(MA)1–PbI3 PSCs. From the IMPS analysis and transient responses, we demonstrated that the addition of GUA improves the charge transport and diffusion coefficient. Lower charge recombination times of ∼4–5 and ∼6–7.5 μs for MAPbI3 and (GUA)0.10(MA)0.90PbI3 PSCs, respectively, were observed. We found that along with the enhanced carrier lifetime, the higher electronic accumulation at the interfaces also significantly contributes to the Voc enhancement. In fact, the incorporation of GUA into the perovskite leads to an improved interfacial and bulk electronic properties due to better charge transport across the bulk and interface and lower electronic defect at energy levels.

Experimental Section

Materials

Guanidinium hydroiodide (TCI, 99.95%), methylammonium iodide (DyeSol), PbI2 (TCI, 99.99%), dimethyl sulfoxide (DMSO, 99.9%), chlorobenzene (95%), titanium(IV) isopropoxide (99.9%), 4-tert-butylpyridine (Sigma-Aldrich, 96%), and lithium bis(trifluoromethylsulphonyl)imide (Sigma-Aldrich, Li-TFSI, 95%) were used. The TiO2 paste (18 NRD) was purchased from Dyesol.

Solar Cell Device Fabrication

PSCs were fabricated according to our previously published paper.[23] Briefly, GUA·HI, MA·HI, and PbI2 precursors were dissolved in anhydrous DMSO (in corresponding molar ratios) at 60 °C together with vigorous stirring overnight to achieve GUAMA1–PbI3 solutions (x = 0, 0.10, and 0.25) with 1.4 M. Then, a two-step program was employed to deposit perovskite solutions, i.e., spin coating at 1000 and 6000 rpm for 10 and 20 s, respectively. To form perovskite, the antisolvent technique was used, and 10 s before the end of spinning, 100 μL of chlorobenzene was poured on the substrates. Finally, the substrates were annealed at 100 °C for 30 min in a dry air box. For the HTL, a solution of spiro-MeOTAD was prepared by mixing the following precursors: 74 mg of spiro-MeOTAD dissolved in 1 mL of chlorobenzene, 17.5 μL of a solution of lithium bis(trifluoromethylsulphonyl)imide (Li-TFSI) dissolved in acetonitrile (520 mg·mL–1), 29 μL of a solution of tris(2-(1H-pyrazol-1-yl)-4-tert-butylpyridine)cobalt(III) bis(trifluoromethylsulphonyl)imide (FK 209) dissolved in acetonitrile (300 mg·mL–1), and 28.8 μL of tert-butylpyridine. Afterward, the HTL solution was spin-coated on the perovskite films at 4000 rpm for 20 s. Finally, the fabrication of the devices was completed by thermal evaporation of 80 nm-thick gold under high vacuum.

Film Characterization

PXRD diffractograms were recorded on an X’Pert MPD PRO (Panalytical) diffractometer (Cu anode, λ = 1.54 060 A°) in an angle range of 2θ = 5–40° with a step of 0.02 degree. UV–vis spectra were recorded by a Varian Cary 5 and PL spectra using Fluorolog 322 (Horiba Jobin Ybon Ltd.). The morphology and smoothness of the films were studied using high-resolution scanning electron microscopy (ZEISS Merlin) and atomic force microscopy (AFM, NanoScope IIIa/Dimension 3100), respectively. UPS was employed to record the valence and Fermi levels of perovskite films using He I (21.2 eV) by AXIS NOVA (Kratos Analytical Ltd., U.K.).

Device Characterization

To measure the J–V curves, a light source (a 450 W Xenon lamp, Oriel) with an intensity of 100 mW cm–2 and a sunlight filter (Schott K113 Tempax, Praezisions Glas & Optik GmbH) were employed. The purpose of using a filter was to adjust the measurement according to the AM1.5G standard condition. J–V measurements were obtained by a Keithley (Model 2400) digital source meter upon applying external bias to the devices. A shadow mask with a black color was used to define the active area, which was 0.16 cm2 in our cases. To perform the electrochemical impedance spectroscopy characterization, a Bio-Logic SP-300 potentiostat together with a frequency response analyzer was used. For impedance measurement, the direct current (DC) bias range was varied from 0 to open-circuit voltage and the alternating current perturbation signal was set to 10 mV. The frequency was changed from 1 Hz to 100 kHz to monitor the device behavior. For IMVS and IMPS measurements, a Bio-Logic SP-300 potentiostat together with a frequency response analyzer was employed and 10% of the DC background illumination intensity was set for modulation current. The galvanostatic mode of Bio-Logic SP-300 with a cool white-light-emitting diode array (12 V, 10 W) as the light source was employed here.