Combining plasmonic trap filling and optical backscattering for highly efficient third generation solar cells

Metal oxide contact layers such as ZnO and TiOx are commonly used in third generation solar cells as they can be solution processed and have a relatively high conductivity. It is well known, that by ultraviolet (UV) light soaking such devices overall device efciency can be boosted. This improvement in efciency is due to high energy UV light exciting hot carriers which then fll trap states in the metal oxide flm. Unfortunately, UV causes degradation of the active layer and thus must be fltered out if long lifetimes are to be achieved. In this work, it is demonstrated that plasmonically excited metal nano-structures embedded in the ZnO metal oxide layer can generate hot charge carriers from visible light alone, thus removing the need for UV light soaking. Furthermore, using this approach the solar cells exhibit simultaneously improved charge transport/recombination properties as well as enhanced light trapping behavior. Consequently, the power conversion efciency of a low-bandgap thieno[3,4-b]thiophene/benzodithiophene (PTB7) based solar cell can be increased from 7.91% to 9.36%.

To characterize the ZnO/AuNRs layer, we performed photoluminescence (PL) measurements to confrm that the AuNRs can generate hot charge carriers after surface plasmon excitation which will in turn fll the ZnO trap states. Figure 3a shows the PL spectrum produced by exciting a pure ZnO flm with a 350 nm source, while Figure 3b, shows the PL spectrum produced after exciting a ZnO flm with embedded AuNRs. In both Figure 3a and 3b, one curve is shown where the sample was not light soaked before the measurement, and one curve where the sample was light soaked with visible light containing no UV component. For the neat ZnO flm (Figure 3a), a broad band emission with maximum at 527 nm can be seen, which is ascribed to defect emission owing to the low processing temperature of our ZnO NPs. This can be attributed to a mid-gap trap level at -5.32 eV that corresponds to the singly ionized oxygen vacancy (V + ). 40 The introduction of AuNRs (2% wt) into ZnO layer didn't signifcantly change the emission level.
Both samples were then placed under solar irradiation (λ > 410 nm) for 5 min as preillumination. It can be seen from Figure 3b (red line) that, the defect emission of the composite flm was efectively quenched after plasmon excitation. While same treatment for neat ZnO flm didn't lead to the reduction of PL intensity. This behavior verifed that the LSPR-excited AuNRs encouraged the trap flling in ZnO without the aid of UV light. We then fabricated two types of solar cells, one with pure ZnO as the ETL (denote as reference device) and one with ZnO embedded with AuNRs (denote as plasmonic device). Figure 4 plots the current density-voltage (J-V) curves produced by the cells under AM 1.5 G illuminations.
Before recording the J-V curves, the cells were placed in front of the aforementioned solar simulation (with no UV component). It can be seen that the maximum power conversion efciency (PCE) of reference devices was 7.91% with an open-circuit voltage (V oc ) of 0.741 V, a short-circuit current density (J sc ) of 16.36 mA cm-2 and a fll factor (FF) of 65.2%, which is comparable to those of previously reported works based on PTB7:PC 71 BM. 41 More detailed statistical performance parameters can be found in Table 1. It can be seen that the plasmonic devices exhibited better performances. The optimal concentration of AuNRs was 2% wt. The best performance was mainly improved by the increase in J sc from 16.36 to 17.76 mA cm -2 and the FF from 64.2% to 70.5% respectively. The V oc was only slightly improved. As a result, a PCE of 9.36% was obtained by the champion device. This is among the highest efciencies reported for the PTB7:PC 71 BM system. [42][43][44] By measuring 10 optimized devices, we obtained the following fgures of merit: V oc at 0.742±0.004 V, J sc at 17.73±0.22 mA cm -2 , FF at 69.8±0.5% and PCE at 9.18±0.13%. We noted that when the AuNRs concentration was further increased to 4%, the cell efficiency decreased. Our experiments indicate that excess AuNRs may act as recombination centers 45 that decrease the charge carrier lifetime and extraction efficiency (vide infra). Quantum efficiency of device containing 4% AuNRs was plotted in SI (see Fig. S2), this shows the 4% device also has a lower EQE across the entire spectrum, supporting the fact that the decrease in efficiency is due to an electrical effect. It is also worth noting that even without the light soaking, the plasmonic device performance was improved, but only by a modest amount mainly due to a slightly improved photocurrent. We attribute this to the enhanced light absorption induced by LSPR efect.  The value in brackets is the best PCE for each device.
In order to obtain a more comprehensive understanding of the photocurrent enhancement, we measured the external quantum efciency (EQE) from the devices, as shown in Figure 5. It can be seen that the EQE was uniformly enhanced over a broad spectral region from 350 to 700 nm with a maximum EQE value for the plasmonic device up to 78%. EQE-integrated J sc s for these devices were comparable to the actual measured J sc s, further confrming our observation. As the EQE enhancement is broad and uniform, this implies that the ZnO/AuNRs composite ETL improved the electrical performance of solar cells, one if the improvement were mainly due to optical efects a more wavelength depended change in the EQE spectra would be expected. The signifcantly higher FF of 69.8±0.5% of plasmonic devices also suggests efcient charge transport and suppressed carrier recombination in the cells, which is consistent with better charge carrier extraction properties.  V is the applied voltage and V bi is the built in voltage. V bi accounts from the diference in work functions between the contact materials and was calculated from the threshold voltage of the diode. V bi was measured at 0.64 V for the hole only device, 0.87 V and 1.02 V for the electron only device using ZnO and ZnO(2wt% AuNRs) as interfacial layer, respectively.
From Figure 6a, it can be seen that ZnO/AuNRs devices exhibit a higher electron current density than their pure ZnO counterparts, suggesting that the presence of AuNRs aids electron extraction. We calculated a corresponding increase in average device mobility from 2.76×10 -4 cm 2 /Vs to 3.81×10 -4 cm 2 /Vs. From Figure 6b, the hole mobility is calculated to be 4.11×10 -4 cm 2 /Vs, this is in agreement with previously reported values. 46 These results suggest that, carrier transport inside the devices became more balanced upon the inclusion of the AuNRs and thus reduced the efect of space charge that usually adversely afects the electrical performance of the device. 47 We take the view that the improved charge transport and extraction can be attributed to the plasmonically excited hot-electrons in AuNRs flling the trap states in ZnO and improving its mobility.
Measuring mobility in disordered materials is notoriously difcult, as the mobility changes as a function of carrier density, and hence as a function of measurement conditions. Diferent measurement techniques are also known to produce diferent values of mobility. Therefore, to check that the ZnO/AuNRs device really does have better charge transport properties and to aid comparison with previously published experimental data we perform Transient Photo Current (TPC) measurements. Figure 7 shows the normalized TPC transients from devices with a ZnO ETL, a ZnO/AuNRs ETL and a device with no ETL. It can be seen that in all cases a typical TPC trace is observed, which is characterized by an initial plateau where charge transport dominates the transient and trap states are flled, followed by a sloped region, which is dominated by charge carrier detrapping from trap states. 48 It is noticeable that the plasmonic device had a shorter pretransit period than the other two non-plasmonic devices, we attribute this to better charge carrier extraction from the device. We estimate the charge carrier mobility using equation where d is active layer thickness, τ i is transit time and V is the build-in voltage. The transit time (τ i ) was found to be 1.37, 1.26 and 1.09 µs for bare Al, ZnO/Al and ZnO(AuNRs)/Al, respectively. ZnO/AuNRs based cells had highest carrier mobility, giving a value of 1.06×10 -4 cm 2 /Vs. We again attribute this to free carriers being able to leave the device more quickly when trap states in ZnO have been flled by plasmonically generated charge carriers. Next, to examine how the addition of the AuNRs afected charge carrier recombination in the device, we performed photoinduced charge carrier extraction by linearly increasing voltage (photo-CELIV) measurements (see Fig. S4 in the SI). 49,50 In the photo-CELIV experiment, an optical pulse is applied to a device, then at time t delay after the pulse the charge carriers are extracted by applying a linearly increasing negative voltage transient. By varying t delay the decay rate of photoexcited charge carriers can be studied. Figure 8 plots the photoexcited charge extracted from the device as a function of t delay . We ft this curve with the carrier recombination dynamic model 51 expressed by the equation in which n(0) is initial photo-generated carrier density and τ is carrier recombination lifetime. We estimated n(0) to be 9.78×10 15 cm -3 for the ZnO device v.s. 1.36×10 16 cm -3 for the ZnO(2wt% AuNRs) device and tau (τ) to be 91.2 µs for the ZnO device v.s. 124.9 µs for the ZnO(2wt% AuNRs) device. The higher n(0) value clearly suggests that solar cells with AuNRs can generate more free-carriers after photo-excitation, which is consistent with its higher observed J sc .
Prolonged τ also suggests that the photo-generated carriers can survive much longer without recombination in plasmonic devices. One reason to embed anisotropic AuNRs into the ZnO ETL was to encourage trap flling in the ETL and boost its mobility. Another motivation for putting this layer in the cell was to use the longer side of the AuNRs to back-scatter light and enhance light trapping. The higher n(0) derived from photo-CELIV analysis has suggested more efcient photon-to-carrier conversion.
Therefore we fnally examine how the nanorods changed the optical properties of the devices.  The refection spectra (R) of plasmonic device was also measured and compared with the pure ZnO one, this is shown in Figure 10a. The plasmonic device showed a relatively lower refectance than the pristine one. The derived absorption spectra (A=1-R) 53 has been plotted in Figure 10b. It can be seen the plasmonic device has enhanced light trapping in a broad range between 370 to 720 nm. This enhancement is more obvious between 480-580 nm than the other regions, which corresponds to the transverse absorption band from longer side of AuNRs. It is possible that the introduction of the nanorods has changed the refractive index of the ZnO layer, and thus the modal distribution of the light within the device, and this could be the root cause for the increase in light absorption. To rule this out, the complex refractive index of the active layer and the ZnO layers (with and without nanorods) were measured and a transfer matrix model used to calculate the change in the distribution of photons in the device caused by the introduction of the nanorods. It was found that the photon density increased by only 1.4% in the active layer of the plasmonic device, which is not enough to account for the observed optical enhancement (full results in the Supporting Information).
If the EQE spectrum of the device is normalized by the absorption spectrum (IQE=EQE/A) one can calculate the internal quantum efciency (IQE). This is a measure of how efcient the generation and transport/recombination processes are within the device while excluding changes in absorption between cells. 54 Figure 10b also plots the IQE spectra for both the plasmonic and non-plasmonic device. The plasmonic device shows a signifcantly higher IQE spectrum than the non-plasmonic device. This further supports our assertion that the main reason for the increase in the device efciency is due to trap flling. The G max values of the reference and plasmonic device (2wt% AuNRs) are 1.12×10 28 m −3 s −1 (J ph,sat =170.3 A·m -2 ) and 1.22×10 28 m −3 s −1 (J ph,sat =184.7 A·m -2 ), respectively. We attribute the noticeably increase in G max to light scattering stems from AuNRs. By normalizing J ph with J ph,sat as shown in Figure 11b, we can estimate charge collection efciency (P c,m ) under maximum power conditions and the exciton dissociation efciency (P c,sc ) at short circuit condition. 56 The P c,m shows an increase from 75.8% to 82.9% by the introduction of the nanorods, comparing well with the measured improvement in IQE.