Type-I alignment in MAPbI3 based Solar Devices with

Type-I alignment in MAPbI3 based Solar Devices with doped-Silicon Nanocrystals Conor Rocks,*,a Vladimir Svrcek,b Tamilselvan Velusamy,a Manuel Macias-Montero,a Paul Maguire,a and Davide Mariotti a a. Nanotechnology & Integrated Bio-Engineering Centre (NIBEC), Ulster University, UK b. Research Center for Photovoltaics, National Institute of Advanced Industrial Science and Technology (AIST), Central 2, Umezono 1-1-1, Tsukuba, 305-8568, Japan

Supporting Information Figure 1a shows transmission electron microscopy (TEM) images of p-Si NCs following ultra-sonication of colloids for 60 minutes before 10 µL of the supernatant was taken and drop-casted onto a holey carbon grid for imaging. The image shows a number of well dispersed p-Si NCs with high resolution image of a single p-Si NC (inset), highlighting clear lattice fringes that confirm the crystalline nature of the particles. Lattice spacing is ̴ 0.19 nm that correlates well with the (220) d-spacing of Si.1 Particle diameters were measured using ImageJ software and plotted in a histogram (Figure 1d) with bin sizes of 0.25 nm and 0.75 nm as the minimum nanocrystal diameter. A log normal fit of the distribution gave the mean diameter equal to 1.79 ± 0.88 nm. In similar form, TEM image of n-Si NCs is presented (Figure 1b) that also shows crystalline particles with clear lattice fringes. The histogram plot with fitting highlights a larger mean particle diameter equal to 2.76 ± 1.72 nm (Figure 1e). Doped Si NCs have been extensively characterized by our group and show that nanocrystals are crystalline in nature and small sized with narrow size distribution. 2,3 Scanning electron microscopy (SEM) was used to image the surface of MAPbI3 films (without Si NCs) that were deposited using a one-step spray method (Figure 1c). Large micro-sized cubic grains can be observed that is typical of pristine perovskite films. Grain

size was measured using ImageJ software and the size distribution is presented in a histogram below (Figure 1f) with mean grain size equal to 2.96 ± 0.79 µm.

Fig 1: TEM image of a) p-type Si NCs with inset showing a single Si NC particle and b) associated histogram of particle size distribution; c) n-type Si NCs with inset showing clear image of a single Si NC particle and d) associated histogram of particle size distribution; e) SEM image of MAPbI3 large grains and f) histogram of grain size distribution.

Figure 2 shows the X-ray diffraction (XRD) pattern for a spray deposited MAPbI3 film that shows (110), (200), (211), (202), (220), (114), (312), (224) and (314) lattice planes corresponding to the 14.05°, 19.92°, 23.47°, 24.48°, 28.37°, 31.84°, 34.96°, 40.48°, 43.11° (2θ) diffraction positions respectively.4 Calculation of unit cell parameters from the peak positions were performed by the relative combination of planar-spacing and Bragg’s law:5 𝜆 = 2𝑑ℎ𝑘𝑙 sin 𝜃ℎ𝑘𝑙

(1)

where λ is equal to 1.54 Å, d is the inter-planar distance between lattice planes and θ is the scattering angle. Calculating the inter-planar spacing d for corresponding diffraction position and knowing MAPbI3 has tetragonal structure we can determine unit cell parameters from 1 𝑑2

=

ℎ2 +𝑘 2 𝑎2

𝑙2

+ 𝑐2

(2)

where Miller indices hkl represent a particular set of equivalent planes with unit cell edge lengths of a and c. First, in order to calculate the lattice constant a, the plane (110) can be used since a = b. Then, inserting the lattice constant a into another plane, e.g. (202), the lattice constant c can be determined. Unit cell parameters are therefore determined to be a = b = 8.01 Å and c = 11.33 Å and matches well with referenced literature;6 The average size of ordered crystallites (D) were calculated using Scherrer’s formula 𝐾𝜆

𝐷 = 𝛽 cos 𝜃

(3)

where K is a dimensionless shape factor and is equal to 0.9, λ is the x-ray wavelength, β is the full width half maximum in radians of the x-ray diffraction peak and θ is the diffraction angle. Implementing the formula for a number of diffraction positions relating to pristine MAPbI3 the average crystallite size is 58.21 ± 8.32. When p-Si NCs are implemented into the MAPbI3 film corresponding MAPbI3 diffraction peaks remain present suggesting that the nanocrystals have not affected grain growth and crystallinity. The average crystallite size is calculated to be 54.96 ± 5.07 for MAPbI3 + p-Si NCs film, which is comparable to the pristine MAPbI3 film.

Fig 2: XRD patterns for a pristine MAPbI3 film and a MAPbI3 + p-Si NCs film.

In order to vary the oxidation degree, p-Si NCs were dispersed into ethanol and DIwater and sonicated for a period of 20, 40 and 60 minutes. This method was previously found to produce Si NCs with a varying degree of oxidation,7 which resulted in different Si0/Si4+ intensity ratios in the Si 2p XPS spectrum. Figure 3 below shows the XPS Si 2p regions for p-Si NCs in ethanol sonicated 20-60 minutes (Figure 3a-c) and for p-Si NCs in DI-water sonicated 20-60 minutes (Figure 3d-e). Each spectrum is labelled with their resulting Si0/Si4+ intensity and is referenced within the main text as Si NC oxidation degree.

Fig 3: XPS Si 2p spectra for p-Si NCs that have been dispersed in ethanol/DI-water and sonicated for 20-60 minutes to give increasing Si0/Si4+ intensity ratios.7

Figure 4a shows the mean work function for p-Si NCs with varying oxidation degrees taking using Kelvin probe scans across a film. Linear fitting of the points allows for interpolation of data to estimate Fermi level of surface oxidised p-Si NCs inside the MAPbI3 film. The fermi level is therefore estimated to be -5.45 eV. Figure 4b shows the Tauc plot of p-Si NCs films with varying oxidation degrees, produced by taking transmission measurements of thin films inside and outside of an integrated sphere in order to account for scattering and reflectance. The bandgap is obtained from the linear dependence of the square root of the absorption coefficient on photon energies and are plotted in figure 4c as a function of oxidation degree (Si0/Si4+). Again linear fitting of the points allows for interpolation of data to estimate the bandgap energy of surface oxidised p-Si NCs within the MAPbI3 film.

Fig 4: a) Plotted average workfunction of p-Si NCs as a function of oxidation degree with interpolated workfunction value for p-Si NCs in MAPbI3 films b) Tauc plot for p-Si NCs at varying oxidation degrees with linear fittings to estimate indirect bandgap energy and c) plotted bandgap energy as a function of oxidation degree with extrapolated bandgap energy for p-Si NCs in MAPbI3 films.

Low binding energy scans from the XPS can give insight into VBM values relative to the Fermi level which displays the energy separation Ef - VBM. Figure 5 shows the XPS Si 2p spectra for p-Si NCs with varying degrees of oxidation present (Si4+/Si0). The linear fitting of the initial signal onset is highlighted for p-Si NCs with limited surface oxidation (0.25) and gives an Ef – VBM value of 0.61 eV. The gradient of onset signal from Si 3p photoelectrons is reduced with increasing oxidation degree along with the appearance of a second onset which arises due to O 2p photoelectrons in SiO2. Continued oxidation however does not majorly affect the Ef - VBM for p-Si NCs and is consistent around 0.63 ± 0.02 eV.

Fig 5: XPS Si 2p spectra for p-Si NCs with varying degrees of oxidation and linear fits to determine onset energy and Ef – VBM estimation.

Figure 6 shows the normalized PL of p- and n-Si NCs (left axis) that have peak PL wavelengths at 640 nm and 700 nm. Also illustrated is the transmittance of MAPbI3 film with absorption occurring around 800 nm (right axis). Down shifting may be also important for future, long lasting and stable perovskite solar cells since high energy photons can cause internal stress and strain due the low thermal conductivity in MAPbI3 films.8 Since MAPbI3 bandgap is relatively small (1.5 eV), a large part of the solar radiation is converted into heat which if not dissipated can result in stresses that lead to large decreases in efficiencies. The increased absorption coefficient in Si NCs coupled with their strong PL at energies close to the perovskites bandgap make them ideal for down conversion.9,10

Fig 6: normalized photoluminescence of p- and n-Si NCs in ethanol (left axis) with transmittance of MAPbI3 film (right axis).

Figure 7a displays the individual mean parameters (Voc, Jsc and FF) for MAPbI3 only devices from 0 minutes illumination to after 8 minutes of continued illumination. Voc and Jsc are initially around 0.81 V and 6.39 mA cm-2 respectively which decrease to lower values of 0.62 V and 4.18 mAcm-2 as a consequence of light soaking. FF values exhibit the same behaviour as Voc and Jsc parameters under light soaking and decreases to 0.31 over a period of 8 minutes. Figure 7b show the individual mean paramters (Voc, Jsc and FF) for MAPbI3 + p-Si NCs devices from 0 minutes illumination to after 8 minutes of continued illumination. The Voc has lowered to around 0.62 V as a consequence of adding p-Si NCs when compared to 0.8 V for perovskite only devices. Additionally, the introduction of p-Si NCs has increased the Jsc to 10.45 mAcm-2. Light soaking effects for MAPbI3 + p-Si NCs devices prompt increases in Voc, FF and Jsc paramters over the illumination period. Figure 7c show the individual mean paramters (Voc, Jsc and FF) for MAPbI3 + p- type Si NCs devices from 0-8 minutes of continued illumination. A increase in Jsc to 10.31 mAcm-2 is observed at 0 minutes when comparing with MAPbI3 only devices. Voc is at the low value of 0.46 V with FF equal to 0.32. Once more, benifical light soaking effects are exhibited for the MAPbI3 + ntype Si NC devices that increase both Voc and Jsc over a period of 8 minutes to values of 0.71 V and 15.8 mAcm-2. Figure 7d presents a summary of all device power conversion efficiency (PCE) and as expected from the analysis of individual Voc/Jsc/FF cell parameters of MAPbI3 devices, PCE showed large instabilities over the short time period decreasing from mean value of 2.64% to 0.87%. In comparison, MAPbI3 composite devices that have included p- and n-Si NCs show substantial increases in average PCE from 2.44% to 4.28% and 1.48% to 4.15%, respectively. The incorperation of p- and n-Si NCs within the MAPbI3 devices exhibit positive light soaking behaviour over the short time period in contrast with MAPbI3 only devices. As previously discussed the I:Pb atomic mass ratio is increased for MAPbI3 films with p- and n-

Si NCs, beyond values typically associated with stoichiometric perovskite (see main text), that implies additional iodide is trapped within the films and allows for slower surface chemistry and degradation.

Fig 7: Corresponding Voc, Jsc, FF mean parameters presented with error bars for SD for a) MAPbI3 only devices b) MAPbI3 + p-type Si NCs devices c) MAPbI3 + n-type Si NCs devices and d) PCE for all devices over illumination period.

Figure 8 shows PCE of all devices as a function of time exposed to atmospheric conditions. After 4 days the MAPbI3 devices show a 40% drop in efficiency, whereas MAPbI3 + p-Si NCs and MAPbI3 + n-Si NCs reduce by around 10% and 20%, respectively. Continued aging of up to 14 days decreases efficiencies of all devices however the linear fits indicate clearly the accelerated reduction for MAPbI3 when compared with the MAPbI3 +p/n-Si NC devices. After 14 days the MAPbI3 device has reduced by near 90% of initial

efficiency and is attributed to the chemical degradation highlighted previously. The addition of p- and n-Si NCs within the perovskite film is advantageous in slowing typical chemical degradation in ambient conditions, as highlighted by XPS analysis. This slowed chemical reduction in atmospheric conditions improved the stability of MAPbI3 + p-/n-Si NC solar devices by a factor of 3 when compared with MAPbI3 only devices.

Fig 8: Power conversion efficiency over time in atmosphere for MAPbI3 and MAPbI3 + pand n-Si NC devices.

References 1

G. Basile, A. Bergamin, G. Cavagnero, G. Mana, E. Vittone and G. Zosi, Phys. Rev. Lett., 1994, 72, 3133–3136.

2

S. Mitra, V. Švrček, M. Macias-Montero, T. Velusamy and D. Mariotti, Sci. Rep., 2016, 6, 27727.

3

V. Svrcek, J. Laser Micro/Nanoengineering, 2007, 2, 15–20.

4

T. Oku, in Solar Cells - New Approaches and Reviews, InTech, 2015.

5

T. Supasai, N. Rujisamphan, K. Ullrich, A. Chemseddine and T. Dittrich, Appl. Phys. Lett., 2013, 103, 183906.

6

Y. Kawamura, H. Mashiyama and K. Hasebe, J. Phys. Soc. Japan, 2002, 71, 1694– 1697.

7

C. Rocks, S. Mitra, M. Macias-Montero, P. Maguire, V. Svrcek, I. Levchenko, K. Ostrikov and D. Mariotti, ACS Appl. Mater. Interfaces, 2016, 8, 19012–19023.

8

A. Pisoni, J. Jaćimović, O. S. Barišić, M. Spina, R. Gaál, L. Forró and E. Horváth, J. Phys. Chem. Lett., 2014, 5, 2488–2492.

9

X. Wen, P. Zhang, T. A. Smith, R. J. Anthony, U. R. Kortshagen, P. Yu, Y. Feng, S. Shrestha, G. Coniber and S. Huang, Sci. Rep., 2015, 5, 12469.

10

L. Pavesi, L. Dal Negro, C. Mazzoleni, G. Franzò and F. Priolo, Nature, 2000, 408, 440–444.