Conductivity Optimization of Mg, Zn and Ag doped TiO2 as Electron Transport Layer for Perovskite Solar Cell

: We report on the modification of structural, optical, electrical and dielectric properties of the electron transport layer (ETL) of Perovskite solar cell (PSC) i.e. Titanium dioxide (TiO 2 ) by adding Magnesium (Mg), Zinc (Zn) and Silver (Ag) impurities in it through Sol-gel method. Modified parameters are studied with the help of X-Ray diffraction (XRD), UV-Visible (UV-Vis) spectroscopy, electrical and dielectric measurements by two probe method respectively. Fascinating results have been obtained in modified Mg-TiO 2 , Zn-TiO 2 and Ag-TiO 2 such as enhanced transmittance (T), reduction in refractive index (n), extinction coefficient (k), Urbach energy (E U ), an increase in DC conductivity, dielectric constant ( ϵ ) and AC conductivity. The conductivity in modified TiO 2 follows the order Ag-TiO 2 >Zn-TiO 2 >Mg-TiO 2 >TiO 2 . The prepared samples are utilized in PSC device fabrication and the increased efficiency following the same order as conductivity suggests the idea of doping in ETL successful in solar cell application .


Introduction
Doped semiconductors have received the attention of the researchers to utilize them in optical, catalytic and electrical applications. The presence of various crystalline polymorphs and morphologies make them suitable in various fields. Size-dependent properties help the researchers to modify their properties to the extent they want in a particular application. In the last decade, the production and consumption of titanium dioxide TiO2 have been considerably increased due to its fascinating characteristics such as large availability, non-toxicity, low cost and optoelectronic properties [1].
TiO2 crystallizes in various phases such as anatase, brookite and rutile where anatase phase possesses high conductivity in comparison to other existing phases. The formation of these phases is a function of temperature. At low temperature, the anatase phase occurs whereas, with an increase in temperature, the rutile phase starts to appear. It finds its application in the various research areas such as photocatalysis, memory devices, solar cells and, MOSFETs. TiO2 material also acts as an electron transport layer (ETL) in case of thirdgeneration Perovskite solar cells (PSCs) to help the electrons generated from light illumination of the active layer to reach the electrode easily [2]. Anatase phase is the desired phase in PSCs. Nanostructured TiO2 can be prepared by a variety of methods such as sol-gel method, hydrothermal method, physical vapor deposition [3], laser chemical vapor deposition [4]. An effective way to modify its electronic properties is through doping [1] i.e addition of impurities to TiO2 lattice. Doping can be done either by replacing O 2− anion or Ti 4+ cation. Cationic doping can be done by metals and anionic doping by non-metals. The lower side of the conduction band (CB) is constituted by Ti 4+ 3d bands therefore replacement of Ti 4+ metal ions tends to bring a change in CB structure. On the other edge, the upper side of the valence band (VB) is formed by O 2− 2p bands, thus anion doping tends to change the VB structure [5].
Here, we have used the Sol-Gel technique to prepare TiO2 nanoparticles.
The properties of prepared nanoparticles have been modified by dopants (Magnesium (Mg), Zinc (Zn) and Silver (Ag)) i.e Mg-TiO2, Zn-TiO2 and Ag-TiO2. The ionic radii of Mg 2+ (0.66 Å), Zn 2+ (0.74 Å) and Ag + (1.15 Å) ions is comparable to Ti 4+ (0.68 Å) ions. Therefore, these dopants can easily substitute Ti 4+ in the TiO2 matrix without any change in its basic crystal structure. The purpose is to examine the effect of doping of Mg, Zn and Ag on pristine TiO2 as the electrical and optical properties of semiconducting materials are substantially modified by impurity doping [6]. Structural properties of prepared films have been analyzed with the help of the XRD technique. UV-Vis spectroscopy is used to study the optical properties. Electrical and dielectric measurements have been done by the two-probe method to study current-voltage (I-V) characteristics and AC conductivity. After these characterizations, materials have been used as an ETL in PSC device fabrication.

Experimental details
The precursor's Titanium butoxide (Sigma Aldrich), Titanium isopropoxide (Sigma Aldrich), Acetic acid (CDH), Sulphuric acid (Qualikems), Hydrochloric acid (HCL), Magnesium nitrate (Loba) and ethanol were used for the synthesis. TiO2 and doped Mg-TiO2, Zn-TiO2 and Ag-TiO2 nanoparticles were prepared by Sol-gel method following the procedure as shown in Fig. 1 [7]. The procedure was the same for other dopants as well. The Perovskite active layer material (CH3NH3PbI3) was prepared following the reported procedure [8]: hydroiodic acid (6 mL, Sigma Aldrich) and methylamine (5.6 mL, Laboratory Rasayan) were stirred in an ice bath at 5 • C for 2 h. The resulting solution was evaporated at 50 • C for one hour to get CH3NH3I. The obtained precipitates were washed with diethyl ether three times and then dried at 100 • C in the oven. To prepare CH3NH3PbI3, CH3NH3I (159 mg) and PbI2 (461 mg, Sigma Aldrich) were dissolved in 1 mL γ-butyrolactone at 60 • C for overnight with stirring. To prepare the hole transport layer (HTL) 6 mg of copper thiocyanate (CuSCN) (Sigma Aldrich) was dissolved in 1 mL of propyl sulphide (CDH).
For film deposition, at first, the substrates were washed sequentially with detergent, acetone and distilled water for 15 minutes each and followed with drying. Now the deposition of TiO2 compact layer was done by the reported procedure [9]: Titanium isopropoxide (2.54 mL) was diluted in ethanol (16.9 mL) and 350 μL HCL was diluted in ethanol (16.9 mL). Two solutions were mixed by adding acidic solution dropwise into the titanium isopropoxide solution with stirring. The resulting solution was spun on to the fluorine-doped tin oxide (FTO) at rpm of 2000 for 60 sec and then the film was annealed at 500 • C. Mesoporous TiO2 layer was deposited by spinning the solution of prepared nanoparticles in polyethylene glycol (PEG) at 5000 rpm for 60 sec [10] and annealed at 500 • C. For device fabrication, CH3NH3PbI3 was deposited on mesoporous TiO2 by spin coating at 2500 rpm for 30 sec and annealed at 50 • C for 10 minutes. After perovskite deposition, HTL was deposited by spin coating the prepared solution at 2000 rpm for 60 sec. Thermal evaporation technique was used to deposit Gold contacts.  Fig. 2 illustrates the XRD profile of pristine TiO2 and Mg-TiO2, Zn-TiO2 and Ag-TiO2 films.

Results and discussion
The peaks observed for the prepared material represented by their hkl values (as shown in Fig. 2) describes the well crystalline behaviour expressed as the desired anatase phase with tetragonal crystal structure according to the JCPDS file 01-089-4921 for TiO2. In addition to it, the peaks contributing to the substrate (FTO) are represented by the symbol '*'. No additional peak of dopants Magnesium (Mg), Zinc (Zn) and Silver (Ag) has been observed. This may be due to the small quantity of dopant or the good dispersion of dopant in pristine material [11] but the existence of dopants is observed in EDS results shown in Fig. 3. A shift in the position of the peaks has been noticed in the modified samples and is maximum for Zn-TiO2 which may be due to the presence of strain introduced by the incorporation of dopant. The induced strain may be related to the broadening of the peak in the diffraction profile. Although several factors are responsible for peak broadening such as defects in crystal, instrumental error, stress, strain, orientation and finite crystal size but instrumental broadening have the major contribution to the broadening of peak [12]. The instrumental broadening should be reduced to its minimum value to have the corrected value of crystallite size. This can be done by considering peak width as the contribution of both samples as well as instrument dependent effects. Where is the corrected FWHM of the sample, is the observed value of FWHM from diffractogram and is FWHM because of instrumental broadening. The average value of crystallite size (D) has been found by applying Sherrer's formula [13] to the principal peak represented by (101) plane.
where k is the shape factor (0.9), λ is the wavelength of Cu radiation (1.54 Å), β is full-width half maxima and θ is Bragg's angle. The calculated average value of D for all the samples is tabulated in table 1. It is observed that D decreases for all the modified samples in comparison to pristine. Defects in the material produce the stress which in turn leads to rearrangement of the lattice structure. This stress induces the strain in the material which is responsible for peak broadening in XRD. The strain induced in the material is given by where θ is Bragg's angle and ϵ is lattice strain. Eq. (2) and Eq. (3) shows that peak width varies as 1/cosθ from crystallite size and strain varies as tanθ. We assume that both strain distribution and particle size contributes independently towards the broadening of peak therefore the line breadth is the sum of Eq. (2) and Eq. (3) [13] i.e.   . 2) in the manner that peaks shift is greater in Zn-TiO2.
The optical parameters of thin films of TiO2, Mg-TiO2, Zn-TiO2 and Ag-TiO2 are studied with the help of transmittance (T) spectra in the spectral range (300 nm -800 nm). Fig. 5(a) shows the T spectra of prepared films. It is observed that T for Mg-TiO2, Zn-TiO2 and Ag-TiO2 increases as compared to pristine TiO2 which make the visible light to pass through this layer and reach the absorber layer (perovskite) leading to the creation of carriers and hence to have an appropriate electron transport layer (ETL) material for device fabrication. From transmittance spectra, we observed that the transmittance spectrum of Ag-TiO2 is different from other modified samples. This difference in the spectrum may be due to grain boundary scattering and also due to surface plasmon effect related to silver nanoparticles [14]. On the other side, T is related to refractive index (n) in a way, as T of the layer increases, the value of n decreases [15].
The decrease in n value predicts a decrease in density of the material and consequently decrease in refraction which tends to pass the light efficiently through the material and reach the next layer. As the refractive index dispersion is an important parameter in designing various optical devices, here we used Swanepoel's method [16,17] to calculate it in which, the sample is assumed of non-uniform thickness deposited on a transparent substrate having refractive index value ′s′. The refractive index of the film is complex and is given by * = − , where n is the refractive index and k is the extinction coefficient. This method is based upon Manifacier's approach [18] i.e in α = 0 regions, n is calculated by Eq. 6 = √( + √( 2 − 2 )) The calculated value of n using Eq. 6 is plotted against λ as shown in Fig. 5(b). A decrease in n with an increase of λ contributes to normal dispersion behaviour. By comparing plots of T and n with λ ( Fig. 5(a), Fig. 5(b)), it is found that Ag-TiO2 having higher T has the lowest n which is in the coordination of having decreasing n with increasing T. The imaginary part of n * , i.e. 'k' calculated by using Eq. 8 tells us the amount of scattering and absorption per unit distance in the medium which the light propagates [19]. k shows the same trend as that of n (Fig. 5(c)) i.e. it reduces with the increment in λ.
Consequently, lesser amounts of scattering and absorption takes place which results in better transmittance and it is lowest for Ag-TiO2. Fig. 6 shows the absorption coefficient (α) versus energy (hν) variation. It is observed that the absorption edge tail is exponential (which is shown by a blue oval in Fig. 6) for all thin films. The presence of this exponential edge indicates the presence of localized states within the bandgap. The extent of tailing can be calculated by applying first approximation to the absorption edge data in terms of an equation given by Urbach [20]. The Urbach rule is written as: Where 0 is a constant and is Urbach energy which gives the exponential edge slope. There is no particular origin known for the existence of exponential dependence of α with hν in amorphous as well as in crystalline materials. The random fluctuations of the internal fields linked with the structural disorder may be the cause of this exponential dependence [19]. is nearly constant or weakly depends on temperature and often represents the width of tail of localized states in the bandgap. Lower means the lower disorder in the materials.
is calculated for all the samples and is given in Fig. 6. We observe that value is lowest for Ag-TiO2 suggesting the lower disorder and is highest for Zn-TiO2 which is in agreement with XRD results. Reduction in leads to a decrease in activation energy and thus decrease in the recombination of carriers which could enhance the efficiency of the ETL material

Where 'I' is current, 'L' is the distance between the electrodes, 'V' is the applied voltage and 'A'
is the area of electrodes. As given in table 2, we have calculated σ at a particular voltage of 2 volts which is found to increase from 6.08 х 10 −9 Ω −1 m −1 for TiO2 to 5.72  To probe the current transport behaviour, temperature-dependent dark conductivity (σdc) measurements are done for all the prepared films (TiO2, Mg-TiO2, Zn-TiO2, Ag-TiO2) in the temperature range (298 K -393 K). It is found that the conductivity increases with an increase in temperature which suggests the semiconducting behaviour of the films. Based on nature of the slope, the overall temperature range (298 K -393 K) can be divided into two regions: higher temperature region (338 K -393 K) and intermediate temperature region (298 K -338 K). Fig. 7(b) shows Lnσdc versus 1000/T plot in two temperature regions (338 K -393 K) and (298 K -338 K). Therefore, σdc can be expressed as Where Ed1, Ed2 are activation energies in (338 K -393 K), (298 K -338 K) regions respectively and k is Boltzmann's constant. We have two activation energies corresponding to two regions which are tabulated in table 2. This decrease in activation energy as compared to pristine helps to have low recombination of charge carriers and leads to having good conductivity. The conduction in these two temperature regions can be explained based on commonly known conduction mechanisms -tunnelling [23] and thermionic emission [24]. Grain boundary trapping model can be used to explain these mechanisms as the films are polycrystalline in nature [25]. According to this model, the films have a large number of micro crystallites separated from one another by grain boundaries. Since the temperature range and Ed2 values are too high to observe hopping so tunnelling has been observed in this region.
Dielectric studies involve the participation of dielectric constant (ϵ′) which is one of the basic parameter giving information about atoms, ions and polarization mechanism. This approach helps us to have the real and imaginary part of the electrical parameters and thus gives information about the properties of the material. As TiO2 exists in three phases (anatase, brookite and rutile), anatase phase has been preferred for its dielectric properties. It has been of great interest for applications in the telecommunications industry due to its unusual high dielectric constant and low dielectric loss. The main difficulty encountered with TiO2 is the high recombination rate of photoexcited electron-hole pairs in the irradiated particles. To deal with this problem, a generic way is to modify TiO2 by doping with suitable material. The study of dielectric behaviour of pristine TiO2 and Mg-TiO2, Zn-TiO2 and Ag-TiO2 has been done over a range of 50 Hz -5 MHz. The parameters, dielectric constant (ϵ′) and dielectric loss (ϵ′′) have been calculated by using Eq. 12 and Eq. 13 respectively [26].
Where C, d, A, G, f and ε0 are the measured capacitance, the thickness of the film, area of electrodes, conductance, applied frequency and dielectric constant in a vacuum (8.85 х 10 −12 Fm −1 ) respectively. The cross-section of the film is shown in Fig. 8 to have the thickness of the film.  The variation of the real and imaginary part of the dielectric constant i.e. (ε′ and ϵ′′) as a function of frequency (50 Hz -5 MHz) at room temperature are shown in Fig. 9(a) and Fig. 9(c). It is observed that ϵ′ value is higher at low frequencies shown separately also in Fig. 9(b) and it decreases with the increase in frequency. The relation of ϵ ′ with binding energy (Eb) of carriers is given by using Eq. 14 [27] Where μ is the effective mass of electron and e is the charge on the electron. One may notice that ϵ′ goes on increasing for doped samples as compared to pristine one ( Fig. 9(a), (b)) which leads to a reduction in Eb of the charge carriers and hence enhancement in their separation takes place leading to increased movement of charge carriers. This result is in good agreement with our I-V results as the conductivity follows the same order as the dielectric constant (Ag-TiO2 > Zn-TiO2 > Mg-TiO2 > TiO2). High ϵ′ suggests the presence of space charge orientation, ionic and electronic polarization. The same trend has been observed for ϵ′′ versus frequency (f) plots ( Fig.   9(c)). ϵ′′ versus frequency (f) for low-frequency range is also shown in Fig. 9(d). ϵ′ versus ′f′ variation can be explained based on Maxwell-Wagner and Koops phenomenological theory [28,29].
According to this model, there are several well-distributed conducting grains separated by partially conducting grain boundaries. There is a great effect on these grains at higher frequencies whereas at low frequencies it is the grain boundaries that play a major role. At low frequency, the dielectric constant is due to the accumulation of charge carriers under the application of electric field which in turn leads to surface charge polarization. As the frequency changes, the field changes at a very fast rate which makes it difficult for the electrons to keep them in pace with the field and hence they lags behind the applied field due to slow movement of electrons with the changing field. This lagging of electrons spread over the whole area and ultimately reaches to grain boundaries which lead to a decrease in ϵ′ with an increase in ′f′. The variation of ϵ′ at low frequency. The high value of ϵ′ at a low value of frequency is because of the good response of polarization (especially surface charge polarization) with the varying field. In the reverse of it, a low value of ϵ′ is due to the poor response of polarization with the applied field. In addition to it, the charge carriers fail to go in coordination with the field at high frequencies which leads to a decrease in ϵ′. Also, there are impurities, defect states and space charge formation between different interface layers of nanomaterials which results in the production of absorption current leading to dielectric loss (ϵ′′). ϵ′′ decreases with the increase in frequency due to the decrease in absorption current. However, charge movement conversion to lattice vibration i.e. photon causes the ϵ′′ at high-frequency [30,31]. The value of σac is higher for doped samples than the pristine one. Zn-TiO2 has higher σac followed by Ag-TiO2 and Mg-TiO2. This increase in σac value helps the researchers to use these materials in various applications to have enhanced conductivity. At low frequency, the low value of σac is due to surface charge polarization whereas, at high frequency, high σac value is due to hopping phenomenon [32,33].

Conclusion
In