Effect of PEDOT:PSS Layer Deposition on Electrical and Photoelectrical Properties of n+-ZnO/n-Si Heterostructure

The results of electrical and photoelectrical characterization of the interface and bulk properties of n+-ZnO/n-Si and poly(3,4-ethylene dioxythiophene):poly(styrenesulfonate) (PEDOT:PSS)/n+-ZnO/n-Si heterostructures are presented. It was found that the PEDOT:PSS layer deposited on the surface of zinc oxide increases the potential barrier at the ZnO/Si interface, leading to higher band bending in the silicon, which is important for solar cell applications. The recombination rate at the interface decreases because of the creation of an inversion layer in the silicon under operational conditions. The increase of the potential barrier in PEDOT:PSS/n+-ZnO/n-Si heterostructures results in the increase of the open-circuit voltage by 54–180%. The external quantum efficiency in PEDOT:PSS/n+-ZnO/n-Si heterostructures increases by 100% at 450 nm.


Introduction
Both inorganic transparent zinc oxide (ZnO) and organic conducting polymer poly(3,4-ethylene dioxythiophene): poly(styrenesulfonate) (PEDOT:PSS) films have recently occupied a wide niche of applications related to photovoltaic cells, which are the key elements of green energetics. Given the intrinsic n-type conductance of ZnO, it was suggested as an electron-collecting layer, 1-3 replacing the expensive indium tin oxide (ITO); intrinsically p-type PEDOT:PSS is often used as a hole-collecting layer in perovskite 4 and hybrid organic silicon heterojunction solar cells. [5][6][7][8] Several more exotic applications of PEDOT:PSS/ZnO structures have also been reported, such as resistive switching devices, 9 optoelectronic sensors, 10 liquified petroleum gas sensors, 11 and ultraviolet light-emitting diodes. 12 A heterojunction consisting of a PEDOT:PSS layer deposited on ZnO nanorods synthesized at the surface of n-type silicon has been suggested as a simple, non-toxic, and low-cost UV light photodetector. 13 Understanding the electronic properties of the PEDOT/ ZnO and ZnO/Si interfaces is of great importance in further development of relevant heterostructures for photovoltaic and optoelectronic applications. It has been reported that deposition of p-type PEDOT:PSS onto n-type ZnO enables one to improve the reliability of ZnO, preventing the adsorption and desorption of oxygen molecules on the ZnO crystal surface. 14 In that paper, non-intentionally doped single-crystalline ZnO was studied, and it was shown that the charge transport between the PEDOT:PSS and bulk ZnO depends on the polarity of the crystal surface.
In the present work we have used highly doped n + -ZnO films deposited by magnetron sputtering on n-type silicon. The doping of ZnO films was performed by adding methane during the deposition process. Recently, it has been demonstrated that methane can be effectively used for deposition of conductive ZnO films. 15,16 The electrical and photoelectrical properties of both ZnO/n-Si heterostructures and hybrid PEDOT:PSS/ZnO/n-Si heterostructures were studied. Capacitance-voltage (C-V) and current-voltage (I-V) measurements and impedance spectroscopy were applied to investigate the effect of PEDOT:PSS deposition on the potential barriers, the density of surface states, and the recombination lifetime at the ZnO/Si interface. The spectral dependences of the external quantum efficiency (EQE) were measured to understand the effect of PEDOT:PSS on the photovoltaic properties of the ZnO/Si heterostructure.

Sample Preparation
Zinc oxide films were deposited on n-type silicon wafers (5 ohm × cm) and glass slides by RF (13.56 MHz) magnetron sputtering. The deposition was carried out in an argon/methane gas mixture at a working pressure of 10 -1 Pa. The ratio of Ar/CH 4 during deposition was kept at 4:1. Further details of the deposition process can be found in Refs. 15,16 The thicknesses of the deposited zinc oxide film were 75 nm and 135 nm. For the electrical measurements, circular Ni dots (1.2 mm diameter) were deposited onto the ZnO surface by DC magnetron sputtering. The second contact was formed by rubbing Ga/Zn eutectic into the back side of the silicon wafer. This procedure was applied to prepare the Ni/ZnO/n-Si heterostructures. For electrical testing of the ZnO layers, several samples of ZnO/glass structures were fabricated in the same as the ZnO deposition.
A hybrid PEDOT:PSS/ZnO/Si heterostructure was formed by PEDOT:PSS solution spin-coating onto the surface of the ZnO film at 3000 rpm for 20 s, followed by thermal annealing at 140°C for 30 min under standard atmospheric conditions. Finally, a gold electrode (circular dot of 1.5 mm diameter) was thermally evaporated on the polymer surface as a top contact and a Ga/Zn eutectic was scratched into the n-type silicon as a back contact.
The choice of metal contact for each type of heterostructure was made taking into account the closeness of the metal workfunction to the position of the Fermi energy for PEDOT:PSS or ZnO to provide a good non-rectifying contact between the metal and the film.

Characterizations
The metal-semiconductor and metal-organic-semiconductor heterojunctions were characterized by C-V, conductance-frequency (G-ω) and I-V measurements with an Agilent 4284 LCR meter and a Keithley 6485 picoammeter. Resistivity and Hall mobility were measured using Van der Paw and Hall effect methods by the ezHEMS Hall effect measurement system (NanoMagnetics Instruments, UK).
Spectral and photoelectric measurements were carried out using the following equipment and standard samples: solar simulator; reference photovoltaic cells; set-up for photoelectrical parameters tests of solar cells; set-up for measuring of the spectral characteristics of photoconverters; and software to automate testing.
The spectral dependences of the external quantum efficiency EQE (λ) were measured using the sample's shortcircuit mode in the wavelength range ∆λ of 360-1200 nm, with the grating monochromator automatically maintaining constant flux of the monochromatic photons at a specified level. For this, the photon flux was divided, using an oscillating beam splitter with a frequency of 20 Hz, into the channel for maintaining the photon flux at a given constant level, and into the channel for measuring the EQE of the sample under investigation. In this case, the photon flux was modulated with a frequency of 20 Hz. The amplification on an alternating signal with a frequency of 20 Hz, followed by a lock-in amplifier, improved the signal-to-noise ratio (see supplementary Figure S1).

Results and Discussion
Determination of the Potential Barrier in the n + -ZnO/n-Si Interface Table I demonstrates the results obtained by the Hall effect and Van der Paw measurements on the ZnO films with different thicknesses. The experiment shows that the ZnO films are highly doped n-type ones with nearly the same doping concentration of about 1.5 × 10 20 cm −3 . The thicker ZnO film has a higher Hall mobility which can be associated with larger grain sizes in such polycrystalline films (see supplementary Figure S2). Figure 1a shows the I-V dependences for ZnO/Si and PEDOT:PSS/ZnO/Si heterostructures with a 75-nm-thick ZnO layer. The I-V characteristics show clear asymmetry reflecting the presence of a potential barrier inside the structure. To make sure that this barrier is not due to a possible Schottky barrier at the metal contact with ZnO (or PEDOT:PSS) film we measured the I-V dependences in co-planar geometry between two contacts on the top of the ZnO or PEDOT:PSS films, and the respective curves are shown in Fig. 1b. They show an ideally linear behavior, which confirms that the potential barrier at the metal-film interface is absent or negligibly small, although the absolute value of the resistance is much higher in the case of PEDOT:PSS film. We attribute this to the poor contact resistance between the metal and the PEDOT:PSS layer, which also leads to much lower currents being measured across the sample for the PEDOT:PSS/ZnO/Si heterostructures (see Fig. 1a). Figure 2 shows the C-V dependences for the ZnO/Si heterostructure with a 135-nm-thick ZnO layer measured at four different frequencies. We should note that a typical decrease of total capacitance towards negative bias, which reflects the widening of the depletion layer in silicon, is observed, either at low (1 kHz) or high (1 MHz) measurement frequencies. The absolute value of capacitance at negative bias drops by approximately an order of magnitude with the increase of frequency from 1 kHz to 1 MHz. The possible reason for this is that, at high frequencies and at depletion (negative) bias, a large portion of the ZnO film is rather more dielectric than a conductor because the lack of free carriers, and the capacitance of the depleted ZnO layer becomes connected in series to the capacitance of the rest of the structure. At low frequencies, minority carriers are able to follow an ac testing signal, 17 and the ZnO layer effectively behaves as a conductor. At intermediate frequency (100 kHz), due to the capacitive response from recharging of interface traps, capacitance behavior becomes more complicated, so we cannot observe the characteristic depletion tail in the C-V plot (see Fig. 2, green curve).
Following a Mott-Schottky approach, the intercept of 1/C 2 versus V plot with the x-axis gives the built-in voltage, V bi , at the silicon-zinc oxide interface. The barrier height, φ b , is connected with the built-in voltage, V bi , (or band bending, in other words) with the equation 18 : where E C is the energy of the bottom of the conduction band, E F is the Fermi energy in silicon, k is the Boltzmann constant, and T is the absolute temperature.
The Mott-Schottky plot for the Ni/ZnO/Si heterostructure is shown in the inset in Fig. 1. Here, the results are presented for the C-V curve measured at 1 MHz, when the contribution of interface traps can be neglected. Insets the scheme of respective measurements for the PEDOT:PSS/ ZnO/n-Si structure; for the ZnO/n-Si structure, the metal contacts lie on the ZnO layer (Color figure online). For an ideal case, taking into account a high doping level of zinc oxide, the donor concentration of 10 15 cm −3 in silicon and the intrinsic conduction band offset of 0.4 eV (due to the difference in electron affinity), at 300 K there should be a small built-in voltage of 0.14 V at the ZnO/Si interface. We derived the built-in voltages of 0.37 eV for 135-nm-thick ZnO and 0.34 eV for 75-nm-thick film from the C-V measurements. These values are higher than the expected theoretical ones, which means that, in our case, there exists some density of a fixed negative charge in the interface native oxide layer between ZnO and silicon, which increases the barrier height. This is also one of the reasons why we could not obtain a good barrier for the ZnO/p-Si system, because the fixed negative charge induces accumulation in the nearsurface layer instead of the expected depletion.
In PEDOT:PSS/ZnO/Si heterostructures, the measured V bi was 0.87 eV for 135-nm ZnO and 0.77 eV for 75-nm ZnO, i.e., the organic layer deposited on top of the ZnO film was found to increase the band bending at the ZnO/Si interface. This can be understood if we assume that, after the contact between the p-type polymer and the n-type semiconductor, the charge redistribution occurs as in a conventional p-n junction, leading to the formation of a depleted layer and the appearance of an additional potential barrier. In Ref. 19 various inorganic/organic hybrid heterojunctions were studied as conventional diodes, and it was shown that, for the ZnO/PEDOT:PSS heterostructure, the resulting barrier equals 0.398 eV. This value is very close to the difference between barriers with and without PEDOT:PSS.
Formation of the potential barrier after deposition of the organic polymer onto the ZnO/Si heterostructure is shown in Fig. 3, where the band diagrams are presented for the PEDOT:PSS/ZnO/n-Si (a) and ZnO/n-Si (b) heterostructures. These diagrams are plotted based on the work function values and band gap widths for PEDOT:PSS and ZnO reported in Refs. 9,20,21. The key question for the PEDOT:PSS/ZnO/n-Si structure is whether the electric field penetrates through the ZnO layer or it is screened due to free carriers, as can be expected for the high electron concentration obtained from Hall measurements. To check the uniformity of the current flow through the ZnO layer, we have carried out scanning spreading resistance microscopy (SSRM) measurements, which give information about the spatial distribution of the current through the film over the film surface. The SSRM revealed a strong inhomogeneity of the current flow, so that points with low resistance are separated by large areas with much higher (up to three orders of magnitude) resistance (see supplementary Figure  S3). This allowed us to draw the band diagram shown in Fig. 3a, and to suggest that, from the capacitance measurements, we get the information about band bending and space charge region width in silicon, and about the recharging of electrically active centers in the ZnO/Si interface. The ZnO film in the PEDOT:PSS/ZnO/n-Si heterostructure appears to be depleted of charge carriers, which also results in its much higher resistivity than in the case of the ZnO/n-Si heterostructure.
The barrier height, φ b , at the heterointerface can also be estimated from the I-V characteristics. For extraction of such parameters of the heterojunctions as barrier height, ideality factor, n, and series resistance, R S , we have used the technique of Cheung and Cheung 22 based on the thermionic emission model of current flow. According to their approach, the forward branch of the I-V characteristic obeys: where R S is the series resistance, n is the ideality factor, k is the Boltzmann constant, T is the absolute temperature, and q is the elementary charge. The dependence dV d(lnI) versus I is a straight line, the slope of which gives the series resistance, and the intercept with the y-axis gives the ideality factor. The ideality factor value can then be used for evaluation of the barrier height, φ b , from: where A is the area of the diode and A* is the Richardson constant.
Obtained in such a way, the values of φ b , R S , and n, together with magnitudes of the built-in voltages, V bi , extracted from the C-V measurements, are presented in  Table II. The difference between the bottom of the conduction band and the Fermi energy (E C − E F ) in silicon with a carrier concentration of 10 15 cm −3 at 300 K equals 0.26 eV. Therefore, for the structures without a PEDOT:PSS layer, we can expect from Eq. 1 slightly (by 0.1 eV) higher barriers, than derived from the I-V characteristics. The lower actual values of φ b can be explained by potential fluctuations at the ZnO/Si interface, the presence of which is also confirmed by the rather high values of the ideality factor. For the structures with a PEDOT:PSS layer, the barriers extracted from I-V are lower than the band bending calculated from the C-V characteristics. It seems that the thermionic emission model does not correctly describe the current through the heterostructure in this case. The detailed analysis of the current flow in such systems needs more study and is beyond the scope of this paper.
It should also be noted that the thicker ZnO film (135 nm) exhibits a much lower value of the ideality factor, reflecting a higher quality of the interface. This can be explained by the fact that, during the ZnO film growth, the film is subjected to hydrogen doping, and the longer this process, the longer the time during which the interface is subjected to hydrogen, which effectively diffuses to the interface and anneals the interface defects, improving the interface quality. The series resistance is also lower for the thicker ZnO film due to the higher carrier concentration and mobility (see Table I). Deposition of a PEDOT:PSS layer on top of the ZnO film results in the dramatic increase of R S because of depletion of the ZnO layer, as discussed above.

Determination of Interface State Density
The interface state density at the ZnO/Si interface was determined from the G-ω measurements using the classical approach of Nicollian and Goetzberger. 23,24 This technique is commonly used to study the electrically active states at the interface between the dielectric and the semiconductor. However, this method was primarily developed for SiO 2 / Si systems with negligible leakage currents. In the case of leaky dielectric layers or heterostructures, the G-F curves needs in correction, as described in Refs. 25,26. Before the conductance related to interface traps is plotted in G/ω versus frequency coordinates, the parasitic dc conductance should be subtracted from the measured value. The value of the dc conductance, G DC , is determined by differentiating the current-voltage dependences. Figure 4 shows the dependences of (G-G DC )/ω versus frequency for the Ni/ZnO/Si heterostructure measured at 300 K with different voltages applied to the top metal electrode. From the maximum value of the G/ω versus the frequency curve, the density of interface states, D it , is determined using the equation D it = 2.5(G/ω) max /(qA), where q is the electron charge and A is the sample area. 24 It can be seen that, at negative gate voltages, corresponding to depletion of the silicon surface layer, the maximum value is lower than at positive gate voltages, corresponding to the accumulation of the silicon surface by majority carriers. With the change in the gate voltage from positive to negative values, the energy position, at which a Fermi level crosses the ZnO/Si interface, is shifted from the edge of the band gap towards its center. The frequency, at which the maximum of G/ω is observed, changes when the gate voltage is swept, reflecting different characteristic time constants of recharging for interface traps situated at different energy levels within the silicon band gap.
It is evident that, at zero applied voltage, the Fermi level crosses the ZnO/Si interface in the upper half of the silicon band gap for the structure without a PEDOT:PSS layer (Fig. 3b), and in the lower half of the band gap (in strong inversion at E C -E F ~ 1 V) for the structure with a PEDOT:PSS layer (Fig. 3a). Figure 5 presents the dependences of interface states density on gate voltage for the Ni/ ZnO/n-Si samples with different thicknesses of ZnO (open symbols, left y-axis). When the bias is swept from negative to positive voltages, the cross-point shifts from the midgap to the bottom of the conduction band for the ZnO/n-Si heterostructure. The density of interface states near the midgap is lower than near the conduction band edge, as could be expected.
The dependences of G/ω max for the samples with a PEDOT:PSS layer are shown in Fig. 5 by solid symbols  (right y-axis). However, for the case of strong inversion, these values are not related to the recharging of the states within the band gap, but reflect the process of minority carrier generation in the space charge region of the silicon. 24 In this case, the G-ω measurements allow us to estimate the minority carrier generation time. 27

Determination of the Effective Carrier Lifetime in Subsurface Si Layer
The analysis of interface states recharging or carrier recombination processes, which are important for the operation of solar cells, can be carried out using impedance spectroscopy. This is a kind of ac electrical measurement technique when the complex admittance, Z, or complex impedance, Y, of the structure is measured at a given bias as a function of ac testing signal frequency. The dependence of Im(Z) = X on Re(Z) = R, or the dependence of Re(Y)/ω = G/ω on Im(Y)/ω = C are called Nyquist or Cole-Cole plots. [28][29][30] In Fig. 6 the dependence of G/ω on C is presented for the Ni/ ZnO/n-Si heterostructure with a 135-nm-thick ZnO layer. Different curves correspond to gate voltages varying from − 0.5 V to 0.5 V with the step of 0.2 V. From these complex impedance plots, the effective carrier lifetime can be estimated, because the highest point of the semicircle in the Cole-Cole plot corresponds to a characteristic carrier recharging frequency, the reciprocal of which is the effective carrier lifetime. 31 Figure 7 shows time constants derived from Cole-Cole plots for four samples. At positive voltages, the time constant is related mainly to the carrier exchange between the interface traps and the silicon conduction band. This is also confirmed by an exponential drop of the characteristic time with increasing the gate voltage. 32,33 At negative biases, two different processes can be distinguished. The first one is related to the interface traps and is observed in all the samples, and the second one, with much shorter characteristic times, is observed in the structures without a PEDOT:PSS layer, and is due to the recharging of the border traps. 32 The impedance analysis shows that, after deposition of a PEDOT:PSS layer, these traps become inactive under operational conditions, and the respective fast recharging process is eliminated. This phenomenon is associated with the formation of the inversion layer and with the Fermi level location near the valence band at the ZnO/Si interface, due to the increase of the silicon surface potential barrier. In this case, the characteristic time determined from the Cole-Cole diagram corresponds to the generation time of the minority carriers.
Photoelectrical Properties of n + -ZnO/n-Si and PEDOT:PSS/n + -ZnO/n-Si Heterostructures. where q is the electron charge, k is the Boltzmann constant, p 0 and n 0 are the concentrations of minority and majority charge carriers, respectively, and n is the ideality factor. For the case of low excitation levels (Δp < < n 0 ), which is the case in our study, the value of V OC is determined by the first term of Eq. 1, and, since Δp/p 0 > > 1: In turn, the effective lifetime of excess minority carriers is proportional to Δp: τ eff ~ Δp.
Thus, a significant increase in the open-circuit voltage indicates a significant increase in the effective lifetime of excess minority carriers in the ZnO/Si samples after the deposition of the PEDOT:PSS layer. Since the effective lifetime is expressed as: where S Σ = S f + S b + V SCR , S Σ is the effective surface recombination rate, which is equal to the sum of the surface recombination rates on the front S f and back S b surfaces and the recombination rate, V SCR , in the space charge region (SCR) of the heterojunction, τ b is the volume lifetime, and d is the sample thickness. When applying the PEDOT:PSS layer, the volume lifetime, τ b , does not change; therefore, a significant increase in the effective lifetime indicates a decrease in recombination losses on the frontal surface and the recombination rate in the SCR. Thus, the deposition of a PEDOT:PSS layer on the frontal surface of samples of ZnO/Si heterostructures effectively electrically "passivates" the near-surface region of these structures, due to the strong increase of the surface potential barrier. In Ref. 35 , it was shown that the surface recombination rate strongly decreases when the silicon interface is under the strong inversion condition.
Finally, the spectral dependences of the EQE were measured, and the resulting spectra are presented in Fig. 8. The decrease of EQE at a wavelength longer than 600 nm is evidently related to the substantial drop of transmittance in this spectral range after deposition of the PEDOT:PSS layer, which limits the light absorption in the silicon. Visually, it is manifested as the dark-blue, almost black, color of the PEDOT:PSS film deposited onto the ZnO surface, although the same films deposited on silicon or glass are transparent. In Ref. 36 , the PEDOT:PSS/ZnO nanocomposites were studied which also had dark-blue and black colors. In Ref. 37 , the electrochromic device was proposed by the addition of Fe 2 O 3 oxide into the PEDOT:PSS film, and it was shown that, applying the reduction potential changes, the color of the film becomes dark blue, and the oxidation potential makes the film transparent. The transmittance spectrum for the reduced state of PEDOT:PSS shows a sharp drop in wavelengths in the 550-750 nm range. Therefore, it is reasonable to suggest that, in our case, after the deposition of PEDOT:PSS onto zinc oxide, the PEDOT:PSS film is being reduced, changing its color and causing effective light trapping. The reduction-oxidation reaction is probably in this case due to the high chemical activity of zinc. Another consequence of this reaction is the additional oxidation of ZnO, which can result in a decrease in the concentration of oxygen vacancies, which are often considered as intrinsic electron dopants. This can further increase the layer resistance and suppress the short-circuit current.
The sharp increase of EQE in the short wavelength range after the deposition of PEDOT:PSS can be explained by the effective passivation of the near-surface region of these structures noted above. An additional contribution can be made by carriers generated and spatially separated in the PEDOT:PSS and ZnO layers under illumination, which is confirmed by the increase of V OC after the PEDOT:PSS deposition. ZnO is sensitive to violet and ultraviolet radiation because of its high band gap, but, in the absence of PEDOT:PSS, the generated carriers do not contribute to the total photogenerated current. The potential barrier at the PEDOT:PSS/ZnO interface drives the carriers to the silicon, and the electrically "passivated" interface suppresses surface recombination of the photoexcited carriers. Therefore, the quantum efficiency in the short wavelength range increases. EQE in the long wavelength range (> 600 nm) mainly originates from the charge carriers generated in the SCR of the silicon substrate, since PEDOT:PSS and ZnO demonstrate weak optical absorption in this spectral range. Therefore, some portion of light can be reflected by the rough PEDOT:PSS layer, causing a slight decrease in the EQE for the PEDOT:PSS/ZnO(75 nm)/n-Si sample compared to the reference sample without polymer coverage. Lower values of EQE for the ZnO(75-nm)/n-Si sample as compared to the ZnO(135-nm)/n-Si sample can be explained by the higher series resistance, R S , and slightly higher built-in voltage, V bi (see Table II). Additionally, the ideality factor for the structure with the thinner film is higher, reflecting the poor interface quality which results in the higher interface recombination rate.

Conclusions
We have studied the effect of the deposition of the PEDOT:PSS layer on the surface of zinc oxide in ZnO/n-Si heterostructures. It was found that deposition of PEDOT:PSS increases the barrier, turning the ZnO/Si interface into strong inversion, which effectively electrically "passivates" the near-surface region. The border traps with a short time constant are not detected after the deposition. Photoelectric measurements have shown that the addition of the organic layer significantly, by 54-180%, increases the open-circuit voltage, but at the same time reduces the short-circuit current due to the low transmission and high resistance of the PEDOT:PSS film. In the blue range, the EQE increases due to both the effective electrical "passivation" of the nearsurface region of these structures and the additional contribution of carriers generated in the PEDOT:PSS and ZnO layers.