Polyaniline on Crystalline Silicon Heterojunction Solar Cells

Organic/inorganic heterojunction solar cells were fabricated on the (100) face of n-type silicon crystals using acid-doped polyaniline (PANI) with widely varying conductivities. For films with conductivities below 10−1S∕cm, the open-circuit voltage VOC increases with increasing film conductivity as expected when VOC is limited by the work function of the film. Extrapolation of these results to the higher conductivity films indicates that PANI could support VOC of 0.7V or larger. VOC measurements for the cells with higher conductivity PANI saturated at 0.51V. We speculate that uncontrolled surface states at the PANI/Si interface are reducing these values.

Heterojunctions to crystalline silicon ͑c-Si͒ have long been of fundamental interest, and amorphous silicon ͑a-Si: H͒ / c-Si heterojunctions are now applied for some commercial "heterojunction with intrinsic thin layer" ͑HIT͒ solar cells. 1 These heterojunction cells have achieved efficiencies as high as 21.8%, which approach the best results ͑24.7%͒ for homojunction c-Si cells; 2 there is a sizable literature on the properties of this type of heterojunction. 3 Conducting polymer/c-Si heterojunctions could also have attractive properties, and early work on polyacetylene/c-Si heterojunctions suggested open-circuit voltages ͑V OC ͒ under solar illumination as large as 0.64 V. 4 While these results were comparable to those for homojunctions at that time, there has since been only sporadic work on polymer/c-Si solar cells and closely related devices. [5][6][7][8][9] In this paper, we report our research on polyaniline/ n-type silicon heterojunction solar cells. Doped polyaniline ͑PANI͒ conducts holes well ͑conductivities as high as 10 3 S / cm have been reported͒, and this near-metallic 10 ͑or nanometallic 11 ͒ character along with its stability in air makes polyaniline interesting for this application. PANI/c-Si heterojunctions have been examined as gas sensors. 5 In the present work, we primarily explored the open-circuit voltage V OC of these heterojunctions under illumination. For PANI films with smaller conductivities ͑Ͻ10 −1 S/cm͒, the dependence of V OC upon conductivity was consistent with limitation by the work functions of the two materials; work function effects on the performance of polymer light-emitting diodes incorporating polyaniline were reported previously. 12 Extrapolation of the measurements to the higher conductivity films suggests that PANI films could support V OC values of 0.7 V or higher, which is about the current limit achieved with a-Si:H/c-Si heterojunctions. The largest V OC we obtained was 0.51 V. Direct measurement of the built-in potential V BI in these cells indicates that PANI/Si surface states are limiting V BI and V OC .
We used "emeraldine salt" dispersions of polyaniline in xylene from Sigma-Aldrich, Inc. ͑film conductivity Ϸ 10 1 S/cm͒ and from Ormecon GmbH ͑ Ϸ 10 2 S/cm͒. 13 To make solar cells, a dispersion was spin coated onto the freshly HF-etched, polished surface of n-type, ͑100͒ silicon wafers 14 with an evaporated aluminum back contact; the area of the PANI film was typically about 1 cm 2 . The films were dried in air following the manufacturer's instructions ͓80°C for 24 h ͑Aldrich͒ and 100°C for 1 min ͑Ormecon͔͒, and a 1 mm diameter gold contact was evaporated onto the PANI film. Films spun from the Ormecon dispersion were 0.35 m thick. 15 We prepared doped PANI films with widely varying conductivities ͑10 −5 to 10 1 S/cm͒ from the Aldrich dispersion by increasing its xylene concentration. This dilution ͑from 2% PANI by weight to 0.3%͒ reduced the thickness of the spun films from 0.4 to 0.1 m, which appears consistent with the effect of the reduced viscosity for diluted films. Scanning electron microscopy indicated that these films were continuous, although with somewhat differing morphologies for the differing dilutions.
The dependence of the current density upon voltage is shown in Fig. 1 for a PANI/n-Si diode ͑Ormecon dispersion͒ for the dark and for tungsten-halogen illuminations through the PANI ͑intensity of 0.1 W / cm 2 ͒. The saturated photocurrent density is 17 mA/ cm 2 , which is reasonably consistent with the photon flux that reaches the c-Si after filtering by the PANI film. The open-circuit voltage V OC is 0.51 V, which is about 0.1 V below the values observed for c-Si based cells with metal back electrodes. 16 The low fill factor a͒ Electronic mail: wwang@phy.syr.edu of 41% is due to the lateral resistance of the PANI film ͑and the absence of a top metallic grid͒. It is of interest that the light and dark curves cross at about 0.6 V; while such "crossover" is not typical for homojunction c-Si diodes, it has been reported previously for heterojunctions of ͓poly͑3,4-ethylenedioxythiophene͒:poly ͑styrenesulfonate͔͒ ͑PEDOT:PSS͒ on a-Si: H. 6,9 In Fig. 2, the open-circuit-voltages V OC ͑at intensity 0.58 W / cm 2 ͒ for each of a series of PANI/c-Si diodes are shown as a function of the conductivity of a PANI film casts onto glass using the same dispersion. V OC was almost constant ͑at V OC sat = 0.52 V͒ as the conductivity of the PANI films varied from 10 −1 to 10 2 S / cm. Below a threshold conductivity c = 0.05 S / cm, the data are fitted satisfactorily by the form where k B T / e = 0.025 V is the ratio of the Boltzmann thermal energy to the electron charge at room temperature. The large range of conductivities-about 10 5 -accessible by fairly modest dilution of emeraldine salt dispersions may be a useful complement to the similarly large effects found in blends of polyaniline with nonconductive ͑filler͒ polymers. 17,18 For Ͻ c , the logarithmic dependence of V OC is consistent with the idea that V OC is limited by the built-in potential across the diode, 19 where E F Si is the Fermi level of the ͑n-type͒ c-Si and E F PANI is the Fermi level of the p-type PANI film. Assuming that the conductivities of the PANI films are mainly determined by their Fermi levels ͑and work functions͒, we write where E V PANI is the transport energy for holes in the PANI films. We obtain for V OC ͑in the built-in potential limit͒ that accounts for measured slope k B T / e = 0.025 V of the dependence of V OC on the natural logarithm of the film conductivity.
We checked that Eq. ͑3͒ does describe the temperaturedependent conductivities of our films. The results are presented in Fig. 3, which show that lower conductivity films have larger activation energies. The fitting curves roughly converge at a negative "Meyer-Neldel" temperature T M-N Ϸ −180 K and a conductivity M-N =2ϫ 10 2 S / cm. Inverse Meyer-Neldel behavior ͑i.e., negative values for T M-N ͒ is typically observed when the ͑temperature-dependent͒ Fermi level lies in a bandtail region. 20 Assuming that M-N estimates the fundamental bandedge conductivity 00 , 21 we can use Eq. ͑4͒ to infer the difference in energy levels between Fermi level in the c-Si and the hole transport level of the PANI; the result E F Si − E V PANI = 0.74 eV is consistent with work function estimates in heavily doped PANI. 22 We now turn to the plateau value for V OC of around 0.5 V that we found ͑Fig. 2͒ for PANI conductivities Ͼ 10 −1 S / cm. This value is of importance for potential applications. V OC is reduced by our use of a metal back contact ͑instead of a back surface field͒ and by photon absorption by the top PANI layer; we estimate these reductions to add to about 0.05 V. The main effect reducing the saturated value for V OC appears to be that the built-in potential V BI did not reach the values of 0.74 V or larger that would be inferred from Fig. 2 for the highest conductivity films ͑from Ormecon GmbH͒. We determined the built-in potential for the heterojunctions incorporating high conductivity PANI using capacitance-voltage ͑C-V͒ measurements and a standard Schottky fitting ͑1/C 2 vs V͒. We obtained V BI = 0.60 V, which is substantially lower than the extrapolated V OC = 0.74 V and indicates that for larger conductivity films, V BI is lower than would be inferred from the conductivity.
We therefore speculate that the Fermi level at the interface of the PANI and the c-Si interface is not as low as would be anticipated for "bulk" PANI, thus leading to reduced values of V BI across the c-Si. Such an effect could be due to a band of surface states that "pin" E F at the interface for the heavily doped material. Angle-resolved ultra- violet photoemission measurements do indicate interfacedependent electronic states for PANI on oxidized c-Si surfaces. 23 It is conceivable that these effects could be mitigated by grafting of PANI onto the c-Si surface 24 or other strategies.
This research has been supported by the National Renewable Energy Laboratory ͑subcontract No. NDJ-2-30630-24͒ and the Syracuse Center of Excellence in Environmental and Energy Systems. We thank Bernhard Wessling ͑Ormecon, GmbH͒ for a critical reading of the manuscript and for samples of Ormecon's polyaniline dispersions, Qi Wang ͑National Renewable Energy Laboratory͒ and Dmitry Poplavskyy ͑Innovalight, Inc.͒ for discussions, and Manfred Schmidt ͑Hahn-Meitner Institut͒ for c-Si wafers.