Time and frequency resolved photoconductivity studies on amorphous semiconductors

• Main Author: Webb, Dennis
• Published: Abertay University 1995
• Subjects: 530.41; Solid-state physics
• Online Access: https://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.259883

Time and frequency domain photoconductivity experiments and analysis assuming the multiple trapping transport model are used in a complementary fashion to investigate the density of states (DOS) and associated trapping and recombination parameters of amorphous semiconductors. The solution of the linearised multiple trapping rate equations is straightforward in the frequency domain, and it is simple to invert the modulated photoconductivity (MPC) response to obtain an accurate DOS. It is proposed that a novel, wide energy range, accurate DOS probe is realisable by applying the Frequency domain inversion to the Fourier transform of transient photoconductivity (TPC) data. A numerical Fourier integral procedure appropriate to the nature of TPC data sets is described and tested. Application of the Fourier transform method to TPC measurements on undoped a-Si:H produces a calculated DOS distribution covering half the mobility gap. The DOS consists of an 0.03 eV exponential band-tail, and a bump of 4.4x10¹⁶cm^-3 states at 0.65eV below E_c, attributed to dangling bonds. The method detects the bump despite there being no related emission feature in the TPC decay. The accuracy of the calculated DOS is demonstrated to depend on the size of the missing contribution to the Fourier integral from the signal at times shorter than the experimentally observed minimum.<br/>An <i>undershoot </i>or sign reversal of TPC measured with optical bias on undoped a-Si:H for steady state generation rate G_ss ≥10¹⁸cm^-3s^-1 is observed at a time t_us which varies inversely and sublinearly with G_ss. From the linearised frequency domain analysis such an undershoot is expected for bimolecular recombination and recombination coefficient C_R>C_n, the band-tail trapping coefficient. The analysis predicts a related sign reversal in the imaginary part of the <i>Fourier </i>transform spectrum at a frequency ω∝=C_R, which is also observed. Values for C_R mostly in the range 1-3x10^-8cm³s^-1 are calculated.<br/>Undoped a-Si:H MPC spectra exhibit a <i>phase peak</i> at a frequency ω_peak which varies superlinearly with G_ss. The phase peak accompanies a transition in the MPC response from DOS dependent to DOS independent behaviour for ω<ω_peak, in which régime the calculated DOS varies as ω^0.43. For a DOS consisting of a single trap species it is expected that the transition frequency be around lOOx smaller than observed and the calculated DOS be proportional to co at low frequency. Assigning a trapping coefficient of 10-100C_n to defect states accounts for the transition frequency but not the variation of the calculated DOS for ω<ω_peak. The calculated DOS for ω<ω_peakmatches the DOS calculatedfrom TPC data.<br/><i>Steady state</i> photoconductivity measurements on a-As₂Se₃ are interpreted to indicate a set of defects sited at 0.66eV above E_v. TPC measurements exhibit no related emission feature. The DOS calculated from the Fourier transform method is a featureless exponential but is unreliable because of the limited measurement bandwidth. The magnitude of the TPC is too small for the DOS to consist solely of an exponential tail extending to the band edge. The DOS calculated from MPC measurements is also a featureless exponential but it is suggested that the MPC response is in the DOS independent régime.