You are on page 1of 6

Issues dealing with the effect of a parallel resistance, and the relation between surface recombination and Voc.

Ive run several simulations assuming infinite mobility for electrons and holes, and varying the values of an external series and a parallel resistance, in order to fit the curve with finite mobilities. These curves are shown in Fig.1 (the units of Rs and Rp are given in m2). As a first approximation, the slope at low voltages can be reproduced adjusting Rp but other parameters like Voc, Jsc, FF and the slope near the open circuit condition cant be easily fitted by simply modifying Rp and Rs. The next step could be adding an external diode and adjusting its current (D0), nevertheless the convergence of the fit with these 3 parameters (Rs, Rp and D0) becomes very cumbersome.

Fig. 1 If the mobility becomes very large, Voc depends on surface recombination. Fig. 2 shows the logarithmic variation of Voc with Pfoto, for two extreme situations, namely without and with an infinite surface recombination. Sc in the plots stands for the surface recombination velocity in m/s. It is also plotted the differential Voc (deltaVoc) for these conditions. It can be seen that deltaVoc doest change with Pfoto, so I dont think that useful information can be extracted from this kind of experiment.

Fig. 2 Perhaps it is better to investigate the relation between bulk and surface recombination by varying the recombination parameter kf or the thickness. Fig. 3 shows Voc against kf for null and infinite surface recombination and finite mobilities. If kf is large enough the surface recombination plays no role, but at as soon as kf decreases, Voc saturates to a finite value (1,07V in this case). This behavior is well known in standard cells when thinking kf as a bulk lifetime or rate.

Fig. 3

The dependance of Voc with Pfoto assuming finite mobility, for the situations of zero and infinite surface recombination but assuming a low kf of 1E3 s-1 is plotted in Fig. 4. Once more, deltaVoc doesnt vary significantly with Pfoto. With the same data, the relationship between Jsc and Voc yields an ideality around 2,3.

Fig. 4 Regarding the possibility that Voc saturates to Vbi (1,4V in this case) when Pfoto increases in the presence of surface recombination, the previous graphs show that there is no evidence of saturation for Pfoto up to 10kW/m2, taking into account either small (kf=1E3) or large (kf=1E6) bulk recombination. Moreover, a linear extrapolation to 1,4V requires a Pfoto of 40MW/m2! Therefore, it seems that the slope of Voc with and without surface recombination remains constant in the practical photon flux range. Perhaps it is missing another condition we must take into account that reveals a saturation regime for Voc. Issues regarding the charge extraction analysis developed by Shuttle et al. (APL 93, 2008)

I tried to reproduce the plots of this paper by simulating a P3HT:PCBM (optical constants obtained elsewhere) solar cell using the same parameters the author describes and assuming typical values for other parameters. As a first step I tried to fit the J-V curves that Kim et al (Nature Materials, Vol 5, 2006) obtained experimentally. Figure 5 shows a simulated J-V with Jsc=7.1 mA/cm2, Voc=0.63, FF=0.51 and eta=2.37%. Kims best cell has Jsc=7.28 mA/cm2, Voc=0.57, FF=0.50 and eta=2.4%

Fig. 5 Figure 6 represents the excess electrons concentration (delta n=n-n0) vs Voc (bottom axis) and Pfoto (top axis). Between 0.57 and 0.64 V it is found a linear relationship in the semilog plot. Shuttle formulated an exponential dependance (at least for Voc between 0.45 and 0.6 V) given by nVoc=n0exp(Voc) and obtained n0=2.8E14 1/cm3 and =9.3. My simulations yielded n0=8.9E13 1/cm3 and =6.4 for Voc bounded by 0.57-0.64V.

Fig. 6 The next plot is delta n vs V in the dark (Fig. 7). The inset shows the linear regime in the semilog scale which corresponds to the voltage range described in the paper, i.e. 0.45-0.7V. Shuttle proposed the following fit ndark=ndark0 exp(V) with ndark0=5E14 1/cm3 and =7.4. My fit gives ndark0=6.6E14 1/cm3 and =2.86.

Fig. 7

The last plot represents the dark current density (Jdark) vs delta n (Fig.8). Shuttle proposed the power law Jdark=dqkdarkndark, where d is the active layer thickness (in this case 150nm), q the elementary charge, kdark and are fitting parameters whose values are 8.9E-20 and 2.58, respectively. Note the lack of units of kdark. My fit for the log-log plot delivers kdark=8.31E-26 1/(sm3) and =2.34. The value of matches the ideality under illumination (obtained by running several simulations at different light intensities and plotting Jsc vs Voc).

Fig. 8

You might also like