s21_Amorphous Silicon

Opto-Electronic Properties of Amorphous Silicon-Germanium Alloys

Ed Nguyen

University of Minnesota, Methods of Experimental Physics, Spring 2021

Introduction

Amorphous semiconductors possess a wide variety of benefits with the major utility being entirely created with thin-film technology. Due to this fact, they are able to be spread over a large surface area on a variety of substrates to create economical solar cells. However, these materials suffer from light-induced degradation, also known as the Staebler-Wronski Effect (SWE). This effect is demonstrated in the following figure:

The sample starts out at state A and after light-soaking (LS), it arrives in state B. It should be noted that the conductivity of the sample was reduced by approximately an order of magnitude after LS. This effect was reversible after annealing to 470 K from room temperature for a couple of hours. Because of the Staebler-Wronski Effect, the efficiency of amorphous solar cells degrades over time with light exposure. A popular approach to minimize SWE involves using tandem solar cells. This method requires using amorphous silicon-germanium alloys to adjust the optical gap. One of the aspects we would like to examine is the opto-electronic properties of these materials.

Theory

    1. Staebler-Wronski Effect (SWE)

    1. Photoconductivity with variation of light intensity

The relationship of photoconductivity and light intensity is:

The Staebler-Wronski Effect [1] [6] was discovered in 1977, and it is essentially the degradation of amorphous silicon due to light exposure. The sample started out in state A, in the dark, after which, it experienced light soaking for a couple of hours. The light was then shut off and the sample was now in state B, with a conductivity lower than state A’s. It was observed that the conductivity of a:Si-H was reversible after annealing at 150℃ for several hours.

, [5]

where

is the photoconductivity, F is the varying light intensity, is the lamp’s intensity, and ‘ℽ’ is the power-law relation between and F. ℽ was determined to be 0.5 for pure a-Si:H and 1 for pure a-Ge:H [5].

ℽ can be obtained as the slope by plotting

vs. on a log-log scale. ℽ was obtained for 6 samples with x = 0, 0.17, 0.48, 0.75, 0.93, and 1


    1. Conductivity with variation of Temperature

In annealing the samples, their conductivity-temperature dependence is dictated by two separate regimes:

    1. Arrhenius Relation

Arrhenius’ Equation states that:

(2)

where σ is the conductivity,

is the initial conductivity, T is the temperature, is the initial temperature, is the activation energy, and the Boltzmann’s constant k.

If we rearrange (2) for 2 points at different conductivities and temperatures, the resulting expression for Ea is:

In addition, since we measured the conductivity’s dependence on temperature,

can easily be determined.

    1. Power Law Relation

The power-law relation states:

[5]

where σ is the conductivity,

is the initial conductivity, T is temperature, is the initial temperature, and the power-law factor is n.

In a similar manipulation to the Arrhenius relation, we can derive an expression for n

Furthermore, n can be easily obtained by plotting the conductivity’s dependence on temperature on a log-log scale for each sample.

Since there are 2 regimes, we would need to determine the appropriate regime for each sample by assessing their activation energies. In addition, illumination would introduce additional defects into the samples, which would result in an increase in photoconductivity and thus, the activation energy [5] [7].

Experimental Apparatus

Figure 2 shows a diagram of the experimental setup. The samples were placed on a heater, which is housed inside a container. Silver paint was applied onto the copper block’s surface to ensure proper thermal contact with the samples. After setting the samples, the top cover was shut and air pumped out of the container until the pressure inside was < 30 mTorr, which was indicated by the vacuum gauge. A circuit is connected across the samples for measuring the conductivity. Both the heater and circuit are software controlled through LabView. After reaching the desired pressure range, the cover was removed and white light of different intensities from a tungsten-halogen lamp (set at 12 V, with an intensity of 75 mW/cm2) was shined on the samples. The intensities were changed by stacking or removing Neutral Density filters. After obtaining data for all intensities, the samples were annealed at 470 K and cooled to 305 K, which put them in the annealed state for repeated measurements if need be. Additionally, the conductivity and temperature will be measured during annealing. This process was repeated for all samples and the relationship between photoconductivity and light intensity can be obtained. Furthermore, the relationship of conductivity and temperature will also be acquired for analysis.

Data

1. Variation with Light Intensity

Thus, we are able to conclude that there isn’t a smooth transition in ℽ with Ge content, but rather two separate regimes. It should be noted the samples marked in red were older and the other samples were made in a series at a later time. Further experiments with more accurate Ge concentration are needed to determine the actual cutoff.

  • Two separate regimes for ℽ:<17% Ge: ℽ is closer to the pure amorphous silicon sample.

  • >17% Ge:ℽ is approximately the same for the high Ge films, especially after 73% Ge.

2. Variation with Temperature

  • Figures 5 and 6 demonstrate a difference in activation energy in state A and state B for pure amorphous silicon. In higher Ge concentration films, there was almost no change in the activation energies with light soaking (indicated by Figure 4), but there was an obvious decrease with the Ge content. This is likely reflecting the decrease in the mobility gap with Ge alloying. In other words, as we introduce more Ge content - more defects in the samples - we see a corresponding decrease in the activation energy. In Figure 4, we can see that for the majority of the samples, the ratio is less than unity, indicating a decrease in conductivity after light exposure due to the SWE. Thus, for the Ge alloys, their activation energy stays the same while their conductivity reduces after illumination. This indicates that Fermi energy is pinned by a large density of states in the mid-gap even though we tried to increase that density with illumination. Therefore, we concluded that the Ge alloyed samples follow the power-law instead and the pure amorphous silicon sample obeys the Arrhenius relation.This postulation is supported if we consider Figure 7:In a-Si:H, there exists a low dangling bond density, and charges are excited from the Fermi Energy to the conduction band edge.

  • In a-Ge:H, the dangling bond density is high, and thus charges travel by means of multi-phonon hopping through the defect states at the Fermi Energy.

In materials with a low dangling bond density, it isn’t feasible for charge carriers to hop through the dangling bonds, and thus, they would require thermal activation. With a high dangling bond density, however, the gaps in vacant bonds are close enough to enable multi-phonon hopping, eliminating the need for thermal excitation. Due to the addition of Ge, the dangling bond density inside our alloys was sufficient to support hopping, and thus the conductivity (σ) is related to the temperature (T) by a factor of n. Conversely, the low dangling bond density in a-Si:H requires the sample to be thermally activated to transport charge. As a result, the conductivity is instead thermally activated.

Conclusion

Our experiment was able to conclude that the high Ge content alloys follow a power-law relation with the conductivity while the Si-rich alloys obey the Arrhenius relation. Not only that, we were also able to determine that there wasn’t a smooth transition of ℽ in the higher Ge content samples, instead, they followed 2 separate regimes, one for samples below 17% Ge and the other after that. With the experiment concluded, we will be looking into other avenues, such as determining the Ge concentration cutoff between the two regimes of ℽ. To do so, we are considering using additional alloys and a finer variation of Ge content, especially in the silicon-rich regime, and possibly extend this experiment to other alloyed films.

References

[1] Agarwal, Satish, and Shobit Omar. “Forty Years of the Staebler–Wronski Effect.”

[2] Almeriouh, Y., et al. “Light Intensity Dependence of the Photoconductivity of Hydrogenated Amorphous Silicon.” Philosophical Magazine B, vol. 63, no. 5, 1991, pp. 1015–1030.

[3] Everitt, B., and J. Kakalios. “Test of the Thermal Equilibration Model for Persistent Photoconductivity in Doping-Modulated Amorphous Silicon.” Physical Review B, vol. 43, no. 8, Mar. 1991, pp. 6820–23. APS.

[4] Janzen, A. F. Solar Energy Conversion II: Selected Lectures from the 1980 International Symposium on Solar Energy Utilization, London, Ont., August 10-24, 1980. Pergamon Press, 1981.

[5] ROSE, A. Concepts in Photoconductivity and Allied Problems. 1978.

[6] Staebler, D. L., and C. R. Wronski. “Reversible Conductivity Changes in Discharge-Produced Amorphous Si.” NASA/ADS.

[7] Leandro R. Tessler, Fernando Alvarez, Temperature and light intensity dependence of photoconductivity in off-stoichiometric hydrogenated amorphous silicon nitride.