Abstract
A theoretical model, combining trapping/detrapping and recombination mechanisms, is formulated to explain the origin of random current fluctuations in siliconbased solar cells. In this framework, the comparison between dark and photoinduced noise allows the determination of important electronic parameters of the defect states. A detailed analysis of the electric noise, at different temperatures and for different illumination levels, is reported for crystalline siliconbased solar cells, in the pristine form and after artificial degradation with high energy protons. The evolution of the dominating defect properties is studied through noise spectroscopy.
Introduction
The study of fluctuation mechanisms is a powerful and nondestructive spectroscopic tool, useful to provide information on the dynamic behaviours and the kinetic processes of the charge carriers in condensed matter^{1}. Lowfrequency electric noise analysis was proposed, twenty years ago, to determine quality of solar cells and photovoltaic modules^{2,3,4}. Subsequently, it was used for the reliability estimation of photovoltaic devices related to microplasma detection^{5} and of iron disilicide heterojunction solar cells^{6}. More recently, the same tool was employed to evaluate transport properties and degradation processes in several other systems, such as organic lightemitting diodes^{7}, polymer:fullerene solar cells^{8,9}, bulk homogeneous semiconductors^{10,11} and polymer/carbon nanotube composites^{12}.
In semiconducting devices, often a 1/ftype noise spectrum is observed. Semiempirical models have been formulated to correlate the current noise to mobility fluctuations^{13,14}, to the number fluctuation of free charge carriers within the space charge region^{15}, or to the semiconductor surface^{16}. However, Hsu^{17} and van der Ziel^{18} were the first to relate the 1/ftype noise in the metaloxidesemiconductor system and in pn junctions to the fluctuation of the defect state population. In this work, it is shown that lowfrequency noise spectroscopy can be used as a sensitive tool for the investigation of the properties of these defect states. In particular, a noise model is proposed to interpret experimental data in the case of pristine and artificially degraded siliconbased photovoltaic devices. The reported results show that the formation of the defects, activated under illumination or charge carrier injection, is related to longterm degradation of the solar cells. Noise analysis can also provide interesting information on radiation damage and can be used for a detailed temperaturedependent electrical characterization of the charge carrier capture/emission and recombination kinetics. This aspect represents an advantage of the fluctuation spectroscopic technique, which gives the possibility to directly evaluate the cell health state.
Noise Model of Solar Cells
Fluctuations in pn junction solar cells strongly depend on the diode geometry^{13}, on the photogeneration processes and on the bias voltage^{15}. In presence of defect states, the trapping and recombination mechanism of the charge carriers produces current fluctuations, that can be the main noise source^{11}. The amplitude of the random current fluctuations Var [I] is, therefore, related to the occupation probability f_{trap} of the traps. Moreover, the recombination pathways from the conduction band to the valence band generate a drop of the minority charge carriers stored in the base material and this effect also influences Var [I]. The main contribution of the traps can be modeled by two energy levels, which act as trapping (E_{T}) and recombination (E_{SRH}) centers. These centers are located above the intrinsic Fermi level E_{i} (see Fig. 1(a)) and are responsible for the transitions between the defect states and the conduction E_{C} and valence E_{V} bands. Under injected or photogenerated charge carriers, the E_{T} energetic states, having density N_{T}, are able to capture and emit electrons from the conduction band. At the same time, the E_{SRH} states, having density N_{SRH}, work as ShockleyReadHall (SRH) type recombination centers with an associated lifetime τ_{SRH}^{19}.
At low frequencies and in forward bias condition, the solar cell ac equivalent circuit, representing the SRH recombination processes, is simply composed by the parallel connection of a differential resistance R_{D} and a capacitance C_{μ}, representing the contribution of the minority charge carriers into the base material^{8}. The fluctuating traps with energy level E_{T} can be modeled by a series RC circuit in parallel to C_{μ}^{20}, as shown in Fig. 1(b) where the cell shunt R_{sh} and series R_{S} resistances are also displayed. Since for a typical solar cell , its effect on current transport can be considered negligible. Therefore, the recombination phenomena, associated to the defect states, can be well described by the time constant τ_{eff} = R_{D}C_{μ}. The density of empty traps into the base material introduces a capacitance contribution C_{trap}, which is related to the variation of f_{trap} with respect to the quasiFermi level under nonequilibrium condition (charge carrier injection). The resistance R_{trap}, instead, represents the kinetic factor of the traps^{20}, which are governed by a characteristic trapping/detrapping time τ_{trap} = R_{trap}C_{trap}. In the frequency region below (2πτ_{trap})^{−1}, the capacitive contribution of the fluctuating trap states becomes significant and the noise spectroscopy is able to follow the charge carrier transfer from the conduction band to empty traps and viceversa (see Fig. 1(c)). This mechanism leads to fluctuations in the number of charge carriers and contributes, therefore, to a current fluctuation that can be measured by the external contacts. The effect of these traps can be modeled as a currentnoise source i_{n}, which represents the charge carriers transfer between C_{trap} and C_{μ} (see Fig. 1(d)). Thermal noise contributions and , related to R_{S} and R_{sh}, respectively, are usually negligible for solar cells, as well as the shot noise present in pn junctions^{1}. As a consequence, the solar cell equivalent noise circuit can be assumed as simply composed by the parallel connection of C_{μ}, R_{D} and i_{n}, as shown in Fig. 1(d). The noise is generated by the change in the number of carriers, that are typically present in the cell, due to the dc bias or light absorption. A characteristic 1/f frequency spectrum is observed, with the presence of a cutoff frequency evaluated as^{8} f_{x} = (2πτ_{eff})^{−1}.
In this framework, the amplitude Var [I] of the 1/f noise component can be defined as the sum of a dark (Var [I_{dark}]) and a photoinduced (Var [I_{ph}]) contribution. Under dark condition and on forward biasing of the solar cell, the Var [I_{dark}] value is influenced by the emission rates e_{n,p} (dimensions of s^{−1}) and volumetric capture coefficients per unit time c_{n,p} (dimensions of cm^{3} s^{−1}) for the electron and hole traps^{19}. When the cell is illuminated, these capture coefficients and emission rates change, as well as the density of filled traps^{21}. As a consequence, the process of carrier photogeneration increases the trap filling until a saturation occurs, meaning that all traps have been activated from the noise point of view. In this situation, the Var [I_{ph}] contribution is dominant, resulting larger than the dark background noise.
The photogenerated current can be expressed as^{22}
where q is the elementary charge, n is the average excess carrier density and A and d are the sample area and thickness, respectively. By taking into account the SRH statistics, the resulting fraction of the filled fluctuating traps is^{19}
Moreover, the fraction of the filled recombination centers is^{19}
where N_{A} is the base doping concentration. In equation (1) the only fluctuating term is related to n, therefore, under the assumption of a binomial probability distribution, Var [I_{ph}] can be expressed as^{1}
Substituting equations (2) and (3) into equation (4) gives
where c_{n}/c_{p} = k is the symmetric ratio^{23}.
By defining
and
equation (5) can be rewritten as
where
represent the amplitudes of the current fluctuations related to the trapping and recombination mechanisms, respectively. Here, is interpreted as the threshold current at which the noise level is almost saturated and I_{A} takes into account the influence of the recombination centers.
Experimental Results and Discussion
The proposed noise model, equations (8) and (9), has been used to analyze experimental results from Sibased solar cells, in their pristine form and after a sequence of proton irradiations, in order to generate an increasing number of traps.
Pristine Sibased solar cells
The study of the fluctuation spectroscopy requires the acquisition of the voltagespectral density S_{V} generated by the device in various operation conditions. In Fig. 2, the lowfrequency S_{V} dependence of a pristine Sibased solar cell is shown at a temperature of 300 K and for different bias points, corresponding to several differential resistance values. Apart from a number of peaks at definite frequencies related to external noise sources, a large 1/f component is evident, followed by a constant amplitude spectrum at higher frequencies. This behaviour has been observed in the whole investigated temperature range, both in dark, see Fig. 2(a) and under illumination, see Fig. 2(b). The measured noise can be well reproduced by^{1}
All the coefficients in equation (10) have been experimentally evaluated with a high level of accuracy and a statistical error lower than 2%. In particular, K is the voltagenoise amplitude, γ = (1.18 ± 0.02) is the frequency exponent and C = 1.04 × 10^{−17} V^{2}/Hz is a frequencyindependent noise component, due to the readout electronics, that can be neglected.
By examining in details the 1/f noise, interesting information on the conduction mechanisms and the dynamics of fluctuations can be extracted. The variance of the voltage can be computed from the experimental data as^{1}
where the frequency interval [ f_{min}, f_{max}] = [1, 100000] Hz is the experimental bandwidth and γ ≠ 1. The Var [V] is characterized by a quadratic dependence on R_{D} at all illumination and temperature conditions, as shown in Fig. 3(a) for 300 K. This result clearly indicates that current fluctuations are the dominant noise source in the photovoltaic system. Their amplitude being given by^{1}
and is independent on R_{D}, as evidenced in Fig. 3(b).
Due to the independence of Var [I] on R_{D}, the photoinduced fluctuations amplitude has been analyzed at a fixed bias condition of R_{D} = 10 Ω and is shown as current noise in Fig. 4 at four different temperatures between 280 and 340 K. The theoretical model of equation (8) has been used to fit the experimental data and is depicted in Fig. 4 as blue solid curves. Under the typical lowlevel injection operating conditions of most silicon solar cells^{24}, the SRH recombination is the dominant mechanism which affects the minority charge carrier lifetime. Therefore, it can be assumed that . Furthermore, for the investigated devices, in the framework of the SRH model, it is possible to express , with τ_{n0} = (c_{n}N_{SRH})^{−1}. Being , varying from (2.04 ± 0.03) × 10^{−2} at 280 K to (3.90 ± 0.06) × 10^{−2} at 340 K, it results that . Since τ_{n0} is weakly temperaturedependent below 340 K, as already verified for borondoped Czochralski (Cz) silicon cells^{23}, similar to those analyzed here, the value of τ_{eff} can be assumed constant. It has been obtained from ref. 25, where photoelectric measurements of the same devices are reported at room temperature. Here, it is worth noting that the effective lifetime can be directly estimated from the noise spectra, when the cutoff frequency occurs within the experimental frequency bandwidth^{8}. Otherwise, alternative techniques have to be used for the evaluation of τ_{eff}^{26}.
From the best values of the fitting coefficients , I_{A}, A_{1} and A_{2} (see Supplementary Table S1 for details), by using equations (6), (7) and (9), it is straightforward to compute the most important defect states parameters. In particular, for low illumination level, I_{A} and A_{2} are such that the main contribution to current noise comes from the first part of equation (8), where the number of fluctuating traps N_{T} plays an important role. The temperature dependence of N_{T} is reported in Fig. 5 (left axis), showing that it does not change within the experimental errors in the whole analyzed temperature range. It is worth noting that, the average value of N_{T} = (8.4 ± 0.1) × 10^{15} cm^{−3} is close to the doping concentration . This is consistent with the formation of unstable boronoxygen defects. Moreover, the energy depth of the traps below the conduction band can be estimated from
where k_{B} is the Boltzmann constant, T the temperature and N_{C} the effective density of states in the conduction band^{22}. The temperature dependence of the trap energy is shown in Fig. 5 (right axis), from which an average value of E_{C} − E_{T} = (0.41 ± 0.04) eV can be evaluated. Again, the estimations of E_{C} − E_{T} and of the symmetric ratio k = (30.9 ± 0.8) at 300 K are in good agreement with what found in literature for a metastable boronoxygen defect in Cz grown silicon, where k = 6.2 − 33.6 and E_{C} − E_{T} = 0.39 − 0.46 eV are reported^{23}.
Proton irradiated Sibased solar cells
As for pristine devices, the same frequency dependence of S_{V} is observed for all the silicon solar cells irradiated with different proton fluences. In particular, in Fig. 6(a) it is shown, at a temperature of 300 K and in dark condition, the presence of a large 1/f noise component followed by a constant amplitude spectrum at higher frequencies. This 1/f contribution is characterized by a quadratic dependence on the differential resistance R_{D}. As a consequence, the currentnoise amplitude Var [I] can be computed from equation (12) and does not depend on R_{D}, as clearly displayed in Fig. 6(b). The overall behaviour results to be completely independent on temperature, proton irradiation and light intensity, thus indicating again current fluctuations as the dominant noise source^{1}.
In the whole investigated temperature range and for all the proton irradiation fluences, the quantity Var [I] is characterized by the same photocurrent dependence already observed for pristine devices. This is shown in Fig. 7. The theoretical model of equation (8) has been applied to the artificially degraded silicon solar cells. Again a good agreement is found between the best fitting curves obtained from equation (8) (blue solid curves in Fig. 7) and the experimental data (full symbols in Fig. 7).
The dependence on irradiation level of the defect states parameters, extracted from the noise analysis, allows to identify the effect of radiation exposure and the corresponding damage of the cells, which is also associated to a decrease of the energy conversion efficiency^{25}. In Fig. 8 (left axis), is shown at 300 K the density N_{T} of active traps contributing significantly to the noise signal, that increases with the fluence of proton irradiation. This behaviour gives a direct indication of the presence of additional degradation processes in the photovoltaic device, also confirmed by the increase of the dark background noise amplitude. It means that the defects created by damaging the samples have the same role in the fluctuation mechanisms both in dark and under illumination. At the same time, a weak decrease of the trap energy depth is displayed in Fig. 8 (right axis) at room temperature. As reported in literature, after proton degradation, the dominant radiationinduced defect in borondoped Cz grown silicon is related to the donorlike defects^{27}. Such a defect state, located at E_{C} − 0.18 eV, acts as a minority carrier trap and influences the trapping/detrapping. This indicates that under low charge carrier injection, as the case here considered, the fluctuating traps closer to the conduction band produce the main contribution to the measured current noise.
The observed effects on N_{T} and on E_{C} − E_{T} are consistent with the reported long wavelength decrease of the external quantum efficiency, due to the increase of recombination centers density caused by proton irradiation^{25,28}. Moreover, the most striking feature of Fig. 8 is an abrupt change evident after the first dose of radiation. This finding suggests that the major disturbance of the structure, leading to a noise enhancement, seems to occur at the beginning of radiation exposure. The values reported in Table 1 are in agreement with this experimental evidence. In particular, the simultaneous decrease of the relevant solar cell photoelectric parameters and the increase of trap density seem to be directly related to an overall reduction in the power conversion efficiency.
Conclusions
It has been shown that noise spectroscopy is a sensitive tool for the electrical temperaturedependent characterization of electronic defects in photovoltaic devices. A theoretical model has been formulated to explain the nature of current fluctuations in solar cells. The applicability of the proposed model has been verified on pristine and artificially degraded Sibased devices. Distinct differences between dark and photoinduced noise have been found and interpreted in terms of a ShockleyReadHall theory. In particular, a combination of trapping/detrapping related processes and charge carrier recombination phenomena has been considered to explain the current fluctuation mechanisms. Characteristic parameters of the dominant defect centers, as extracted by noise analysis, have been compared with literature data. In pristine cells, the energetic position and the symmetric ratio are in good agreement with the values obtained by using alternative techniques and well consistent with the boronoxygen complex defect. On the other hand, in degraded devices, the total trap density increases with the fluence of proton irradiation and seems to be directly related to a decrease of the power conversion efficiency. Application of this noise model to other photovoltaic materials, such as organic and perovskite compounds, is currently in progress.
Methods
The solar cells, used to test the validity of the proposed electric noise model, were realized by SOLARTEC, type “SC2140Z824”. The investigated area A of the photovoltaic devices was 1 cm^{2} with a wafer thickness d of 320 μm. The initial efficiency of pristine samples was measured to be 15% under AM 1.5 G conditions, with a shortcircuit current I_{sc} = 37.45 mA, opencircuit voltage V_{oc} = 581 mV and fill factor FF = 68.96% (see ref. 25 for details). Degradation effects were artificially induced on the devices with proton irradiation, realized at the HelmholtzZentrum Berlin für Materialien und Energie (Germany), with a proton energy of 65 MeV in air. A homogeneous defect distribution, with this energy value, was verified by using a SRIM (Stopping and Range of Ions in Matter) code. Specific information on the analysis performed can be still found in ref. 25. The samples were irradiated with three different proton fluences: 2 × 10^{11} protons/cm^{2}, 1 × 10^{12} protons/cm^{2} and 5 × 10^{12} protons/cm^{2}.
All the measurements were carried out by varying the temperature from 280 to 340 K with a thermoelectric cooler and by stabilizing it with a computercontrolled PID loop to better than 0.1 K. A commercial cool white Light Emitting Diode (LED) “KLC8 Edixeon K series”, from Edison Opto, was used as light source. This choice of adopting a LED source was done in order to have a lownoise source and to discriminate intrinsic photoinduced mechanisms from extrinsic ones, potentially related to the complex electronics of the solar simulator.
The setup for the electrical transport and noise characterizations was arranged in order to minimize the presence of spurious components in the measured spectral traces. A lownoise Keithley dc current source was used for the biasing of the samples. The dc voltage drop was recorded with a digital multimeter, while the ac voltage signal was amplified with a lownoise PAR5113 preamplifier and analyzed by a dynamic signal analyzer HP35670A. A specific procedure, based on a sequence of fourprobe and twoprobe noise measurements^{29}, allowed the estimation and removing of contact noise contributions, a possible source of artifacts in this type of experiments. External noise contributions, such as the photon number fluctuation, light source bias circuit noise and light source emission noise, were also experimentally evaluated and their influence was systematically excluded (see for details page 3 in Supplementary information and Supplementary Fig. S1).
Additional Information
How to cite this article: Landi, G. et al. A noise model for the evaluation of defect states in solar cells. Sci. Rep. 6, 29685; doi: 10.1038/srep29685 (2016).
References
Kogan, S. Electronic Noise and Fluctuations in Solids (Cambridge University Press, Cambridge, 1996).
Jones, B. K. Electrical noise as a measure of quality and reliability in electronic devices. Adv. Electron. Electron Phys. 87, 201–257 (1993).
Chobola, Z. Impulse noise in silicon solar cells. Microelectron. J. 32, 707–711 (2001).
Chobola, Z. Noise as a tool for nondestructive testing of singlecrystal silicon solar cells. Microelectron. Reliab. 41, 1947–1952 (2001).
Chobola, Z., Lunak, M., Vanek, J. & Barinka, R. Lowfrequency noise, microplasma and electroluminescence measurements as faster tools to investigate quality of monocrystallinesilicon solar cells. Opt. Eng. 52, 051203 (2013).
Bag, A., Mukherjee, C., Mallik, S. & Maiti, C. K. Low frequency noise in iron disilicide heterojunction solar cells. In: Proc. 19^{th}IEEE International Symposium on the Physical and Failure Analysis of Integrated Circuits, 1–4 (2012).
Rocha, P. R. F. et al. Lowfrequency noise as a diagnostic tool for OLED reliability. In: Proc. 22^{nd}Int. Conf. on Noise and Fluctuations, 1–4 (2013).
Landi, G., Barone, C., De Sio, A., Pagano, S. & Neitzert, H. C. Characterization of polymer:fullerene solar cells by lowfrequency noise spectroscopy. Appl. Phys. Lett. 102, 223902 (2013).
Barone, C., Landi, G., De Sio, A., Neitzert, H. C. & Pagano, S. Thermal ageing of bulk heterojunction polymer solar cells investigated by electric noise analysis. Sol. Energy Mater. Sol. Cells 122, 40–45 (2014).
Paggi, M., Berardone, I., Infuso, A. & Corrado, M. Fatigue degradation and electric recovery in Silicon solar cells embedded in photovoltaic modules. Sci. Rep. 4, 4506 (2014).
Palenskis, V. & Maknys, K. Nature of lowfrequency noise in homogeneous semiconductors. Sci. Rep. 5, 18305 (2015).
Barone, C., Landi, G., Mauro, C., Neitzert, H. C. & Pagano, S. Universal crossover of the charge carrier fluctuation mechanism in different polymer/carbon nanotubes composites. Appl. Phys. Lett. 107, 143106 (2015).
Kleinpenning, T. G. M. 1/f noise in pn junction diodes. J. Vac. Sci. Technol. A 3, 176–182 (1985).
Schiebel, R. A. A model for 1/f noise in diffusion current based on surface recombination velocity fluctuations and insulator trapping. IEEE Trans. Electron Devices 41, 768–778 (1994).
Macku, R. & Koktavy, P. Analysis of fluctuation processes in forwardbiased solar cells using noise spectroscopy. Phys. Status Solidi A 207, 2387–2394 (2010).
Anderson, W. W. & Hoffman, H. J. Surfacetunnelinginduced 1/f noise in Hg1−xCdxTe photodiodes. J. Vac. Sci. Technol. A 1, 1730 (1983).
Hsu, S. T., Fitzgerald, D. J. & Grove, A. S. Surfacestate related 1/f noise in pn junctions and MOS transistors. Appl. Phys. Lett. 12, 287 (1968).
van der Ziel, A. Noise in solidstate devices and lasers. Proc. IEEE 58, 1178–1206 (1970).
Shockley, W. & Read, W. Statistics of the recombinations of holes and electrons. Phys. Rev. 87, 835–842 (1952).
Sah, C.T. The equivalent circuit model in solidstate electronicsPart I: the single energy level defect centers. Proc. IEEE 55, 654–671 (1967).
Yau, L. D. & Sah, C. T. Measurement of trappedminoritycarrier thermal emission rates from Au, Ag and Co traps in silicon. Appl. Phys. Lett. 21, 157 (1972).
Sze, S. M. & Ng, K. K. Physics of Semiconductor Devices (WileyInterscience, New Jersey, 2007).
Rein, S. & Glunz, S. W. Electronic properties of the metastable defect in borondoped Czochralski silicon: unambiguous determination by advanced lifetime spectroscopy. Appl. Phys. Lett. 82, 1054 (2003).
Gogolin, R. & Harder, N. P. Trapping behavior of ShockleyReadHall recombination centers in silicon solar cells. J. Appl. Phys. 114, 064504 (2013).
Neitzert, H. C. et al. Electroluminescence efficiency degradation of crystalline silicon solar cells after irradiation with protons in the energy range between 0.8 MeV and 65 MeV. Phys. Status Solidi B 245, 1877–1883 (2008).
Bellone, S., Neitzert, H.C. & Licciardo, G. D. Modelling and characterisation of the OCVD response at an arbitrary time and injection level. Solid. State. Electron. 48, 1127–1131 (2004).
Khan, A. et al. Role of the impurities in production rates of radiationinduced defects in silicon materials and solar cells. J. Appl. Phys. 90, 1170 (2001).
Meillaud, F. et al. Protoninduced degradation of thinfilm microcrystalline silicon solar cells. J. NonCryst. Solids 352, 1851–1854 (2006).
Barone, C. et al. Experimental technique for reducing contact and background noise in voltage spectral density measurements. Rev. Sci. Instrum. 78, 093905 (2007).
Acknowledgements
The authors would like to thank S. Abate of CNRSPIN Salerno for his valuable technical support. A. Denker of the HelmholtzZentrum Berlin für Materialien und Energie (Germany) is also acknowledged for the proton irradiation of the photovoltaic devices.
Author information
Affiliations
Contributions
G.L. formulated the noise model for solar cells. C.B., C.M. and S.P. performed the voltagenoise measurements of all devices. H.C.N. contributed to the detailed characterization of irradiated solar cells. All authors wrote the main manuscript text, discussed the results and implications and commented on the paper by reviewing it accurately.
Ethics declarations
Competing interests
The authors declare no competing financial interests.
Electronic supplementary material
Rights and permissions
This work is licensed under a Creative Commons Attribution 4.0 International License. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in the credit line; if the material is not included under the Creative Commons license, users will need to obtain permission from the license holder to reproduce the material. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/
About this article
Cite this article
Landi, G., Barone, C., Mauro, C. et al. A noise model for the evaluation of defect states in solar cells. Sci Rep 6, 29685 (2016). https://doi.org/10.1038/srep29685
Received:
Accepted:
Published:
Further reading

Understanding the thermal degradation mechanism of perovskite solar cells via dielectric and noise measurements
Nanotechnology (2020)

IronBased Superconducting Nanowires: Electric Transport and VoltageNoise Properties
Nanomaterials (2020)

An analysis of carrier dynamics in methylammonium lead triiodide perovskite solar cells using cross correlation noise spectroscopy
Applied Physics Letters (2020)

Influence of Charge Transport Layers on Capacitance Measured in Halide Perovskite Solar Cells
Joule (2020)

Toward reliable high performing organic solar cells: Molecules, processing, and monitoring
APL Materials (2020)
Comments
By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.