A Comparison between p+n and n+p GaAs Displacement Damage Coefficients Following Proton Irradiation

Jeffrey H. Warner 1,2, Scott R. Messenger 1, Robert J. Walters 1, Geoffrey P. Summers2

1U.S. Naval Research Laboratory, Washington, DC 2University of Maryland, Baltimore County, Baltimore, MD

ABSTRACT

In this paper, the photovoltaic response of p+n and n + p GaAs solar cells is monitored as a function of proton fluence at different proton energies. The energy dependence of displacement damage coefficients (DCs) describing the photovoltaic degradation for these devices are compared with calculations of nonionizing energy loss (NIEL). The short circuit current DCs for both device types follows the same energy dependence. In contrast, the open circuit voltage DCs follows a different energy dependence at higher proton energies (E> 10 MeV).

INTRODUCTION

Satellite environments contain energetic protons, neutrons, electrons, heavy ions and gamma rays that arise from an interplay of the solar wind, the cosmic background radiation, and Earth's geomagnetic field. The most common particles in the Earth's Van Allen radiation belts are electrons and protons having energies up to several hundred MeV. Exposure to these particles degrades the performance of the electronic and optoelectronic devices. The ability to predict how these devices will respond in a particle radiation environment is important in predicting the expected mission lifetime.

One important parameter used in predicting the survivability of devices in space is called the damage coefficient, or DC. A damage coefficient can be generated from any parameter whose dependence on particle fluence is known for a given particle energy. Presently, the process of determining the energy­dependent values of damage coefficients is laborious and expensive because devices must be irradiated at many different particle energies and fluences. It is desirable to be able to analytically calculate the energy dependence of a given damage curve because measurements would only then need to be made at a single energy for future studies. For almost two decades researchers at the Naval Research Laboratory (NRL) have been using the concept of nonionizing energy loss (NIEL) to do displacement damage correlation analyses [1].

A number of damage correlation studies have been performed for proton-irradiated GaAs optoelectronic devices. A comparison of the proton energy dependences derived from various device parameters is shown in Fig. 1. The ordinate axis (the normalized DC) is plotted on a relative scale in which all coefficients have been normalized to unity at 10 MeV. Plotting the data in this way allows the functionalities of the curves to be more easily observed and compared. The solid line represents the total NIEL while the dashed line represents only the Coulombic portion of the NIEL. The

978-1-4244-1641-7/08/$25.00 ©2008 IEEE

total NIEL calculation includes all interactions, i.e. elastic (Coulombic and nuclear) and nuclear inelastic.

It can be seen in Fig. 1 that for proton energies below about 10 MeV the two NIEL curves are very much the same, and that the data correlate strongly with both curves. Above 10 MeV, the damage coefficients generated from the different devices either follow the Coulombic NIEL, the total NIEL, or lie in a region bounded by the two NIEL curves. The solid squares in the figure represent the proton energy dependence of the minority carrier lifetime DCs as measured in amphoteric GaAs LEDs [2]. These data fall well below the total NIEL curve and suggest that the total NIEL significantly overestimates the damage produced. The solid triangles in the figure represent damage coefficients for InGaAs/GaAs OW LEDs [3]. These data also fall below the total NIEL but to a lesser extent. The solid circles represent data from Lee et al. which clearly track the total NIEL [4]. Perhaps the most interesting observation is that the damage coefficients for the GaAs solar cell short circuit current (J se , solid diamonds) appear to follow the Coulombic portion of NIEL, while the open circuit voltage (Voe, open diamonds) appears to follow the total NIEL [5]. A second example showing two parameters measured in similar devices that show different high energy behavior of the damage coefficients is that of the LED data (solid squares represent the carrier lifetime; open inverted triangles represent the light output intensity [6]). The DCs follow

i o ...

101' ............ ....., .................. ,.........,......,. ........... ....,.--.--.-........... _

GaAs NIEL

• _ TotalNIEL ---- Coulombic NIEL

• LEO Carrier Lifetime (Barry et al.) V LEO Light Output Intensity (Warner et at) • AIGaAslGaAs MQW Laser (Lee et at) ... OW LED Light Output Intensity (Walters et at) Q Photovoltage (Anspaugh et at) • Photocurrent (Anspaugh et al.)

10.2 ......................................... _ .............................. _ ........................... ....

1~ 1~ 1~ Proton Energy (MeV)

Fig. 1. Damage coefficients derived from various GaAs optoelectronic devices plotted versus incident proton energy E, and normalized to unity at E = 10 MeV for convenience. Solid line: Total NIEL. Dashed line: Coulombic portion of the NIEL only. Some data correlate with the Coulombic NIEL; other data correlate with the total NIEL; still others fall in between.

1.0 .....-,....,...,..,.,.,.",......,...,....,........,---.....,..,,.,.,.,...,.......,....,.....,........,......,. ....... .,.,.,..,..

0.9

:;: 0.8 .., ~ 'ii 0.7

E o Z 0.6

0.5

Proton Energy __ 2MeV

-+- 3.8MeV ___ 53 MeV

......... 227 MeV

NRL p+n GaAs

0.4 ............................................ """""---' .................................... .....,. .............. .......... 1010 1011 1012 1013 1014

Proton Fluence (p+/cm 2)

Fig. 2. Normalized short circuit current as a function of proton fluence measured in p + n GaAs solar cells after irradiation by protons. These data were normalized to the pre-irradiation value.

the same energy dependence until -100 MeV, where they diverge. Hence, the results depicted in Fig. 1 demonstrate that, at higher proton energies, additional (or at least different) damage mechanisms may arise. If this is true, the solar cell data in [5] suggests these mechanisms also influence electrical parameters differently. The solar cell data only extends out to 50 MeV, so it is uncertain if the trend would continue for an increase in proton energy.

In summary, NIEL has been successful in correlating proton-induced displacement damage for GaAs for proton energies below 10 MeV independent of the parameter or device type. Despite these successes, differences have been observed in damage correlation studies for proton-irradiated GaAs-based devices when the proton energy exceeds -10 MeV. One main objective of this paper is to generate damage coefficients on p + nand n + p GaAs solar cells and compare them with the energy dependence of NIEL. It also needs to be determined if the J se and Voe do indeed follow different energy dependences and if this dependence is influenced by the device polarity.

EXPERIMENTAL DETAILS

The GaAs solar cells used in this study were both p + nand n + p polarities grown on GaAs substrates by molecular beam epitaxy (MBE). The samples were grown and processed at the Naval Research Laboratory (NRL). A detailed description of the solar cell structures, doping concentrations, and thicknesses are given in [8, Table I]. A schematic diagram for an NRL GaAs solar cell very similar to the one used in this study can be seen in [7]. Both polarity solar cells were designed such that the beginning-of-life (BOL) efficiency was nearly identical. However, the different cell types have different contributions of the photocurrent generated in the emitter, depletion and base regions because of the difference in layer thicknesses required to obtain similar BOL efficiency. The BOL performance predictions were accomplished by modeling the external quantum

978-1-4244-1641-7/08/$25.00 ©2008 IEEE

1.0 .....-,....,...,..,.,.,.",......,...,....,........,---.....,..,,.,.,.,...,.......,....,.....,........,......,. ....... .,.,.,..,..

0.9

0.7

Proton Energy __ 2MeV

-+- 3.8MeV ___ 53 MeV

......... 227 MeV

NRL p+n GaAs

1011 1012 1013 1014

Proton Fluence (p+/cm 2)

Fig. 3. Normalized open circuit voltage as a function of proton fluence measured in p + n GaAs solar cells after irradiation by protons. These data were normalized to the pre-irradiation value.

efficiency (QE) of the solar cell using the Hovel equations, which also included the Alo.2oGao.sAs window layer in the calculations [9]. There was good agreement between the measured and modeled external quantum efficiency results as shown in [8, Fig. 2].

The p+n samples were irradiated with 2, 3.8, 53, 160, and 227 MeV protons while the n+p samples were irradiated with 3.8 and 227 MeV protons. The proton irradiations were performed at various different facilities to accommodate the proton energy range of interest, which include NRL (2 and 3.8 MeV), TRIUMF (53 MeV), and the Francis H. Burr Proton Therapy Center (227 MeV) with proton fluences ranging from 6x10 10 to 1x10 14

cm- 2. All the irradiations were performed at room temperature with low enough beam currents to avoid any significant increase in sample temperature. Typically, three solar cells were irradiated at each fluencelevel.

Illuminated current-voltage (Ltv) measurements were performed both before and after irradiation using an Oriel 1000 W solar simulator (AMO, 1 sun, 25°C conditions). The average pre-irradiation values for both cell types were nominally identical. The short circuit current Jse, open circuit voltage Voe, fill factor FF and efficiency Eft were approximately 25.51 mAlcm 2, 0.975 V, 0.832 and 15.13 %, respectively

EXPERIMENTAL RESULTS

The semi-empirical equation representing the variation of the photovoltaic parameters with particle fluence for solar cells following irradiation is given by [10]:

(1 )

where N( C/J) represents the normalized parameter of interest (such as Jse) normalized to the pre-irradiation

1.0 """""""""""""""""""'''''''''''....."r--''''''''''''''''''''"",,,,'''''''''''''''''''''''''''''''''''''''''''''''''''

0.9

:il .., "C

~ 0.8 .; E 0 Z

0.7

1011 1012 1013 1014

Proton Fluence (p+/cm 2)

Fig. 4. Normalized short circuit current as a function of proton fluence measured in n + p GaAs solar cells after irradiation by protons. These data were normalized to the pre-irradiation value.

values, <I> is the particle fluence value for a given proton energy, and C and <l>x are fitting parameters to be determined at each proton energy. Previous measurements have shown that constant C should be independent of the incident proton energy for a particular photovoltaic parameter of a given technology [12]. This implies the constant <l>x will change with incident proton energy. <l>x will be used as a DC in this study. The normalized DCs can be derived from experimental data by forming the ratio given by

(2)

where <l>x is the measured change in one of the solar cell's parameter and Es is the standardized particle energy. By convention for protons incident on GaAs, Es is 10 MeV. Theoretically, the data can be normalized to any incident proton energy but an energy should be chosen that is no greater than 10 MeV since deviation with NIEL have been observed for energies above this value. Also, the energy should be sufficient to create uniform damage in the active region of the device. Equation (2) is used to generate the normalized energy dependence of the DCs so that device performance can be made for any given radiation environment, which will be compared with the energy dependence of the calculated nonionizing energy loss (NIEL). This will provide incite as to whether the NIEL is expected to accurately predict the radiation response of the photovoltaic parameters.

and shows the degradation of J se and Voe as a function of proton fluence normalized to pre-irradiation values for p + n GaAs solar cells. For the most part, the degradation curves in the figure are shifted to higher fluence as the proton energy increases, which qualitatively agree with the energy dependence of the calculated NIEL. Figs. 4 and 5 show the degradation of the J se and Voe as a function of proton fluence normalized to pre-irradiation values for n + p GaAs solar cells. Each curve shown in the figures were fit using (1)

978-1-4244-1641-7/08/$25.00 ©2008 IEEE

1.0,-.....,....,...,.,.,...,.........,......,...,...,..,.....,....-.-.,....,..,.,..,.."---....,...,..,.,..,...,....-.,......,..,...,..,,mo

0.9

0.7

0.6 ':-:-................ ~ ................. .....,.. ................. .......,.---' .................. 011..... ................. ...J 1010 1011 1012 1013 1014

Proton Fluence (p+/cm 2)

Fig. 5. Normalized open circuit voltage as a function of proton fluence measured in n + p GaAs solar cells after irradiation by protons. These data were normalized to the pre-irradiation value.

Table 1. Mean values for fitting parameter C for each photovoltaic parameter for p + nand n + p GaAs solar cells.

J sc

0.302 0.182

Mean values for parameter C (JJsc Voc avoc Pmax

0.068 0.061

0.128 0.105

0.013 0.018

0.281 0.250

(JPmax

0.033 0.009

to determine the fitting parameters C and <l>x. There was some variation in the fitting parameter C for a given photovoltaic parameter and cell type when comparing at different proton energies. The deviation about the mean for a given parameter and cell type was typically -10 %. There was no apparent trend for the value of the fitting parameter C as a function of proton energy which suggests the fitting parameter C is indeed independent of proton energy. The damage coefficient <l>x was finally determined by re-fitting each dataset using a one parameter fitting routine while the parameter C was held fixed at the mean value specified in. The normalized DCs were calculated using (2) in order to compare with the energy dependence of N I EL.

The energy dependence of the damage coefficients describing the degradation of J se and Voe for p+n and n+p GaAs solar cells are compared with NIEL in Fig. 6 and Fig. 7, respectively. For p+n GaAs solar cells, the energy dependence of Voe follows more closely the total NIEL while J se appears to more closely track the Coulombic NIEL. This data suggests that there is indeed a separation between Voe and J se for proton energies above 10 MeV. The 53 MeV data point in this study confirms Anspaugh's previous data at 50 MeV and shows the separation to continue for higher proton energies. In contrast, the energy dependence of Voe and J se follow the same energy dependence with NIEL for the n + p GaAs solar cells. These data fall slightly higher than the Coulombic NIEL. It should be mentioned that the energy dependence of the maximum power was determined for both cell types (although not shown here) and fall directly between the two NIEL curves.

DISCUSSION

The NIEL approach has been shown to give an unsatisfactory correlation of the energy dependence of some device characteristics in GaAs devices as previously mentioned. One of the first data sets to display this is the one produced by Barry et al. [2]. In [7] measurements performed on nominally identical LEOs showed the energy dependence of the light output intensity to have different energy dependence than the carrier lifetime data. These data again show that different parameters associated with similar devices follow different energy dependences with NIEL. These data are consistent with the new results presented here for p+n GaAs solar cells where the Voc and J sc have different energy dependence with NIEL, which has also confirmed the previous findings from Anspaugh's data. However, the energy dependence of all the damage coefficients (J sc , Voc, and Pmax ) for the n p GaAs solar cells followed the same ener2Y dependence and closely tracked that of J sc for the p n GaAs solar cells. This suggests that the energy dependence of Voc might depend on whether the material is n- or p-type. A possible explanation might be attributed to different radiation-induced defects controlling the degradation of Voc especially since these solar cells are one-sided, abrupt junction devices, which means that most the depletion region extends into the base of the device. For example, the depletion region for the p+n cell extends into the n-type base. In contrast, similar radiation-induced defects in n- and p-type material are responsible for degradation of J sc causing the Des to have a similar energy dependence.

It was recently shown for p + n GaAs solar cells that defects are introduced having completely different electrical response for proton energies ;::: 10 MeV compared with protons having lower energy [14]. In particular, very active recombination centers are introduced and their recombination efficiency also changes for certain irradiation conditions. Therefore, it is not surprising that Des do not vary linearly with the calculated NIEL at the higher proton energies. Currently, there is minimal data in the literature for the radiation-induced defects following proton irradiation on n + p GaAs solar cells, so a comparison of the defects introduced cannot be made at this time with those produced in p + n GaAs solar cells at the higher proton energies.

The ability to predict the radiation response of device characteristics in complex multi-particle energy radiation environments using monoenergetic particle ground based measurements can be extremely useful for spacecraft designers. Usually in the past, to determine the energy dependence of a device degradation characteristic, measurements at multiple particle energies were performed. A set of Des were thus generated for a particular device. This process is often accomplished at great expense and time. We have found that this process can be greatly simplified using results of an analytical calculation that generates the energy dependence of the damage coefficients known as the NIEL. Applied to displacement damage effects, the NIEL approach is becoming more widely used in the radiation effects community to correlate the effects of different energetic particles to that of a

978-1-4244-1641-7/08/$25.00 ©2008 IEEE

101.,...........,.......,....,.... ....... ......,.-...,.......,.......,...,. ......... '"'"I""-...,........,......,...,....~ NRL p+n GaAs

- Total ············i !

......... i ".

---- Coulombic • short circuit current • open circuit voltage ' . .................

10-2L.. ....................................... - ................................. ~2-................................. 103 101 10 Proton Energy (MeV)

Fig. 6. Comparison of proton damage coefficients (J sc and Voc) measured on p+n GaAs solar cells with the calculated NIEL. All of the data have been normalized to the value at 10 MeV.

101.,...........,......,.....,... ....... ......,.-....,.......,......,...,....,.....'"'"I""-...,........,..... ........ ~

'Ii ......

' . .......... - Total ---- Coulombic • short circuit current • open circuit voltage

NRL n+pGaAs

............... t ........

................

10-2L ............................... ~1-........................ '":2~ ........................... 103 10 10 Proton Energy (MeV)

Fig. 7. Comparison of proton damage coefficients (Jsc and Voc) measured on n+p GaAs solar cells with the calculated NIEL. All of the data have been normalized to the value at 10 MeV.

common energy. This is easily accomplished for proton energies below 10 MeV since all the damage coefficients are seen to always follow the NIEL independent of the device and parameter measured. Although deviations are observed for proton energies above 10 MeV, this does not mean NIEL cannot be used with great confidence because the Des are always bound between the Coulombic and total NIEL. Therefore, the curves represent a lower and upper bound for damage predictions. In [16], calculations were performed for the total displacement damage dose using both NIEL curves to determine the range of dose expected for solar cells shielded with various thickness of fused silica coverglass. Those calculations were performed for two different incident proton spectra and the overall effect of using the total or Coulombic NIEL for calculating the total displacement dose deposited is minimal until extreme thicknesses of shielding are used.

Although, there can be a 30% difference in the expected total displacement damage dose when the Coulombic or total NIEL is used, this only translates to a difference in the expected maximum power degradation of -3%. This provides confidence that, although deviations are observed, NIEL is still useful for making damage predictions.

It should be mentioned that the energy dependence of the maximum power for both cell types studied here were identical. The maximum power is usually the parameter of interest when making device performance predictions. Although, the DCs for the p+n cells follow different energy dependence at the higher energies, damage predictions for the maximum power would be identical for the cell structures studied here.

CONCLUSION

The data presented here has shown deviations with NIEL for proton energies;::: 10 MeV but the damage coefficients are bound between the two NIEL curves which represent a lower and upper limit for damage predictions. The damage coefficients for the p + n GaAs solar cells do not always follow the same energy dependence at the higher proton energies but do for the n+p GaAs solar cells. The 53 MeV data point for the p+n cells confirms the previous data point in earlier work and also extends the damage coefficients to higher proton energies [5]. The energy dependence of the maximum power for both cell types was identical and falls between the two NIEL curves. These data are generally supporting the theory that the displacement damage induced by proton irradiation can be modeled based on NIEL. In the high energy range, the correlation appears to vary depending on the device and actual parameter under study.

A clear conclusion for the radiation effects community presented here is that one cannot assume that the degradation characteristics of a given parameter within a given device will be the same for the device in general. In doing the testing, one should measure the device characteristic of immediate interest. That is, it should not be expected that the energy dependence of the short circuit current is the same as the maximum power.

[1]

[2]

[3]

REFERENCES

S. R. Messenger, G. P. Summers, E. A. Burke, R. J. Walters, and M. A. Xapsos, "Modeling Solar Cell Degradation in Space: A Comparison of the NRL Displacement Damage Dose and the JPL Equivalent Fluence Approaches", Progress in Photovoltaics: Research and Applications 9, 103 (2001). A. L. Barry, A. J. Houdayer, P. F. Hinrichsen, W. G. Letourneau, and J. Vincent, "The Energy Dependence of Lifetime Damage Constants in GaAs LEOs for 1-500 MeV Protons", IEEE Trans. Nucl. Sci. 42, 2104 (1995). R. J. Walters, S. R. Messenger, G. P. Summers, E. A. Burke, S. M. Khanna, 0 Estan; L .S. Erhardt, Hui Chun Liu; Gao Mae; M. Buchanan,

978-1-4244-1641-7/08/$25.00 ©2008 IEEE

[4]

[5]

[6]

[8]

[9]

[10]

[12]

[14]

[16]

A. J. SpringThorpe, A. Houdayer, C. Carlone, "Correlation of Proton Radiation Damage in InGaAs-GaAs Quantum Well Light-Emitting Diodes", IEEE Trans. Nucl. Sci. 48,1773 (2001). S. C. Lee, Y. F. Zhao, R. D. Schrimpf, M. A, Neifeld, and K. F. Galloway, "Comparison of Lifetime and Threshold Current Damage Factors for Multi-Quantum Well (MQW) GaAs/GaAIAs Laser Diodes Irradiated at Different Proton Energies", IEEE Trans. Nucl. Sci. 46, 1797 (1999). B. E. Anspaugh, "Proton and electron damage coefficients for GaAs/Ge Solar cells," in Proc. 22nd Photovoltaic Specialist Cont. (PVSC), 1991, pp.1593-1598. J. H. Warner, R. J. Walters, S. R. Messenger, G. P. Summers, S. Khanna, D. Estan, L. Erhardt, and A. Houdayer, "High-energy proton irradiation effects in GaAs devices," IEEE Trans. Nucl. Sci. 51,2004,pp.2887-2895. Jeffrey H. Warner, Scott R. Messenger, Robert J. Walters, Geoffrey P. Summers, Justin R. Lorentzen, David M. Wilt, and Mark A. Smith, "Correlation of Electron Radiation Induced­Damage in GaAs Solar Cells," IEEE Trans. Nucl. Sci. 53, 2006, pp. 1988-1994. H. J. Hovel, "Solar Cells," Semiconductors and Semimetals, vol. II, 1975, pp. 17-20. B. E. Anspaugh, GaAs Solar Cell Radiation

Handbook, JPL 96-9,1996. S. R. Messenger, G. P. Summers, E. A. Burke, R. J. Walters, and M. A. Xapsos, "Modeling Solar Cell Degradation in Space: A Comparison of the NRL Displacement Damage Dose and the JPL Equivalent Fluence Approaches", Progress in Photovo/taics: Research and Applications 9, 2001, pp.103-121. Jeffrey H. Warner, Scott R. Messenger, Robert J. Walters, Geoffrey P. Summers, Manuel J. Romero, and Edward A. Burke, "Displacement Damage Evolution in GaAs Following Electron, Proton and Silicon Ion Irradiation," IEEE Trans. Nucl. Sci. 54, 2007, pp. 1961-1968. J. H. Warner, S. R. Messenger, R. J. Walters, and G. P. Summers, "The effects of high energy protons on shielded space cells," in Proc. World Cont. Photo voltaic Energy Conversion, Waikoloa, HI, pp. 1854-1857, May 2006.