1. Introduction

Significant advancements have been achieved in the performance of quantum cascade lasers (QCLs) that operate in the mid-infrared spectral region (3–12 μm) since the first demonstration by Faist et al. [1]. QCLs now emit high output powers in continuous-wave mode at room temperature [2 –4]. Measurements by several different groups [5 –7] have demonstrated that QCLs are robust and reliable with long-term operation enabling advanced applications, such as space-based missions. No studies, however, have been reported in the literature about the performance of QCLs after radiation exposure. To withstand the harsh environment of space, QCLs must be insensitive to ionizing radiation that would be encountered during a typical mission.

The most common sources of space radiation are from solar cosmic rays, which consist mainly of protons and are ejected sporadically from the sun during solar flare events; magnetically trapped protons and electrons; and galactic cosmic rays [8]. Thus, cumulative damage to QCLs from electron and proton exposure must be evaluated to determine their radiation susceptibility before they can be used in a space mission.

There have been several studies examining the radiation degradation mechanisms in laser diodes [9 –12] as well as quantum dot lasers [13,14]. Radiation damage generally increases the optical losses and decreases the electrical efficiency in diode lasers so that an increase in the lasing threshold current is observed. In particular, ionizing radiation can produce high local concentrations of electron-hole pairs that can result in an increase in the number of nonradiative recombination centers. Electrons and protons can also cause internal defects within the semiconductor structure due to atomic displacement, which can be particularly damaging to semiconductor devices. Gamma rays are often used for radiation testing since they generate high-energy electrons as they traverse the semiconductor material, which is identical to the behavior of Bremsstrahlung x rays produced when high-energy electrons are stopped in shielding. Gamma rays produced from the radioactive decay of Cobalt-60 are a more convenient MeV photon source than x-ray sources.

Studies of diode lasers using radiation doses that are typical for a space mission have shown a minimal effect on performance. Although the actual dose received by electronic components is a function of satellite orbit, mission duration, and the protective shielding in which the components are enclosed, total radiation doses are typically between 15 and 100 krad for a 7-year mission [15]. Troupaki et al. [16] have demonstrated that proton-irradiated laser diode arrays at 808 nm exhibit a rise in threshold current of less than 5% when exposed to 200 MeV protons for doses up to 60 krad(Si); the threshold current also recovered over time and was attributed to self-annealing. They did not observe any discernable effects for the laser diode arrays after gamma irradiation using doses up to 200 krad(Si).

QCLs, which are fundamentally different than diode lasers, do not rely on electron-hole recombination for photon generation. Instead, QCLs require only injection of electrons into the conduction band for laser emission. Photons are generated as the injected electrons cascade between discrete energy levels in the conduction band of a multiple quantum well heterostructure consisting of a few hundred layers with thicknesses of typically 1–5 nm. The filled valence bands in QCLs, i.e., no holes are present, eliminates Auger recombination. Carrier relaxation time in QCLs is dominated by optical phonon scattering and is of the order of a few picoseconds. Thus, unlike interband lasers or diode lasers, the dominant nonradiative loss mechanism is optical phonon scattering. Diode lasers, on the other hand, require confinement of electrons and holes in the active region for efficient radiative recombination so nonradiative recombination processes are important loss mechanisms. These fundamental differences may lead to different sensitivities of laser performance to the various types of radiation damage. For example, radiation-induced defects that facilitate nonradiative electron-hole recombination would not impact QCLs, which are unipolar and lack holes. In contrast, the hundreds of thin layers in a QCL may be more sensitive to radiation damage, which could degrade the carefully engineered electronic band structure.

In this paper, we report on work to characterize the performance of QCLs after exposure to two types of ionizing radiation: Cobalt-60 gamma rays (1.17 and 1.33 MeV) and 64 MeV protons. Although low-energy protons dominate the trapped proton fluxes, protons below 30 MeV are effectively shielded so that a 64 MeV proton beam is a good choice for testing. The total radiation doses used in this study vary between 20 and 46.3 krad(Si). Most of the QCLs were exposed to both a 10 krad(Si) dose of protons as well as gamma radiation over three equal doses for a total accumulated gamma dose of 26.3 krad(Si). For a comparison, the total daily dose calculated for a spacecraft with an aluminum shield thickness of $0.2\mathrm{g}/{\mathrm{cm}}^2$ at an altitude of 500 km and 60° inclination is calculated to be 13.0 rad(Si); over a 7-year timeframe the cumulative dose would be 33.2 krad(Si) [8]. Thus, the dose amounts used in this paper are within the range of total cumulative doses typically experienced by electronic components in a space environment.

2. Experimental

We tested seven Fabry–Perot (FP) QCLs from two vendors. Two of the devices labeled Ham-Ce and Ham-Cf are from Hamamatsu and only limited design details are known. These two QCLs lase nominally at 5.3 μm and are mounted epi-side up on a copper c-mount with as-cleaved facets.

Five of the devices are from Maxion Technologies/Thorlabs and additional design details have been provided. All five QCLs as well as an additional QCL, M577G, which was used as a control and not exposed to any radiation, are based on AlInAs/GaInAs heterostructures and incorporate an active region based on a four quantum-well active region using a double-phonon resonance for the lower level to provide a large population inversion and minimize thermal backfilling [17]. Fabrication of the devices is similar to [18] except two of the QCLs (i.e., M664O and M577G), which were grown and fabricated much earlier than the other devices, used wet etching instead of dry etching to form the laser ridges. M664O and M577G lase at 8.2 and 5.4 μm, respectively. Both QCLs have a highly reflective (HR) coating (${\mathrm{Y}}_2{\mathrm{O}}_3/\mathrm{Ti}/\mathrm{Au}$) on the rear facet. The other four lasers were processed from the same wafer and lase at 4.8 μm and are labeled M1244. Two of the QCLs (M1244C and D) have an all-dielectric HR coating on the rear facet whereas the other two QCLs (M1244F and G) have simply as-cleaved facets. The Maxion QCLs are mounted epi-down using AuSn solder onto an AlN submount and then either Pb/Sn or Ag/Sn soldered onto a copper c-mount.

Earlier measurements have shown that QCLs exhibit a small burn-in effect in which a lower threshold current and a higher slope efficiency is observed as current flows through the device [5]. Thus, all of the QCLs used in this study have been burned in by operating the QCLs for at least 100 h prior to these experiments.

A. Proton Irradiation

The proton irradiation testing was performed at the University of California at Davis Crocker Nuclear Laboratory using a 64 MeV proton beam. Each exposure involved a proton fluence of $7.46\times {10}^{10}\mathrm{p}/{\mathrm{cm}}^2$ for a dose of 10 krad(Si). This fluence was selected to be higher than the total flux expected from trapped protons. For example, with minimal aluminum shielding and at an altitude of 500 km with 60° inclination, the proton flux is estimated to be $3\times {10}^4\mathrm{p}/{\mathrm{cm}}^2$-day [8]. Three sets of exposures were performed over different time periods, and due to the logistics of the proton exposure tests, the characterization before and after exposure was separated by about 1–2 months. Two of the QCLs (Ham-Ce and Ham-Cf) were exposed to two proton irradiation steps for a total proton dose of 20 krad(Si); the period between the two exposures was 6 weeks.

B. Gamma Irradiation

The gamma irradiation work was performed at Pacific Northwest National Laboratory using a Cobalt-60 source. The QCLs are mounted in a thin, plastic box that is irradiated with gamma radiation in three steps with an exposure of 10,000 roentgen (R) and an exposure rate of approximately 5000 R/h for each step. This exposure results in an absorbed dose of 8.77 krad(Si); thus, the total accumulated dose for all three steps is 26.3 krad(Si). The characterization before and after exposure was separated by 1–2 days, and the period between each exposure was typically 2–3 weeks.

C. Measurement Details

Each QCL was characterized prior to the radiation exposure to establish the baseline performance. The changes in output power at set current levels along with the threshold current and slope efficiency were used to monitor any effects from radiation exposure. The measurement system, which includes a calibrated thermopile detector with a 19 mm aperture and a temperature-controlled laser mount, is enclosed in a Styrofoam cooler as described in [5] to minimize thermal fluctuations. The QCLs are attached to the laser mount, which is maintained at 20°C, with a 2–56 cap screw, as shown in Fig. 1. A custom current controller [19,20] is used to drive the QCLs; the lasers are operated quasi-CW using a 40 kHz square wave with 100% modulation depth and a 50% duty cycle. A modulated drive current provides a lower variability in the output average power; with continuous-wave (cw) operation, the FP-QCLs show mode-hopping when the laser is cycled on and off, causing larger power fluctuations [5]. The above-threshold currents were set initially to produce optical powers from the QCL of approximately 6, 12, 18, 24, and 29 mW. Each current step was 15 s long, with two steps below threshold followed by the five lasing steps and finally two steps with the current off. In each step, the first 10 s were allowed for power readings to stabilize, which accounts for the slow response of the power meter, and only the power readings for the last 5 s were used for analysis. The two initial steps were used to preheat the QCL to near its operating point and to provide a measurement of the power meter offset, which was subtracted from the power readings. The final two steps were also used to measure the power meter offset at the end of the cycle and to observe any drift. The output power at each current level was typically measured for 26 cycles, and these values were averaged together to provide a mean output power at each current level, which was used to generate power versus current (L-I) curves for each of the QCLs. By performing a linear regression on the data points of the L–I curves, the threshold current and slope efficiency can be calculated. Since only minimal changes in threshold current and slope efficiency ($<0.03\%$) were observed after 24 h of operation, the measurements in this paper do not include a warm-up period in order to minimize testing time. Additional characterization involved a Bruker Vector 22 FTIR spectrometer to record the spectral output and a Keyence digital microscope to provide photomicrographs of the QCLs.

 

Fig. 1. Expanded view of schematic for the commercial mounting fixture used in the testing station. Before a c-mount is attached, both surfaces are cleaned with methanol, and then a 2–56 mounting screw is torqued to ensure thermal contact between the two surfaces. A thermistor is integrated into the laser mount to control the temperature. Variations in thermal contact can lead to changes in the active region temperature of the QCL, which is attached to the top of the c-mount.

Download Full Size | PDF

1. Control QCL

A control QCL, M577G, was added to the testing protocol after the first Co-60 exposure and was evaluated in tandem with the QCLs exposed to radiation. This QCL did not leave the measurement laboratory and was not subject to the stresses of radiation dose or transportation and provided an indicator for changes in the test measurement setup.

2. Measurement Variability

Since the measurement setup contains one temperature-controlled mount, the QCLs were dismounted when testing multiple devices and were also dismounted when they were sent to the radiation facility for exposure. Repeated measurements, however, in which the laser was dismounted and remounted to the temperature-controlled mount, showed this action to be a major source of variability in the output power even at a set current and temperature. This increased variability in output power was much higher than that measured when the QCL was left attached to the mount [21] and is most likely due to imperfect reproducibility of the thermal contact between the QCL and the laser mount. Because the QCL temperature was regulated by controlling the temperature of a thermistor embedded in the plate of the laser mount, changes in thermal contact between the QCL and the mount will result in changes to the temperature of the QCL active region. Minor changes in the active region temperature will cause fluctuations in the output power as well as threshold current. Thus, in order to decouple the effect of laser mounting and dismounting from potential effects of radiation, we measured the variability from dismounting and remounting five different QCLs prior to any radiation exposure for a set of 10 measurements each. For these measurements, a set procedure was followed: the laser mount surface and the back of the c-mount were cleaned with methanol, and the QCL was attached to the laser mount using a torque wrench to tighten the mounting screw to 32 oz-in. The average uncertainty for the threshold current over these 10 measurements for five different QCLs was $\pm 0.2\%$, and the uncertainty for the slope efficiency was $\pm 2\%$. On the other hand, the uncertainty was much lower, by at least an order of magnitude, when the QCL was not removed from the mount. Thus, minor changes in threshold current ($<0.2\%$) or slope efficiency ($<2\%$) are below the measurement uncertainty and are not detectable.

Most of the measurements in this paper are an average of 10 measurements to better account for the uncertainty that is observed with dismounting and remounting the QCL. The measurements involving Ham-Ce and Ham-Cf, as well as the gamma exposure tests for M664O, however, have a fewer number of measurements. The error bars reported in this paper represent twice the standard deviation ($2\sigma$) from the average. For data that show the fractional change, the error bars are calculated using propagation of errors; thus the average uncertainty for the fractional change in the threshold current and slope efficiency is $\pm 0.3\%$ and $\pm 3.0\%$, respectively.

The spectral output also showed minor variations in the broadband emission as the laser is dismounted and remounted. In particular, the intensity of different longitudinal modes within the linewidth changes slightly when the laser was dismounted and reattached to the temperature-controlled mount. Thus, the spectral measurements were primarily used to look for large spectral changes from radiation exposure.

3. Results and Discussion

Ham-Ce and Ham-Cf were exposed to 10 krad(Si) of high-energy 64 MeV protons in two steps for a total accumulated proton dose of 20 krad(Si). Figures 2(a) and 2(b) show the measured changes in the threshold current for Ham-Ce and Ham-Cf, respectively, as a function of proton dose. The characterization prior to the first proton exposure involved only one set of measurements; since the largest variability results from dismounting and remounting the QCL to the temperature-controlled mount, the error bars for this initial measurement are assumed to be the average uncertainty calculated in the prior section (i.e., $\pm 0.2\%$). Furthermore, for this initial characterization, we discovered that the thermopile was not mounted in a manner that maintained precise alignment between the laser and the thermopile when it was moved in and out of position; thus, only the threshold current is reported because it is less sensitive to the thermopile position than the slope efficiency. After the first proton exposure, test fixtures were changed to provide precise mechanical alignment between the laser and the thermopile. Measurements using these test fixtures showed good reproducibility when only the thermopile was moved in and out of position; the threshold current showed a variability of $\pm 0.02\%$ and the slope efficiency showed a variability of $\pm 0.2\%$; these uncertainties are an order of magnitude better than the uncertainties due to dismounting and reattaching the QCL to the temperature-controlled mount.

Figure 2(a) shows a small increase in threshold current after the first proton exposure for Ham-Ce although it is unlikely it was due to radiation exposure because no measurable change was observed when the device was exposed a second time with an equivalent proton fluence. This small increase in threshold current is most likely due to measurement uncertainty from the limited number of measurements as well as the lack of using appropriate test fixtures to improve the precision of the thermopile position. Furthermore, Ham-Cf, which is very similar to Ham-Ce, does not show a measurable change in threshold current after the first proton exposure, as shown in Fig. 2(b). Mechanical damage to Ham-Cf from mishandling prevented further testing with this QCL and only two measurements were completed after the second proton exposure.

 

Fig. 2. Plot of the calculated average threshold current for (a) Ham-Ce and (b) Ham-Cf as a function of proton dose. The error bars are shown as $2\sigma$. The initial characterization prior to radiation exposure involved only one measurement so the error bars are assumed to be $\pm 0.2\%$. After the first proton irradiation, five and six measurements are used for Ham-Ce and Ham-Cf, respectively. The measurements after the second proton irradiation involve three and two measurements for Ham-Ce and Ham-Cf, respectively.

Download Full Size | PDF

Ham-Ce along with a different QCL, M664O, were then exposed to gamma rays in a series of three steps for an accumulated dose of 26.3 krad(Si). Figures 3(a) and 3(b) show the threshold current as a function of gamma dose for both Ham-Ce, which had already been exposed to protons prior to these measurements, and M664O. No measurable change in threshold current is observed after each exposure. The percent change in threshold current after an accumulated dose of 26.3 krad(Si) is only $+0.03\%\pm 0.22\%$ for Ham-Ce and $+0.1\%\pm 0.15\%$ for M664O. These small changes are within measurement error.

 

Fig. 3. Plot of the calculated average threshold current for (a) Ham-Ce and (b) M664O as a function of gamma dose. The error bars are shown as $2\sigma$ and involve 3–4 measurements for each data point.

Download Full Size | PDF

The proton irradiation tests were repeated using five QCLs (M664O and the four M1244 devices). These five QCLs were exposed to 10 krad(Si) of high-energy 64 MeV protons. For these tests, an average of 10 measurements were used to better account for the variability that results from dismounting and remounting the QCL to the temperature-controlled mount. Figures 4(a) and 4(b) show the percent change in threshold current and slope efficiency, respectively, for the QCLs after the proton exposure. A control QCL was also tested in tandem and is shown in Fig. 4 for comparison. Although the error bars differ for each QCL and for each set of measurements, the uncertainty for the fractional change in threshold current is generally around $\pm 0.3\%$ and the uncertainty for the fractional change in slope efficiency is generally around $\pm 3\%$, as discussed earlier. No measurable change is observed and all of the QCLs, except for M1244D, show a slight decrease in threshold current (i.e., a decrease in losses in the QCL) after proton irradiation. The change in threshold current varied from $+0.1\%$ to $-0.4\%$, and the change in slope efficiency varied from $+0.3\%$ to $-1.2\%$; these changes are within the measurement uncertainty. These tests clearly show that QCLs are robust to proton exposure at these radiation levels, which are typically used for space qualification.

 

Fig. 4. Measured change in (a) threshold current and (b) slope efficiency for the various QCLs after a proton dose of 10 krad(Si). The control QCL, M577G, remained in the laboratory and was not exposed to any radiation. The error bars were calculated using propagation of errors and are shown as $2\sigma$.

Download Full Size | PDF

The QCLs from the M1244 series were then exposed to gamma rays. M664O was not used in this series of tests because it had already been exposed to gamma rays prior to the proton exposure. Figures 5(a) and 5(b) show the percent change in threshold current and slope efficiency, respectively, after an accumulated gamma dose of 26.3 krad(Si). The change in threshold current varied from $+0.1\%$ to $+0.3\%$ showing a bias toward a higher threshold current, including for the control QCL. The change in slope efficiency varied from $+0.3\%$ to $-0.3\%$. These changes in threshold current and slope efficiency are within the measurement uncertainty. Higher uncertainties for the threshold current and slope efficiency were usually observed for both M1244F and M1244G, which do not have an HR coating on the rear facet; this larger variability is presumably due to half of the light exiting the rear facet, which can lead to scattering off of the laser mount and increase variability with small changes in mounting position and laser temperature.

 

Fig. 5. Measured change in (a) threshold current and (b) slope efficiency for the various QCLs after an accumulated gamma dose of 26.3 krad(Si). The control QCL, M577G, remained in the laboratory and was not exposed to any radiation. The error bars were calculated using propagation of errors and are shown as $2\sigma$.

Download Full Size | PDF

The bias toward a slight increase in threshold current for all of the QCLs including the control QCL may be due to the excessive handling of the QCLs during the testing. Over the duration of these measurements, the QCL was dismounted and reattached to the laser mount 40 times. Furthermore, throughout the entire testing program, at least 60 measurements have been completed for the M1244 QCLs and control QCL. The laser chip is exposed for QCLs on c-mounts and is particularly vulnerable to damage while being mounted and dismounted. Micrographs of all of the QCLs (including the control QCL) did show minor chips and fractures in the InP substrate after the testing was completed although the laser facets did not appear damaged. Furthermore, inspection of the back of the c-mounts for all of the QCLs showed discoloration and the gold coating appeared to be wearing off from the repeated mounting; this excessive wear of the mounting surface may affect the thermal contact causing a slight increase in threshold current. Further evidence for the degradation of the mounting surface with repeated mounting is shown in Fig. 6; a blank gold-coated copper c-mount has been dismounted and remounted 20 times with a similar mounting procedure used for the tests in this paper. Signs of wear and discoloration were observed on the back of the c-mount, as shown by the photograph on the right in Fig. 6. Further tests, however, would be needed to confirm that excessive handling or degradation of the mounting surface is the cause for the slight bias in the threshold current measurements. Regardless, excessive handling such as used in these experiments is not recommended for QCLs on c-mounts and provides the greatest threat to laser damage since the laser chip and facet are both exposed. For most applications, however, this hazard is not relevant since the QCLs usually remain on a mount or in a package for the entire test or experiment.

 

Fig. 6. Micrographs of the back of the c-mount before it has been mounted (left) and of the same c-mount after it has been mounted and dismounted 20 times (right).

Download Full Size | PDF

Spectral measurements using an FTIR were also recorded prior to the radiation testing and after the completion of the radiation exposure tests. The spectra did not show significant changes after the radiation exposure for any of the QCLs and differences in the intensity of different longitudinal modes within the broad spectral linewidth are similar to changes observed with dismounting and reattaching the QCL to the temperature-controlled mount as observed prior to any radiation exposure.

4. Conclusion

This paper has demonstrated that FP-QCLs from two different vendors continued to operate after exposure to radiation levels that are likely to be encountered for most space systems. The accumulated dose varied from 20  to 46.3 krad(Si) and two different types of radiation were used. Six of the QCLs were exposed to both high-energy 64 MeV protons and gamma rays; one QCL was exposed to just the 64 MeV protons.

No measurable changes in output power, threshold current, and slope efficiency were observed. These studies show that FP-QCLs continue to demonstrate excellent performance after both proton and gamma irradiation making them a viable choice for use in space missions. These tests suggest much higher radiation doses are required to damage the energy levels in the electronic band structure of the QCL and degrade performance.

Mission requirements may require different operating conditions or QCLs be used. This work tested devices from two different vendors, but due to the properties of QCLs, we expect QCLs from other vendors to also be insensitive to radiation damage at similar exposure levels. In addition, this work only investigated FP-QCLs although single-mode distributed feedback (DFB) QCLs are important for applications that require a narrow spectral linewidth. Thus, future studies may involve DFB-QCLs to determine if they are more susceptible to radiation damage. Finally, this work tested the QCLs at an operating temperature of 20°C to avoid condensation or having to perform the tests in a vacuum environment, but space-based missions may require operating the QCLs at a lower temperature to reduce the heating requirements. Operation at lower temperatures is not expected to have a greater impact on radiation sensitivity; the threshold current, slope efficiency, and output power are actually improved at lower temperatures.

Radiation doses were selected to match typical radiation doses that are encountered for most space-based missions over a 7-year timeframe. These tests clearly show that QCLs are expected to function although additional radiation testing of QCLs may be required depending on the mission requirements. Future testing will involve higher doses that may lead to radiation damage to provide a better understanding of the radiation degradation mechanisms in QCLs.