In situ ultraviolet shock radiance measurements using GaN-on-sapphire photodetectors

Planetary exploration is constrained by the capabilities of atmospheric entry technologies, including thermal protection systems (TPSs). Spacecraft experience high heat fluxes during atmospheric entry and thus require a TPS to protect the payload.¹ Prediction of the entry heat load and subsequent sizing of the TPS are often focused on convective heating. However, for large vehicles and high entry velocities, shock-layer radiation can be a major contributor.^2–4 As of now, few in-flight shock-layer radiation measurements have been made. The FIRE II vehicle in 1965, Apollo 4 in 1967, the Planetary Atmosphere Experiments Test (PAET) vehicle in 1971, and the Orion Exploration Flight Test-1 (EFT-1) vehicle in 2014 used on-board radiometers to measure radiation during Earth re-entry.^5–8 These radiometers consisted of a sensing element (often a thermopile) behind a window embedded in the TPS. Due to the window and sensor materials used by the radiometers on-board Apollo 4 and PAET, measurements were limited to visible to near infrared radiation, neglecting ultraviolet radiation, which can account for half of the radiative heating.⁵ 

Spacecraft heatshields including those on PAET, Apollo, Space Shuttle, and Mars Science Laboratory (MSL) have been successfully instrumented for measuring temperature and pressure.^7,9 These in-flight measurements have enabled aerothermal model refinement and improved TPS design. However, the advancement of atmospheric entry sensing technologies has been limited due to the high heat fluxes and complexity of instrumenting the TPS. Current state of the art seen on the Mars Entry Descent and Landing Instrumentation (MEDLI) sensor suite on the MSL heatshield consists of thermocouples embedded in a plug of TPS material (Phenolic-Impregnated Carbon Ablator, PICA) and pressure transducers connected to pressure ports in the TPS.^10–12 By exploring wide bandgap semiconductor materials, such as gallium nitride (GaN), aluminum nitride (AlN), and silicon carbide (SiC), the operating regime for TPS sensors and electronics can be extended.^13–23 Sensors using wide bandgap materials have shown thermal, mechanical, and chemical stability, allowing for potential operation in extreme temperature and radiation rich environments, such as in space and during atmospheric entry.²⁴ In particular, GaN-based photodetectors have been shown to be highly responsive to incident ultraviolet (UV) radiation as well as robust to harsh environmental conditions.^13–19 GaN has a direct and tunable bandgap through modification of the doping content; thus, the optical properties of GaN-based UV photodetectors can be tuned to fit to the specific application.¹⁹ GaN-based sensors may be placed closer to the surface of the TPS allowing for more accurate measurements to be obtained.

In this study, the feasibility of implementing GaN-based photodetectors to detect shock-layer UV radiative heating through in situ measurements in a shock tube is explored. GaN-on-sapphire metal-semiconductor-metal (MSM) photodetectors were installed in the Electric Arc Shock Tube (EAST) facility at NASA Ames Research Center and characterized in a flight-relevant Titan entry environment (98% N₂, 2% CH₄). Accurate characterization of the UV shock-layer radiation is especially important for a Titan aeroshell as the peak radiative heat flux for a Titan aerocapture mission is predicted to be four times greater than the peak convective heat flux with the majority of the radiative heating in the ultraviolet regime due to CN radicals formed by the shock wave.^3,4

Although the detailed microfabrication process for the GaN-on-sapphire photodetectors can be found elsewhere,²⁵ a brief description is repeated here. First, the GaN film was mesa etched via reactive ion etching for device isolation. Next, 50 nm of SiO₂ passivation was deposited via atomic layer deposition (ALD) and subsequently etched to create the active area for each photodetector. Semi-transparent electrodes with 30%–50%¹⁹ UV transmittance were formed using a standard metal lift-off procedure with 3 nm Ni/10 nm Au. Lastly, electrical contact metal was deposited (20 nm Ti/200 nm Au). The photodetector active area is 250 μm × 250 μm with an electrode length of 240 μm, a width of 5 μm, and a spacing of 5 μm. The GaN-on-sapphire chip used in this study is 5 mm × 5 mm and has nine equally spaced photodetectors. Two of the nine photodetectors on this chip were characterized in this preliminary study and will be referred to as photodetector 1 and photodetector 2. An optical image of a microfabricated GaN-on-sapphire UV MSM photodetector is shown in Fig. 1(a).

To enable spectral response characterization and in situ shock tube testing, the GaN-on-sapphire chip was epoxied (Devcon, 5 Minute) and wirebonded to a transistor outline (TO) header (Spectrum Semiconductor Materials, Inc., TO-8). To protect the wirebonds during shock testing, they were encapsulated in epoxy while leaving the photodetectors exposed to the shock wave. An image of the packaged GaN-on-sapphire chip is shown in Fig. 1(b).

To characterize the spectral response of the GaN-on-sapphire photodetectors, a laser driven light source (EQ-1500, Energetiq) with nearly constant radiance across the ultraviolet spectra and spectrometer (McPherson 218) with diffraction grating (246.16 grooves/mm and 226 nm blaze) setup was utilized as shown in Fig. 2. To begin, the spectrometer count to power relationship was determined using a NIST-traceable calibrated reference photodetector (918D-UV-OD3R, Newport) with a power meter (841-P-USB, Newport). The inlet and exit slit widths of the spectrometer were set to 2 mm and 0.75 mm, respectively, to maximize the applied optical power with a reasonable wavelength resolution (10% of peak at ±20 nm). Next, the calibrated photodetector was used to measure the power of the laser driven light source from 500 nm to 200 nm in increments of 10 nm. In order to reduce second order diffraction effects, a long pass filter was used for wavelengths greater than 340 nm. Finally, current-voltage measurements were taken for the GaN-based photodetectors under illumination from 500 nm to 200 nm in 10-nm increments. The inset of Fig. 2 shows the power incident on the GaN photodetectors as a function of wavelength as measured at the imaging plane of the spectrometer by the reference detector with the difference in detector size between the reference (0.75 mm × 1.13 cm) and GaN (250 μm × 250 μm) photodetectors accounted for. Two separate measurements were taken above and below 340 nm with the filter in place. Differences in source alignment cause the applied power to differ between the two measurements. In the results shown below, the spectral variation in the applied optical power is accounted for as will be discussed. All electrical measurements were taken using a semiconductor parameter analyzer (B1500, Agilent Technologies).

The current-voltage responses for the two separate GaN-on-sapphire MSM photodetectors are shown in Fig. 3 under dark (unilluminated) and illuminated (500 nm, 365 nm, and 200 nm wavelengths) conditions. From the figures, it is easily seen that the photodetectors are most responsive to 365 nm illumination which directly corresponds to the bandgap of GaN (3.4 eV).

To further quantify the photodetector performance, the photocurrent-to-dark current ratio (PDCR),^20,21 which is defined as

where I_photo is the photocurrent and I_dark is the dark current, is often quoted. Since the photo-generated current (I_photo − I_dark) for GaN-based photodetectors is proportional to the square root of the applied optical power,²⁶ a more objective metric would be the normalized photocurrent-to-dark current ratio (NPDR)^27,28 defined as

When defined in this way, the NPDR accounts for the different applied optical powers at different wavelengths which are shown in the inset of Fig. 2. The NPDR as a function of illumination wavelength for photodetectors 1 and 2 are shown in Fig. 3. The NPDR peaks around 365 nm and quickly drops off in the visible wavelength region. Another photodetector sensitivity factor which is directly proportional to external quantum efficiency¹⁹ is responsivity,

where once again the square root dependence of photo-generated current with applied optical power is taken into account. The spectral responsivities for both tested GaN-on-sapphire photodetectors are shown in Fig. 3. The applied optical power used for these calculations was that measured using the calibrated reference photodetector as shown in Fig. 2. For both GaN-based photodetectors, the 365 nm to visible rejection is about one order of magnitude, and the responsivity at 200 nm is about one-third of the responsivity at 365 nm.

The slight differences in the current-voltage response, NPDR, and spectral responsivity between the two photodetectors shown in Fig. 3 can be attributed to variations during fabrication (defects in the GaN, differences in metal contact resistance, misalignment during photolithography, etc.) as well as variations in packaging (wirebonding, TO header, etc.). While the magnitude of the photo-response may not be the exact same between the two photodetectors, the overall trends seen in Fig. 3 are the same. Furthermore, these results are consistent with other DC spectral responsivity measurements of GaN-based photodetectors reported in the literature.^19,29 The relatively low UV/visible rejection ratio can be attributed to persistent photoconductivity effects due to impurities, dislocations, and defects trapping photo-generated carriers for long times after the illumination source has been removed.^{13,14,19,29,30} Most often, spectral responsivity measurements are reported using a chopper with a lock-in amplifier, in which all phenomena occurring at rates slower than the chopper frequency are eliminated. Using this technique results in UV/visible contrasts greater than three orders of magnitude. However, the chopping frequency affects the absolute value of the responsivity, its variation with incident optical power, and its variation with the excitation wavelength.^19,26,29,30 For implementation on an aeroshell (or any other real-world application without a pulsed illumination source), it is most important to understand the DC characteristics of the device to accurately quantify the UV shock-layer radiation.

The EAST facility at NASA Ames Research Center is a ground-based facility that produces flight relevant radiation. The facility uses four separate spectrometers to observe and characterize radiation from the shock wave from vacuum ultraviolet (120 nm) to near infrared (1700 nm).^2,3,31 A schematic of the EAST facility is shown in Fig. 4 as well as the test setup for characterizing the performance of the GaN-based photodetectors in the shock tube. The packaged photodetectors were soldered to wire leads in the vacuum-tight shock port holder and installed in the test section of the shock tube just downstream of the spectrometers used by the facility. The two GaN-on-sapphire MSM photodetectors (photodetector 1 and photodetector 2) used for spectral response characterization were used to measure the UV radiance during two separate shock tests. A source meter (2400 Series, Keithley) was used to take current-voltage measurements pre- and post-shock as well as apply a voltage and measure the transient current response during the shock test. Due to limitations of the equipment, the sampling rate during the shock test was only about 8 Hz. Since the Titan entry conditions and test parameters dictated that the shock speeds should be around 5-6 km/s, a separate high frequency (5 MHz) data acquisition (DAQ) system was also used to acquire data during the shock test. The DAQ was triggered to record 4 ms of data when the burst disk ruptured, whereas the source meter collected data for several minutes before and after the shock event. The high frequency measurements were taken as voltage measurements (V_meas) across a resistor from which the photodetector current was calculated. For all tests, the applied voltage (V_app) was 200 mV.

Table I lists the parameters for the two shock tests conducted, with photodetector 1 active during the first shock test and photodetector 2 active during the second. The first shock test melted the front edge of the TO header and broke the wirebonds of photodetector 1, rendering the low frequency transient current measurements useless. The transient current response of photodetector 2 during the second test, measured with the source meter (8 Hz sampling frequency), is shown in Fig. 5. The shock event is clearly visible in this data, and the slow current decay after the shock has passed is consistent with the persistent photoconductivity phenomena described in Sec. III B. The unilluminated (dark) current-voltage response for photodetector 2 is unchanged from pre- to post-shock as shown in the inset of Fig. 5. A change in the dark current would indicate material damage (displacements, dislocations, etc.) and/or electrical degradation (Schottky barrier height). Since the unilluminated current-voltage response is unchanged, this indicates that there is little degradation in the device performance after being subjected to the shock wave.

The 5 MHz sampling frequency transient current response during the shock test is shown in Fig. 6(a) for photodetector 1 and Fig. 6(b) for photodetector 2. A moving average filter with a window size of 100 samples was applied to the collected data to reduce the noise. The filtered signal displays a distinct increase in current after the shock passes the detector giving a direct measure of the ultraviolet shock-layer radiation. In Fig. 6, the time when the shock and contact surface pass the detector is easily identifiable due to the large spikes in current. These spikes can be attributed to the exposed contacts collecting free electrons in the partially ionized gases or forming a ground loop between the measurement circuit and the facility.

For GaN-based photodetectors, the photo-generated current is proportional to the square root of the applied optical power.²⁶ Applying this relation to the measurements from the spectral response characterization at 365 nm (termed “spectral response data” in Fig. 7) results in the fits (termed “P^1/2 fit” in Fig. 7) shown in Fig. 7. The photo-generated current from the shock tests can be calculated from Fig. 6 as the average current before the shock subtracted from the average of the maximum 100 filtered samples between the shock and contact surface. Overlaying this photo-generated current from the shock tests (termed “shock test 1” and “shock test 2” in Fig. 7) on top of the square root of power fits in Fig. 7 results in a measured optical power of about 0.30 μW for both shock tests. From the EAST facility spectrometers, the average UV (190 nm–370 nm) intensity during the two shock tests was measured to be about 49.5 μJ/cm² and 45.0 μJ/cm². Multiplying these values by the active area of the respective GaN-based photodetector and the characteristic time gives 0.31 μW and 0.28 μW which are in excellent agreement with the 0.30 μW result from the square root of the power fit to the data from GaN-based photodetectors.

Using NASA’s chemical equilibrium with applications (CEA) software,³² the temperatures behind the shocks were calculated to be 5589.53 K and 5324.28 K for tests 1 and 2, as listed in Table I. These very high temperatures persist for less than 1 ms, and thus the photodetectors are not heated substantially. Therefore it is reasonable to neglect temperature effects in the proceeding analysis. However, the descent through a planetary bodies’ atmosphere could be on the order of several minutes. To protect the photodetectors from these very high temperatures over this relatively long duration, they could be implemented behind a window or on the backside of the aeroshell, similar to the implementation of previous heatshield sensors. However, due to the temperature tolerance of GaN, these sensors can be placed closer to the surface of the TPS than previously possible, thus increasing the understanding of the UV radiative heat load.

Atmospheric entry conditions for many planetary bodies have been predicted to have high heat fluxes dominated by shock-layer radiation rather than thermal convection. In order to design thermal protection systems to enable exploration missions to these places, this heat load needs to be well understood. GaN-on-sapphire photodetectors were shown to be highly responsive to ultraviolet radiation, especially at 365 nm, demonstrating an order of magnitude contrast from visible radiation. The GaN-on-sapphire ultraviolet photodetectors were successfully used to detect ultraviolet radiance behind a shock in a simulated Titan entry environment in an electric arc shock tube. The calculated optical power from the transient current response measurements made by the GaN photodetectors was 0.30 μW, which is in agreement with the UV radiance measured by EAST facility spectrometers. Furthermore, the electrical response of the photodetectors was not measurably degraded by the shock as well as there was no visible damage to the photodetectors post-shock. These results demonstrate the feasibility of small-scale GaN-based sensors operating in the harsh atmospheric entry environment.

The authors would like to thank Mark McGlaughlin and Rick Ryzinga for maintenance and operation of the Electric Arc Shock Tube facility. Additionally, Caitlin Chapin and Ateeq Suria are thanked for their fabrication assistance and insightful discussion. This work was supported by a NASA Graduate Aeronautics Scholarship (Grant No. NNX15AV94H). Fabrication work was performed in part at the Stanford Nanofabrication Facility (SNF) and Stanford Nano Shared Facilities (SNSF).