Resonant Cavity–Enhanced Photodiodes for Spectroscopy of CH Bonds

Resonant cavity‐enhanced photodiodes targeted within the spectral region of absorption by CH bonds are demonstrated. The 3.0 – 3.3   μ m region of the infrared spectrum contains many substances that are useful to measure spectroscopically. However, the measurement of individual substances requires a high spectral specificity, that is achieved by the resonant cavity photodiodes with spectral response widths of < 40   nm . Two material systems are investigated for detection at this wavelength range—an InAs absorber on an InAs substrate and an InAsSb absorber lattice‐matched to a GaSb substrate. The resonance wavelength of the InAs‐based device responds at ≈ 3.3   μ m , closely tuned to an absorption peak of methane to allow precise sensing of this gas. At 300   K a quantum efficiency of 52 % is achieved, with a specific detectivity of 2.5 × 10 10 cm Hz / W . The InAsSb‐based device is sensitive at ≈ 3.7   μ m , but the structure could be tuned to the methane absorption peak. Devices could be simply created to target other substances in the C−H absorption region by altering the layer thicknesses in the structure. Both structures can be used for spectrally specific gas sensing in this region of the infrared.

DOI: 10.1002/pssa.202100056 Resonant cavity-enhanced photodiodes targeted within the spectral region of absorption by C─H bonds are demonstrated. The 3.0-3.3 μm region of the infrared spectrum contains many substances that are useful to measure spectroscopically. However, the measurement of individual substances requires a high spectral specificity, that is achieved by the resonant cavity photodiodes with spectral response widths of < 40 nm. Two material systems are investigated for detection at this wavelength range-an InAs absorber on an InAs substrate and an InAsSb absorber lattice-matched to a GaSb substrate. The resonance wavelength of the InAs-based device responds at %3.3 μm, closely tuned to an absorption peak of methane to allow precise sensing of this gas. At 300 K a quantum efficiency of 52% is achieved, with a specific detectivity of 2.5 Â 10 10 cm ffiffiffiffiffiffi Hz p =W. The InAsSb-based device is sensitive at %3.7 μm, but the structure could be tuned to the methane absorption peak. Devices could be simply created to target other substances in the CÀH absorption region by altering the layer thicknesses in the structure. Both structures can be used for spectrally specific gas sensing in this region of the infrared.

Design
GaAs and GaSb substrates are usually favored over InAs for RCE-PDs in part because they allow for simpler distributed Bragg reflectors (DBR) to be used-AlGaAs/GaAs for GaAs substrates and AlAsSb/GaSb for GaSb substrates. These DBRs both include a binary layer, which is relatively simple to grow. In contrast, GaAsSb/AlAsSb ternary DBR layers are required for lattice-matched growth on InAs. This increases the complexity because both layer compositions need to be carefully controlled. The structure, as shown in Figure 1a, uses a DBR mirror above and below the cavity. A 12-pair DBR is used below the cavity and a 5.5-pair DBR above. The AlAs 0.16 Sb 0.84 and GaAs 0.08 Sb 0.92 layers used have an %20% lower refractive index contrast than AlAsSb and GaSb, although it would be simple to add more layers to counteract the loss in reflectivity.
The cavity consists of an nBn structure and a cavity spacer layer beneath. InAs is used for the n layers and AlAs 0.16 Sb 0.84 is used for the barrier and filler layers. The use of a bulk binary absorber simplifies the cavity growth compared with other RCE-PDs and is more likely to produce a high-quality crystalline layer. InAs-based RCE-PDs have previously been demonstrated, [11] but it is not clear from previous research how the performance of InAs RCE-PDs compares with the devices on GaSb. Devices were fabricated from the structure, grown by molecular beam epitaxy on an n-type InAs substrate (see Experimental Section).

Optical Characterization
The two detectors use similar designs but with the key differences in the absorber and substrate materials. The designs are otherwise similar, allowing for comparisons to be made based on optical measurements. A Fourier-transform infrared (FTIR) spectrometer was used to characterize the spectral response. The responsivity and quantum efficiency (QE) were calculated with reference to a commercial Vigo MCT detector of known responsivity. The normalized response of each detector is shown in Figure 2a The exact wavelength of the resonance peaks is determined by a combination of the optical path lengths of the DBR layers and the cavity. These two structures could be fabricated with the same 3.3 μm resonance wavelength by simply altering the thicknesses of the DBR layers and the total cavity thickness.
It has previously been reported that the resonance peak wavelength shifts with temperature, with the majority of the shift attributed to changes in refractive indices of the cavity layers. [14,15] Figure 2b shows the temperature dependence of the resonant wavelength for the two devices. The gradient of the two fit lines shows clearly that the peak shift due to temperature is very similar for both devices-0.27nm=K for the InAsSb-based device and 0.28nm=K for the InAs-based device. The temperature dependence might be expected to reflect the resonant wavelength, dλ=dT ∝ jλj. However, this is not what is seen. This indicates that the layers in the cavity have lower temperature coefficients of the www.advancedsciencenews.com www.pss-a.com refractive index for the InAsSb-based device. This is as expected from literature values of the temperature coefficients. [16][17][18] Figure 2c shows the temperature dependence of the full width half maximum (FWHM) of the resonance peak for the two devices. Both devices show a linear FWHM decrease of %0.04 nm=K. We attribute this effect to increases in the absorption coefficients with temperature. The dependence of the FWHM on the absorption coefficient has been modeled for the InAs-based structure and is overlaid on the same plot. The increase in FWHM above 250 K can be explained by an increase in absorption coefficient of %18 cm À1 =K. The similarity in the FWHM variation for the two devices is expected, due to the similarity of the absorber materials. Below 250 K the InAs-based device shows a far stronger change of 0.16 nm=K, attributed to a higher rate of change of the absorption coefficient close to the band edge. This is not seen in the InAsSb-based device as the resonant wavelength is not close to the band edge.
The peak responsivity for both devices is also significantly dependent on the temperature of the device, as shown in Figure 2d. The temperature of the highest peak responsivity impacts the real-world applications of the devices. The peak temperatures differ significantly-225 K for the InAsSb-based device and 350 K for the InAs-based device. The temperature dependence of the spectral response can also be largely attributed to the absorption coefficients of the absorbers. The InAs absorption coefficient at the resonant wavelength decreases rapidly below 275 K as the bandgap increases, reducing the overall absorption. Both devices show a slight decrease in responsivity above a certain temperature, which could be attributed to an increased absorption coefficient damping the resonant enhancement.

Electrical Characterization
The full optoelectronic characterization is presented for the InAsbased RCE-PD only. Select comparisons of the electrical properties of the two devices are made; however, it is important to note that these results do not necessarily represent the ultimate performance of each material system. Therefore, it is not possible to state that the performance of one is better than the other. The comparisons do offer insight into the temperature dependencies of each structure and allow educated expectations about the limitations of each material system to be formed. www.advancedsciencenews.com www.pss-a.com The InAs-based RCE-PD was designed to target the absorption peak of methane at %3.3 μm. With reference to figure 3a, it can be seen that the resonance peak coincides closely with the targeted absorption peak. [19] This demonstrates the focused spectral targeting that RCE-PDs can achieve with appropriate calibration of the layer thicknesses. The targeted narrow linewidth allows for the measurement of methane concentrations with high spectral specificity. The quality factor of the InAs-based cavity was found to be in the range 80À137, increasing with temperature. Both devices demonstrate similar quality factors at 300 K, of 88 and 84, for the InAs and InAsSb-based devices, respectively, indicating that the higher-complexity DBRs in the InAs-based device did not compromise material quality. Figure 3b shows a coupled high-resolution X-ray diffraction (XRD) scan of the sample. Good lattice matching of the ternary mirror layers was confirmed, with all layers within 200 sec of the InAs substrate.
The external quantum efficiency was calculated from FTIR spectral response measurements. Temperature and voltagedependent results are shown in Figure 4a. Results for the InAs-based device show that at 200 K and above there was a measurable response that monotonically increased with temperature up to 350 K, with a slight decrease at 375 K. Peak QE was measured to be 67% at 350 K, with a 300 K value of 52%-both measured with an applied bias voltage of À0.5 V. Below 200 K there was no resonant response attributed to the cutoff wavelength of the InAs absorber falling below the resonant wavelength. [20] For comparison, the InAsSb-based device achieves a peak QE of 66% at 225 K and À1.9 V. The QE decreases slowly with temperature to a 300 K value of 57% at the same bias voltage. Both devices can achieve high room temperature quantum efficiencies, but only the InAsSb-based device can maintain these values at low temperatures. The significant decrease in QE at lower temperatures seen by the InAs-based device at this resonant wavelength is intrinsic to the absorber material. For shorter resonant wavelengths, the resonant response would be further from the cutoff wavelength and the QE would not decrease at low temperatures.
Leakage currents for the InAs-based device were measured between 100 K and 375 K at 25 K increments, as shown in figure 4b. From these measurements, an Arrhenius plot was derived at an applied bias voltage of À0.3 V, as shown in figure 4c. Fitting to this plot reveals an activation energy of %300 meV, slightly lower than the bandgap of InAs of %354 meV. This activation energy indicates that the ShockleyÀReedÀHall (SRH) mechanism does not dominate and Auger current is the most significant. A conventional InAs nBn detector with a 2 μm-thick absorber by Pedrazzani et al. [20] is shown in the same figure, demonstrating a more than 50-fold decrease in leakage current that the resonant cavity structure achieves at room temperature. Also presented are measurements for the InAsSb-based device, which show an activation energy of 330 meV-approximately the bandgap of the absorber. It is likely that the InAs-based material system could also demonstrate an activation energy similar to the bandgap of the absorber with further refinement.
The responsivity and leakage current measurements were used to calculate the voltage and temperature dependence of the specific detectivity at the resonant wavelength, as shown in Figure 4d. The calculated values take into account both Johnson and shot noise. For the InAs-based device, the highest values of 8 Â 10 10 cm ffiffiffiffiffiffi ffi Hz p =W were achieved at both 200 and 225 K and an applied bias voltage of À0.3 V. At higher temperatures, the specific detectivity decreases monotonically, whereas a room temperature (300 K) value of 2.5 Â 10 10 cm ffiffiffiffiffiffi ffi Hz p =W was achieved with the same applied bias voltage. Select measurements for the InAsSb-based device are shown for comparison, exhibiting a slightly lower 300K value of % 1.5 Â 10 10 cm ffiffiffiffiffiffi ffi Hz p =W. However, at lower temperatures, the InAsSb-based device demonstrates significantly higher performance; at 200 K, a value of % 5 Â 11 10 cm ffiffiffiffiffiffi ffi Hz p =W was measured. A similar comparison is also made to a commercial InAs photodiode available from Teledyne Judson. [21] The InAs-based RCE-PD outperforms the commercial detector at room temperature but does not match the highest performance achieved at 188 K. www.advancedsciencenews.com www.pss-a.com

Conclusion
A high-quality RCE-PD was grown on an InAs substrate with the purpose of spectral sensing of methane and other compounds with C─H bonds. The resonance wavelength of %3.3 μm lines up well with the targeted absorption peak and would allow for measurement of methane concentrations with high spectral specificity. A maximum quantum efficiency of 67% was measured at 350 K and a maximum specific detectivity of 8 Â 10 10 cm ffiffiffiffiffiffi ffi Hz p =W was measured at 200 K. At 300 K, a quantum efficiency of 52% and a specific detectivity of 2.5 Â 10 10 cm ffiffiffiffiffiffi ffi Hz p =W demonstrate the high sensitivity. The combination of high specificity and high sensitivity shows the ideal properties for spectral sensing that the RCE-PD design possesses.
Comparisons with a previously grown RCE-PD with an InAsSb absorber showed that both material systems would be suited to spectral sensing in the absorption region of the C─H bond. Comparable quality factors indicate that the more complex DBRs in the InAs-based device do not compromise device performance. The InAs-based device had favorable high-temperature performance, with a spectral response that increased with temperature up to 350 K. Significantly, the high performance of this device could be utilized without cooling.
A similar FWHM for both devices indicates that both have high spectral specificity. The InAsSb-based device exhibited the highest sensitivity of %5 Â 11 10 cm ffiffiffiffiffiffi ffi Hz p =W, at 200 K, but would require thermoelectric cooling to achieve it.

Experimental Section
Device Fabrication: Both device structures were grown by molecular beam epitaxy (VEECO GENxplor) on InAs and GaSb substrates, respectively. The substrates were first degassed at 350 C before being placed in the growth chamber. Immediately before growth, surface oxides were removed by heating the substrates above the growth temperaturedependent on the substrate. An %1 μm-thick buffer layer of the same material as the substrate was grown prior to the structure growth. SUMO cells provided the group III fluxes, whereas valved cracker cells provided the fluxes for As 2 and Sb 2 . All layers were grown at %1 ML s À1 . The layers in the cavity were n-type doped with GaTe, whereas the DBRs were not intentionally doped. The exact compositions and thicknesses of the layers were calibrated by test samples prior to the growth of the final devices.
Postgrowth, XRD analysis was used to determine lattice matching for all ternary alloys-measurements were carried out on a Bruker D8 Discover X-ray diffractometer. The top DBR was etched to allow titanium/gold contacts to be placed on the top layer of the cavity. The top cavity layer was then etched to define the device mesas. Standard lithography and etchants were used throughout. Leakage currents for the InAs-based device, À25 K increments. c) Arrhenius plot of the leakage currents for the InAs-based device, applied bias voltage of À0.3 V (squares), the InAsSb-based device, applied bias voltage of À1.9 V (circles), and an InAs-based nBn (triangles). [20] Activation energies of %300 meV (solid black line) and 330 meV (solid blue line). Rule 07 with a 3.5 μm cutoff wavelength (dashed line). d) The peak specific detectivity for the InAs-based device (squares), InAsSb-based device (circles), and a commercial InAs photodiode (triangles). [21] www.advancedsciencenews.com www.pss-a.com FTIR Characterization: The optical properties of the devices were studied by a Bruker Vertex 70 FTIR spectrometer. The spectral response was measured with the device held in place inside a cryostat outside of the FTIR, with the beam from the FTIR directed onto the device. The responsivity and spectral response were calculated with reference to a commercial HgCdTe detector (Vigo PVMI-2TE-12-1 Â 1), with known responsivity. The photosensitive area of the device was considered to be the area of the top DBR.
CurrentÀVoltage Measurements: The currentÀvoltage characteristics of the devices were measured inside a cryogenic probe station (Lakeshore TTPX). The devices were surrounded by an integrated cold shield, to reduce thermal radiation onto the device. The current was measured as the voltage was swept from zero in the positive and then negative direction consecutively.