1. Introduction

The GaNAsSb material system has attracted great interest for potential photodetector applications in the near infrared region (0.9-1.6µm) [1]. By keeping the ratio of nitrogen (N) content to antimony (Sb) content at 1.0 to 2.6, the GaNAsSb material can be tailored to lattice-match GaAs at wavelengths from 0.9µm to 1.3µm. Compared with the incumbent InP-based technology, GaNAsSb offers significant advantage due to the use of lower cost GaAs substrate and availability of larger GaAs substrate.

Achieving high photoresponsivity is one of the key challenges in GaAs-based dilute nitride photodetector research. So far, reported results of GaAs-based dilute nitride photodetectors can be categorized into two groups: one based on quantum well (QW) absorption layers [2-4] and the other one based upon bulk absorption layers [5-10]. For devices based on a quantum well absorption layer, a thin dilute nitride layer (<10nm) is used. While such QW devices enable the utilization of highly strained dilute nitride layers and thus offer a photo-response up to 1.6µm, their photoresponsivity is generally low [2, 3] (typically less than 0.03A/W) due to the thin QW photoabsorption layer. To overcome this limitation [4], a resonant cavity has been incorporated into the device structure. On the other hand, photodetectors based on bulk dilute nitride absorption layers (>0.4µm thick) suffer from reduced photo-response at long wavelengths. So far, the highest reported cut-off wavelength is ~1.4 µm [10]. This is due to the difficulty in incorporating more than 3.5% of nitrogen into the material. Nevertheless, photodetectors based on bulk dilute nitride absorption layers exhibit a higher photoresponsivity compared to QW-based devices. Recently, photoresponsivities of up to about 0.1A/W have been reported for bulk GaNAsSb/GaAs devices [5, 6, 11]. These photoresponsivity values are still much lower as compared to those of commercial InGaAs photodetectors, with a typical photoresponsivity of up to ~0.9A/W at 1.3µm.

In this paper, we report on a significant improvement in the photoresponsivity of GaNAsSb/GaAs photodetectors with a GaNAsSb bulk photoabsorption layer at 1.3µm wavelength. The devices exhibit characteristics which strongly suggest the presence of photogenerated carrier multiplication due to the avalanche effect.

2. Technology

The device structure shown in Fig. 1 was grown using a molecular beam epitaxy (MBE) system in conjunction with a radio frequency (RF) N plasma-assisted source and a valved Sb cracker source. The i-GaNAsSb (bulk) photoabsorption layer was 0.5µm-thick for i-GaNAsSb layer grown at 350°C and 400°C, and 2µm-thick for the i-GaNAsSb layer grown at 440°C and 480°C. The RF nitrogen plasma power was 180W and the beam equivalent pressure (BEP) of the Sb flux was ~1×10^-7 torr. Under these conditions ~3.3% of N and 8% of Sb were incorporated into the i-GaNAsSb layer, which was confirmed by x-ray diffraction (XRD). Using the band anti-crossing (BAC) model [12], the optical bandgap of the i-GaNAsSb layer was estimated to be ~0.9eV. The doping concentrations of the p-type (C-doped) and n-type (Si-doped) GaAs contact layers were approximately 2×10¹⁹cm^-3 and 5×10¹⁸cm^-3, respectively, and the growth temperature of these layers was 600°C. The use of carbon as p-type dopant minimizes the out-diffusion, compared to beryllium

The devices have a diameter of 80 µm. The devices were fabricated using standard photolithography and wet etch process. After defining the mesa patterns by photolithography, an acid-based solution, NH₄OH: H₂O: H₂O₂ (5: 250: 2) was used to etch away GaAs and GaNAsSb layers, which were not protected by the photoresist, at room temperature. The etch depth was measured using a surface profiler. The Ohmic p- and n-contacts were formed by Ti (50nm)/Au (200nm) and Ni (5nm)/Ge (25nm)/Au (100nm)/Ni (20nm)/Au (100nm), respectively. In addition, the n-contact was annealed at 380°C for 60s. These contacts were connected to metal banding pads using air bridge metallization technology. There is no passivation for the devices.

 

Fig. 1. Schematic diagram of GaAs/GaNAsSb/GaAs p-i-n photodetector structure.

Download Full Size | PDF

3. Results and discussion

The photoresponsivity measurement was carried out using a quartz tungsten halogen lamp as the light source, in conjunction with a monochromator. Furthermore, the light source was calibrated using a commercial InGaAs photodetector to measure the power of light arriving at surface of the devices. The photocurrent was measured using combination of a low noise pre-amplifier and a lock-in amplifier. Figure 2(a) shows the plot of photoresponsivity at a reverse bias of 3V vs. wavelength for the devices whose i-GaNAsSb layers were grown at 350°C to 480°C. The photodetectors show a photo-response up to wavelength of 1350nm. Figure 2(b) shows the photoresponsivity at different reverse biases measured at the wavelength of 1300nm.

From Fig. 2(b), it is interesting to note that the photodetector with GaNAsSb layer grown at 350°C shows an extremely high photoresponsivity value of ~12A/W under 4.8V reverse bias at 1300nm. This is more than 2 orders higher than previously reported results. Assuming a unity quantum efficiency, a photodetector at 1300nm exhibits a maximum responsivity of ~0.75A/W, taking into account 29% incident power reflection due to refractive index difference at the air/GaAs interface. Thus, a photoresponsivity value of 12A/W implies a quantum efficiency value significantly larger than 1, possibly due to the presence of an avalanche carrier multiplication effect. From Figs. 2(a) and 2(b), it can also be seen that the photoresponsivity of the devices increases as the growth temperature of the i-GaNAsSb photoabsorption layer decreases, except for the device with the i-GaNAsSb layer grown at 480°C. As can be seen from Fig. 2(b), the photoresponsivity of the device grown at 480°C is the lowest at reverse biases below 1V, confirming that the responsivity generally decreases with increasing growth temperature. As the reverse bias is increased, the responsivity of the 480°C rises much stronger as compared to the other devices. This behavior will be further explained below.

 

Fig. 2. (a) Plot of photoresponsivity vs. wavelength measured at a reverse bias of 3V for devices with i-GaNAsSb layer grown at 350°C, 400°C, 440°C and 480°C. (b) Plot of photoresponsivity vs. reverse bias measured at 1300nm for devices with i-GaNAsSb layer grown at 350°C, 400°C, 440°C and 480°C.

Download Full Size | PDF

The photodetectors with GaNAsSb layer grown at different temperatures have different depletion widths under the same reverse bias due to different unintentional doping concentration in the i-GaNAsSb layer. From capacitance-voltage (C-V) measurement, the unintentional doping concentrations in the i-GaNAsSb layer grown at 350°C, 400°C, 440°C and 480°C were experimentally determined to be 2×10¹⁶cm^-3, 6×10¹⁶cm^-3, 3×10¹⁷cm^-3 and 1.5×10¹⁸cm^-3, respectively. These unintentional doping is p-type and is induced by defects states, especially nitrogen related defects. Based on these unintentional doping concentrations, the depletion region width in the i-GaNAsSb layer at different reverse biases can be calculated. From our previous report [13], the absorption coefficient α was measured using a spectroscopic ellipsometer and has a value of 1.3×10⁴cm^-1 at the wavelength of 1300nm. Using the measured photoresponsivity values, calculated depletion region widths and values of α, the photocurrent multiplication factor M for all devices at different reverse biases are calculated and shown in Fig. 3. From Fig. 3, it can be seen that the photodetector with i-GaNAsSb layer grown at 350°C has a M value of ~30 under 4.5V of reverse bias. This high value of M confirms our earlier suggestion of the presence of a photogenerated carrier multiplication due to the avalanche effect. As the carrier avalanche effect is directly dependent on the electric field strength at the depletion region, the values of M in Fig. 3 are re-plotted against the average electric field strength at the depletion region and shown in Fig. 4.

As mentioned earlier, the photodetector whose i-GaNAsSb layer was grown at 480°C showed a different characteristic in that its photoresponsivity rises much stronger with increasing reverse bias as compared to the other devices. At high reverse voltages (>1.5V) it thus exhibits a higher responsivity compared to the device with i-GaNAsSb layer grown at 440°C. This can be explained by the high electric field in the depletion region of this device as shown in Fig. 4. Due to the high unintentional doping concentration of 1.5×10¹⁸cm^-3 in the i-GaNAsSb layer grown at 480°C, the depletion region is comparably thin thus resulting in a high electric field strength of about 200-400KV/cm in the depletion region. This is in contrast to the other devices which exhibit an average electric field strength of <200kV/cm in their depletion regions.

 

Fig. 3. Plot of multiplication factor M as function of reverse bias voltage measured at 1300nm for devices with i-GaNAsSb layer grown at 350°C, 400°C, 440°C and 480°C.

Download Full Size | PDF

It is interesting to note that as the growth temperature of the i-GaNAsSb layer deceases from 440°C to 350°C, the devices showed a higher value of M, even at much lower electric fields. The photodetector with i-GaNAsSb layer grown at 350°C exhibits a high carrier multiplication factor at average electric field strengths of <100kV/cm. Even when considering a non-uniformly distributed electric field in the depletion region, the maximum electric field strength is ~100kV/cm and 180kV/cm at reverse bias of 1V and 5V, respectively. This electric field strength is unexpectedly low, considering the fact that GaAs or InGaAs based avalanche photodetectors only show carrier multiplication at electric field strength higher than ~200kV/cm [14, 15]. These results suggest that the decrease in growth temperature of the i-GaNAsSb layer leads to a higher impact ionization coefficient in the material, resulting in initiation of the carrier avalanche process at low electric field.

The high ionization coefficient and initiation of the carrier avalanche process at low electric field in photodetector with low temperature grown i-GaNAsSb layer could be explained by the existence of mid-gap As antisite defects (As_Ga) in the material. It is known that dilute-nitride materials contain As_Ga defects [7, 16] as they are grown at non-equilibrium low temperature (<500°C) growth conditions. We expect that the i–GaNAsSb layer grown at 350°C has the highest amount of As_Ga defects as content of these defects increases proportionally in response to the decrease in the growth temperature of dilute-nitride material [16].

Generally, carriers in a p-n junction require energy of $\frac{3}{2}{E}_g$ to start an impact ionization and thus avalanche process [17]. E_g is the bandgap of the material. Mid-gap defects, such as As_Ga are reported [18] to enhance the impact ionization process by lowering the energy required in the impact ionization process. Instead of energy of $\frac{3}{2}{E}_g$ , the impact ionization process through the mid-gap defects states requires only energy of $\frac{E_g}{2}$ [18]. By lowering the required energy, the existence of mid-gap defects enables a more efficient impact ionization and carrier multiplication process at a lower electric field. This explains our observation that photodetectors, which have i-GaNAsSb layer with more As_Ga defects, have higher carrier multiplication and initialize the impact ionization process at a lower electric field.

 

Figure 4 Plot of multiplication factor M as function of average electric field measured at 1300nm for devices with i-GaNAsSb layer grown at 350°C, 400°C, 440°C and 480°C.

Download Full Size | PDF

The detectivity, D* of the devices at 1300nm were estimated by ${\mathrm{D}}^{*}={\mathrm{\Re }}_{\lambda }\sqrt{\frac{R_{o}A}{\mathit{KT}}}$ , where ℜ_λ is the responsivity of devices at zero bias at 1300nm, R_o is the dark impedance at zero bias and A is the detector area. To calculate the R_o, the dark current-voltage (I-V) data was fitted using I=a(e^bv−1)+c(e^dv−1) [19]. By taking the derivative (dV/dI) of the fitted curve equation at zero bias, the value of R_o can be obtained. The value of R_o for device with i-GaNAsSb layer grown at 350°C, 400°C, 440°C and 480°C are 3×10⁶, 5×10⁶, 5.5×10⁶ and 9×10⁵Ω, respectively. The low value of R_o could be due to the defect states in the GaNAsSb layer. Using the fitted value of R_o, the detectivity at 1.3µm for device with i-GaNAsSb layer grown at 350°C, 400°C, 440°C and 480°C are estimated to be 2.6×10⁹, 8.5×10⁹, 1.7×10⁹ and 9.5×10⁷ cm√Hz/W, respectively.

4. Conclusion

In conclusion, this paper reports on the RF nitrogen plasma-assisted MBE growth of four P-i-N photodetectors whose the i-GaNAsSb layer were grown at 350°C, 400°C, 440°C, and 480°C. The device with the i-GaNAsSb photoabsorption layer grown at 350°C exhibited extremely high photoresponsivity of 12A/W corresponding to a photogenerated carrier multiplication factor of 30. This high photoresponsivity and multiplication factor in this photodetector is considered to be due to the high impact ionization coefficient in the i-GaNAsSb layer grown at low substrate temperature.