Heterojunction phototransistors (HPTs) have garnered considerable attention for their applications in optical receivers within lightwave communication systems [1,2] and as integral components for optoelectronic mixers/modulators in microwave photonic systems [3]. Notably, HPTs have found utility in the realm of infrared camera pixels for low-light imaging systems [4,5]. The inherent advantages of HPTs, such as high optical gain without impact ionization, intrinsic non-linear optical performance, and seamless on-chip integration with electronic circuits, underscore their appeal in diverse applications. However, as collector current levels escalate, there is an evident decline in the optical gain (G_opt) of HPTs when the base pushout occurs in the base–collector (B-C) junction. Particularly, HPTs with a B-C homojunction undergo a swift degradation of G_opt at elevated collector current densities. To address this challenge, type- II heterostructures have been implemented at the interface between the base and collector layers, as evidenced in heterojunction bipolar transistors (HBTs) [6], demonstrating the high current capability of InP/GaAsSb/InP double heterojunction bipolar transistors (DHBTs) [7]. Nevertheless, InP’s transparency to 1.55 µm wavelength precludes its use as the collector layer in HPTs designed for near-infrared radiation.

To overcome these challenges, we have devised DHPTs employing type-II heterojunctions of GaAsSb/InP and GaAsSb/InGaAs. The substantial valence band discontinuity at the InP/GaAsSb emitter–base (E-B) and GaAsSb/InGaAs B-C junctions imparts a robust hole blocking property, facilitating elevated power output characteristics at minimal bias voltage. The valence band offset inherent in the type-II B-C heterostructure delays the onset of base push-out at increased collector current levels. Although Memis et al. [8] reported a highly sensitive photon detector featuring an InP/GaAsSb nano-injector and an InGaAs absorption layer, capable of detecting optical power below 10 nW, experimental studies on the InP/GaAsSb/InGaAs type-II heterostructure remain relatively scarce despite its superior performance.

To optimize the optical performance of the DHPTs, we implemented an emitter-ledge passivation process in the device fabrication. The InP emitter-ledge structure, transparent to 1.55 µm lightwave, effectively mitigates base surface recombination current [9]. This study presents a demonstration of the high-current characteristics of InP/GaAsSb/InGa-As DHPTs operating at low bias voltages. Comparative analyses between DHPTs with and without the ledge structure are conducted to scrutinize the impact of the emitter-ledge structure. The observed maximum optical power level at which G_opt diminishes was identified, and a comparative assessment was conducted between the current gain (β) of DHBTs and the G_opt of DHPTs.

The epitaxial layers constituting the DHPTs were meticulously grown on an n⁺-InP substrate through the precision of a molecular beam epitaxy system. Table I presents the schematic of the device structure. Notably, an InP emitter serves as the ledge structure for surface passivation. The design incorporates a 1 µm-thick InGaAs collector layer strategically engineered to attain an external quantum efficiency, reaching 50 % at a wavelength of 1.55 µm when the HPTs operate in p-i-n mode. Schematic depictions of the DHPTs-with the ledge layer (WL) and without the ledge layer (WOL) are presented in Figs. 1(a) and 1(b), respectively. A SiN_x thin film, employed for anti-reflection coating in both devices, additionally functions as a passivation layer for the exposed base of the DHPTs-WOL. Fabrication of DHPTs with a 50 µm-diameter optical window was accomplished through standard optical contact lithography and a selective wet etching process. Nonalloyed Ti-Pt-Au (20/30/200 nm) metallization was evaporated and lifted off for the emitter contact on the top surface and for the collector contact on the backside of the substrate. An InGaAs cap layer and an InP emitter layer were etched with solutions of H₃PO₄:H₂O₂:H₂O and H₃PO₄:HCl, respectively. To ensure passivation of the surface in DHPTs-WL, the InP layer was deliberately preserved on the extrinsic base. A 200 nm-thick SiN_x film, deposited by plasma-enhanced chemical vapor deposition at 300 °C, played a pivotal role in this configuration. Mesa isolation etching was executed from the SiN_x film down to the InGaAs collector layer, following the protection of the optical window using a photoresist. In a comparative assessment of the G_opt of the DHPTs with the β of DHBTs, DHBTs (emitter area, AE: 80 × 80 µm²) were fabricated utilizing the same device structure. Non-alloyed Pd/Ir/Au metallization was employed for the base electrode, and the sheet resistance of the base approximated 7 kΩ/□. The optical performance evaluation of the fabricated DHPTs was conducted using a 1.55 µm laser diode and an HP 4155A semiconductor parameter analyzer at approximately 295 K. Laser power calibration was achieved by measuring the incident optical power through a lensed single-mode optical fiber, utilizing a commercial InGaAs-based p-i-n photodiode module within a dark box.

Table 1 . Layer structure of type-II InP/GaAsSb/InGaAs DHPTs..

Layer	Material	Dopant	Doping (cm−3)	Thickness (nm)
Emitter cap	n+ InGaAs	Si	2 × 1019	100
Emitter	n+ InP	Si	3 × 1019	50
Emitter	n− InP	Si	3 × 1017	70
Base	p+ GaAsSb	C	1 × 1019	20
Collector	n− InGaAs	Si	3 × 1016	1,000
Subcollector	n+ InP	Si	1 × 1019	300
n+ InP Substrate

Figure 1. Schematic cross-sectional representation of DHPTs (a) with and (b) without a ledge emitter.

Figure 2(a) presents the collector dark current (I_Cdark) of DHPTs-WL and DHPTs-WOL with a 50 µm diameter. At a V_CE of 0.5 V, DHPTs-WL exhibited an I_Cdark of 4.8 nA, whereas DHPTs-WOL recorded 133.9 nA, indicating a remarkable 30-fold reduction in I_Cdark by leveraging the emitter ledge structure. The injection of electrons directly from the emitter sidewall onto the exposed base surface of DHPTs- WOL results in an elevated surface recombination current [10]. Moreover, the interface states induced between InGaAs and SiN_x during the plasma deposition step potentially contribute to the augmented surface recombination current in DHPTs-WOL. Conversely, the InP ledge structure in DHPTs-WL obstructs direct electron injection from the emitter sidewall into the extrinsic base surface and prevents plasmainduced damage during the anti-reflection coating process.

Figure 2. (a) Collector dark current (I_Cdark) and optical output (I_Cph) characteristics at different illumination powers of 7, 14, 22, 27, and 35 µW. (b) Optical gain and collector photocurrent at V_CE = 0.5 V with respect to the incident optical power for DHPTs-WL and DHPTs-WOL.

Figure 2(a) further illustrates the collector photocurrent (I_Cph) as a function of V_CE at optical power levels ranging from 7 to 35 µW at a wavelength of 1.55 µm. I_Cph increased proportionally with incident optical power as photo-generated holes in the collector migrated into the base, reducing the barrier energy of electrons at the E-B junction and facilitating electron injection from the emitter to the collector. The optical gain, defined as G_opt = hvΔI_c/qP_in, where hv is the energy of an incident photon, and ΔI_c is the difference between I_Cph and I_Cdark [1], is depicted in Fig. 2(b) alongside I_Cph for DHPTs-WL and DHPTs-WOL across various optical power levels up to 690 µW. DHPTs-WL exhibited higher G_opt and ICph than DHPTs-WOL. At V_CE = 0.5 V, the measured I_Cph for DHPTs-WL (DHPTs-WOL) was 36.3 (4.5) and 61.6 (24.2) mA under P_in of 140 and 690 µW, respectively, resulting in G_opt values of 207.8 (25.6) and 71.6 (28.1) corresponding to the respective optical input power. During the conduction process under illumination, within the DHPTs-WOL, a greater number of electrons undergo recombination owing to increased leakage. These electrons cease to contribute to the electron current, resulting in a subsequent decrease in G_opt. A comparative analysis with previously reported lateral InP/InGaAs DHPTs featuring a 50 × 50 µm² detection area (L-DHPTs) [11] reveals that the I_Cph of DHPTs-WL is comparable to that of L-DHPTs, falling within a comparable range of several milli-amperes at the same incident optical power level. Notably, our devices operate at 0.5 V of V_CE, while L-DHPTs operate at 2 V of V_CE. The large valence band offsets in the type-II band alignment at the E-B and B-C junctions allow DHPTs-WL to confine holes effectively, inducing higher collector photocurrent at a lower bias voltage. The maximum G_opt of DHPTs-WL is 207.8 at 140 µW of P_in, with no degradation from the Kirk effect observed at this optical power level. Conversely, L-DHPTs showed a maximum G_opt of 163 at a P_in of 44 µW, above which G_opt started to decrease. We attribute this to the substantial valence band discontinuity at the GaAsSb/InGaAs junction of DHPTs-WL, impeding accumulated holes from entering the collector under high optical power levels, ultimately resulting in the observed fall-off in optical gain. The decline in optical gain ultimately occurs beyond an optical power threshold of 140 µW. This reduction can be attributed to the formation of an electron barrier at the base side of the B-C heterojunction. This phenomenon arises owing to the accumulation of holes in that specific region without extending into the collector depletion region [6,12]. Consequently, the heightened base recombination in this configuration leads to a consequential decrease in optical gain.

Figure 3 plots the β of the DHBT against the optical gain, G_opt, of DHPTs-WL as a function of collector current density (J_C). G_opt values were measured under an incident optical power ranging from 1.3 nW to 690 µW at V_CE = 0.5 V. The β of the DHBT is 598.2 at a J_C of 10³ A/cm². Notably, G_opt remains lower than β within the investigated range of J_C. This disparity is attributed to the non-ideal external quantum efficiency (η_ex) of the devices, stemming from the thin InGaAs absorption layer and imperfections in the anti-reflection coating. The relationship between G_opt and β is expressed as β = (G_opt/η_ex)−1 [13]. The computed average η_ex of DHPTs-WL is 41 %. The quantum efficiency can be approximated by η_ex = η_i(1 − R)(1 − e^−ad), where R is the optical reflectivity between air and the SiN_x film, a is the absorption coefficient of the intrinsic region, d is the thickness of the absorption layer, and η_i is the internal quantum efficiency. Considering the structure of DHPTs-WL, η_ex can be increased to 49 %. Future optimizations in surface passivation and anti-reflection coating hold promise for further enhancing η_ex.

Figure 3. Comparison of the current gain of DHBT and optical gain of DHPTs-WL.

In summary, our study successfully demonstrated a DHPT employing a type-II heterostructure composed of InP/GaAsSb and GaAs-Sb/InGaAs. At low bias voltages, the fabricated DHPTs operate effectively, exhibiting a robust collector current condition attributed to the commendable confinement of holes within the base layer. Particularly, DHPTs-WL effectively averted the degradation of optical gain until P_in reached 140 µW. The inclusion of the InP emitter-ledge layer contributes to a substantial reduction in I_Cdark by suppressing the base surface recombination current. Our devices exhibit an impressive G_opt of 207.8, translating to a responsivity of 259.3 A/W, recorded at 1.55 µm wavelength, P_in = 140 µW, and V_CE = 0.5 V. This achievement underscores the efficacy of the DHPTs in harnessing the advantages of type-II heterostructures for enhanced device performance under lowbias conditions.

This research was supported by the National Research Foundation of Korea (NRF) grant funded by the Korean government (MSIT) (NO.2020R1C1C1004971).

The authors declare no conflicts of interest.