Engineering carrier lifetimes in type-II In(Ga)Sb/InAs mid-IR emitters

The growing applications interest in the mid-infrared (mid-IR) wavelength range (∼3–30 μm) can in many ways be attributed to the dramatic improvement in semiconductor-based light sources in this technologically vital frequency band. This new class of sources offers the potential for new compact sensing systems, taking advantage of the mid-IR's position as the spectral location of the distinct vibrational and rotational absorption signatures for a wide range of molecular species. At the same time, such emitters could be used to mimic the mid-IR thermal radiation emitted by all biological and mechanical objects, for a range of thermal scene projection, tagging, signaling, or countermeasure applications.

The remarkably rapid development of the quantum cascade laser (QCL)¹ offers a potential technology for many of the above applications. The QCL is a unipolar laser emitting light via designed intersubband (ISB) transitions for electrons traveling through complex semiconductor heterostructures. These lasers offer increasingly high powers and efficiencies, as well as impressive wavelength flexibility, by quantum engineering of the semiconductor heterostructure. The QCL emits across much of the mid-IR, with continuous wave (CW), room temperature (RT) operation for the majority of wavelengths essential for mid-IR sensing systems.² However, QCLs have historically struggled to reach shorter wavelengths (3–4.5 μm) using the InGaAs/AlInAs lattice-matched to InP material system. Approaches such as strain-balanced QCLs on InP substrates^3–6 have extended the wavelength range of QCLs, though at the expense of additional design and growth complexity. Alternatively, new materials systems for QCLs, for instance, InGaAs/AlAsSb,⁷ InAs/AlSb,⁸ or II/VI material systems, such as the ZnCdMgSe alloys,⁹ can extend QCL emission to shorter wavelengths, but these less technologically mature material systems lag behind the more traditional (In/Ga/Al)As materials in both growth and processing.

The ISB transitions responsible for the QCL's wavelength flexibility are also the source of the QCL's greatest weaknesses. First, the ISB dipole matrix element ensures that only TM-polarized light is emitted from the QCL active region, preventing surface emission without the additional fabrication of outcoupling structures. Second, and perhaps most importantly, the QCL suffers from extremely fast nonradiative recombination lifetimes, a result primarily of phonon scattering between electronic states in the QCL active region.^10,11 These ∼ps timescale nonradiative lifetimes ensure large threshold current densities, limiting QCL wall-plug efficiency,¹² and result in weak subthreshold emission. While QCLs are ideal sources for applications requiring a compact, coherent, mid-IR light source, they are rather less than ideal as light emitting diodes (LEDs), due to both the surface emission and subthreshold efficiency limitations discussed above. Thus, the QCL may not necessarily be the ideal choice for all mid-IR applications, in particular, those which might require narrow bandwidth (though not monochromatic), incoherent light, in a surface emitting configuration.

Many of these challenges associated with the QCL are overcome in the interband cascade laser (ICL), where light is generated by interband, and not ISB, transitions, in materials with type-II band-alignments.^13–15 The electron–hole recombination of the ICL results in excited state lifetimes much longer than those of the QCL,¹⁶ as well as the potential for surface emission from interband cascade active regions. The former positions interband cascade emitters as increasingly attractive sources for applications requiring high wall-plug efficiency and low operating powers, particularly in the 3–6 μm wavelength range,^17,18 while the latter opens the door to the development of interband cascade-based LEDs and/or vertical cavity surface emitting lasers.¹⁹ 

While ICL and QCL performance has continued to improve, the complex bandstructure required for efficient mid-IR light emission from these devices can make the design, modeling, and growth of cascade lasers time consuming and difficult. Because of this significant complexity, it would not be hard to imagine the potential appeal of a straightforward quantum well emitter in a pn-junction, similar to the active regions of the optoelectronic devices that form the basis of our modern telecommunication (coherent) and solid state lighting (incoherent) infrastructure, to give only two prominent examples. To reproduce such near-IR and visible devices at longer wavelengths would require narrow bandgap barrier material surrounding even narrower bandgap material to form type I quantum wells. Diode lasers have been demonstrated at the short wavelength edge of the mid-IR (1.9 μm < λ < 3.4 μm) using GaInAsSb quantum wells (QWs) in AlGa(In)AsSb matrices with RT and CW operation.²⁰ Alternatively, InAs QWs grown on a multifunctional metamorphic buffer layer have demonstrated strong emission and recently room temperature lasing,^21–23 with emission limited to less than ∼3 μm due to the fundamental limitation imposed by the InAs band gap. Relying on narrow bandgap materials imposes additional, significant limitations, most notably resulting from Auger recombination, which increases exponentially with decreasing band gap and increasing temperature, and greatly complicates the development of efficient narrow bandgap mid-IR emitters.

A compromise between the complexity of the cascade lasers and the limitations on the emission spectra and carrier lifetimes of narrow band semiconductor material is offered by the material system consisting of strained In(Ga)Sb layers deposited on InAs surfaces. In(Ga)Sb has a type-II broken band alignment with InAs, as can be seen in Fig. 1(a), ²⁴ such that thin layers of In(Ga)Sb in InAs results in hole, but not electron, quantum confinement. Controlling In(Ga)Sb layer thickness controls the energy position of the holes in the In(Ga)Sb which recombine with electrons at the InAs conduction band edge. In addition, In(Ga)Sb will have a significant lattice mismatch with InAs, as high as 6.5% for pure InSb, and decreasing with increasing Ga content. Because additional Ga in the In(Ga)Sb layer primarily changes the In(Ga)Sb conduction band offset, but not the valence band position, this material system offers an intriguing opportunity to independently engineer both strain and energy states in thin mid-IR emitting quantum structures.

A recent work has investigated the growth of thin In(Ga)Sb layers on InAs, using, as a foundation, previous efforts focused on InAsSb/InAs superlattices and quantum wells.^25–27 In particular, there has been interest in leveraging the strain in the In(Ga)Sb/InAs system, which is remarkably close to that of the InAs/GaAs material system (well-studied for its formation of self-assembled quantum dots, or SAQDs), for demonstration of both InSb/InAs SAQDs (Refs. 28–31) and submonolayer (SML) InSb quantum dots (QDs).^32,33 The ability to form type-II QDs is particularly important for the engineering of carrier lifetimes in these emitters, as the increased quantization of energy offers a bulwark against the overwhelming Auger recombination limiting narrow band gap emitters. While a number of InSb/InAs QD and SML sources, emitting near the InAs bandedge (due to weak hole confinement),^31,32,34 and thicker InSb layers emitting further into the mid-IR,³⁵ have been demonstrated, there has been little in the way of efforts to characterize and engineer carrier lifetimes in these material systems. In this work, we investigate techniques for controlling carrier lifetimes in In(Ga)Sb/InAs mid-IR emitters. We study carrier lifetimes as a function of emitter spacing, and compare emission with and without InGaAs carrier-confining heterostructure barriers. We then investigate the growth conditions required for the formation of In(Ga)Sb nanostructures on InAs surfaces. Finally, we compare In(Ga)Sb QW structures to QDs formed in the same material system. Our materials are characterized by temperature-dependent Fourier transform infrared (FTIR) photoluminescence (PL) spectroscopy, time-resolved PL (TRPL) measurements, and structurally by atomic force microscopy (AFM). We observe a strong correlation between our materials' temperature performance and carrier lifetimes, and demonstrate a significant improvement in carrier lifetimes using conduction bandstructure engineering or alternatively, by the formation of In(Ga)Sb QDs in our InAs material system.

In our efforts to engineer the optical performance of In(Ga)Sb insertion layers in InAs matrices, we focused on two primary sources of inefficiency in our emitters. First, poor confinement of holes can shift emission from our insertion layers to the InAs, especially at higher temperatures, as holes are thermally excited out of the In(Ga)Sb insertion layers, an effect observed in our previous work.³⁶ Similarly, the lack of confinement for electrons in our system will decrease the efficiency of our emitters, as electrons are able to diffuse away from the light emitting insertion layers in our device. For this reason, we first investigate techniques for better confinement of electrons to the active region of our emitters. Subsequently, we investigate the effects of higher dimensional quantization in our insertion layers. Three-dimensional confinement has been proposed as a potential mechanism for minimizing Auger recombination in narrow bandgap materials,³⁷ and we explore the growth parameter space for In(Ga)Sb quantum confining structures on InAs and characterize the optical properties of these emitters.

Our emitters were grown in a SVT molecular beam epitaxy system on semi-insulating GaAs substrates. Because of the large lattice mismatch between our InAs matrix and the GaAs substrate, we use an intermediate GaSb layer to provide a growth surface for the InAs. Following the substrate oxide desorption, 200 nm of undoped GaAs buffer layer is grown at 610 °C, after which the substrate temperature was lowered to 510 °C and the sample annealed for 10 s without arsenic overpressure, so that the growth surface becomes gallium-terminated. We then grow a thin (10 nm) GaSb layer at 510 °C using monomeric antimony from a cracking source, followed by 90 nm of GaSb grown after increasing the growth temperature to 545 °C. The growth of the GaSb layer is performed following the approach Huang et al.³⁸ to induce in-plane rather than out-of-plane dislocations at the GaAs/GaSb interface. The GaSb layer then serves as a nearly lattice-matched growth surface for our InAs, in order to reduce defects in subsequent layers. After the 100 nm GaSb growth, we grow a thick (0.5–1 μm) buffer layer of undoped InAs at 470 °C, at which point we grow our emitter structures. The samples grown in this work are shown in Fig. 1 and can be divided into two basic sets.

In the first set of samples, we investigate techniques for controlling the carrier lifetime and temperature performance of InSb insertion layers in InAs matrices. In this set of samples, we grow three layers of thin (1.75 ML) InSb insertion layers with undoped InAs spacers, and then capped with a 30 nm undoped InAs spacer, with all layers grown at 470 °C. An additional InSb layer was grown on the top surface of each sample, for AFM measurements. For our study of these InSb insertion layers, we first look at the effect of InSb layer spacing, using a series of samples with 30, 10, and then 5 nm spacing between the three InSb insertion layers. Our last sample in the first set consists of three InSb insertion layers separated by 10 nm of InAs, with an InGaAs blocking structure on either side of our three InSb layers, in an effort to confine the photoexcited electrons to the device active region.

In a previous work, we observed the formation of quantum nanostructures in our material system with the addition of Ga in the InSb insertion layers.³⁶ The use of quantization to decrease the effects of Auger recombination has been demonstrated in a wide range of mid-IR devices and structures.^39,40 For this reason, in our second set of samples, we look to understand the growth parameters required for In(Ga)Sb quantum nanostructure formation on InAs surfaces, and to study the temperature performance and carrier lifetimes of the grown samples. Our structures consist of In(Ga)Sb insertion layers in InAs, spaced by 20 nm of undoped InAs, with a 30 nm undoped InAs cap layer and a surface layer for AFM studies. In our study of the growth parameter space, we investigated the effects of In/Ga ratio and equivalent deposition thickness for our In(Ga)Sb layers, as well as the effects of growth temperature and Sb pressure. The basic device structures investigated in this work are shown in Fig. 1, as well as the band alignment of the different material systems investigated, with band positions and alignment adapted from Vurgaftman et al.²⁴ 

Our samples are first characterized by FTIR amplitude modulation step scan PL spectroscopy. Samples are mounted in a low temperature cryostat and pumped with a 75 mW 980 nm laser diode modulated at 10 kHz, with 50% duty cycle. Emitted light from our samples is collimated with a Ge lens and directed into the FTIR. The signal from the FTIR's internal MCT detector is demodulated with a lock-in amplifier and fed back into the FTIR to form the interferogram. PL spectra are collected for each sample from temperatures of 77 up to 300 K (or the highest temperature PL is observed). For each sample, we determine the integrated PL signal from the insertion layers and plot this integrated intensity as a function of temperature, which serves as a measure of the temperature performance of the emitters. In addition, we also perform TRPL measurements on all of our materials. In these measurements, the sample is again housed in a low temperature cryostat and excited with a pulsed (1 ns) 1064 nm laser with 14 μJ pulse energy and 10 kHz repetition rate. Emission is collected with a parabolic mirror and then focused onto a high speed Kolmar MCT detector. The InAs band edge emission is filtered with a 3.5 μm long pass filter before the detector. Pulse energy is controlled using neutral density filters. The detector signal is collected and averaged using a LeCroy 12-bit high speed oscilloscope.

TRPL measurements of mid-IR detector material typically utilize a single exponential fit to accurately capture the minority carrier lifetimes at low-level injection (the typical operating conditions for mid-IR detectors). For emitters, however, a more complicated fitting process must be used in order to capture the carrier dynamics at higher injection levels (commensurate with emitter operation). Here, we follow an approach similar to that used by Olsen et al.⁴¹ to model our material response, writing the instantaneous carrier lifetime as

where n_o is the equilibrium majority carrier concentration (assuming n-doping), δn is the excess carrier concentration (for optical excitation δn = δp), τ_no and τ_po are the majority and minority Shockley–Read–Hall (SRH) lifetimes, B is the bulk radiative coefficient, and C_n the Auger recombination coefficient. For each sample, at a fixed temperature, we collect multiple TRPL scans for a range of pulse energies. We then generate a modeled PL emission versus time plot with a set of A, B, and C coefficients. In the description of the modeled data, we consider the A coefficient to be the entire first term of Eq. (1) at low injection levels corresponding to δn < n_o, so that A ≈ τ_po⁻¹. We match the modeled data to our experimental data at the lowest pulse energy, including a term to take into account the time response of our detector, and then scale our modeled results using the increase in our experimental pump energy. Then using a fitting algorithm, we adjust our A, B, and C coefficients to find the best fit to our data for all of our pump energies. The extracted A and B coefficients provide insight into both the radiative and SRH recombination in our material. However, due to the limited response time of our detector, the choice of the C fitting parameter has little effect on the quality of the fit.

The normalized low temperature PL spectra from our InSb insertion layers is shown in Fig. 2(a). The samples with 10 and 30 nm spacing all show very similar PL spectra, with spectral emission peaks within Δλ = 0.05 μm of λ = 4.47 μm, with similar emission bandwidth. However, the sample with the 5 nm spacing shows a clear blueshift in emission wavelength, emitting with a spectral peak near λ = 4.23 μm (taking into account the spectral distortion from the λ = 4.25 μm atmospheric absorption peak). The blueshift of the emission from the 5 nm spacing sample suggests a possible quantization of energy states in the InAs conduction band, provided by the InSb insertion layers. This explanation for the change in the PL spectra of the 5 nm spacing sample is consistent with the observed temperature performance of the four InSb insertion layer samples, shown in Fig. 2(b). Here, we see improved temperature performance with decreasing spacer thickness, again suggesting a potential quenching of nonradiative recombination due to increased quantization in the conduction band. In particular, we note that the InSb insertion layers with the 5 nm InAs spacers provide improved temperature performance despite their blueshifted emission, suggesting that emission blueshift is not a result of increased quantization in the valence band (which typically leads to poorer confinement and weaker temperature performance³⁶). Additionally, we note that the InSb layers with the InGaAs blocking barriers show the most improved temperature performance of all the samples in this set, again, pointing toward improved confinement of electrons in the InAs conduction band as key to the improved temperature performance of our emitters. All told, we observe an almost order of magnitude improvement in the PL quenching as a function of temperature through the engineering of the conduction band structure.

Our InSb insertion layer samples are also characterized by TRPL, and the results shown in Fig. 3 for the four sample structures investigated, with the extracted A and B coefficients for each sample noted in the TRPL plots. As can be clearly seen from this data, the 5 nm spacer sample shows the smallest A (corresponding to the least SRH recombination) but also the largest B, corresponding to the strongest radiative recombination. On the other hand, the samples with 30 nm spacing and 10 nm spacing + InGaAs barriers, both show poor performance in the TRPL data, with a particularly large A, which we attribute to the additional strain, and potential defects, introduced by the InGaAs layers. The poor performance of the sample with the InGaAs barriers is readily observable from the raw TRPL data, showing no emission tail whatsoever and a weak TRPL signal even at its peak, again attesting to the trade-off offered by the InGaAs barriers (improved electron confinement at the cost of degraded emission efficiency).

Previous work by our group has demonstrated improved temperature performance from InGaSb insertion layers when compared to those with pure InSb.³⁶ AFM data indicated that the InGaSb insertion layers, when deposited upon InAs, form nanoscale quantum structures, which would increase confinement of holes, and potentially improve nonradiative (Auger) losses in our emitters. However, the growth window for the formation of InGaSb quantum dots is remarkably narrow. In the following, we describe our parameter space investigation for InGaSb QD formation on InAs substrates, and correlate the optical properties of our InGaSb insertion layers to the surface topography of our layers. In the following investigation, we used the same growth structure for all samples, consisting of the same thin GaSb layer and thick InAs buffer as our InSb samples, followed by three layers of InGaSb, separated by 20 nm of undoped InAs, followed by a 30 nm undoped InAs cap layer and a surface layer of InGaSb (with the same growth parameters as the buried layers) for surface studies. A 15 s anneal in Sb overpressure is performed after the growth of each InGaSb layer. While we investigated a range of growth parameters, we made an effort to focus on those which might have a significant effect on the adatom surface mobility of our InGaSb layers. The lack of QD formation in our InSb layers we believed to be a result of the high mobility of In on Sb-terminated surfaces, which would favor the formation of smooth, highly strained layers of InSb, but not QD formation, due to the lack of In nucleation sites on the Sb surface. Thus, we performed multiple parameter space investigations (growth temperature, In/Ga ratio, and Sb overpressure) focused on decreasing adatom mobility on our growth surface.

Our first growth study looked to understand the effect of the In/Ga ratio in our InGaSb layers. While increasing Ga content decreases the average adatom mobility for our surface layer (due to the larger binding energy between Ga and Sb-terminated surfaces),⁴² increasing Ga content will also decrease the lattice mismatch between the InAs growth surface and our insertion layer, which is a primary driver for QD formation. Three samples were grown at 470 °C, with Sb pressure of $2.5 × 10 − 7$ Torr, each using 1.75 ML of In_1-xGa_xSb with x = 0 (InSb), x = 0.2, x = 0.4, and x = 0.5. The AFMs of the surface layers are shown in Fig. 4, from which we observe that for the x = 0 (InSb) sample, a smooth surface is obtained, corresponding to a planar, strained InSb deposition, resulting in QW formation for the buried layers. Interestingly, at x = 0.2, our surface AFM shows the formation of nanoholes, as opposed to a planar surface [Fig. 4(b)]. However, for x = 0.4, we begin to see the formation of large, interconnected features on the sample surface. At our highest Ga content, x = 0.5, we see the formation of more isolated, smaller-sized structures, suggesting that the addition of Ga to our InSb layers indeed affects the surface mobility of the adatoms of our insertion layers and assists in the formation of nanostructures. It should be mentioned here that the continued addition of more Ga will not necessarily result in better QD formation, as increased Ga fraction will decrease strain in our insertion layers, which will ultimately limit QD formation.

The low temperature PL and temperature-dependent PL from the In/Ga ratio study are shown in Figs. 4(e) and 4(f), respectively. Interestingly, all of the Ga-containing layers show very similar PL spectra, while the pure InSb insertion layer shows a red-shifted emission. Because the addition of Ga to the InSb insertion layer should only, to first order, change the conduction band of the insertion layer, the observed PL shift would appear to result from the change in the structure of the deposited material, and not the active regions' band structure. We see improved temperature performance with increasing Ga concentration in our InGaSb layers, with the exception of the x = 0.5 sample, which shows a slight decrease in the integrated PL at high temperatures, compared to the x = 0.4 sample. Taking into consideration the AFM and PL data, we narrowed our parameter space to focus on InGaSb insertion layers with x = 0.4.

Following the In/Ga ratio study, we investigated the effects of the effective deposition thickness of our InGaSb layers. Three samples were grown for this study, keeping all other growth parameters constant (In_1-xGa_xSb with x = 0.4, growth temperature 470 °C, and Sb pressure $2.5 × 10 − 7$ Torr). The traditional Stranski–Krastanov (SK) InAs QDs grown on GaAs begin to form after depositing ∼1.7 ML layer of InAs, and given the similarities in lattice mismatch between our InAs/InSb material system and the GaAs/InAs system, we grew samples with InGaSb thickness less than (1.25 ML), approximately equal to (1.75 ML), and greater than (2.25 ML), this critical thickness. Figure 5 shows the AFMs from each of these three samples. The sample with 1.25 ML of InGaSb shows no evidence of QD formation, the 1.75 ML deposition shows a nonuniform surface of interconnected features, but not distinct, isolated nanostructures, while the 2.25 ML sample shows clear evidence for nanostructure formation, though not the lenslike QDs generally associated with SK QD formation. The low-temperature PL shows the expected red-shift in emission energy with increasing thickness of InGaSb.

Finally, six samples were grown with a range of substrate temperatures (380, 400, 430, 450, 470, and 490 °C) in order to study the effects of the growth temperature on QD formation. For this set of samples, we held our In_1-xGa_xSb ratio at x = 0.4, our Sb flux at $2.5 × 10 − 7$ Torr, and our equivalent InGaSb deposition thickness to be 2.25 ML, parameters extracted from the growth studies described above. Figures 6(a)–6(c) show AFM images from representative samples grown with substrate temperatures 380, 450, and 490 °C, respectively. At low growth temperatures (380–430 °C), we observe the formation of large clusters on our sample surfaces [Fig. 6(a)], with average dimensions for diameter D ∼ 150 nm and height h ∼ 15 nm. At intermediate temperatures (450, 470 °C), we see nanostructures more similar in dimension to traditional QD structures [Fig. 6(b), with D ∼ 30 nm and h ∼ 1.5 nm], though they do not possess the smooth, lens-shaped quality traditionally associated with pseudomorphic QD formation. At higher temperatures (490 °C), the sample surface is smooth, and no evidence for QD formation is observed.

Figure 6(d) shows the low temperature PL from all of the samples in the substrate temperature study. No insertion layer emission is observed from the samples grown at substrate temperatures below 450 °C, though InAs band edge emission is seen from these samples, with band edge PL intensity increasing with growth temperature (presumably reflective of the improving quality of the InAs material). Insertion layer emission is observed for samples grown with substrate temperatures above 430 °C, with peak PL intensities again increasing with increasing temperature. Thus, at low growth temperatures, only large structures form on the surface and only weak band-edge emission is observed from the InAs. At higher temperatures, strong PL is observed from the insertion layers, but the lack of nanostructure formation suggests that this emission results from planar InGaSb layers (QWs) and not QDs. The combined results of our PL and AFM studies suggest a narrow growth window between 430 and 470 °C which can sustain nanostructure formation. The combined results of our growth studies point to the importance of controlling adatom mobility on our InAs surfaces. Insertion layers grown at lower temperatures and with increased Ga content show a greater propensity for forming 3D structures on the growth surface, while those with less Ga content, and at higher temperatures appear to form planar, QW-like structures. The weak binding energy of In atoms to Sb-terminated surfaces, an effect well-described in the literature⁴² is the most likely culprit preventing nucleation of In(Ga)Sb nanostructures.

For group V-limited growth, Sb-As exchange⁴³ at the growth surface will result in a surface poorly suited to the formation of 3D structures. The effect of Sb was further investigated by limiting the Sb-overpressure during the In_0.6Ga_0.4Sb insertion layer growth, in an effort to move toward group-III limited growth of our insertion layers. To do so, we decreased the Sb overpressure by 40% to $1.5 × 10 − 7$ Torr, holding our growth temperature at 470 °C, with x = 0.4 and t = 2.25 ML. AFMs of the two samples with identical growth conditions, with the exception of the Sb pressure, are shown in Figs. 7(a) and 7(b). Here, we observe a striking difference between our surfaces. The sample with the decreased Sb pressure shows a transition from the irregularly shaped nanostructures observed in our previous growths to a clear quantum dotlike lens shape. These structures show a bimodal size distribution, with one set of QDs forming with D = 30 ± 3 nm and h ∼ 1.5 nm and the other having D = 10 ± 5 nm and h ∼ 1.5 nm. PL from the QD sample shows a strong insertion layer PL with extremely weak quenching as the temperature is increased. The temperature dependent PL of the QD sample, as well as our x = 0.0, t = 1.75 ML; x = 0.4, t = 1.75 ML; and x = 0.4, t = 2.25 ML samples all grown at T_Sub = 470 °C (the latter three grown with P_Sb =  $2.5 × 10 − 7$ Torr) are presented in Fig. 7(d). A clear improvement in temperature performance is observed for all of the samples with nanostructure formation. Moreover, the QD sample shows the weakest PL quenching, decreasing by only a factor of 5 as the temperature is increased from 77 to 275 K, an improvement of almost 2 orders of magnitude when compared to the planar InSb insertion layers. Figure 7(e) also shows fitted low temperature TRPL data for the same samples as shown in Fig. 7(d) (with the exception of the 1.75 ML InGaSb sample). An improvement in the PL emission decay is clearly observed for the QD sample (triangles), as well as for the sample with nanostructure formation (circles), especially when compared to the pure InSb (QW) sample (squares).

The results from the integrated PL emission, as well as the TRPL studies are shown in Table I, which presents the cumulative data from both our InSb insertion layer band engineering study as well as the InGaSb nanostructure growth efforts. From this data, we see a number of very clear trends. While adjustments in conduction band engineering can offer a path toward improved temperature performance and carrier lifetimes, the advantages of this technique, at least when applied to the InAs/In(Ga)Sb material system, are limited. Band-structure engineering requires the addition of (strained) alloys to our emitter structures, which can have a deleterious effect on active region material quality, and adversely affect the emitter efficiency. However, samples with nanostructure formation show clear improvement in temperature performance as well as improved radiative and weaker nonradiative (SRH) recombination, suggesting the significant advantage of 3D confinement of charge carriers in narrow bandgap materials.

Though the improvements in temperature performance and carrier lifetimes are pronounced in the InGaSb nanostructure, and particularly QD, systems, the growth window for nanostructure formation in the InAs/In(Ga)Sb material system is rather narrow, and even narrower for the formation of InGaSb QDs. Our growth studies suggest that controlling adatom mobility on the InAs surface, either by reducing growth temperature, increasing Ga content, or alternatively, by reducing the Sb-overpressure, is key to facilitating QD formation in the InAs/In(Ga)Sb system.

In summary, in this work we explored two separate approaches to engineering lifetimes in InAs/In(Ga)Sb type II mid-IR light emitting systems. In the first, we demonstrated the effects of conduction band engineering in InAs/InSb type II quantum wells. In particular, we show that decreasing spacing between InSb layers and the addition of InGaAs blocking layers can improve the temperature performance of our emitters. However, the InGaAs layers appear to introduce additional defects to our material, and despite the improved temperature performance with the barriers, only weak emission from this sample was observed, results commensurate with the coefficients extracted from our TRPL data. Second, we investigated the growth parameters required for the formation of nanostructures in our In(Ga)Sb insertion layers, as well as the temperature dependence and carrier lifetimes of emission from these nanostructures. We demonstrate significant improvements in both carrier lifetime and temperature performance with the formation of nanostructured InGaSb films, and additional improvements when the more traditional, lenslike quantum dots are formed on the growth surface. The results presented here provide insight into the carrier dynamics of the InAs/In(Ga)Sb material system, and offer a potential roadmap toward the development of efficient and straightforward type-II emitters for mid-IR applications.

The authors (L.Y. and D.W.) gratefully acknowledge support from the NSF CAREER Program (Award No. ECCS-0925542). Y.Z. acknowledges support from the Illinois Postdoctoral DRIVE fellowship.