InGaP-based quantum well solar cells: Growth, structural design, and photovoltaic properties

III-V compound semiconductors have been used in many optoelectronic devices because the flexibility of their bandgap structure results in material systems with interesting electrical and optical properties.¹ The ability to tune the band gap while maintaining lattice matching by growing III-V alloys has promoted their use in the last four decades in solar cells, light-emitting diodes, lasers, photodiodes, optical modulators, etc. Specifically, the efficient tuning of semiconductors' bandgaps has allowed for more flexibility in the design of III-V multijunction solar cells (MJSCs),² which are the most efficient photovoltaic system. The maximum efficiency of MJSCs is limited by the number of junctions (subcells) in the structure and the current matching requirement in series connected subcells. The efficiency of MJSC has been improved dramatically in the past few years with four junction cells reaching 46.0% using wafer-bonded structures^3,4 and 45.7% using an inverted metamorphic structure.⁵ In order to achieve efficiencies higher than 50% which is crucial for the cost effectiveness of concentrator and space systems,⁴ the structures with five and six junctions will have to be used.^6–10 However, not all the desired (optimal) bandgaps required for the next generation five and six junction solar cells are available with the current growth technologies, while maintaining the lattice-matched condition to Ge or GaAs.⁴ For example, semi-empirical modeling of five and six MJSCs indicates that subcells of band gaps of 1.6–1.8 eV should be used. The available material system lattice matched to GaAs(Ge) that can used to obtain these desired bandgaps is Al(In)GaAs, which may suffer poor diffusion length and long-term reliability problems associated with the reactivity of aluminum with oxygen unless it is grown at temperatures higher than 750 °C.¹¹ Such high temperatures will compromise the tunnel junction performance. InGaAsP quaternary alloys in the bandgap range (1.4–1.9 eV) suffer from a miscibility gap which prevents growth of high quality bulk material when grown lattice-matched to GaAs.^12,13 Therefore to reach higher efficiencies, it is desirable to obtain an alternate aluminum-free structures which provide an optimal band gap configuration for next-generation MJSC design.

One approach for band-gap tuning while maintaining lattice-matching is strain balanced multiple quantum wells (SBMQWs).¹⁴ The SBMQWs offer wide range of achievable bandgaps; however, careful design is important where the quantum confined Stark effect,¹⁵ the quantum size effect (QSE),¹⁶ and the strain effects¹⁷ play an important role in determining the absorption energy of QWs. One example of SBMQWs is the InGaAs/GaAsP structure grown on GaAs substrate to tune the bandgap of GaAs (1.42 eV) to ∼1.2 eV.^18–20 Alternating layers of compressive InGaAs and tensile GaAsP are grown with an in plane lattice constant equal to that of GaAs substrate.²¹ InGaAs/GaAsP superlattices have been used in many other optoelectronic applications, such as light-emitting diodes,¹⁴ lasers,²² modulators,²³ and photodiodes,²⁴ and have been used as a buffer layer to reduce defects in GaAs epitaxial layers grown on Si substrates.²⁵ 

We report on the growth conditions and the fabrication details of new SBMQW solar cell structures used to tune the bandgap of In_0.49Ga_0.51P (1.87 eV) solar cells, as shown in Fig. 1. This research will help develop subcells of 1.6–1.8 eV which are important for future five and six junction cells. In the device structure, the MQW structures are included in the unintentionally doped intrinsic layer (i layer) of InGaP p-i-n structures. Two SBMQW structures are fabricated in this work and grown on GaAs substrates. In both of these structures, the compositions and the thicknesses of the layers are chosen such that the compressive stress in the wells is balanced by the tensile stress in the barriers. The active region of the first demonstrated SBMQW device consists of alternating In_xGa_1−xAs_1−zP_z wells and In_yGa_1−yP barriers as shown in Fig. 1(b), where x > y. InGaAsP and InGaP may also be grown lattice matched²⁶ to tune the bandgap of In_0.49Ga_0.51P to lower energy values, whereas the use of the InGaAsP/InGaP SBMQWs presented in this work offers more flexibility in tuning the device into a wider range of bandgaps, while maintaining efficient carrier transport than the lattice matched structures. The second SBMQW device's active region consists of alternating In_xGa_1−xP wells and In_yGa_1−yP barriers, where x > y, Fig. 1(c). In this work, an analysis of both structures material and device characteristics is presented. We show that careful control over the well and barrier compositions and thicknesses is required to efficiently balance the stress in the two layers. The InGaP-based quantum well solar cells extend the absorption edge compared to the standard In_0.49Ga_0.51P non-MQW cell.

In this section, the growth conditions, the structural design, and the fabrication details of the proposed quantum well structures are presented.

The samples were grown in a Thomas Swan MOCVD system fitted with a custom reactor, designed, and built inhouse.²⁷ The growth pressure was fixed at 200 Torr, and purified N₂/H₂ were used as the carrier gases. The precursors used in this work are tertiarybutylarsine (TBAs), tertiarybutylphosphine (TBP), trimethylindium (TMIn), trimethylgallium (TMGa), dimethylzinc (DMZn), and disilane for arsenic, phosphorus, indium, gallium, zinc, and silicon, respectively. The samples studied were grown on (100) GaAs mis-oriented by 2° towards 〈110〉 direction. Due to the use of organometallic sources, we found that several steps are required for the growth of InGaP in our reactor that might not be included in case of using arsine and phosphine. Each sample employs a standard GaAs buffer layer grown at 640 °C followed by a 50 Å GaAsP interfacial layer with the As ramping down through 3 steps each of 4 s, before the In_0.49Ga_0.51P base growth. This interfacial layer has minimized the arsenic carryover from GaAs to InGaP and prevented the forming of compositional mixing at the GaAs/InGaP interface. The nucleation of the In_0.49Ga_0.51P base layer starts at 575 °C to avoid three-dimensional nucleation, and the temperature is subsequently ramped up in a minute to 650 °C with a growth rate of 5.0 Å/s. One minute before the end of the base growth, the temperature is ramped down to 585 °C for the SBMQW growth. Details on the choice of quantum wells' growth temperature are mentioned later. Similarly, the temperature is ramped up to 650 °C to grow the emitter after growing the SBMQWs. The base and emitter thicknesses are 1.0 and 0.1 μm with doping concentration of 1 × 10¹⁷ and 1 × 10¹⁸ cm⁻³ using silicon and zinc, respectively.

Two SBMQW structures are grown in a p-i-n solar cell device, as shown in Fig. 1.

(I) In_xGa_1−xAs_1−zP_z/In_yGa_1−yP (x > y) SBMQWs consist of alternating In_xGa_1−xAs_1−zP_z wells and In_yGa_1−yP barriers that are under compressive and tensile stress, respectively, Fig. 1(b). Several fundamental issues related to the growth of InGaAsP quaternary alloy have to be discussed first. These issues make the choice of the quantum wells' growth temperature more challenging. Two competing factors affect the choice of the growth temperature: phase separation and three-dimensional nucleation. InGaAsP quaternary alloy has an immiscible region that is more thermodynamically favored at low growth temperature for comparable As and P contents.^12,13 But for sufficiently thin InGaAsP films, the driving forces for immiscibility is reduced, and miscible films can be grown.²⁸ Spinodal decomposition occurs by migration at the growing surface, and the homogenous alloy is stabilized by the strain with respect to the substrate.²⁸ Thus, there is a limit to the thicknesses of miscible quaternary layers that can be grown, and this limit is affected by the composition growth temperature. In order to reduce the miscibility gap range, InGaAsP should be grown at high temperatures. However, it becomes more difficult to grow strained structures at high temperatures as three dimensional nucleation and surface morphology issues start to appear. In order to grow miscible InGaAsP thin films and maintain a good surface morphology, the quantum wells are grown at low temperature (585 °C). However, this temperature is relatively low to grow miscible InGaAsP film with high arsenic content (>10%), which is important to achieve the required bandgap. We compensated for that by increasing the indium content in the wells to achieve the required bandgap. The V/III ratio of bulk In_0.49Ga_0.51P, In_yGa_1−yP barrier, and In_xGa_1−xAs_1−zP_z well is 50, 60, and 150, respectively. The relatively high V/III ratio of InGaAsP wells is necessary to obtain arsenic composition of 5%. The In_xGa_1−xAs_1−zP_z/In_yGa_1−yP SBMQWs have been grown with compositions of 0.60 < x < 0.75, 0.35 < y < 0.43, and 0.9 < z < 0.98 and periods ranging from 80 Å to 230 Å. In this work, the indium (x) and phosphorus (z) compositions in the wells are set at 0.70 and 0.95, respectively. In the barrier, the indium composition (y) is set at 0.40.

(II) In_xGa_1−xP/In_yGa_1−yP (x > y) SBMQWs:

In order to avoid the phase separation issues related to the growth of InGaAsP quaternary alloy, we have investigated the growth of SBMQWs made of two InGaP ternary compounds, with different indium compositions. This structure consists of alternating In_xGa_1−xP wells and In_yGa_1−yP barriers that are under compressive and tensile stress, respectively, Fig. 1(c). In both the wells and barriers, the Ga and P flows are kept the same, while the indium flow is ramped up and down to grow the well and barrier alternatively. In_xGa_1−xP/In_yGa_1−yP SBMQWs have been successfully grown with compositions of 0.6 < x < 0.75 and 0.30 < y < 0.45 with periods ranging from 100 Å to 250 Å. The V/III ratio of In_xGa_1−xP well and In_yGa_1−yP barrier is 30 and 40. For the SBMQW reported here, the indium compositions in the wells (x) and barriers (y) are set to 0.75 and 0.40, respectively.

X-ray diffraction (XRD) was used to determine the indium and phosphorus compositions of alloy calibration samples. The XRD superlattices' satellite peaks of the MQW structures coupled with the growth rate measurements were used to ensure the strain balance condition had been achieved and to find the thicknesses of the individual layers. Solar cell devices were fabricated as 2.5 mm × 2.5 mm etched mesas using standard photolithographic procedures. Ge/Pd/Ti/Ag/Au n-type and Ti/Pd/Ag/Pd/Au p-type ohmic contacts were deposited via e-beam evaporation. InGaAsP/InGaP and InGaP/InGaP mesa devices are etched using H₂O₂:HBr:H₂O (1:10:20) and HCl:H₃PO₄:H₂O₂ (5:5:1), respectively. No antireflection coatings or windows are used in this study. Optical microscopy was performed using an Olympus BX41microscope fitted with a Nikon D3000 camera. Optical emission of MQW structures was evaluated by photoluminescence (PL) using a 532 nm frequency doubled Nd:YAG laser. The external quantum efficiencies (EQEs) of the solar cell devices were measured using a quartz tungsten halogen lamp with a Newport 1/4-m monochromator, a lock-in amplifier, and calibrated Si photodiode. The illuminated current density-voltage (J-V) characteristics were measured with an Oriel 1-kW solar simulator with an AM1.5G filter and calibrated with an InGaP reference cell with a known AM1.5G J_SC.

In this section, the material and device characteristics of InGaAsP/InGaP and InGaP/InGaP SBMQWs are presented and analyzed. First, the calibrations of the wells and barriers are presented. Then, the model used to balance the quantum wells coupled with the (004) XRD results of the two structures are discussed. Next, a comparison between the experimental and modeled PL emissions of the two structures is carried out. Then, the quantum efficiency of the two quantum well devices is compared. Finally, the effect of well number on InGaAsP/InGaP SBMQW device is discussed.

The compositions of bulk InGaP films (>0.5 μm) grown at 585 °C were determined by XRD applying Vegard's law. The indium composition in InGaP is plotted versus the ratio of the input partial pressure, [TMIn]/[TMGa], as shown in Fig. 2(a). It is noted that for [TMIn]/[TMGa] ratios less than 2.5, the indium incorporation in InGaP is linear with the input partial pressure ratio. For higher [TMIn]/[TMGa] ratios, the incorporation of indium increases slowly implying that indium evaporation or surface segregation is taking place.

The indium and arsenic contents of the quaternary alloy (InGaAsP) are estimated using XRD of ternary InGaP and GaAsP calibration curves shown in Fig. 2. We were not able to use PL to determine the bandgap of the quaternary alloy by growing bulk epitaxial films because the films of these compositions grown at 585 °C will be immiscible as mentioned earlier,¹² and we could not get a good surface morphology for thick bulk layers. Fig. 2(b) shows the arsenic composition in GaAsP versus the input partial pressure ratio, [TBP]/[TBAs] at 585 °C. For convenience, the V/III ratio is shown too on the same plot. The incorporation of arsenic is linear with the change of flow at 585 °C, as shown in Fig. 2(b).

To balance the SBMQWs, a zero-stress balance model is used where the average in-plane stress due to biaxial strain in the quantum wells/barriers should be zero²⁹ 

where $aGaAs$⁠, $aw$⁠, and $ab$ are the lattice constants of GaAs substrate, well, and barrier, respectively. $tw$ and $tb$ are the well and barrier thicknesses, as shown in Fig. 1. $ϵw$ and $ϵb$ are the strain in wells and barriers, respectively. $Aw$ and $Ab$ are the stiffness parameters for wells and barriers, respectively, which are calculated as follows:

where $C11$ and $C12$ are the elastic stiffness coefficients which are obtained from Ref. 30 for the binary compounds and interpolated linearly for the ternary and quaternary compounds. The thicknesses of wells and barriers obtained from the zero-stress model for the two quantum well structures are shown in Fig. 3. It is noted that the values obtained from this model serve as a better measure of the balancing of the two structures compared to the thicknesses obtained from the average-lattice method or the thickness-weighted method.²⁹ 

The XRD was used to ensure the stress balance condition on GaAs substrates was met. The (004) XRD scans of the two SBMQWs are shown in Fig. 4. The diffraction pattern consists of the (004) GaAs substrate reflection peak and several weaker satellite peaks due to the periodicity of the QWs. The number shown above each peak is the order of the satellite. The presence of clear peaks indicates that the MQW regions of well-defined periodicity were formed, and the presence of higher order satellite peaks indicates abrupt interfaces. The n = 0 peak corresponds to the MQW average lattice constant perpendicular to the surface and differs from that of the GaAs substrate due the difference in the elastic properties of the two alloys.²⁹ Periodicity calculations based on the satellite peaks and growth rate measurements for the In_0.70Ga_0.30As_0.05P_0.95/In_0.40Ga_0.60P SBMQWs indicate that the wells are around 40 Å and the barrier around 120 Å. The corresponding well and barrier thicknesses for In_0.75Ga_0.25P/In_0.40Ga_0.60P structure are 65 and 170 Å, respectively. These values are close to the initial estimate from the zero-stress balance model and indicate that we have good control over the growth conditions.

The critical layer thickness (CLT) of the In_0.70Ga_0.30As_0.05P_0.95 wells grown at the previously mentioned conditions is 90 Å, by our estimate using the Matthews-Blakeslee model.³¹ The lattice constant of InGaAsP used in the model is based on the linear interpolation of that of binary compounds. Nomarski differential interference microscopy images of the surfaces of the InGaAsP/InGaP SBMQWs are shown in Fig. 5. A mirror like surface for InGaAsP/InGaP SBMQW of well thickness of 45 Å (<CLT) is shown in Fig. 5(a), indicating that the MQW is closely lattice matched to the GaAs substrate. If the quantum wells are not strain-balanced or if the individual layers' thickness is above the CLT, crosshatching will appear as shown in Fig. 5(b) for a well thickness of 100 Å (>CLT). A similar behavior was observed for InGaP/InGaP structures (not shown).

As we mentioned earlier, the motive of this work is to tune the bandgap of In_0.49Ga_0.51P (1.87 eV) based structures to lower energy values (1.65–1.8 eV), while maintaining lattice matching conditions to GaAs substrates. As an illustration of the bandgap tunability in these QW structures, the photoluminescence characteristics of the two QWs are depicted in Fig. 6. Modeling of the QWs emission is shown in the same figure too. The model is based on estimating the conduction and valence bands offset using the Anderson's rule³² based on the electron affinities values given for the binary compounds³⁰ and interpolated linearly to calculate the corresponding values for ternary and quaternary compounds. Strain effects³⁰ and quantum confined Stark effects¹⁵ are included in the model. The Kronig-Penney model is used to calculate the first electron and hole states.³³ The thicknesses of wells and barriers used in the model are inferred by total period thickness measured by XRD and growth rate measurements. As shown in Fig. 6, the QWs emission is tuned by varying the thickness of the well from ∼30 Å to 90 Å for the two QW devices. The thicknesses of the barriers in the two structures are adjusted accordingly to balance the structure. The peak PL wavelength emission increases as the thickness of the well increases. This can be attributed to a reduced QSE associated with thicker wells, leading to decrease in emission energy since the compositions are fixed in this study. The peak PL emission of In_0.70Ga_0.30As_0.05P_0.95/In_0.40Ga_0.60P SBMQWs can be tuned from 1.85 to 1.65 eV. The average of the full width half maximum (FWHM) of these structure is ∼25 meV and increases to ∼40 meV for the thickest well device which cross the CLT. The wide tunable range of InGaAsP/InGaP promotes their use as efficient aluminum-free subcells for five and six junction's cell. The alternative material system for that wavelength-range is Al(In)GaAs, which may suffer from growth and long-term reliability problems related to oxygen contamination.

On the other hand, the peak PL emission shift of In_0.75Ga_0.25P/In_0.40Ga_0.60P SBMQW saturates at ∼1.82 eV for the thickest well devices, as shown in Fig. 6. It is noteworthy that the modeled values deviate greatly from the experimental values for the InGaP/InGaP MQWs, as shown in Fig. 6. For example, the relaxed bandgap of In_0.75Ga_0.25P well is ∼1.63 eV, while the modeled strained bandgap is around 1.72 eV for a well thickness of 65 Å, which is lower than the achieved experimental value (1.82 eV). The reason behind this discrepancy between the theoretical value and the experimental result is not clear. However, a possible issue might be indium segregation (surface accumulation) taking place during the growth of the In_0.75Ga_0.25P well. Recent work has shown indium segregation initiates at the start of lattice matched In_0.49Ga_0.51P growth and saturates when the grown layer thickness is ∼150 Å.³⁴ We are using well thicknesses of less than 90 Å to maintain efficient carrier transport and to make sure the well thickness is less than the CLT. Accordingly, it is likely that this surface accumulation smears out the profile of incorporated indium in the well. Thus, the profile is not a rectangular shape, as usually assumed for calculating subband states.³⁵ Another possible issue is the composition pulling effect taking place at the well growth resulting in effectively reduced indium composition than expected from bulk calibration. It is also possible that the graded quantum well interfaces are formed. We have previously shown by modeling that the formation of graded interfaces may lead to thinner wells than expected and increases the bandgap of InGaAs/GaAsP quantum wells.³⁶ These aforementioned effects will modify the electronic states of the quantum wells and lead to a blue shift in the transition energies compared to the modeled quantum well with abrupt interfaces, as shown in Fig. 6.³⁷ It is worth pointing out that the limited minimum achievable effective bandgap for the InGaP/InGaP may limit its use an efficient subcell for next generation photovoltaic devices.

The study of minority carrier transport across the MQWs is important for the understanding and optimization of the performance of these devices. The transport properties of QW devices can be dominated by tunneling, for thin barriers, or thermionic emission. Fig. 3 shows that in order to balance the two QW structures presented here, the average thickness of the barriers is around 2–3 times thicker than that the well. For example, the barrier thicknesses for In_0.70Ga_0.30As_0.05P_0.95/In_0.40Ga_0.60P and In_0.75Ga_0.25P/In_0.40Ga_0.60P are 135 Å and 110 Å, respectively, for a well thickness of 40 Å. This implies that carrier tunneling across barriers cannot take place in this structure even with very thin wells where the QSE effects are dominant. The thermionic emission will thus be the dominant transport mechanism if the carriers at the ground state are able to surmount the effective barrier height and are transported across the depletion region. We estimate the barrier height for electrons and holes at n = 1 state to be 0.1672 and 0.2152 eV for In_0.75Ga_0.25P/In_0.40Ga_0.60P SBMQW for 65 Å well. The corresponding values for In_0.70Ga_0.30As_0.05P_0.95/In_0.40Ga_0.60P are 0.1497 eV and 0.2089 for 45 Å well.

Fig. 7 shows the quantum efficiency of the two SBMQW structures in comparison to standard InGaP p-i-n structures lattice matched to GaAs. The i layer of InGaAsP/InGaP SBMQWs is made of 30 period quantum wells of 45 Å In_0.70Ga_0.30As_0.05P_0.95 wells and 140 Å In_0.40Ga_0.60P barriers. The i layer of InGaP/InGaP SBMQWs is made of 30 period quantum wells of 65 Å In_0.75Ga_0.25P wells and 170 Å In_0.40Ga_0.60P barriers. The i layer of standard device is made of undoped In_0.49Ga_0.51P and with thickness same as that of the QW region in both the two structures. We first note that the EQE of standard device falls rapidly at around 663 nm (1.87 eV), which is the band-edge of In_0.49Ga_0.51P. In addition, the quantum efficiency for light in the bulk absorption range of both the two QW structures is close to the standard device, this implies that the thermionic carrier transport is efficient. Due to absorption in i-SBMQW region, the EQE of the QW devices extends beyond the In_0.49Ga_0.51P band-edge approaching 682 and 750 nm for InGaP/InGaP and InGaAsP/InGaP MQWs, respectively. The absorption edge obtained from the quantum efficiency measurements is consistent with that obtained from the PL emission in Fig. 6 for both QW structures. It is noteworthy that in InGaP/InGaP the achieved red-shift is low relative to InGaAsP/InGaP, which may limit its use in five and six MJSCs. However, the red-shift achieved using the latter makes this structure a promising candidate for next generation photovoltaic subcells.

Fig. 8 shows the illuminated J-V characteristics of the InGaAsP/InGaP SBMQWs in addition to an InGaP standard device. The open circuit voltage and the short circuit current density of the InGaAsP/InGaP SBMQW device are 1.11 eV and 8.6 mA/cm², respectively. The corresponding values for the standard device are 1.23 eV and 7.425 mA/cm². The improved short circuit current of SBMQW over the standard device is due to extension in the absorption threshold, as indicated by the EQE of Fig. 7(a). It is worth noting that the InGaAsP/InGaP SBMQW device exhibits a relative efficiency increase of about 4% over the standard device.

Since the minimum achievable effective bandgap of InGaP/InGaP MQWs is around 1.82 eV using the current growth conditions, we will focus more on the analysis of InGaAsP/InGaP SBMQWs. We experimentally investigate the effect of increasing the number of quantum wells on the performance of In_0.70Ga_0.30As_0.05P_0.95/In_0.40Ga_0.60P SBMQWs. In this study, only the number of periods is altered: 10 (SBMQW10), 20 (SBMQW20), 30 (SBMQW30), and 45 (SBMQW45). The well and barrier thicknesses are kept at 45 Å and 140 Å, respectively, across all these devices. These four SBMQW samples have the same PL emission, 750 nm (1.65 eV). Data about these samples are summarized in Table I. Fig. 9 depicts the XRD scans taken across (004) reflection for samples SBMQW10, SBMQW20, SBMQW30, and SBMQW45. The vertical dotted lines on the superlattices peaks indicate that the period thickness and compositions of the layers in the InGaAsP/InGaP SBMQW active regions were identical for all four samples. The strong satellite peaks on the high number of quantum well devices (SBMQW30 and SBMQW45) signify that no relaxation has been observed with higher number of periods.

The EQE of the four samples is depicted in Fig. 10. As the period number increases from 10 to 30, the excitonic quantum efficiency of the structure is improved. This can be attributed to greater light absorption taking place due to inclusion of more wells in the intrinsic layer of the device. The enhanced EQE at wavelengths below the bulk GaInP band edge is due to drift assisted carrier collection. As the number of periods increases further (30 to 45), the EQE is reduced. This is accompanied by degradation in the quantum efficiency of the device across the bulk part of the spectrum, especially the blue region. This can be attributed to the thickness of the grown MQW region (0.83 μm) is approaching or exceeding the depletion region of the device. The background doping in the MQW region is 4 × 10¹⁵ cm⁻³ as measured using Hall measurements, which corresponds to about 0.8 μm depleted region. This will result in the formation of an undepleted region with a few QWs, which will inhibit the transport of minority carriers through this region due to the absence of a significant electric field.

To better understand the earlier assumption, the current-voltage is measured under one sun illumination. The short circuit current density (J_sc), the open circuit voltage (V_oc), and the fill factor (FF) are summarized in Table I. The effect of number of period on J_sc is shown in Fig. 11. It is noted that the J_sc increases with the increase of period number for period range of 10 to 30, correlating with the EQE shown in Fig. 10. For higher number of periods, the J_sc degrades as might be expected if current from the emitter and base is inhibited. This suggests that the optimum well number for this MQW structure and growth conditions is about 30 for the unintentional background doping present in the intrinsic region.

This research deals with the bandgap engineering of III-V semiconductor materials in order to tune the bandgap of InGaP-based structures, while maintaining the lattice matching conditions to GaAs substrates. Two strain balanced multiple quantum well structures are proposed. In both the two structures, In_yGa_1−yP is used as the barrier material. The well is made of In_xGa_1−xAs_1−zP_z (x > y) for the first quantum well structure and of In_xGa_1−xP (x > y) for the second device. We were able to grow thin InGaAsP films in the composition range where bulk films of the same compositions suffer from the lack of miscibility. We use the zero-stress balance model to estimate the well and barrier thicknesses. The emission and absorption wavelengths are controlled by varying the well thickness resulting in a tunable-QW structure. The InGaP/InGaP SBMQW structure exhibits a red shift of only 20 nm beyond the In_0.49Ga_0.51P band-edge in comparison to ∼80 nm achieved using the InGaAsP/InGaP structure. The effect of the number of periods on the InGaAsP/InGaP SBMQWs is studied, which shows that the unintentional background doping in the MQW region limits the number of periods that can be grown with efficient light absorption and carrier collection.

This work was funded by the National Science Foundation (NSF) GOALI program under Contract No. 1102060, and by the Department of Energy (DOE) under Contract No. DE-EE0005403.