1. Introduction

The most challenging issues to address in nitride based devices are poor p-type region conductivity and difficulties with low resistance ohmic p-type contacts processing. Recently, increased attention has been dedicated to the interband tunnel junctions (TJs) [1] for the efficient conductivity conversion from p-type to n-type in III-nitride devices [2–6]. Application of TJs eliminates the need for p-type contact deposition and creates more freedom in device design [4,7–9]. It was clearly demonstrated that TJs resistance for wide bandgap semiconductors can be effectively reduced by making use of the piezoelectric fields in the region of the junction [2,6]. However, for metal organic vapor-phase epitaxy (MOVPE) it is difficult to activate the p-type conductivity in the (In)GaN:Mg layers that are buried below n-type layers due to the fact that diffusion of hydrogen is completely blocked through n-type region [5,10–12]. The devices with TJ grown by MOVPE suffer from a substantial increase in operating voltage [13]. To reduce the voltage activation of magnesium through side walls of the device has been proposed [14]. However, such an approach is inefficient in devices with large dimensions limiting the freedom of design. Recently, new approaches of in situ activation of p-type to form buried TJ by MOVPE have been demonstrated with success [15]. On the other hand, for plasma-assisted molecular beam epitaxy (PAMBE), the p-type doping is achieved without the need of any activation process. Therefore PAMBE overcomes the limitations of MOVPE for practical realization of the vertical devices with buried p-type layers [16].

High power pulsed LDs are attractive for many applications like gas sensing, printing, environment pollution control. Recently one of the widely growing field is the light detection and ranging (LIDAR) in cartography, automotive and industrial systems [17]. The LIDAR systems require high power light (10-100 Watts) and very short light pulses – for the safety reasons. The coupling of the light coming from stack of LDs with external optics is much easier than from arrays of LDs since the spatial separation between devices can be smaller roughly two orders of magnitude (<1 µm vs. >100µm). The simultaneous operation of cascade of n LDs increases slope efficiency (SE) of the full device n-times, which makes high power lasing conditions accessible for smaller currents. In addition, the level of catastrophic optical damage (COD) is n times higher in comparison with a single LD. Therefore stack of LDs operating in pulse mode interconnected by TJs can be a viable compact alternative for high power laser sources.

There is one report on a stack of two III-N LDs grown by MOVPE which shows very weak evidence of simultaneous laser action from both active regions [18]. In particular no extra peak in lasing spectrum nor improvement in SE was observed for that device. This is probably due to difficulties with activation of Mg acceptors in buried p-type layers discussed above.

In this work we demonstrate a stack of two III-N LDs grown by PAMBE which simultaneously lases at two distinct wavelengths and shows a substantial increase in SE.

2. Experimental

2.1 PAMBE growth

The epitaxial structure of the LD stack presented in this work was grown entirely by PAMBE on bulk (0001) GaN. The substrate was a commercially available Ammono-GaN crystal with threading dislocation density 10⁵ cm⁻² [19]. The structure consists of two LDs interconnected by a TJ, as shown schematically in Fig. 1. At the top of the structure another TJ is placed which makes it ready for growth of a subsequent LD. The TJs are placed far away from the waveguides of the LDs to avoid generation of additional optical losses. The calculated optical mode overlap with the TJ is extremely low in the order of 10⁻⁷. Assuming even a high absorption loss of α = 4000cm⁻¹ the resulting optical loss is 0.01 cm⁻¹ which is negligible.

 

Fig. 1 Schematic image of the processing design of stack of two LDs grown by PAMBE.

Download Full Size | PDF

2.2 LD structure characterization

Details of epitaxial structure are presented in Fig. 2(a) where the sequence of layers is marked on the scanning transmission electron microscopy (STEM) image obtained for the studied LD stack. The active regions of both LDs as well as the tunnel junctions is schematically depicted in Figs. 2(b)-2(d). The bottom LD (LD1) contains Al_0.05GaN claddings, 220 nm In_0.04Ga_0.96N waveguide, and 25 nm In_0.17Ga_0.83N single quantum well (SQW) [20]. The top LD (LD2) is an Al-cladding free to reduce the tensile strain in the structure [21–23]. It has GaN claddings, 120 nm In_0.08Ga_0.92N waveguide and 25 nm In_0.18Ga_0.82N SQW. We intentionally designed a slightly different In content in the SQWs of both LDs to be able to verify their lasing via observation of two peaks in the lasing spectra. The used growth conditions were the same as for single PAMBE LDs and are described elsewhere [24].

 

Fig. 2 (a) STEM image of the stack of two LDs grown by PAMBE with the layer sequence. Details of (b) top LD2 active region lasing at 459 nm, (c) tunnel junctions used to interconnect the LDs and grown on top of the LD stack and (d) bottom LD1 active region lasing at 456 nm.

Download Full Size | PDF

The TJ region consists of 10 nm In_0.17GaN QW and 60 nm In_0.02GaN:Mg and 20 nm In_0.02GaN:Si barriers as presented in Fig. 2(c). The barriers are doped with Mg and Si at the levels of 5x10¹⁹ cm⁻³ and 4x10¹⁹ cm⁻³ respectively. First 5 nm of the QW is heavily doped with Mg at a level of 1x10²⁰ cm⁻³ while the following 5 nm of the QW is n-type doped at a level of 1.8x10²⁰ cm⁻³. The Mg and Si doping profiles in TJs were optimized to achieve atomically flat surface without defects which is essential for the growth of the subsequent devices on top of the stack.

3. Results

3.1 TJ optimization

Figure 3(a) presents the band diagram of the stack of two LDs interconnected by TJ. The electrons are injected into LD1 and recombine in the active region with holes injected from the opposite side. The holes are generated by tunneling of the electrons out of the valence band of LD1 into the conduction band of LD2 taking place in the TJ. These electrons are then injected into the active region of LD2, recombine with holes and so on making the whole structure acting as a cascade. We found that the heavy doping of the QW in the TJ is essential to obtain low voltage drop across TJ. SiLENSe 5.4 package was used to calculate the band diagrams of two TJs consisting of: (i) an undoped InGaN QW taking advantage of built-in piezoelectric fields proposed by Krishnamoorthy et al. [6] and (ii) a heavily doped InGaN QW shown in Fig. 2(c). The doping in the QW surrounding was kept the same. Figure 3(b) presents the calculated band diagrams. As one can see the doping of the QW reduces the depletion width of junction in comparison to the undoped QW which should reduce the turn-on voltage and differential resistance. The simple reason for better performance of the doped InGaN QW is that the electric field from the p-n doping adds up to the field generated by the piezoelectric. Every doped TJ, in which the electric field arising from the p-n junction is in the same direction as the piezoelectric field, should therefore perform better. To test this prediction we have grown two test LED structures with TJs on top. The structure of the LEDs were similar to the LD2 designs without the thick GaN claddings. The TJ consisted of an undoped InGaN QW and a doped InGaN QW as it the theoretical calculations. The measured IV characteristics are presented in Fig. 3(c). The LED with the heavily doped TJ has a much lower turn-on voltage and series resistivity than the undoped counterpart. The experimental data is consistent with the theoretical findings presented in Fig. 3(b).

 

Fig. 3 (a) Band diagram of the stack of two III-nitride laser diodes interconnected with a tunnel junction, (b) band structures of undoped and doped tunnel junctions, (c) I-V characteristics of LED test structures with undoped and doped InGaN QW tunnel junctions.

Download Full Size | PDF

The atomic force microscopy (AFM) image presenting the surface morphology of the whole stack of two LDs is shown in Fig. 4. Atomic step edges are visible. The surface roughness obtained for 5x5 μm² area in the middle of the wafer is 0.25 nm. Smooth growth front has been maintained during the whole epitaxial process as confirmed by high quality interfaces between subsequent layers studied in TEM, cf. in Fig. 2(a). Importantly, the presence of heavily doped TJs does not introduce extended defects. The atomically flat surface after the growth of stack of two LDs enables the epitaxy of the next LDs on the top.

 

Fig. 4 AFM image showing the surface morphology of the stack of two LDs grown by PAMBE. Atomic step edges are visible.

Download Full Size | PDF

3.2 Demonstration of LD stack

The 15 μm x1000 μm LDs mesa was formed by reactive-ion etching with a depth of 1.9 μm through both active regions of the LD stack. The LDs had cleaved uncoated mirrors. Upper part of mesa was covered partially by dielectric to better isolate LDs walls from metal coating. This is feasible because the structure ends with the highly conductive n-type layer which provides a uniform current spreading over the whole LD stripe area.

The LDs were operated with 200ns long pulses and a repetition rate of 1kHz. The light-current (L-I) characteristics of the cascade of two LDs is shown in Fig. 5. Two lasing thresholds had been observed. The first one at a current density of 2.8 kA/cm² with a SE of 0.7 W/A. The second one occurred at 4.4 kA/cm² and the observed slope efficiency increased up to 1.4 W/A. As it will be shown later the LD2 is the first to start lasing. The higher threshold current density of LD1 is attributed to a different design with a lower optical confinement factor (Γ) [23]. The calculated values of Γ are 2.4% and 3.1% for LD1 and LD2, respectively. We had previously reported a lower threshold current density of LD with higher indium content in the waveguide [25]. The doubling of the SE indicates that the same electrons (holes) are used two times to generate light in both LDs. Obtained SE exceeds the theoretical limit for a single laser diode which is an undeniable proof that we observe lasing from two LDs. The maximum value of SE for a wavelength of 460 nm is equal to 2.7 W/A assuming internal losses and injection efficiency equal to 0 cm⁻¹ and 100%, respectively. Our devices are prepared without facet coatings therefore the light is emitted half through the front and half through the back facet leading a SE of 1.35 W/A for one facet. Furthermore, to verify the observation of lasing from both LDs the emission spectra were collected at 3.7 kA/cm² and 5.3 kA/cm² and are presented in Figs. 6(a) and 6(b), respectively. Inserts to Figs. 6(a) and 6(b) show near-field patterns collected at these current densities using a Gaussian telescope setup [26]. Strong filamentation is observed as expected for a wide-ridge LDs [27]. At j = 3.7 kA/cm² there is only one peak in the spectrum and a single near-field pattern visible. Above the second threshold a second peak in the spectrum and a second near-field pattern appear. Observation of one peak in the spectra before the doubling of SE and two peaks after the doubling is a clear indication that both LDs operate simultaneously for current densities above 4.4 kA/cm². The maximum optical power obtained for the studied structure was 2.2 W per laser facet and can be further increased by the use of dielectric coatings. Application of this design for n- LDs interconnected by (n-1) TJ allows to increase SE n-times.

 

Fig. 5 Light-Current characteristics of the stack of two LDs structure grown by PAMBE. Two lasing thresholds are observed. The slope efficiency is doubled after the second LD starts to lase.

Download Full Size | PDF

 

Fig. 6 Lasing spectra of the stack of two LDs obtained for (a) 3.7 kA/cm2 and (b) 5.3 kA/cm2. Inserts show the collected near-field patterns.

Download Full Size | PDF

The voltage drop across device is presented in Fig. 7. We compare DC measurements of LDs up to 4 kA/cm² current density for different PAMBE LDs. The voltage across device for 4 kA/cm² for single LD devices reaches 4.8V and 5.6V for standard (with metal p-type contact) and TJ LD, respectively [16]. For the same current density the stack of the two LDs reaches to 14.5 V with the turn-on voltage of 7.3V (3.5V turn-on voltage for TJ LDs). This indicates much higher series resistance of TJs used in for stack of two LDs. We found that heavily Si doping of n-type part of InGaN QWs in TJ decreases the turn on voltage and series resistance of TJ. However, Si doping levels above 2x10²⁰ cm⁻³ causes dislocation formation, which is not acceptable for LDs stack. Therefore, for stack of LDs presented in this work we have chosen Si doping level at 1.8x10²⁰ cm⁻³. This resulted in good crystal morphology required for growth of subsequent LD, as can be seen from AFM and TEM images (Figs. 2 and 4). However, the TJ series resistance and turn-on voltage increased. For single TJ LD, the Si doping level was much higher (5x10²⁰ cm⁻³). Further work on the increase of the tunneling probability and preserving high structural quality will be essential for practical demonstration of stack of LDs in continuous wave mode. Very promising path will be the replacement of donor dopant in the junction from Si to Ge what should allow for higher n-type doping without deterioration of crystal quality [28].

 

Fig. 7 Voltage-current density diagram for PAMBE LDs.

Download Full Size | PDF

4. Conclusions

In conclusion, we demonstrated a stack of two III-nitride laser diodes operating at different wavelengths grown by plasma-assisted molecular beam epitaxy. The first LD starts to operate at 2.8 kA/cm² with SE equal to 0.7W/A and the wavelength of 459 nm. The lasing threshold of the second LD is at the current density of 4.4 kA/cm² and the emission wavelength is 456 nm. Doubling of the slope efficiency was observed above 4.4 kA/cm² indicating a simultaneous lasing of two LDs. This construction paves a way to achieving III-nitride high power pulse laser diode stacks for LIDAR applications.