The feasibility of the active layers and DBRs discussed in previous two chapters is examined by the performance of optically pumped LW-VCSELs. This chapter reports the structures and the characteristics of LW-VCSELs for optical pumping, including the InP-lattice-matched and wafer-fused structures.
The Structure of LW-VCSEL
To further validate the viability of the DBR structures for long wavelength VCSELs, we have grown a laser structure based on the InP/InGaAlAs DBRs using the growth interruption technique. Figure 5.20 shows the laser structure. The epitaxial layers were grown on an n-type InP substrate. The first step was to grow the 5/4λ thick InGaAlAs and InP calibration layers followed by the 35 pairs InGaAlAs/InP DBRs. The interruption time was 0.2 minute. The laser reflectometry monitored the epitaxial growth. The 5/4λ thick InGaAlAs and InP calibration layers were first grown to check the growth rate and growth conditions before the entire DBRs were grown. The 2λ thick periodic gain cavity was grown in the second step. At the same time, another InP dummy wafer was loaded in MOCVD for photoluminescence (PL) measurement. The laser structure has InGaAlAs cladding layers of 2λ thick with a band edge emission peak at 1.1 μ m. Three sets of strain compensated multi-quantum wells (SCMQWs) were placed at the anti-nodes of the electric standing wave field within the 2λ thick cavity to increase the enhancement factor of MQW active regions. Each set of SCMQWs consisted of five InGaAlAs (strain = 1.37%, thickness = 5.5 nm) quantum wells and four InGaAlAs (strain = -0.6%, thickness = 9.3 nm) barriers and the PL wavelength was tuned at 1.51 μm, which is 40 nm blue-shifted from the target emission wavelength to insure the proper operation of the VCSELs at room temperature [24]. Two InP space layers of half λ thickness grown on the top and the bottom of the cavity were served to protect the InGaAlAs layer from being oxidized during the post processing. Finally, the wafer was coated with 10 pairs SiO2/TiO2 top dielectric mirrors to form a complete VCSEL structure.
141
The Results of Optical Pumping
Figure 5.21 shows the reflectivity and PL measurements during every process step. All of the reflectivity results were normalized to the reflectivity of Au film.
Figure 5.21 (a) shows the reflectivity curve of 35 pairs InGaAlAs/InP DBRs. The center wavelength of the stop-band is located at 1555 nm. Figure 5.21 (b) shows the reflectivity of the half cavity VCSEL. The PL spectrum of the active regions grown on the InP dummy wafer in the same run is shown in Figure 5.21 (c). The PL peak is 1510 nm and the full width half maximum (FWHM) is 54 nm. The peak of shorter wavelength was the signal belonged to the previous grown layers on the InP dummy wafer. Figure 5.21 (d) shows the reflectivity of the complete VCSEL structure and the Fabry-Perot dip is located at 1558 nm. Figure 5.21 (e) demonstrates the PL spectrum of the complete VCSEL structure. The peak wavelength is coincided with the Fabry-Perot dip and the FWHM is measured to be 3.3 nm. The equivalent quality factor, Q, is estimated about 470 in the vertical direction.
The complete VCSEL structure was placed in an optical pumping system as shown in Figure 5.22. The pumping source was a continuous wave operated Ti:sapphire laser. The wavelength of the Ti:sapphire laser was tuned at 990 nm although the pumping wavelength at slightly beyond 1.1 μm would be more desirable to avoid absorption of the InP (λg = 0.9 μm) space layers and InGaAlAs (λg = 1.1 μm) cladding layers. The diameter of the pumping beam entered from the
top dielectric DBRs was estimated to be 30 μm. Figure 5.23 shows the pumping result of the VCSEL. The threshold pumping power is 30 mW at room temperature.
The wavelength of the output beam is 1562 nm. The minimum linewidth above threshold is 1 nm limited by the resolution of the spectrometer. The red shift of the peak wavelength from the Fabry-Perot dip was attributed to the local heating caused
by the strong absorption. The equivalent threshold current density is calculated to be 2 kA/cm2 when taking into account the absorption of the pumping light in the cladding layers and the reflection at the surface. Compared this number with the best threshold current density obtained in chapter 4, which is about 1.45 kA/cm2 shown in Figure 4-11, the relatively large threshold current density might be attributed to two reasons. One is the non-optimized quantum well structure in terms of the amount of net strain in periodic gain structure. The other is the relatively large cavity loss such as the absorption of the pumping power by the 2λ thick cladding layers in the laser structure. All in all, the quality of the DBRs and the active region have been basically qualified in this demonstration for the first step in LW-VCSEL process.
5-4 Electrically Pumped LW-VCSELs
5-4-1 Selectively etched undercut apertures in InP-based LW-VCSELs
Unlike GaAs-based VCSELs, there is no natural oxidizable material in InP-based monolithic VCSEL from which an oxide aperture can be formed. In wafer-fused devices, an AlAs aperture placed in the GaAs-based mirror is typically used. In the InP-lattice-matched materials, there are been some reported on oxide apertures with AlInAs, but the lateral oxidation rate is low. We have demostrated an undercut aperture is employed instead of the oxide aperture, and this is formed by selectively etching an InGaAlAs-based active region with H2SO4 solution. An advatage of the undercut aperture is that this constrains current exactly into the desired area of the active region, and there is no current spreading between the current aperture and active region as can be seen in oxide aperture. Surface-recombination appears to be low in the 1.55μm multiple quantum-well (MQW) active regions.
Based on the success of the optically pumped LW-VCSELs, a half
143
intracavity-contacted VCSEL with n-doped (Si) and p-doped (Zn) for electrically pumping was grown by MOCVD in a single step. A schematic of this structure is shown in Fig. 5.24. This structure is fabricated into devices of diameters 20μm by CH4 reactive ion beam etching (RIE). The mesas were selectively etched by H2SO4
solution. For these devices, a high etching selectivity existed between the active region and InP cladding layers. The undercut apertures could be controlled by different etching time. The top dielectric DBR was deposited with 5 pairs of an electron-beam-evaporated SiO2/TiO2 stack. The half intracavity-contacted VCSEL allow the VCSEL to utilize undoped top DBR, reducing free-carrier absorption in conventinal p-DBR. However, there was no lasing operation due to leakage currents in these selectively etched undercut devices.
5-4-2 Fabrication of LW-VCSELs by Ion-implantations
Implantation technique is a common approach for current confinement [25,26].
Ion-implantation has long been an easy and stable fabrication method in 850 nm VCSEL industry. In this section, we’d like to discuss the structure and experimental results of the InP-based LW-VCSEL with a Si-implantation current aperture.
The Structure of LW-VCSEL
Figure 5.25 (a) shows the laser structure. The epitaxial layers were grown on an n-type InP substrate. The first step was to grow 42 pairs InGaAlAs/InP DBRs. The interruption time was 0.2 minute. The quarter-wavelength thickness of InP and InGaAlAs was 122 nm and 110 nm, respectively. To avoid the absorption of DBRs while the operating wavelength was 1.55 µm, the lattice matched In0.53Ga0.39Al0.08As was used with a band gap emission wavelength of 1.42 µm. The laser reflectometry
in-situ monitored the epitaxial growth. Then, the 3/4λ thick n- and p-type InP spacers for phase matching were grown. The half-wavelength thick cavity was sandwiched between InP spacers. The cavity consisted of n- and p-type InAlAs inner cladding layers and strain compensated multiple quantum wells (SCMQWs). The SCMQWs consisted of seven InGaAlAs (strain = 1.4%, thickness = 6 nm) quantum wells and eight InGaAlAs (strain = -0.8%, thickness = 7 nm) barriers and the photoluminescence (PL) wavelength was 1551.5 nm with the FWHM of 108.3 nm.
The wafer was then sent to ion-implantation with Si after aperture mask patterning.
The diameter of the aperture was 28 µm. The Si ions can compensate the p-type dopant and even make the InP layer to become n-type outside the aperture to form current blocking regions. The energy of the implanting ions is 100 KeV. The peak concentration of the Si in InP spacer layer located at about 100 nm away from the wafer surface measured by the secondary ion mass spectrometry (SIMS). Next, the wafer was passivated with SiNx and coated with metal contacts for both sides. The emission aperture was 10 µm in diameter. Finally, the wafer was coated with 10 pairs SiO2/TiO2 top dielectric mirrors to form a complete VCSEL structure. The schematic of LW-VCSEL structure with a Si-implanted current aperture is shown in Figure 5-25 (b).
Results and Discussion
The voltage and emission light output versus driving current characteristics are shown in Figure 5.26. The solid lines in Figure 5.27 are the reflectivity and PL curves measured with only the half-cavity, which is the as-grown structure shown in Figure 5.25 (a). The PL peak is 1547.4 nm with FWHM of 38.6 nm in comparisons to 108.3 nm for PL curve of the original MQW. The shrinkage of the FWHM demonstrated the
145
increased quality factor of the cavity provided by the high reflectivity bottom DBR.
However, this device did not show stimulated emission under the continuous-wave (CW) operation. The emission output light saturated above 8 mA. The dashed line in Figure 7-3 is the electro luminance (EL) curve for the device. The FWHM of the EL spectrum is 36.9 nm with peak of 1518.0 nm for the full LW-VCSEL structure. The nearly unchanged FWHM and the quality factor of the full LW-VCSEL structure represented the ill function of the top mirror and large optical loss inhibiting the lasing operation. We have found that the top dielectric mirror degraded rapidly after the forward current applied. The degraded top dielectric mirror also changed the original phase matching condition in the half cavity structure, and thus the emission peak of EL spectrum for the full VCSEL structure has shifted. The relatively high operation voltage might arise from the poor p-type contact and caused more heat in the small aperture. To further modify this type of structure, we need to find more stable material conditions for dielectric DBRs to prevent the mirror degradation and improve the p-type contact by inserting a p-type InGaAs for contact layer to reduce the operation voltage.
5-4-3 Long Wavelength Light Emitting Diodes with Buried Tunnel Junctions
Introduction
The buried tunnel junctions have been applied in many opto-electronic devices, such as multi-junction solar cells [27], multi-layer GaAs lasers [28-31], vertical cavity surface emitting lasers [32] and GaN based blue light emitting diodes [33].
Introducing buried tunnel junctions in LW-VCSELs has two main advantages. First, the buried tunnel junction allows selective tunneling current injection and the areas
without the tunnel junction automatically serve as the current blocking function. In comparisons to the ion-implantation, the buried tunnel junction provides precise aperture location inside the LW-VCSEL structure. Second, the buried tunnel junction exhibits higher effective refractive index for the current aperture providing index-guiding effect as discussed in chapter 2. In this subsection, we’d like to discuss the initial attempt to fabricate the lone wavelength light emitting diodes (LEDs) with buried tunnel junctions before we fabricated the LW-VCSELs with tunnel junctions.