R. L. Naone Materials Department,
University of California,
Santa Barbara, CA 93116
E. R. Hegblom*, D. Lofgreen, L. A. Coldren
ECE Department, University of California,
Santa Barbara, CA 93116
Improved performance of smaller vertical cavity lasers is important for applications which require arrays of vertical cavity lasers operating at low power such as free-space optical interconnections between computer boards. If one obtains ideal scaling in which threshold current density and slope efficiency remain constant as device size shrinks, overall threshold currents should scale inversely with the device area. However, it is not a simple matter to achieve this scaling due to electrical and optical losses in etched pillar[1] or dielectric apertured devices[2]. By engineering the shape of intra-cavity dielectric apertures it is possible to better confine the current and provide low-loss guiding of the optical mode, thereby approaching the desired scalability. More importantly, the power conversion efficiency is higher at the lower output powers because the voltage is lower despite increased resistance due to current transport through the aperture.
The lateral optical confinement by oxide aperturing in VCLs is not perfect.
Theoretical analysis in which the optical cavity was modeled as a periodically apertured waveguide[3] supported experimental evidence that in thick (~80nm) oxide apertured devices, the performance for sizes less than 4µm was limited by excess optical scattering losses from the large abrupt index step. The modeling explained the reduced optical scattering loss observed in devices with a much thinner oxide [4]. Furthermore, the analysis showed that by tapering the aperture tip and positioning it in the optical cavity at the first null, the excess optical loss can be eliminated.
By utilizing the strong suppression of oxidation rate by adding GaAs to AlAs, we can grade the AlGaAs composition in such a way as to obtain the desired tapered oxidation profile[5]. We fabricated a device using an aperture layer design which gives a 0.7µm linear oxidation profile ending at the 1st standing wave null. From external efficiency measurements of these devices we observed much lower optical scattering losses than abrupt oxide apertures particularly at sizes less than 2µm in diameter. We were thus able to fabricate sub-micron diameter devices while maintaining good slope efficiencies, as shown in Fig. 1. Even though resistance increases inversely with radius, the devices exhibited power conversion efficiencies exceeding 20% at only 150µW for 2-3µm size devices, and the highest efficiency breaks 30% at ~1mW for the 4-5µm ones (Fig. 2). In addition to the low optical loss of our aperture design, we measured a 1/e^2 width of the near field intensity of 2.8µm in a 2-3µm diameter aperture device, demonstrating the strong waveguiding of the tapered oxide. Unlike the low-loss but weak guiding of a simple thin aperture, this design should provide stable mode behavior under dynamic operation.
Figure 1. L-I curves are shown for three very small diameter devices. Maximum power conversion efficiencies are 20% @ 150µW for the 2-3µm.
Figure 2. Power conversion efficiency vs. output power for three different aperture diameters indicated next to the curves.
We compared the scalability of threshold current density, Jth, of different device designs and observed that in 2µm radius tapered aperture devices, Jth is 3 times the broad area value while apertures farther away from the active region showed 5 times the broad area value. The first standing wave null in our devices is about 50nm away from the active region which means that our aperture design provides better current confinement than apertures placed in the 1st DBR period. Our calculations suggest that current spreading under the aperture before it is injected into the active region alone cannot account for the increased Jth at smaller sizes, and we believe once in the active region it is the lateral carrier diffusion which dominates the current leakage in this size regime. In the ideal case in which all leakage current is eliminated, a significant improvement in the power conversion efficiency at low output powers and smaller sizes is possible. With no current leakage, the device performance extrapolates to an improvement in wall plug efficiency from 20% to 35% at 150µW of power.
Our main focus in improved scaling of the threshold current is preventing the carriers in the active region from diffusing away from the center of the device. We are investigating techniques which form a buried heterostructure (BH) that acts as a barrier to lateral carrier diffusion.
For device aperture sizes less than 2µm, we expect at least a 50% reduction of threshold current even if our BH is 2µm larger, due to the large diffusion length of the carriers.
One approach involves intermixing the laser structure using high temperature annealing. The bandgap of the intermixed InGaAs QW active region increases due to out-diffusion of In and in-diffusion of Al from the SCH regions. Fig. 3 shows the projected improvements in threshold current by lateral carrier confinement using a BH approach.
Figure 3. Projected Ith improvement of the existing
devices (squares) by lateral carrier confinement. The dashed and solid lines
show the threshold reduction
by using a buried heterostructure (BH) 2µm larger than and the same size as
the aperture, respectively. The dotted line shows the ideal scaling of the
broad-area threshold current density.
This work was funded by NSF through QUEST and DARPA through the Heterogeneous Optoelectronics Technology Center.
[1] B.J. Thiebault, T. A. Strand, T. Wipiejewski, M. G. Peters, D. B. Young, S. W. Corzine, L.A. Coldren, and J.W. Scott, J. Appl. Phys., 78, 5871 (1995).
[2] K. L. Lear, S. P. Kilcoyne, R. P. Schneider, K. D. Choquette, and G. R. Hadley, presented at 1995 IEEE LEOS Annu. Meet., San Francisco, CA, paper SCL 16.7.
[3] E. R. Hegblom, D. I. Babic, B. J. Thibeault, and L. A. Coldren, Appl. Phys. Lett., 68, 1757 (1996).
[4] B. J. Thibeault, E. R. Hegblom, P. D. Floyd, R. Naone, Y. Akulova, and L.A. Coldren, IEEE Photon. Tech. Lett., 8, 593 (1996).
[5] R. L. Naone, E. R. Hegblom, and L. A. Coldren, Electron. Lett, 33, 300 (1997).
[6] P. D. Floyd, B. J. Thibeault, E. R. Hegblom, J. Ko, and L. A. Coldren, IEEE Photon. Technol. Lett., 8, 590, (1996).