Mesa-Size Dependence Characteristics of Vertical Surface-Emitting Lasers

Journal of Electronic Materials,  Sep 2004  by Das, N C,  Chang, W

• E-mail
• Print
• Link

The vertical-cavity surface-emitting laser (VCSEL) is the most suitable light source for many optoelectronic applications because of its planar nature. The design of large VCSEL arrays requires accurate modeling of device characteristics. In this paper, we present a thermal model to analyze the dependence of VCSEL threshold current and light-output characteristics on aperture size. For both 850-nm and 980-nm VCSELs, a linear dependence of threshold current on device area is observed for oxidized aperture sizes with diameters between 5 µm and 25 µm. Good agreement between theoretical and experimental light-output characteristics is observed using a simple thermal model.

Key words: VCSEL, oxidation, surface-emitting laser

INTRODUCTION

Vertical-cavity surface-emitting laser (VCSEL) devices are attractive light sources for various applications including optical recording, optical communications, and optical computing. Recently, many new design concepts including the use of a narrow trench on the top mirror1 and a photonic crystal structure2 have been proposed to achieve VCSELs with single-mode operation and better quantum efficiency. Large two-dimensional VCSEL arrays (32 × 32) have been produced for optical communication circuits.3 To predict the performance of VCSEL arrays, it is required to model the individual device performance accurately. The increasing complexity of these vertically integrated structures, together with the precision required in their epitaxial growth, demands accurate simulation and characterization tools. Accurate modeling allows the device designer or experimental physicist to propose new structures with reasonable confidence and to more readily incorporate VCSEL devices into optoelectronic circuits. Several efforts have been made to model the VCSEL device performance analytically4 and through computer simulations.5,6 It is also well known that computer simulations involve solutions of many differential equations with certain boundary conditions and based on many assumptions. There exist discrepancies between simulated results and experimental data7 mostly because of various assumptions used in the theory. We used a thermal model in conjunction with experimental data to predict the VCSEL performance with different mesa sizes for both 850-nm and 980-nm devices. This model is useful to predict the device performance of different geometries of a particular VCSEL wafer from the experimental results of a device of one mesa size. In this paper, we compared the experimental and theoretical light versus current (L-I) characteristics from different mesa-size devices.

Generally, mesa boundary conditions, ion implantation,8 or oxidation9 are used for current confinement in VCSEL devices. The oxide-confined VCSEL has several advantages over conventional ion-implanted and mesa-confined VCSELs.10 They are: (a) low resistance of the top distributed Bragg reflector (DBR) is fully used, (b) elimination of sidewall nonradiative recombination near optical cavity, (c) optimization of gain to desired laser mode, and (d) a smaller refractive index of the Al-oxide layer induces index-guided optical confinement. Hence, the oxide-confinement VCSEL has low threshold current, low voltage drop, and the highest electrical-to-optical power conversion efficiency. We used the wet oxidation technique for current confinement for both 850-nm and 980-nm devices. Good agreement between theoretical modeling and experimental results is obtained for the limited range of mesa sizes used in our experiment.

DEVICE FABRICATION

The 850-nm top-emitting VCSEL structure was grown by the metal-organic chemical vapor deposition (MOCVD) technique on an n-plus substrate. It has a bottom DBR consisting of 39 pairs of Si-doped Al^sub 0.2^Ga^sub 0.8^As-Al^sub 0.9^Ga^sub 0.1^As, ando a λ laser cavity consisting of three pairs of 70-[Angstrom] Al^sub 0.26^Ga^sub 0.74^As-GaAs quantum wells (QWs). The top DBR consists of 23 pairs of Al^sub 0.75^Ga^sub 0.25^As-Al^sub 0.2^Ga^sub 0.8^As p-doped layers. The heavy-hole excitation resonant energy was designed to account for the bandgap narrowing at higher concentration. This ensures a good match between the gain spectrum and the cavity characteristics. A 3000 -[Angstrom] high Al-content (Al^sub 0.98^Ga^sub 0.02^As) layer for selective lateral oxidation was placed outside the optical cavity region and at the null position of the optical field in the VCSEL structure.

The VCSEL device processing started with a ringcontact Ti/Pt/Au metal deposition as the top contact layer. The detailed fabrication procedure is given in an earlier publication.11 An inductively coupled plasma (ICP) etching technique with a Cl^sub 2^ and BCI^sub 3^ gas mixture is used for mesa etching. The mesa diameter varies between 15 µm and 34 µm, and the mesa height after ICP etching was 4 µm. Wet oxidation was carried out at 425°C for 30 min with nitrogen carrier gas bubbled through water at 95°C. The oxide width around the edge of the mesa structure caused by the preceding oxidation is approximately 5 µm. Hence, the aperture diameter is 10 µm less than the corresponding mesa diameters of different devices. The device mesas were passivated by low-temperature, plasma-enhanced chemical vapor deposition (PECVD) of a 2,000-[Angstrom] SiO^sub 2^ layer. Spincoated cyclotene (BCB) resin was used for planarization. After complete curing in a nitrogen environment, the BCB film was back-etched in CF^sub 4^/O2 plasma until the BCB was etched completely away from the top of the mesa area. We used an offset interconnect metal contact for flip-chip bonding so that the pressure caused by flip-chip bonding would not be applied on the VCSEL mesa structure. A Ge/Ni/Au metal film was deposited on the back side of the substrate as the n-contact layer. Rapid thermal annealing was done at 410°C for 60 sec in a nitrogen environment to reduce the contact resistance. The schematic diagram of the cross-sectional view of the VCSEL is shown in Fig. 1. After mesa etching, both oxidation layers are exposed and oxidized by a wet oxidation technique. Current flows through the cross-section area where no oxidation occurs.