Menu
Cart 0

Vertical-Cavity Surface-Emitting Lasers (VCSELs)

This is a continuation from the previous tutorial - quantum well semiconductor lasers.

 

1. Background Information

Semiconductor lasers described in the previous tutorials have cleaned facets that form the optical cavity. The facets are perpendicular to the surface of the wafer and light is emitted parallel to the surface of the wafer.

For many applications requiring a two-dimensional laser array or monolithic integration of lasers with electronic components (e.g., optical interconnects), it is desirable to have the laser output normal to the surface of the wafer. Such lasers are known as surface-emitting lasers (SEL).

There is a class of surface-emitting lasers which have their optical cavity normal to the surface of the wafer. These devices are known as vertical-cavity surface-emitting lasers (VCSEL). In this tutorial the fabrication and performance characteristics of these vertical-cavity types of surface-emitting lasers are described.

In the late 1970s, Soda et al. reported on an SEL fabricated using the InGaAsP material system. Their device structure is shown in Figure 10-1. The surfaces of the wafer form the Fabry-Perot (FP) cavity of the laser.

Fabrication of this device involves the growth of a double heterostructure on an \(n\)-type InP substrate. A circular contact is made on the \(p\)-side using an SiO2 mask.

The substrate side is polished making sure that it is parallel to the epitaxial layers, and ring electrodes (using an Au-Sn alloy) are deposited on the \(n\)-side. The lasers had a threshold current density of \(\sim11\text{ kA/cm}^2\) at \(77\text{ K}\) and operated to output powers of several milliwatts.

 

Figure 10-1.  Schematic cross section of a surface-emitting laser. Surfaces of the wafer form a Fabry-Perot cavity.

 

Compared to this early work in the late 1970s, the advances in surface-emitting laser fabrication technology have been phenomenal. Present state-of-the-art SELs have threshold current of \(\lt1\text{ mA}\) at room temperature and can be modulated at high bit rates.

Although there are many differences between the early work (Figure 10-1) and the recent devices, one key difference is the development of very high reflectivity (\(\gt99\%\)) mirrors by using Bragg reflectors composed of multilayer semiconductor stacks. The following section discusses this technology.

 

 

2. Mirror Reflectivity

A generic SEL structure utilizing multiple semiconductor layers to form a Bragg reflector is shown in Figure 10-2.

The active region is sandwiched between the \(n\)- and \(p\)-type cladding layers which are themselves sandwiched between the two \(n\)- and \(p\)-type Bragg mirrors.

The Bragg mirrors consist of alternating layers of low-index and high-index materials. The thicknesses of each layer is equal to one quarter of the wavelength of light in that medium.

Such periodic quarter-wave-thick layers can have a very high reflectivity. For normal incidence, the reflectivity is given by

\[\tag{10-2-1}R=\left[\frac{1-\frac{\mu_l}{\mu_1}\left(\frac{\mu_2}{\mu_3}\right)^{2N}}{1+\frac{\mu_l}{\mu_1}\left(\frac{\mu_2}{\mu_3}\right)^{2N}}\right]^2\]

where \(\mu_2\) and \(\mu_3\) are the refractive indices of the alternating layer pairs, \(\mu_l\) and \(\mu_1\) are the refractive indices of the medium on the transmitted and incident sides of the DBR mirror, and \(N\) is the number of pairs. As \(N\) increases, \(R\) increases. Also for a given \(N\), \(R\) is larger if the ratio \(\mu_2/\mu_3\) is smaller.

 

Figure 10-2.  Schematic illustration of a generic SEL structure utilizing distributed Bragg mirrors formed by using multiple semiconductor layers. DBR pairs consist of AlAs (71.1-nm thick) and Al0.1Ga0.9As (60.5-nm thick) alternate layers. The active layer could be either a quantum well or have a thickness similar to a regular double-heterostructure laser.

 

For an SEL to have a threshold current density comparable to that of an edge-emitting laser, the threshold gains must be comparable for the two devices. The threshold gain of an edge-emitting laser is \(\sim100\text{ cm}^{-1}\). For an SEL with an active-layer thickness of 0.1 μm, this value corresponds to a single-pass gain of \(\sim1\%\). Thus, for the SEL device to lase with a threshold current density comparable to that of an edge emitter, the mirror reflectivities must be \(\gt99\%\) to ensure that cavity losses are smaller than the gain available during a single pass.

Instead of alternating low- and high-index layers of semiconductors, SELs have also been fabricated using alternating low- and high-index layers of dielectric materials such as SiO2-TiO2 or Si-SiO2. These dielectric materials are useful for SELs emitting near 1.3 μm or 1.55 μm (made by using the InGaAsP material system), because it is more difficult to fabricate semiconductor stacks for the InGaAsP material system compared with the AlGaAs material system.

Central to the fabrication of low-threshold SELs is the ability to fabricate high-reflectivity mirrors. The scanning electron photomicrograph of an SEL structure is shown in Figure 10-3.

 

Figure 10-3.  Scanning electron photomicrograph of a GaAs-AlGaAs SEL wafer.

 

It consists of alternating layers (22 pairs) of \(n\)-Al0.9Ga0.1As and \(n\)-Al0.1Ga0.9As grown over an \(n\)-GaAs substrate followed by the GaAs active region sandwiched between \(n\)- and \(p\)-type Al0.3Ga0.4As cladding layers.

The top cladding layer is further followed by 15 pairs of \(p\)-Al0.9Ga0.1As layers and \(p\)-Al0.1Ga0.9As layers and a thin (~ 20 nm) \(p\)-GaAs contact layer.

The two stacks of alternating layers on each side of the active layer form the mirrors of the laser cavity. The thickness of each layer in the stack equals \(\lambda\)/4, where \(\lambda\) is the wavelength of light in the medium.

As the number of layer pairs increases, the effective reflectivity of the stack increases. The calculated peak reflectivity at 0.85 μm of an alternating stack of Al0.9Ga0.1As and Al0.1Ga0.9As layers as a function of the number of pairs is shown in Figure 10-4.

Figure 10-4.  Calculated reflectivity of an Al0.9Ga0.1As-Al0.1Ga0.9As multilayer semiconductor Bragg reflector as a function of the number of pairs.

 

The reflectivity spectrum of an SEL structure is shown in Figure 10-5. The reflectivity is \(\gt99\%\) over a 10-nm band. The drop in reflectivity in the middle of the band is due to an FP mode.

Figure 10-5.  Reflectivity of an SEL wafer consisting of top and bottom DBR mirrors similar to that shown in Figure 10-3.

 

The number of pairs needed to fabricate a high-reflectivity mirror depends on the refractive index of layers of a pair. For large index differences fewer pairs are needed. For example, in the case of alternating layers of ZnS and CaF2 for which the index difference is 0.9, only six pairs are needed for a reflectivity of 99%. By contrast, for an InP-InGaAsP (\(\lambda\sim1.3\) μm) layer pair for which the index difference is 0.3, more than 40 pairs are needed to achieve a reflectivity of 99%.

While the large index difference between the alternating semiconductor layers is useful for high reflectivity, the accompanying energy-gap difference results in potential barriers at heterointerfaces. These potential barriers impede the flow of carriers through a Bragg-reflector mirror stack, resulting in a high-resistance device.

It turns out that the use of graded heterobarrier interfaces in pace of an abrupt interface reduces the series resistance significantly without compromising the reflectivity.

 

 

3. GaAs-AlGaAs and InGaAs-GaAs Surface-Emitting Lasers

Much of the work on surface-emitting lasers has been carried out using the GaAs-AlGaAs material system. The GaAs SELs (\(\lambda\sim0.85\) μm) have either a thick (~ 0.1 μm) or thin (quantum-well) active region.

The InGaAs SELs (\(\lambda\sim1\) μm) grown over a GaAs substrate generally have a multiquantum-well active region. Several types of lasers have been fabricated by using this system. They generally differ in the way in which high-reflectivity mirrors and current confinement are achieved.

Five principle structures used for SEL fabrication are

  1. Etched-mesa structure
  2. Ion-implanted structure
  3. Dielectric isolated structure
  4. Buried heterostructure
  5. Metallic-reflector structure

Schematics of these devices are shown in Figure 10-6 for the case of a GaAs active region. Since the substrate is absorbing in this case, light is generally emitted from the top. It can also be emitted from the bottom by using a structure in which the substrate near the emitting region has been etched away.

 

Figure 10-6.  GaAs-AlGaAs SEL structures: (a) etched-mesa structure, (b) ion-implanted top-emitting structure, (c) dielectric isolated structure, (d) buried heterostructure, and (e) metallic-reflector structure.

 

In the case of InGaAs (\(\lambda\sim1\) μm) grown over a GaAs substrate, light is not absorbed by the substrate, and can therefore be collected from the substrate side (Figure 10-7).

Jewell and coworkers first reported excellent performance for such lasers using the etched-mesa structure with an InGaAs active region. Etched mesas are typically a few micrometers in diameter (3 to 5 μm) which allows fabrication of a large number of lasers on a single substrate.

Critical to the fabrication of this device is mesa etching with damage-free surfaces. Jewell and coworkers used a reactive-ion etching process for their devices. However, for practical use a suitable bonding scheme is required. A polyimide can be used to fill up the region around the etched mesas, to allow a practical bonding scheme and subsequent CW and high-speed testing.

 

Figure 10-7.  InGaAs-GaAs SEL structures. The \(n\)-GaAs substrate is transparent to the emitted light. Hence the light can be collected from the substrate side. (a) Etched-mesa structure and (b) ion-implanted structure.

 

In addition to very high reflectivity mirrors, the key element necessary for the fabrication of good etched-mesa SELs is the use of an etching process that produces surfaces with very little nonradiative recombination.

Threshold currents of 0.7 mA and 2 mA have been reported for InGaAs-GaAs (\(\lambda\sim1\) μm) and GaAs-AlGaAs (\(\lambda\sim0.85\) μm) etched-mesa lasers respectively.

The \(L-I\) characteristics of a low-threshold etched-mesa-type SEL are shown in Figure 10-8. Modulation bandwidths in excess of 2 GHz have been realized for this structure.

 

Figure 10-8.  \(L-I\) characteristics of a low-threshold InGaAs-GaAs SEL. Calculated top and bottom mirror reflectivities of this structure were 0.99998 and 0.9998 respectively. The light is emitted from the bottom.

 

Top-emitting GaAs (\(\lambda\sim0.85\) μm) SELs of the type shown in Figure 10-6(b) have been studied extensively. Proton implantation is generally used to produce a region of high resistivity around a 10-μm diameter opening. This scheme reduces current spreading, producing a region of high gain at the center of the opening where the lasing action takes place.

Typically, laser structures are designed to have 99.9% reflectivity for the substrate-side multilayer mirror and 99-99.5% reflectivity for the top mirror. These lasers have operated over a wide temperature range. CW and pulsed \(L-I\) characteristics of top-emitting GaAs SELs at different temperatures are shown in Figure 10-9.

Also shown is the single-wavelength output spectrum of the laser output showing a single longitudinal mode with a mode-suppression ratio in excess of 30 dB. The single-mode behavior arises from the short FP cavity. The cavity length is so short (~ 1 μm) that only one longitudinal cavity mode fits within the gain spectrum.

Under pulsed operation both the output power and the operating temperature are higher, indicating the importance of heat dissipation in SELs. The rollover in the \(L-I\) curve of SELs is primarily due to heating which arises from a high series resistance (typically > 100 Ω) and a voltage drop at threshold of these devices.

 

Figure 10-9.  \(L-I\) characteristics under (a) CW operation and (b) pulsed operation, and (c) spectrum of a proton-implanted GaAs-AlGaAs surface-emitting laser. The higher output power of this device compared to that shown in Figure 10-8 is due to the lower reflectivity of the output mirror.

 

For some SELs, the threshold current under pulsed operation is found to be higher than that for CW operation. This effect has been explained by heating-induced self-focusing of the optical mode which results in higher modal gain at a given current for CW operation than that for pulsed operation.

The ion-implanted planar SELs can be modulated at high speeds. The measured small-signal modulation response at different output powers is shown in Figure 10-10. As discussed in the modulation response of semiconductor lasers tutorial, the bandwidth is found to be proportional to the square root of the output power. Under gain-switched conditions pulses as short as 22 ps have been obtained.

 

Figure 10-10.  Small-signal modulation characteristics of a proton-implanted GaAs-AlGaAs SEL.

 

A variant of the SEL structures makes use of a combination of metallic reflector and a multilayer Bragg mirror on the epitaxial side for achieving high reflectivity. These devices have the advantage of somewhat lower resistance and are more tolerant of epitaxial growth variations simply because fewer layers are needed to form the mirror. Pulsed operation using a silver metal film as a reflector (without any multilayer Bragg mirror) on the \(p\)-side has also been achieved [see Figure 10-6(e)].

In a modification of the SEL structure, the quantum-well gain regions are located at the maximum of the standing-wave pattern in the FP cavity. This structure is known as the resonant periodic-gain structure. It allows for the maximum coupling between the gain region and the optical field, resulting in a very low threshold for such devices.

 

 

4. InGaAsP-InP Surface-Emitting Lasers

As mentioned previously, the refractive-index difference between InGaAsP (\(\lambda\sim1.3\) μm) and InP layers is smaller than in GaAs SELs, hence InGaAsP-InP SELs use more pairs (typically 40-50) to produce a high-reflectivity (> 99%) mirror.

Such mirror stacks have been grown by chemical-beam epitaxy (CBE), metal-organic chemical vapor deposition (MOVPE), and molecular-beam epitaxy (MBE) growth techniques.

The measured reflectivity of a 45-pair quarter-wave stack grown by CBE is shown in Figure 10-11. The mirror stack was designed for maximum reflectivity around 1.6 μm.

 

Figure 10-11.  Measured reflectivity spectrum of a 45-pair InGaAsP (121.6 nm thick, \(\lambda_\text{g}=1.45\) μm) and InP (134.8 nm) DBR stack.

 

Such mirror stacks along with a dielectric Si-SiO2 stack on the \(p\)-side have been used to fabricate InGaAsP SELs. The structure and the \(L-I\) characteristics of such a device are shown in Figure 10-12.

The threshold current of these laser sis generally higher than that for cleaved-facet lasers. As mirror fabrication technology develops further, a reduction in threshold current is likely to follow.

 

Figure 10-12.  (a) Structure and (b) \(L-I\) characteristics of an InGaAsP-InP SEL.

 

 

5. Laser Arrays

Edge-emitting phase-locked laser arrays have been extensively studied for high-power operation [refer to the laser arrays tutorial]. An often-encountered problem in the performance of an edge-emitting laser array is that the far field along the junction plane has two lobes when neighboring emitters oscillate out of phase. This out-of-phase oscillation is favored for gain-guided arrays because there is no gain between emitters.

It is possible to design surface-emitting laser arrays such that the neighboring emitters oscillate in phase by design, and hence the device has a single-lobe far field. In such a device (see Figure 10-13), current flows in the region between the mirrors and hence the gain is maximum between the mirrors which favors the in-phase operation.

 

Figure 10-13.  Schematic of an SEL array.

 

For many applications of SEL arrays, each laser in the array should be biased and controlled individually. Independently addressable laser arrays have been fabricated. It is even possible to design multi-wavelength SEL arrays such that each laser operates at a slightly different wavelength using a growth technique where the Bragg wavelength of mirror stacks is varied across a wafer. Such laser arrays have been used in transmission experiments.

 

 

The next tutorial discusses about optical amplifiers.


Share this post


Sale

Unavailable

Sold Out