Selectively oxidized VCSELs go singlemode

Th 0008lwfea13f1

Manipulating modal gains or losses allows selectively oxidized vertical-cavity surface-emitting lasers to reach single-transverse-mode operation.

Kent D. Choquette, Sandia National Laboratories

Vertical-cavity surface-emitting-laser (VCSEL) sources have been adopted into Gigabit Ethernet applications in a remarkably short time. Due to their reduced threshold current, circular output beam, and inexpensive and high-volume manufacture, VCSELs are particularly suitable for multimode optical-fiber local area networks (LANs). Selectively oxidized VCSELs contain an oxide aperture within the laser cavity that produces strong electrical and optical confinement, enabling high electrical-to-optical conversion efficiency and minimal modal discrimination and allowing emission into multiple transverse optical modes.1 Such lasers make nearly ideal LAN sources.

In addition, VCSELs that emit into a single optical mode are also increasingly sought for emerging applications including data communication with singlemode optical fiber, barcode scanning, laser printing, optical read/write heads, and modulation spectroscopy. Achieving singlemode operation in selectively oxidized VCSELs is a challenging task, because the inherent index confinement within these high-performance lasers is very large.Th 0008lwfea13f1

Figure 1. The active region and optical cavity of a vertical-cavity surface-emitting laser are sandwiched between distributed-Bragg-reflector (DBR) mirrors. Structures intended to achieve single-transverse-mode optical output from a VCSEL include etched DBR surface relief (top left), tapered oxide aperture (top right), extended optical cavity (bottom left), and gain aperture (bottom right).

Typical VCSELs have optical cavity lengths on the order of one wavelength and thus operate with a single longitudinal optical mode. However, because of their relatively large cavity diameter (roughly 5-20 microns), these lasers usually operate in multiple transverse optical modes. Each transverse mode possesses a unique wavelength and transverse spatial intensity profile. For applications requiring small spot size or high spectral purity, lasing in a single optical mode-usually the lowest-order fundamental mode-is desired. Pure fundamental-mode emission in a selectively oxidized VCSEL can be attained by increasing the optical loss of higher-order transverse modes relative to that of the fundamental mode. Creating mode-selective loss increases modal discrimination, leading to singlemode VCSEL operation.

Strategies for producing singlemode VCSELs have recently been developed.2 The techniques are based either on introducing loss that is relatively greater for higher-order optical modes or on creating greater gain for the fundamental mode (see Figure 1). Increased modal loss for higher-order modes has been demonstrated by three different techniques.

The first approach uses an etched-surface relief on the periphery of the top facet that selectively reduces the reflectivity of the top mirror for the higher-order modes. An advantage of this technique is that the etched ring around the edge of the cavity in the top mirror can be produced either during VCSEL fabrication by conventional dry etching, as described by researchers at the University of Ulm (Ulm, Germany), or it can be postprocessed on a completed VCSEL die, as demonstrated by workers at the University of Bristol (Bristol, England) using focused ion-beam etching. A disadvantage of etched-surface relief is that it requires careful alignment to the oxide aperture and can increase the optical scattering loss of the fundamental mode, as manifested by the relatively low (<2 mW) singlemode output powers that have been reported.Th 0008lwfea13f2

Figure 2. Selectively oxidized 850-nm-emitting single-transverse-mode VCSEL contains a 4x4-micron tapered oxide aperture that produces greater than 30 dB of side-mode suppression at 3.5-mW optical output.

It would be desirable to introduce mode-selective loss into the VCSEL epitaxial structure to avoid extra fabrication steps and to provide self-alignment. Two such techniques are the use of tapered oxide apertures and extended optical cavities within the VCSELs. The first approach, pursued at Sandia National Laboratories (Albuquerque, NM), is predicated on designing the profile of the oxide aperture tip to preferentially increase the higher-order mode loss. The aperture-tip profile is produced by tailoring the composition of the aluminum gallium arsenide layers, which are oxidized to create the aperture within the VCSEL. A VCSEL containing a tapered oxide whose tip is vertically positioned at a null in the longitudinal optical standing wave can produce greater than 3 mW of singlemode output and greater than 30 dB of side-mode suppression (see Figure 2). However, this structure requires a detailed understanding of the oxidation process and also produces additional loss for the fundamental mode.

A second way to increase modal discrimination is to extend the cavity length of the VCSEL and thus increase the diffraction loss of higher-order modes. Researchers at the University of Ulm reported singlemode operation up to 5 mW using a VCSEL with a 4-micron-thick cavity spacer inserted within the optical cavity.

Using even-longer cavity spacers introduced multiple longitudinal modes, but single-transverse-mode-operation up to nearly 7 mW was demonstrated. It is interesting to note that VCSELs containing multiple wavelength cavities do not appear to suffer any electrical penalty, although careful design is required to balance the tradeoffs between the modal selectivity of transverse and longitudinal optical modes.

Finally, manipulating the modal gain rather than loss also can produce singlemode VCSELs. A technique to spatially aperture the laser gain independent of the oxide aperture has been developed at Sandia. The key aspect of these VCSELs is the lithographically defined gain region that is produced by intermixing the quantum-well active region at the lateral periphery of the laser cavity.

The fabrication process begins with the growth of the bottom Bragg mirror and optical cavity containing the quantum-well active region. The active region is homogenized by ion implantation around masked regions that form the laser cavities. Following the quantum-well intermixing, the top Bragg mirror is grown epitaxially. The resultant VCSEL has a central quantum-well region that preferentially provides gain for the fundamental mode. Singlemode output of more than 2 mW with a side-mode-suppression ratio greater than 40 dB has been obtained.

Although this approach requires greater fabrication complexity, it is anticipated that higher performance can be reached with further refinement of the process parameters.

Motivated by the new demands of emerging VCSEL applications, single-mode VCSELs are currently under development at numerous laboratories around the world.

The techniques demonstrated to date introduce modal discrimination by increasing the optical loss of the higher-order modes or by increasing the relative gain of the fundamental mode. The comparative merits of the various approaches are still being sorted out; the criteria are the degree of increased fabrication complexity, tradeoffs to the overall laser performance, and maximum singlemode output that can be achieved. Currently, 5 mW represents the best singlemode output from a VCSEL, whereas 10 mW or greater of singlemode power would likely fuel the next generation of VCSEL applications.

  1. K. D. Choquette, K. L. Lear, and R. P. Schneider Jr., Efficient Semiconductor Light-Emitting Device and Method, US Pat. # 5,493,577 (Feb. 20, 1996).
  2. Proc. SPIE Vertical Cavity Surface Emitting Lasers V, K. D. Choquette and S. Lei, eds. (2000).

Kent D. Choquette is a principal member of technical staff at Sandia National Laboratories (Albuquerque, NM). He can be reached at e-mail:

This article appeared in the May 2000 issue of Laser Focus World, Lightwave's sister publication.

More in Components