In the past decade, diode lasers have made great strides in both power and reliability. The performance improvements have enabled new applications, particularly in areas where optical brightness is a key performance metric. The fiber-coupled diode pumping of fiber lasers is a critical technology in commercial, industrial and defense applications. Materials processing, once the singular province of fiber lasers, can now be addressed in some respects by diode laser beams directly delivered by optical fiber.

Diode Lasers

FIGURE 1 (Left). A schematic representation of the structure of a laser diode. Figure 1 (Right). The optical cavity and light output in a laser diode.
Diode lasers are typically constructed of molecularly layered binary, ternary, and quaternary semiconductor crystals grown by molecular beam epitaxy or metal organic chemical vapor deposition. The job of the diode laser engineer is to properly select the constituent elements and structure the layers to achieve the requisite electro-optical properties of a diode laser – efficient control of the flow and capture of electrons and the proper guiding of the laser light through the structure.

Semiconductor crystals are deliberately grown to incorporate impurities into the lattice. These impurities serve to control the conductivity of the semiconductor, with one side of the structure receiving extra electrons and the other side positive carriers known as holes. Near the center of the diode laser structure are quantum wells which serve as the diode laser’s active region. The quantum wells are only several tens of angstroms across, narrow enough to create quantized energy states for both the electrons and holes. The energy difference between the quantized energy states of the carriers determines the diode laser’s emission wavelength. Careful engineering of the strain in the quantum wells allows the engineer to further tailor such properties as the threshold current and the gain characteristics of the device.

When the epitaxial structure is processed, metalized layers deposited on the semiconductor surface serve as the injection area of the electrical carriers that, upon transport in sufficient quantity to the active region, create the population inversion necessary to achieve lasing. The lasing structure is completed by cleaving either side of the semiconductor to create the reflective facets needed for the final element necessary to achieve lasing, that of optical feedback.

In practice, the facets of the diode laser are coated with dielectrics that enhance reflectivity at one end and diminish it at the other, ensuring most of the light exits one end of the laser. The cleaving of the semiconductor to create the facets leaves dangling atomic bonds at the crystal/air interface, which serve as electrical carrier recombination sites, creating heat rather than light. The facets have historically been the limiting factor in achieving both long life and high power from diode lasers with the facet failing due to catastrophic optical damage enabled by the carrier recombination. Advances in treatments and coatings at the facets are responsible for the rapid advance in recent years of maximum achievable power and concomitant improvement in reliability that can now extend to tens of thousands hours of usable life. Figure 1 depicts representations of the diode structure and optical cavity of a diode laser.

FIGURE 2. Optical concept that combines 8 laser bars after fast and slow axis collimation for a total of 300W output power from a 200um core fiber at 50A and greater than 50% efficiency.
In the plane normal to the diode laser plane, the coherent optical field generated through the recombination of the electrons and holes in the active region is guided by the optical index variation introduced by the various layers of the semiconductor. It is one of the rare, fortuitous gifts of nature that the same semiconductor layers that guide the laser optical field so well also have the characteristic of confining the electrons where they need to be confined in the diode. As the optical guiding layers are very thin, just a few microns in depth, and the confinement is diffraction limited, this axis of the optical field diverges very rapidly, oftentimes in the range of more than 80 degrees for full width at 90% power enclosure. As such, it is referred to as the fast-axis of the diode.

Parallel to the plane of the diode, the optical field is either very weakly optically guided or guided only by the region of the semiconductor where sufficient injection current exists to achieve optical gain. The width of the injected region as defined by the metallization is typically 50 to 200 microns. Given such a broad, optically multi-mode profile, the light emitted in this direction is far from diffraction limited and is typically referred to as the slow-axis of the device. Divergence in the slow-axis is typically 10-15 degrees at full-width, 90% power enclosure.

It is the properties of the fast- and slow-axes that the optical engineer must manipulate to efficiently couple the laser light into an optical fiber. A characteristic of the diode is its brightness – the ratio of the output power to the beam parameter product (BPP), where the BPP is the product of the beam waist size and its divergence. The brightness of the diode laser is at best a conserved quantity, meaning it can’t be improved upon with only passive means, but that it can be diminished through power loss and beam distortion. The art of high brightness fiber-coupled diode laser module design is the development and manufacture of the highest brightness diode lasers possible and then manipulating the laser light into the optical fiber while losing as little of the original brightness as possible.


The emitted laser light must be rapidly collimated after exiting the diode facet so that it is not lost through absorption or scattering in the module. The fast-axis is typically captured by a cylindrical lens of 500-1000 micron focal length. The diffraction limited nature of this axis results in a residual fast-axis divergence in the low, single digits of milliradians.

In the slow axis, each diode laser emitter on the laser bar has its own slow-axis lenslet. To achieve best collimation it is desirable to fill as much of the lenslet aperture as possible, resulting in scattering losses at the edges where each of the lenslets abut. Due to the multi-moding and a beam well above the diffraction limit, these lenslets can only achieve a residual divergence of 10-15 milliradians while keeping losses in check. Other sources of power loss and, hence, brightness are losses at optical interfaces and through non-ideal lens shapes.

A common measure of the optical quality of light is M2. An M2 of 1 means the beam is a perfect Gaussian in that axis. Any increase in the value M2 in excess of 1 means an ever worsening beam quality. For diode lasers, the fast-axis is reasonably close to a M2 value of 1 but the slow axis is closer to 10 or more for each emitter. It is therefore desirable to transform the beam such that one axis retains the poor M2 of just a single beam in the slow axis direction, while stacking beams in the other axis such that the near ideal fast axis from each emitter is additive. The net result is a much more symmetric beam quality profile that can more easily be coupled into the optical fiber.

FIGURE 3. IS46, tailored laser bar (T-Bar) module.
Further down the optical pathway, optical telescopes are used to trade beam size for beam divergence. If the residual beam divergence is not adequately controlled then a trapezoidal effect is created, with beams originating from more distant diodes creating larger spot sizes that do not efficiently couple into the fiber. The use of telescopes is also necessary to match both the spot size and numerical aperture of the beam to the core size and numerical aperture of the fiber.

In order to keep the beam within the design constraints of the module dimensions, mirrors are often used to redirect the beam path as well as in the combining of the beams. Close optical stacking of the beams is critical to achieve high-brightness in the module, but practical limits prevent precise edge-to-edge alignment of the beams. Figure 2 shows optical stacking of multiple beams in the fast axis direction at left and a full optical simulation of a module to include all optics at right.

Finally, polarization multiplexing can be used to double the brightness of a diode laser array by rotating the polarization of half of the emitted light and then physically overlaying it with the other half of the emitted light. As the polarization purity of the diode laser is typically 95% or less, this method results in some loss of power in exchange for increased optical areal density. Careful consideration must be given to the application of this technique.

Optical Fiber

In its simplest incarnation, an optical fiber is a flexible, cylindrical pairing of a higher optical index core with a lower optical index cladding. The fiber guides the light through total internal reflection. Presently, typical fiber core diameters are 100, 200, 400, 800 and 1000 microns but there is no particular restriction on core size and other diameters can be readily found.

FIGURE 4. Electro-optical characteristics of IS46 module.
Typical numerical apertures (NA) of the fiber or, equivalently, the greatest angle of incidence at which the fiber can accept light, are 0.15, 0.22 and 0.44 although, here again, other values are certainly possible and available. Any light impinging upon the fiber with a spot size (beam waist) greater than the core diameter or an angle greater than the NA is either coupled into the cladding layer or reflected outright. In either case, any light outside the core must be dealt with by the module engineer. The smaller the core diameter and the lower the NA the higher is the brightness of the fiber for a given power.

Due to the geometry of the emitted light from the diodes, the beam delivered to the fiber is rectangular in shape in both physical and NA spaces. With the geometry of the fiber being cylindrical in both spaces, it is often necessary to deliberately overfill both spaces of the aperture of the fiber, with overfilling of the NA being more dangerous to the survivability of the fiber.

In these cases, a mode stripper is frequently used to strip the excess light from the cladding before it can propagate and damage the fiber or be delivered to the end of the fiber where it can damage the load.

A Design Example

DILAS manufactures a diode laser module that delivers 300W to the exit end of a 200μm/0.22NA fiber but weighs only approximately 300 grams (Figure 3) while achieving greater than 50% electrical-input power to optical-output power (E-to-O) efficiency. The diode laser chip engine is a tailored laser bar, or T-Bar, which maximizes the brightness of the diode laser. An efficiency of 65-70% E-to-O for bars coming from the manufacturing line is typical. A relatively long diode cavity aids in heat removal, extends diode life, and reduces slow-axis divergence. The emitter spacing all but thermally isolates one emitter from another permitting optimal power generation while, here again, extending diode life and reducing divergence.

Not covered in the above sections is one of the more practical but difficult design challenges — that of removing waste heat from the module. For the IS46, the diodes are mounted to a direct-copper-bond (DCB) heat sink that flows water through the structure. The design is easily mass produced at low cost, permits the water to get very close to the diode surface for efficient heat removal, and weighs just 30 grams. A total of four T-Bars can be mounted to a single DCB, resulting in an optical building block of 200W raw optical power. Only two such DCBs are needed in the module. Most of the module weight results from the backbone needed to support the boards, making it easily scalable in power without concomitant weight increase.

The T-Bars are then micro-lensed with fast- and slow-axis lenses. The subsequent optical train expands the beam and includes several folding mirrors to keep the module envelope smaller than the volume enclosed by a soup can. The diode mounting and optical alignments are performed on automated machinery that increases throughput, repeatability and yield while reducing cost. Figure 4 provides the optical performance of the IS46.


The ideal fiber coupled module is one where a single, 100% efficient diode laser of perfect beam quality is abutted and coupled without loss to an optical fiber without the need for any intermediary beam combining or lensing operations. Diode laser engineers continue to push the state of the art in high-power, high-brightness, high-efficiency diode lasers with each year bringing us ever closer to that idealized single emitter.

In the meantime, optical engineers find ever more clever ways to combine beams with ever decreasing loss. In order to house such modules, mechanical engineers invent new ways to manage the inevitable lost power with greater efficiency at reduced cost and weight.

This article was written by Tobias Koenning, David Irwin, Steve Patterson, PhD., and Jens Biesenbach, PhD., DILAS Diode Laser, Inc. (Tucson, AZ). For more information, contact Dr. Patterson at This email address is being protected from spambots. You need JavaScript enabled to view it., or visit .