Advanced optics expand the applications of high power diode lasers
High power diode laser bars can replace conventional Nd:YAG lasers in several applications, but only if special optics are used for the shaping of their emission beam. The beams are notoriously asymmetric and astigmatic, as shown in Figure 1. When focused, their elliptical beam profiles are greatly elongated, instead of circular, which is undesirable for most applications.
Beam shaping (BS) technology strives to address this problem by improving the optical quality of the diode laser beams. Most BS techniques use several complex optical elements, hard to mount and align. Using a non-classical design, we have recently developed a system that achieves beam equalization with only two elements.
A widespread solution to the BS problem in continuous wave (CW) bars is to couple each emitter with cylindrical microlens pairs.1 This popular and compact solution works efficiently for CW diodes that have up to 20 emitters per one-centimeter bar. It is, however, not suitable for quasi-CW (QCW) bars that contain about three times as many emitters. Recently, the output power of QCW bars has also been increased to 200W per bar, with the result that they have become more attractive for direct applications. Other solutions,2–5 such as the stepped mirror2 and mirror-pair techniques,3 address the BS problem efficiently, but also require several components for complete beam correction.
We have solved the BS problem with a compensated beam transformation of a unique design. Our scheme is completely independent of emitter geometry, and can therefore be used with both CW and pulsed diode bars. Moreover, to increase the pulse energy, it can be used in series with multiplexing devices, such as polarization and wavelength coupling, or with beam compression stages, to add more laser sources. As in the case of stepped mirrors, the beam emission is split into several parts in the elongated axis (fast axis), and recomposed in such a way that the overall brightness is the same in both fast and the orthogonal slow axes.
The number of subdivisions is chosen so as to equalize the beam parameter product (a laser beam quality index) of each axis.2 The transformed beam thus consists of several virtual emitters, each lying in the same plane of the slow axis and infinitely conjugated in the fast axis.
Our solution has the advantage of keeping the optical paths of each sub-beam constant. In this way, the brightness of the spot is preserved by avoiding the defocusing effects that occur with other BS techniques. The core of our optical system has only two elements: a divider and a recombiner. Each element has a tilted micromirror array in which each mirror face is plane and parallel to one in the other element (see Figure 2).
The effect of the plane-parallel micromirrors is to displace the sub-beam axis of propagation that lies in the slow axis to the fast axis. The general condition for the conservation of spot brightness is therefore achievable with this system (see Figure 3). We built an inexpensive prototype by gluing together layers in which the plane parallel micromirrors were made by grinding glass on a mandrel with calibrated grooves (see Figure 4). As seen in Figure 3, we obtained a beam spot size of 320 × 250μm2 with a numerical aperture of 0.28.
Smaller spots can be obtained either by changing the demagnification factor (at the cost of a higher numerical aperture), or by improving the production technique by using diamond turning and plastic molding. In this case, the expected improvement is a spot size of 280 × 120μm2. The size of the complete system (ca. 6cm) depends on the focal length of the focusing elements, and is one third smaller than devices of similar efficiency.
The applications of high power diode lasers may be expanded by achieving higher intensities and smaller beam spots. Several areas may benefit from our BS system, such as microelectronics, materials spot-welding, laser surgery and dentistry6 (see Figure 5).
These studies were financed by the CNR-INFM bridge project, and by the Department of Information Engineering of the University of Padova. Patent Pending TO2005A000727.
