The coherent superposition of laser beams – at least in principle – can solve two major challenges in one go:

1. the scaling of the output power while conserving diffraction-limited beam quality, and
2. the fast – non-mechanical – stirring of the (resulting) laser beam.

This is of particular interest for diode lasers, which can be arranged in large arrays. In this way the limited output power of a single-mode diode can be increased substantially and beam stirring with a large number of resolvable pixels becomes possible.

The coherent beam combining relies on the control of the relative phases of the individual laser beams. Once this is accomplished, the beams can be combined into one beam, e. g., by using beam splitters in reversed direction, or by putting the beamlets side by side in the near field. The latter is a kind of realization of Huygens’ principle, the superposition of elementary waves, originating from a wavefront. From there, it is only a small step to stirring the beam by adjusting the phases of the individual beams (phased array).

Up to now, large phased arrays have mostly been demonstrated starting with a single laser beam followed by passive elements like LCDs for deflecting the beam. Sayyah et al. now report on a phased array based on VCSEL arrays. Injecting a small amount of power from a master laser into each VCSEL will lock it (within a certain limit, the locking range) on the frequency of the master laser. The phase difference between the master and the VCSEL beam depends on the frequency difference between the free-running and the locked VCSEL. So for a given phase (difference) one has to maintain the free-running frequency (difference) of the VCSELs. But what seems to be a complication actually is an advantage, since it allows to easily change the phases of the individual lasers simply by adjusting the injected currents.

Two problems come along with this approach: the maximum phase shift within the locking range is π instead of the ideally required 2 π for beam stirring, and – since the distance of the individual sources is larger than the wavelength – there will be higher order peaks. The authors solved these problems by using two VCSEL arrays in series, thereby realizing a combined phase shift of 1.6 π, and by arranging the VCSELs on the chip pseudo randomly, which eliminates the secondary peaks but, of course, not the power loss.

The authors used a stack of two 64-element VCSEL arrays of 0.3 x 0.3 mm² with an average single-mode output power of 2.5 mW per element. Fifty elements could be used for the superposition, 30 mW (24 % of the available output power) could be found in the nearly diffraction-limited central beam, the background stayed below -7 dB. 2-D beam stirring over a field of 2.2° x 1.2° was realized with a subset of 16 elements. Again, 24 % of the available output was found in the stirred beam.

Scaling to much larger arrays using this approach is limited

1. by the power of the master laser – 50 mW should be sufficient for 10,000 lasers,
2. by the necessity of addressing each VCSEL of the array (from the edge), and
3. probably by an increased accuracy and complexity of the phase control – if the brightness and the number of resolvable pixels is to scale with the number of elements.

If, depending on the application, more power or brightness is needed, one could think of cascading several of these VCSEL arrays. Also, with additional beam forming, it should be possible to increase the percentage of power in the stirred beam, if needed.

---

---

 Add Comment

You must log in to add comments.