Scientists at the Soreq Nuclear Research Center in Yavne, Israel, have previously demonstrated that cylindrically polarized light can be amplified in solid-state rod amplifiers with higher efficiency than can linearly polarized light. When linearly polarized light is amplified in a solid-state rod, as much as half of the amplified light is converted to the orthogonal polarization by the rod’s thermal birefringence. To obtain a linearly polarized output, this orthogonally polarized light must be filtered out and discarded.Cylindrically polarized light, on the other hand, can be amplified in a rod with much less polarization scrambling because the E-vector in cylindrically polarized light lies along a radial or azimuthal direction and is always aligned either parallel or perpendicular to the rod’s thermal gradient — and therefore does not see the rod’s thermally induced birefringence. In the past, the problem has been how to efficiently create the cylindrically polarized input light for the amplifier.The scientists in Israel have now shown a straightforward and efficient extracavity technique to convert a linearly polarized Gaussian beam into a cylindrically polarized doughnut beam. (The doughnut beam, a linear combination of TEM01 and TEM10 beams, is often identified as the TEM01* beam.) They are not the first to address this issue, but their result represents a significant advance over previous work.Figure 1. The pie-shaped segments of a half-wave plate were assembled with their slow axes oriented as shown by the arrows (left). The fabricated disk was antireflection-coated at 1.06 μm (right). Images reprinted with permission of Optics Letters.Their converter consists of eight pie-shaped segments of a half-wave plate forming a complete disk (Figure 1). The slow axis of each segment is oriented as shown in the figure, so that vertically polarized light, for example, is not rotated at all in the segments at 6 and 12 o’clock. Vertically polarized light passing through the segments at 3 and 9 o’clock is rotated by 90°, and vertically polarized light passing through the other four segments is rotated by ±45°.Figure 2. When illuminated with vertically polarized light, the disk in Figure 1 converts light to a close approximation of radially polarized light (a). In the far field, the approximation becomes even better (b).In other words, the light emerging from the half-wave disk is polarized in an approximation to radial polarization (Figure 2a). In the far field, the polarization distribution resembles true radial polarization even more closely (Figure 2b).In an experimental demonstration of the device, the scientists fabricated a disk of antireflection-coated half-wave-plate segments and illuminated it with 50 mW from a linearly polarized Nd:YAG laser. By Fourier transforming the near-field intensity, they had calculated that the far-field intensity distribution would be very close to a doughnut mode, and that is precisely what they observed (Figure 3a). Rotating a linear polarizer in the far field reveals that the light is radially polarized (Figures 3b-e). About 10 percent of the transmitted light is present in the eightfold star halo around the beam, as can be seen in Figure 3a. Spatially filtering the light after it passes through the half-wave disk cleans up most of this stray light (Figure 3f).Figure 3. In the far field, the light transmitted through the disk in Figure 1 closely resembles a doughnut beam (a). Viewed through a vertical polarizer (b), a polarizer oriented at +45¼ from vertical (c), a horizontal polarizer (d) and a polarizer oriented at –45¼ from vertical (e), the light is strongly radially polarized. Spatialfiltering cleans up most of the light in the eightfold star halo in the first image (f).The incoming linearly polarized beam had an M2 value of ∼1.3, and the converted beam’s M2 was ∼2.51, reasonably close to the calculated value of 2.19. Eighty-six percent of the incoming light was converted to radial polarization, close to the predicted maximum value of 89 percent. Conceptually, two of these half-wave disks could enable efficient amplification of linearly polarized light in rod amplifiers. The low-level linearly polarized incoming light would be converted to a cylindrical polarization, which — unlike linear polarization — could be amplified without scrambling in a rod whose thermal gradients had perfect cylindrical symmetry. After amplification, a second half-wave disk would convert the high-power beam back to a linear polarization. The second half-wave disk will have to sustain multikilowatt power levels. Issues associated with fabricating such a disk are currently being addressed at Soreq.Optics Letters, June 1, 2007, pp. 1468-1470.