Single-crystal Fibers


The IR absorption of most oxide crystals is well documented in the literature. In Fig. 1 we show the IR absorption edge for four different oxide crystals: sapphire, YAG, GGG, and yttria. The absorption at about 4.8 µm in YAG is due to multiphonon absorption and, therefore, an intrinsic property of the material. Yttria also has a weak multiphonon peak at 5 µm. As expected the IR edge is shifted toward longer wavelengths for oxides with the heaviest ions. For example, if we extrapolate from the IR absorption data the expected absorption for a fiber made from these materials, we would find that the loss for pure sapphire, YAG, GGG, and yttria is 18, 5.5, 0.5, and 0.15 dB/m at 4 µm, respectively. Based on these results it is clear that the only viable SC fibers for use at 4 µm from one of these four crystals would be GGG or yttria. The difficulty with yttria is that it has a phase transition at about 100°C below the melting point and thus it is hard to grow a single crystal of this material. We have tried to grow an yttria fiber starting with a polycrystalline (ceramic) source rod. In all cases we obtained small cracks in the outer diameter of the fiber indicative of the effects of the phase change in this material.

Figure 1 - IR absorption edge for bulk oxide crystals. Note the excellent transmission of GGG and
yttria at longer wavelengths.

Loss for SC sapphire fiber

Loss measurements of our SC sapphire fibers were made at discrete wavelengths with Ar-ion, He–Ne, and Er:YAG lasers. All the lasers (except for the 1.15-mm He–Ne) had single-mode outputs or were apertured to produce nearly a single mode. Broadband, spectral loss measurements were made with a combination of an Ando spectrum analyzer, a Perkin-Elmer Lambda 9 spectrometer, and a Perkin-Elmer Model 1725X FTIR spectrometer with a LN2-cooled InSb detector. All spectral and laser measurements were made by the cut-back method to remove coupling losses.

The spectral attenuation of one of our best 300-mm diameter, 1-m long sapphire fibers is shown in Fig. 2. This fiber was grown under computer control at 2 mm/min in air and had diameter fluctuations less than ±0.5%. In addition, the fiber was annealed in air at 1000°C for 12 h after growth. The losses are higher than the bulk data of Innocenzi et al. below 3 mm. However, at 2.94 mm the loss in this fiber is ~0.3 dB/m, which is approaching the theoretical limit of ~0.13 dB/m. This is one of the lowest losses obtained at 2.94 mm in a SC sapphire fiber. We attribute this low loss to the elimination of crystal imperfections caused by instabilities in the growth process and good diameter control.

Figure 2 - Three micrometer absorption peaks in our lowest-loss sapphire fiber.

Loss for SC YAG fiber

The spectral loss for our YAG fiber is shown in Fig. 3. The loss measurement was taken on a Nicolet Protégé 460 FTIR spectrometer. The fiber diameter was 400 µm and the length 5 cm. As mentioned above, the absorption near 4.8 µm is due to multiphonon absorption. The two small peaks around 3.5 µm are associated with OH defects. Specifically, the YAG structure in the fiber has distortions or oxygen-defects (vacancies) that give additional locations for OH producing additional hydrogen-bond stretch lengths (frequencies). The two strong absorption lines at 3.45 and 3.54 µm are very near hydrogen-bond absorptions for the dodeca-octahedral and dodeca-tetrahedral bond lengths which are not observed for a bulk YAG crystal source rod. In addition, the absorption lines at 2.944 and 2.972 µm in both the YAG fiber and feed crystal are well documented to be due to hydrogen-bond absorption. The losses for a 35-cm length of YAG fiber measured using green and red visible lasers and an Er:YAG laser at 2.94 µm are summarized in Table 1. These measurements were using a cutback technique to remove reflection and coupling losses. The loss of 3 dB/m at 2.94 µm is much higher than the measured loss of 0.4 dB/m for sapphire at the same wavelength. We would not expect the loss for YAG to be this high and the work continues to improve the loss.

Figure 3 - Spectral loss for SC YAG fiber.

Loss for SC spinel fiber

Spinel fibers composed of a 50/50 mix of MgO and Al2O3 were grown into fibers as long as 20 cm. The spectral response of a 400 µm diameter by 4 cm long spinel fiber is given in Fig. 5. In the case of spinel we see strong water absorption at 3 µm and similar OH contamination in the 3.5 µm region as seen in Fig. 4 for YAG. Not surprisingly, the loss for the spinel fiber at 2.94 µm is a very high 9 dB/m.

Figure 4 - Spectral loss for SC spinel fiber.

Loss for SC GGG source rod

GGG is an attractive SC fiber crystal. It is cubic and, as may be seen in Fig. 1, it has good long wavelength transmission. The source bars of GGG that we obtained for fiber growth had dimensions of 2 mm × 2 mm × 5 cm in length. The spectral loss for a GGG source bar is shown in Fig. 5. From this data we see the excellent IR transmission also displayed in the data in Fig. 1. The material seems relatively free of water absorption but again we see evidence of the OH-related defects near 3.5 µm. So far we have not been successful in growing GGG fibers but this work continues.

Figure 5 - Spectral loss for bulk GGG source rod.

Return to SC Fibers: Background | Continue to SC Fibers: LHPG