The magneto-optical (MO) oxide layer consists of (Bi,Y)3Fe5O12 or BiYIG, bismuth garnet. This material was selected because it has a better figure of merit than the CeYIG previously used, especially at lower wavelengths (1310 nm vs. 1550 nm). A top-down deposition process was developed in which BiYIG/YIG stacks are grown on the Si waveguide with YIG on top. The stack is annealed at 800°C/5 min to crystallize both layers, with the YIG templating the BiYIG leading to garnet phases rather than other oxides, and the BiYIG is directly on the Si waveguide. Initial attempts led to a film with Bi oxide phases, because the Bi was in excess and could not escape during the anneal as occurs in Si/YIG/BiYIG stacks. Hence the composition was adjusted to include slightly more Fe, which yielded films with only garnet peaks.

Conditions were also developed for growth of Bi-substituted iron garnet (BiYIG) on GGG garnet and Si substrates by combinatorial pulsed laser deposition.

Figure 1. Single Crystal BiYIG/GGG. Top left: XRD of two recent films on (111) GGG showing the effect of adding Fe. Right: Magnetometry data showing in plane (IP) and out of plane (OP) hysteresis loops for 100 nm thick Bi.97Y2.47Fe5O12 grown at 560oC, 20mTorr, 400mJ, on GGG (111). Lower left: OP Faraday loop of same film. Right: FR vs wavelength, in agreement with measurements on bulk BiYIG

For single crystal films on GGG (Figure 1), the conditions for growth that produced films with the best saturation magnetization (Ms) and surface topography were found to be at higher temperatures of 520-560°C and at oxygen pressures of 10 to 20 mTorr. Structural characterization revealed the growth of epitaxial BiYIG film on GGG without any secondary phases. This result was further confirmed by compositional analysis that showed the ratio of Bi+Y/Fe, as expected, was approximately 0.6 (in the range 0.62-0.65) suggesting no formation of secondary ferrous phases. The FR was 1.5 °/μm, which is comparable to other work considering the Bi content. The saturation field for out of plane hysteresis or Faraday loops is ~2 kOe which is close to that expected just from shape anisotropy, i.e. magne-tocrystalline or magnetoelastic contributions to anisotropy are probably small.

For polycrystalline films on Si (Figure 2) top-down crystallization of BiYIG using a YIG seed layer on topographical substrates was carried out to promote the crystallization of BiYIG on photonic substrates. A bilayer was grown (YIG/BiYIG/substrate) at 650°C, then annealed at 800°C. With the top seed layer, Bi escape during annealing was suppressed and the composition had to be adjusted (less Fe was added) to avoid secondary phases.

Figure 2. Polycrystalline Films on Si. Left: XRD of polycrystalline BiYIG film on Si grown with YIG bottom seed layer, showing the characteristic garnet peaks. Right: Scanning electron microscopy (SEM) of BiYIG grown on the sidewall of a SiN waveguide in a TE-mode isolator. The fabrication of the TE mode isolator is shown in the schematic.

X-ray diffraction (XRD) showed crystallization to the garnet structure and the saturation magnetization was consistent with the film thickness and the bulk magnetization of YIG and BiYIG (which are similar). However, films on Si had much weaker FR than expected.

During this work, the pulsed-laser deposition (PLD) system was reconfigured leading to a higher intensity of light incident on the target and higher growth rates, which led to a change in composition of most materials deposited by PLD. Compositional analysis showed that more recent BiYIG films contained less Bi than before, and this may account for the lower FR. The target, which contained Bi:Y:Fe = 0.8:2.2:5, yielded films of 0.5:2.5:3.8, or 0.5:1.9:4.2 when additional Fe oxide was code-posited. (In the latter case, Bi+Y/Fe = 0.57 which matches the stoichiometric ratio of 3/5 = 0.6.)

The results of growth experiments indicate that films grow with garnet crystal structure on GGG, even if the Bi+Y/Fe stoichiometry is not exactly correct. However, growth on Si is less forgiving, and making good quality garnet requires a closer control of stoichiometry. The Bi:Y ratio is controlled mainly by temperature, and the Bi+Y/Fe ratio is sensitive to laser power.

This work was done by Caroline Ross and Juejun Hu, Massachusetts Institute of Technology for the Air Force Research Laboratory. AFRL-0264

This Brief includes a Technical Support Package (TSP).
Integrated Magneto-Optical Devices for On-Chip Photonic Systems

(reference AFRL-0264) is currently available for download from the TSP library.

Don't have an account? Sign up here.

Aerospace & Defense Technology Magazine

This article first appeared in the September, 2018 issue of Aerospace & Defense Technology Magazine.

Read more articles from this issue here.

Read more articles from the archives here.