When fiber Bragg gratings (FBGs) are written into all 120 cores of a multicore optical fiber, great things can happen, as researchers from the University of Bath (Bath, England), Universidad de Valencia (Burjasot, Spain), and the Waterford Institute of Technology (Waterford, Ireland) know. Specifically, such an arrangement can become a “photonic lantern” filter that serves as an ultranarrowband line blocker for astronomy.1
Photonic lanterns couple the functions of single-spatial-mode optical devices (in this case, FBGs) to multimode fiber inputs and outputs. To do this, they combine many singlemode fibers together (or in some cases, rely on integrated photonics). Photonic lanterns based on the combination of many singlemode fibers with FBGs have been constructed with as many as 61 identical FBGs, with at least one already under test with a telescope.2
But these existing photonic lanterns consist of many identical devices fabricated separately and then placed in parallel, spliced to singlemode fiber pigtails. The new device avoids all this trouble, being made instead as a single large fiber with many singlemode cores, all with individual FBGs that are written in one process. The two ends of the large fiber are then tapered to match the input and output multimode-fiber ports (see figure).
Mode-number matched
For low loss, the fiber device must contain at least as many singlemode cores as there are modes in the input multimode fiber (note: the singlemode cores are nonpolarizing, so each actually supports two polarizations). For low loss at the outgoing end, the output fiber must support at least as many modes as there are singlemode cores in the fiber device. As a result, for input and output fibers with identical diameters (thus supporting identical numbers of modes), there is a certain number of single-mode cores that minimize mismatch losses: This is called a mode-number-matched design.
The experimental design, which operates in the 1550 nm spectral region, contains 120 germanium-doped, 3.9-μm-diameter cores with a 0.22 numerical aperture and a hexagonal grid spacing of 16.9 μm, all encased in a 230-μm-diameter silica fiber.
Single-spectral-notch FBGs were created in a single operation using a 244-nm-wavelength, 100 mW laser beam passed through a phase mask with a 1067 nm period to perpendicularly illuminate the multicore fiber, which had been hydrogen-loaded for two weeks beforehand under a 20 bar pressure.
The resulting single spectral-loss notch was at about 1549.1 nm. Because the line-center wavelength for each singlemode fiber core depends on the core’s diameter, the diameter tolerance of the cores was kept at 1.3%, which results in a wavelength spread of 160 pm or less.
The researchers fabricated numerous multicore-fiber samples and tested the FBG notch depth and notch wavelength for every core in every sample, with two of them having a wavelength-spread standard deviation of 100 and 67 pm, respectively. Samples with stronger gratings had more wavelength spread, showing that the spread was due more to imperfect FBG fabrication than to differences in core diameters.
Because the fabrication process tended to result in some singlemode cores not having FBGs, a new multicore fiber was made with a larger uniform silica outer cladding, which ensured that the UV light used to write the FBGs did indeed reach every single-mode core, although some cores were shadowed by other cores, thus reducing the quality of some FBGs.
Finally, the multicore fiber was tapered at both ends to transition to the input and output multimode fibers. At a small enough taper, the light leaves the singlemode cores and bounces around in the cladding; thus, a lower-refractive-index additional outer cladding was added to the tapers to keep the light confined.
After mounting, a spectral scan was taken across the region of interest, showing a spectral notch with a depth of about 7 dB. The relatively small size of the dip was chalked up to the fact that not all FBGs were of good quality in these first experimental prototypes. The coupling loss between the test device and the corresponding input and output multimode fibers was tested and found to be less than 0.5 dB.
In an interesting side note, the researchers found that a similar device, but with no FBGs, can serve as an effective mode scrambler, being far more effective than an equivalent length of simple multimode fiber.
REFERENCES
1. T. A Birks et al., Opt. Exp., 20, 13, 13996 (June 18, 2012).
2. J. Bland-Hawthorn et al., Nat. Comm., 2, 581 (2011).
