Mode-Matching Technique May Enable Broad Telecom Bandwidths

Breck Hitz

Erbium-doped fiber amplifiers (EDFAs) are widely credited with enabling the huge growth of fiber optic communications during the last decade of the 20th century. Because they could boost the amplitude of optical pulses traveling in fiber without converting those pulses to electrical signals, they extended the range of fiber optic links to many hundreds of kilometers. But the EDFA gain bandwidth, limited primarily to the C-band between 1530 and 1565 nm, restricts the bandwidth of most of today’s optical networks.

New fiber can be put in place to meet the ever-growing demand for bandwidth, but it would be much more efficient to increase the capacity — the bandwidth — of existing fiber, which transmits effectively at wavelengths between 1300 and 1600 nm. And several years ago, researchers in Taiwan demonstrated chromium-doped fiber amplifiers whose gain bandwidth covered essentially the same 1300- to 1600-nm range as optical fibers’ transmission window.

But the laser-heated pedestal growth method they used to fabricate the fiber made it nearly impossible to fabricate fiber cores smaller than 10 μm or to maintain a uniform core diameter over a length of the fiber. As a result, the chromium-doped fibers caused huge insertion losses when inserted into conventional single-mode telecom fibers. These losses were the deal-stopper: They made it unrealistic to use the chromium-doped fiber in telecom systems.

Figure 1. The mode propagating in the single-mode fiber (SMF) excites many modes in the multimode chromium-doped fiber (MMCDF), and these modes interfere destructively, resulting in an enormous loss. This loss makes it impractical to integrate chromium-doped fiber amplifiers into conventional optical networks. Images reprinted with permission of Optics Letters.

Now, scientists at National Sun Yat-Sen University in Kaohsiung and at National Taiwan University in Taipei, both in Taiwan, together with others at the University of California, Santa Barbara, have demonstrated a technique of inserting the multimode, chromium-doped fiber into single-mode fiber link without the losses of previous methods.

Figure 2. A numerical calculation showed that coupling efficiency greater than 90 percent across a broad spectral region is possible if the chromium-doped fiber core is approximately 14.5 μm.

The insertion loss that usually results when the chromium-doped fiber is coupled to single-mode fiber is the result of mode mismatch between the fibers. The mode propagating in the single-mode fiber excites many modes in the larger chromium-doped fiber and, as those modes propagate, they get out of phase with each other and interfere destructively (Figure 1). To avoid this, the scientists calculated numerically the coupling efficiency between the two fibers as a function of the diameter of the chromium-doped fiber. They found that a certain diameter resulted in only one of the many possible modes being excited in the chromium-doped fiber. As a result, good coupling efficiency was possible with the appropriate core diameter (Figure 2).

Figure 3. Experimental data showed that the coupling efficiency varied between 60.3 and 87.7 percent between 1300 and 1600 nm. The experimental data for a chromium-doped fiber with a 15.5 ±3.4-μm core are shown as individual points here. The solid and broken lines are calculated results for 12- and 19-μm cores, respectively.

To confirm their calculations, the scientists sandwiched a 1-cm length of the chromium-doped fiber between two sections of single-mode fiber, as diagrammed in Figure 1. Rather than splice the fibers, they used an index-matching fluid between them to reduce the Fresnel reflections. The 1-cm length of chromium-doped fiber had a core diameter of 15.5 μm, with an approximate variation of ±3.4 μm along its length. They observed that the coupling efficiency varied between 60.3 and 87.5 percent across the 1300- to 1600-nm spectral range (Figure 3).

Optics Letters, April 15, 2008, pp. 785-787.