Index-matching gel

Index-matching gel is a viscous, transparent material, usually a silicone or hydrocarbon compound, with a refractive index close to that of the glass it touches. Placed between two fiber ends or optical surfaces, it replaces the air gap and removes most of the index step, and with it most of the reflection and loss. Gels for silica fiber are made with indices of about 1.45–1.46 near 1550 nm; liquids (index-matching oils) and UV-curing epoxies of similar index serve the same purpose where a gel is impractical.

Reflection with and without gel

At normal incidence the reflectance of an interface between indices $n_1$ and $n_2$ follows from the Fresnel equations:

$$R = \left(\frac{n_1 - n_2}{n_1 + n_2}\right)^{2}$$

A fiber core of phase index 1.45 against air gives $R$ = 3.4%, a return loss of 14.7 dB. Against a gel of $n$ = 1.46 the same face gives

$$R = \left(\frac{0.01}{2.91}\right)^{2} = 1.2\times10^{-5},$$

a return loss of 49.3 dB. The phase index is the relevant one; the group index of standard fiber, about 1.468 at 1550 nm, does not enter the reflection. Even a poor match, 1.40, reflects only 0.031% (35.1 dB).

The gain in transmission is smaller but useful. An air gap between two cleaved ends has two silica–air faces; adding their reflected powers, the pair transmits 93.4%, a loss of 0.30 dB. When the gap is short compared with the coherence length of the source, the two reflections interfere as in a Fabry–Perot etalon and the reflected fraction swings between nearly zero and about 13% as the gap changes by a quarter wavelength. Gel removes both the loss and this sensitivity to the gap.

Uses

• Mechanical splices. Two cleaved fibers are aligned in a V-groove or sleeve with gel between them. The joint loses a few tenths of a dB and needs no splicer; fusion splicing gives lower loss and better long-term stability for permanent links.
• Temporary connections. Bare fibers butted in a gel-filled groove allow quick measurements without connectors, for example in a cutback measurement.
• Terminating open fiber ends. An unused cleaved end immersed in gel, often after a tight bend or an angled cleave, no longer sends 3.4% back toward a laser or amplifier. In optical time-domain reflectometry this suppresses the large end reflection and the dead zone that follows it.
• Fiber-to-chip coupling. A drop of index-matching liquid at a chip facet or under a fiber array removes the gap reflection and its etalon ripple.
• Permanent attachment. In pigtailing and fiber-array attach, UV- or heat-curing optical epoxies with indices chosen near the glass fill the optical path and hold the parts. The cured index differs somewhat from the liquid value.
• Microscopy. Immersion oil of $n$ = 1.515 matches cover glass and lets an oil microscope objective reach a numerical aperture above 1. Against N-BK7 ($n$ = 1.5168) the residual reflection is about 3.5 × 10⁻⁷, compared with 4.2% for N-BK7 in air.

Pitfalls

Temperature. The index of a gel changes far faster with temperature than that of silica: of order −3 × 10⁻⁴ to −4 × 10⁻⁴ per kelvin for silicone gels, against about +1 × 10⁻⁵ per kelvin for fused silica. Taking −3.5 × 10⁻⁴ K⁻¹, a gel matched at 1.450 at 25 °C drops to about 1.429 at 85 °C and rises to about 1.473 at −40 °C, and the reflection grows from essentially zero to about 6 × 10⁻⁵ (42 dB return loss) at either end. Datasheets accordingly quote the gel index at one stated temperature and wavelength.

Contamination and voids. Trapped dust scatters and absorbs, air bubbles bring back two glass–air faces, and gel left on connector end faces attracts dirt.

Aging and outgassing. Gels can dry, migrate or yellow over years and some outgas in vacuum, so gel-filled joints drift more than fused or epoxied ones.

Hidden reflections. Gel removes the facet reflection, which defeats measurements where that reflection is the quantity of interest, such as optical return loss qualification of a connector.

Common questions

What refractive index should an index-matching gel have for optical fiber?

Close to the phase index of the silica core near the working wavelength, about 1.45–1.46 at 1550 nm. A mismatch of 0.01 leaves a reflection near 10⁻⁵, and even a mismatch of 0.05 keeps it below 0.04%.

Is index-matching gel the same as immersion oil?

They work the same way, but immersion oil is formulated with $n$ = 1.515 to match cover glass and microscope optics, and with controlled dispersion and low fluorescence. On a silica fiber core ($n$ = 1.45) it reflects about 5 × 10⁻⁴ (33 dB return loss), well below the 3.4% of air though roughly forty times more than a gel of 1.46.

Does index-matching gel reduce insertion loss?

Yes, mainly by removing the two Fresnel reflections of an air gap, about 0.3 dB. It does not correct lateral offset, angular misalignment or mode-field mismatch between the two fibers.

References: E. Hecht, Optics, 5th ed. (Pearson, 2017); B. E. A. Saleh and M. C. Teich, Fundamentals of Photonics, 3rd ed. (Wiley, 2019); G. P. Agrawal, Fiber-Optic Communication Systems, 4th ed. (Wiley, 2010).