Published by: Research & Development Department, Technologie Optic.ca Inc., March 2026

Introduction

Optical-fiber attenuation (loss, typically expressed in dB/km) is a primary design constraint in fiber-optic communications because it governs link power budgets, maximum span lengths, and the need for amplification or regeneration. Modern silica fibers can achieve losses below ~0.2 dB/km in favorable wavelength regions, which is why standardized telecom windows and band plans are central to network engineering. A historically significant limitation is the "water peak"—an absorption-driven loss feature near ~1.4 μm that reduced usable spectrum in early single-mode deployments.

Origin of the Water Peak

The water peak originates from hydroxyl-related species (commonly described as OH⁻ ions or OH groups bonded into the silica network, e.g., Si–OH) that absorb optical power via molecular vibrational transitions. These extrinsic absorption mechanisms produce relatively narrow spectral loss peaks superimposed on the intrinsic loss background of silica (Rayleigh scattering at shorter wavelengths and infrared absorption at longer wavelengths). OH-related absorption bands in silica fibers occur near ~1.39 μm, with the fundamental OH vibration in the mid-infrared near ~2.73 μm; the ~1.39 μm feature is an overtone/combination band that intrudes into telecom-relevant wavelengths.

From a standards and measurement perspective, the water-peak region is referenced near 1383 nm; for example, the ITU-T water-peak specification for G.652.D is explicitly anchored at 1383 nm (±3 nm) in its attenuation requirements. Manufacturing pathways that introduce moisture or hydrogen-bearing species are therefore critical: water/hydroxyl incorporation can occur during production steps, embedding OH into the glass and increasing attenuation at the OH absorption peaks. The combined result is a distinctive loss spike around ~1.4 μm (often centered near ~1383 nm), superimposed on the broader attenuation curve, as illustrated in Figure 1.

Impact on Telecommunication Systems

Telecom operation is commonly organized into wavelength bands that partition the low-loss region—O, E, S, C, and L bands spanning roughly 1260–1625 nm. In early single-mode fibers, the OH-induced loss peak around ~1.4 μm effectively split practical operation between the ~1.3 μm window (historically attractive for low chromatic dispersion) and the ~1.5–1.6 μm window (widely used for low loss and high-performance amplification). As a result, the E-band (1360–1460 nm) was less utilized historically because residual OH impurity made its attenuation comparatively high among the O/E/S/C/L bands.

This spectral penalty reduced the usable continuous spectrum for wavelength-division multiplexing. Wavelengths near 1383 nm lie within the broader telecom range yet were uneconomic for wide-range transmission in the presence of high OH loss, limiting multichannel expansion across the full band. The second and third telecom windows were originally separated by a pronounced loss peak around ~1.4 μm due to OH absorption, motivating avoidance of that region until low-OH fibers became available.

Low Water Peak Fibers

Low water peak fibers (LWPF) mitigate this limitation primarily by suppressing OH contamination during preform fabrication and consolidation. A widely documented approach is chemical dehydration during soot-based fabrication (e.g., OVD/VAD-type processes): porous glass is consolidated at high temperature in controlled atmospheres that include chlorine, which reacts with Si–OH to form Si–O–Si linkages and volatile HCl (and oxygen), thereby purging hydroxyl from the glass. This dehydration step can reduce hydroxyl content from several hundred ppm to below ~0.1 ppm.

Comparable principles apply in inside-deposition manufacturing such as MCVD, where dedicated dehydration steps employing chlorine-containing gases reduce OH concentration in deposited soot layers before full densification. The net effect is strong suppression of the OH absorption feature: low water peak fibers can exhibit attenuation near 1383 nm approaching the values seen near 1310 nm (e.g., ~0.35 dB/km).

LWPF performance is formalized in standards. ITU-T G.652.D specifies a maximum attenuation of 0.40 dB/km at 1383 nm (±3 nm) after hydrogen ageing, and constrains attenuation across the 1310–1625 nm range—formalizing "low water peak" behavior for wideband system compatibility.

Implications for Modern Networks

By restoring practical usability of wavelengths that include (and surround) the historical water-peak region, LWPFs enable broader-band operation for coarse wavelength division multiplexing (CWDM) and related access/metro architectures. The ITU-T CWDM grid (Recommendation G.694.2) defines nominal central wavelengths spanning approximately 1271 nm to 1611 nm with 20 nm spacing, supporting up to 18 channels in principle across that range. A key engineering motivation for CWDM is cost: coarse WDM allows simultaneous transmission of several widely spaced wavelengths with separation sufficient for uncooled sources—reducing component complexity relative to tighter grids.

In parallel, fiber standardization aligns with extended-band utilization. ITU-T G.652.D explicitly includes constraints at 1383 nm and sets wideband attenuation specifications consistent with operation through the O/E/S/C/L landscape (approximately 1260–1625 nm when suitable cable and component constraints are met). However, deployed infrastructure heterogeneity remains important: fibers installed before ~2000 can still exhibit high E-band attenuation, and network planners may need to verify water-peak suitability—via attenuation profiling near 1383 nm—before committing to full-band CWDM upgrades.

Conclusion

The water peak is an attenuation feature in silica optical fibers caused by hydroxyl-related vibrational absorption near ~1.4 μm, often referenced around ~1383 nm. Historically, it created a practical spectral gap between the widely used ~1.3 μm and ~1.5–1.6 μm telecom windows, reducing the exploitable wavelength range for multiwavelength capacity expansion. Low water peak fibers address this constraint through improved dehydration chemistry and process control (notably chlorine-based removal of Si–OH during preform consolidation), and their performance is formalized in modern ITU-T standards via explicit water-peak attenuation requirements.

This manufacturing-and-standards evolution is directly relevant to contemporary fiber infrastructure because it underpins cost-effective CWDM channel expansion across a broader fraction of the 1260–1625 nm low-loss region. For network planners and engineers, the key practical implication is that low water peak compliance—as specified in ITU-T G.652.D—must be verified when designing or upgrading systems that require operation across the E-band or the full O-to-L-band range.