What separates a coating that holds its spectral curve for years from one that drifts after a humidity cycle — and how to specify the difference.

Short answer

An AR coating works by interference: alternating high- and low-index layers a quarter-wavelength thick cancel reflection at the design wavelength. Whether that stack survives depends less on the design than on how it was deposited. Coatings deposited at 380–420 °C are dense, hard, and spectrally stable. Low-temperature coatings are porous, absorb water vapour, and shift their curve in service — which reads as unexplained sensor drift. When you compare suppliers, ask for deposition temperature, monitoring method, and a humidity-shift test result.

What the coating does, numerically

Sapphire's refractive index of ~1.77 creates about 7.7% reflection per uncoated surface. Two surfaces cost roughly 15% of the light before any absorption.

An ideal single-layer coating needs an index of √1.77 ≈ 1.33 — remarkably close to magnesium fluoride (1.38). That is why single-layer MgF₂ is the classic sapphire AR, and why it delivers roughly ≥89% single-sided.

Production stacks go further, using alternating high- and low-index materials:

A well-controlled multilayer stack brings double-sided transmission to ≥95% across the visible, and higher in narrow bands.

Why substrate temperature is the real specification

Deposition temperature determines film density, and film density determines almost everything else:

A porous film absorbs water into its voids. Water has an index of 1.33, so the effective index of the layer rises, and the quarter-wave condition moves — the curve shifts toward longer wavelengths. In a device measuring at 940 nm near the edge of a coating band, that shift is a direct transmission loss.

This is why "≥95%" without a deposition temperature is an incomplete specification. Two suppliers can meet the same initial number and behave completely differently after six months in a humid climate.

Process control: what to ask about

A supplier using both crystal and optical monitoring can hold the curve on the substrate rather than on a witness tool — a meaningful difference for curved or small parts.

The tests that reveal coating quality

Ask for these as reports with numbers, not as "passed". A test without a measured before/after value is not evidence.

Coating design versus measurement band

A stack optimised for "visible 400–700 nm" is a common default and a common mistake for PPG:

• It covers 525 nm and 660 nm well

• It typically rolls off before 940 nm

• SpO₂ depends on 940 nm

So specify the bands explicitly and ask for transmission at those bands, at the AOI range. If the supplier's stack is broadband to 1000 nm, ask what that costs in peak transmission at 525 nm — there is always a trade.

Masking, edges, and the parts coating cannot fix

Coating is applied after shaping and polishing. It cannot compensate for:

• Edge chips — they scatter regardless of coating

• Subsurface damage from coarse polishing — it scatters from beneath the film

• Poor flatness — it changes the AOI distribution across the aperture

If a window fails optical validation after coating, the cause is often upstream. This is the practical argument for a supplier that controls growth, shaping, polishing, and coating in one plant: the feedback loop between polishing quality and coating performance stays inside one process owner.

Cost, honestly

Coating is a small share of the piece price for a large window and a significant share for a tiny one, because loading, masking, and handling dominate at small sizes. Two cost levers matter more than the stack itself:

1. Clear aperture definition — allowing cosmetic tolerance outside the optical zone can cut polishing and inspection time.

1. Masking method — printed ink versus mechanical masking changes both yield and cycle time.

Neither is visible in a price comparison unless you ask.

FAQ

How many layers does a PPG window AR coating have? Typically 4–8 layers for broadband visible, more when the design must hold 940 nm as well. Layer count is a design choice; density and adhesion are process choices.

Does AR coating reduce scratch resistance? A dense oxide stack is harder than bare sapphire's surface in abrasion terms, but a thin film can still be worn through locally. AF on top reduces friction and cleaning damage, which is why the AF layer is specified for skin-contact windows.

Can AR and AF be combined? Yes — AR first, AF as the outermost layer. The AF film is thin (nanometre scale) and does not materially change the AR curve, but it does change surface energy, which is the point.

Will the coating survive ultrasonic cleaning? Depends on adhesion and density. Specify the cleaning process in the RFQ; a dense, high-temperature coating generally survives standard IPA and mild ultrasonic cycles, while a porous coating may not.

Why does my supplier's transmission differ from the sample report? Common causes: different AOI, different band definition (peak vs average), measurement on a witness rather than the part, or a coating batch drift. Ask for the curve and the measurement conditions.

Talk to the sapphire engineers

Sun Yin Crystal Industry Company Limited has coated sapphire since 1994 — AR deposited on Japanese Showa optical coating equipment with dual crystal and optical monitoring plus ion assistance, at 380–420 °C. Typical result: ≥95% double-sided transmission, with AF on top at ≥115° water contact angle, still ≥100° after 15,000 abrasion cycles.

https://www.sunyinsapphire.com/
SUN YIN CRYSTAL INDUSTRY COMPANY LTD