Optical glass covers silicate, borate, phosphate, fluoride and chalcogenide families, with refractive index, Abbe number, thermal expansion, transmission band and refractive-index homogeneity (Δn) as the five levers that drive every lens, prism and window callout [S3].
Fused silica dominates any application above 900 °C or below 200 nm, while BK7-style crown glass handles 90% of visible-band lab and imaging optics at a fraction of the cost; the wrong pick usually surfaces as thermal runaway, chromatic blur or wavefront error on the first production lot [S1][S3].
Refractive Index and Abbe Number: The Core Optical Trade
Refractive index (n<sub>d</sub>) and Abbe number (V<sub>d</sub>) move in opposite directions across the Schott/Sumita/O-Hara catalogs, and the right pair is fixed by the wavelength band and the maximum allowable longitudinal chromatic aberration [S3].
For visible-band camera and microscope objectives the working set is typically BK7 (n<sub>d</sub> ≈ 1.517, V<sub>d</sub> ≈ 64) paired with an SF-type dense flint (n<sub>d</sub> ≈ 1.78, V<sub>d</sub> ≈ 26) to doublet-correct longitudinal color; index deviation per melt is held inside ±2 × 10⁻⁴ for precision lenses and looser bands (Δn ±5 × 10⁻⁴) for illumination windows [S3]. UV-grade fused silica drops index to n<sub>d</sub> ≈ 1.458 with V<sub>d</sub> ≈ 68, which is why it is chosen for achromatic UV optics where dispersion control outweighs raw index [S1].
Thermal Expansion and Service Temperature Envelope
Thermal expansion coefficient (α, × 10⁻⁶ /K) and softening point set the temperature envelope, and fused silica at α ≈ 0.55 × 10⁻⁶ /K is the benchmark — roughly one order of magnitude below borosilicate crown (α ≈ 7.1 × 10⁻⁶ /K) [S1].
That single fact drives most of the high-temperature selections: fused silica softens above 1600 °C and survives thermal-shock ΔT > 900 °C on 2–4 mm window sections, which makes it the default for semiconductor wafer inspection, laser cutting heads and furnace viewports [S1]. Borosilicate glasses (Pyrex / BK7 / B270) are rated for continuous service at roughly 200–230 °C with thermal-shock ΔT on the order of 150–180 °C, and that is usually the upper limit for uncooled LED collimator and projector optics.
Transmission Band: UV, Visible, IR or Multi-Spectral
Transmission band sets which glass family is even in the running; lead-free silicate and phosphate glasses cover the 400–2000 nm band, fused silica opens 200–2500 nm, and chalcogenide glasses (As₂S₃, GeAsSe) take over from 2 µm to 12 µm for thermal imaging and CO₂-laser windows [S3].
Internal transmittance at the operating wavelength is the metric to spec, not a generic transmission percentage. A 10 mm thick fused-silica window holds > 90 % internal transmittance at 200 nm and > 85 % at 2500 nm, whereas a 10 mm BK7 plate drops below 80 % below 350 nm and below 30 % above 2400 nm [S1][S3]. For multi-spectral payloads the selection often layers a fused-silica corrector plate ahead of a CdSe or GaAs refractive group, accepting the cost step to keep the full 0.4–12 µm channel.
Homogeneity, Striae and Bubble Grade for Imaging Optics
For lenses, prisms and laser-grade windows the material spec must include refractive-index homogeneity class, striae grade and bubble/inclusion grade — not just n<sub>d</sub> and V<sub>d</sub> [S3].
Schott/Ohara grade "H4" homogeneity (Δn ≤ 2 × 10⁻⁶) is the entry point for interferometric test optics, while "H1" (Δn ≤ 1 × 10⁻⁶) is reserved for high-power laser beam expanders and photolithography stepper lenses; a single striae line in a 50 mm clear aperture can deflect a transmitted wavefront by λ/2 or worse [S3]. Bubble/inclusion grade (0 / 1 / 2 per ISO 12123-style conventions) is the gate for laser damage: 1064 nm Nd:YAG optics typically call for grade 0 (no bubble > 0.1 mm within the clear aperture) to keep damage threshold above 15 J/cm², while illumination optics tolerate grade 2.
Comparison of Main Optical Glass Options
Seven glass families are commonly listed in supplier catalogs, and the selection comes down to four decision criteria: spectral band, max service temperature, cost band and typical application. [S1]
Fused silica wins on UV/IR transmission, thermal shock and chemical purity, at a cost band typically 8–15× BK7 per kg of finished optic [S1]. BK7-type crown glass is the cost baseline for visible imaging. SF-type dense flint handles high-index achromatic doublets. Borosilicate (B270 / Pyrex) is the cheap window and sight-glass default. Phosphate laser glass is specialized for high-peak-power laser rods. Lead-arsenate and chalcogenide glasses cover mid-wave IR; fluorophosphate and fluorosilicate glasses sit in low-index / ultra-low-dispersion niches for apochromatic UV lenses [S3]. For process engineers, the same selection logic maps onto other engineering materials — the industrial ceramic selection criteria reference covers the same cost-versus-temp-versus-purity tension, and a parallel breakdown of nickel alloy grades and service envelopes is a useful template when corrosion or temperature is the gating constraint rather than optics.
Chemical, Mechanical and Radiation Resistance
Acid resistance, alkali resistance, Knoop hardness and radiation darkening set the service envelope where the optical metrics alone do not decide the call [S1].
Fused silica holds acid class SR 1 (DIN 12116) and alkali class AR 1 (ISO 695) thanks to > 99.9 % SiO₂ purity, and that purity is what makes it the default for semiconductor wet-etch bath windows, pharmaceutical sight glasses and high-purity fluid sight-glass bodies [S1]. BK7 sits in the mid-range for chemical resistance and is often ruled out of pharmaceutical skids in favor of fused silica or quartz because of leaching risk on long-term steam-in-place (SIP) exposure. For radiation-rich environments (nuclear cell windows, space optics, high-energy physics beamlines) cerium-stabilized radiation-resistant silicate glass is the call, with radiation-induced transmission loss typically held under 2 % after 10⁶ rad gamma dose; this is the same selection logic used for sight glass bodies in pressurized or irradiated service.
Sourcing, Standards and Lead-Time Reality (2026)
Three reference documents govern most optical-glass procurement: ISO 12123 (test methods for homogeneity, striae, bubbles), ISO 13384 (dimensional tolerances on optical components) and MIL-G-174 (military glass classification, where relevant). For flame-fused or wet-chemical process control, the Schott TIE series catalogs and Ohara/HOYA technical sheets are still the engineering reference; China-based suppliers (C-Laser, Suzhou Hugongshan Trading and similar vendors) now carry the standard BK7 / B270 / UV-fused-silica blanks with 4–8 week lead times, while custom n<sub>d</sub>/V<sub>d</sub> pairs still run 12–20 weeks [S2].
Verify each shipment on three datapoints: catalog part number with melt lot, measured n<sub>d</sub> and V<sub>d</sub> from the supplier's certificate, and a bubble/striae inspection report. For high-end projects, run a 633 nm interferometer wavefront check on a witness sample per ISO 12123 before blank release; the cost of that gate is trivial compared with a re-polish on a 200 mm optic.
The most likely next move is the gradual substitution of BK7 by low-auto-fluorescence fused silica in 266 nm / 355 nm laser optics, driven by lifetime-cost data, and tighter Q4 2026 stock availability on SF-type dense flints as several Chinese blank-makers consolidate melt capacity. Track the Schott, Ohara and HOYA quarterly catalogs and the China Optics Valley (Wuhan / Changchun) trading-hub stock lists for lead-time shifts above 4 weeks.