Surface roughness of glass explained for optical and industrial finishes

What surface roughness of glass means

The surface roughness of glass refers to the small height variations left on a glass surface after forming, grinding, polishing, etching, coating or handling. It is different from visible scratches, chips and waviness, although those conditions can also affect performance. For buyers, engineers and finish inspectors, the useful question is not whether the glass is simply smooth or rough. It is which roughness parameter is being measured, over what area, with which instrument and for which function.

Optical windows, display cover glass, laboratory substrates, bonding surfaces and decorative frosted glass may all need different roughness levels. A good specification connects the finish to the job the surface must perform, such as limiting light scatter, supporting coating adhesion, improving sealing, controlling cleanliness or creating a specific tactile feel.

In surface finish work, roughness is the fine-scale part of surface texture. Larger shape errors are usually treated as form, while intermediate spacing errors may be treated as waviness. ISO 10110-8:2019 is used for specifying surface texture on optical element drawings, and ISO 25178-2:2021 defines terms and parameters for areal surface texture measurement. (iso.org)

For more articles on finish selection and inspection language, see the Surface Finishes section.

Why glass roughness matters in real applications

Glass is often selected because it can be transparent, chemically durable, dimensionally stable and capable of taking a very smooth finish. The right finish, however, depends on the application. A lens or laser window usually benefits from low roughness because fine height variation can increase light scatter and reduce optical efficiency. A bonding surface may need a controlled texture that helps wetting or mechanical interlocking, while still avoiding contamination traps.

In optical parts, roughness is closely related to scatter, but it should not be treated as the only cause of scatter. Surface contamination, coating defects, subsurface damage and digs can also influence optical behavior. ISO 10110-8 notes that polished optics can be specified by statistical texture methods and that roughness is associated with high spatial frequency errors, while waviness belongs to longer spatial scales. (iso.org)

Industrial glass also has non-optical reasons to control roughness. On sealing faces, high peaks can create leak paths or uneven gasket compression. On coated glass, surface texture can affect coating uniformity and perceived haze. In decorative glass, intentional roughness produces a matte appearance and changes the way the surface feels. In laboratory or electronic substrates, nanoscale roughness can influence film growth, cleaning behavior and inspection repeatability.

Common roughness parameters used for glass

Roughness values are useful only when the parameter is named. A phrase such as low roughness or polished finish is too vague for production or inspection. Even a number can cause disagreement if the parameter, unit and measurement method are missing.

Parameter	What it describes	When it is useful	Main limitation
Ra	Arithmetic average roughness along a 2D profile	General comparison of machined, ground or polished surfaces	Can hide isolated peaks, pits and directional texture
Rq or RMS	Root mean square roughness along a profile	Optical and precision surfaces where peaks and valleys matter more	More sensitive to outliers and measurement noise
Rz	Peak-to-valley style height over defined sampling lengths	Functional limits for peaks, grooves or sealing contact	Definitions vary by standard and setup, so the standard must be named
Sa	Arithmetic mean height over a 3D measured area	Areal inspection of glass panels, substrates and textured finishes	Depends strongly on scan area, filtering and missing-data handling
Sq	Root mean square height over a 3D measured area	Optical profiling, AFM and areal surface comparison	May not match a 2D Rq value taken from one line
Sz	Maximum height over the measured area	Detecting extreme peaks or pits	Can be dominated by one defect or artifact

ISO 25178-2:2021 is especially relevant when the surface is measured as an area rather than as a single line profile. It defines areal texture parameters such as Sa, Sq and Sz, which are often more representative for glass when the surface texture is non-uniform. (iso.org)

A common mistake is to compare Ra and Sa as if they are interchangeable. They are related concepts, but they come from different measurement geometries. A single 2D line can miss local pits, coating islands or polishing marks that a 3D areal scan would capture. Conversely, a large areal scan can include waviness or form if filtering is not properly applied.

Measurement methods and where they fit

Contact stylus profilometry

A stylus profilometer drags a fine tip across the surface and records a height profile. It is familiar in metal finishing and can be useful for ground or moderately rough glass. Its limitations become important on very smooth optical glass, soft coatings or delicate surfaces. Tip radius, contact force, filtering and scan length all affect the result. A stylus may also miss narrow features that are smaller than the tip can enter.

Optical profilometry

Optical instruments, including coherence scanning interferometry, phase-shifting interferometry and confocal methods, are widely used because they are non-contact and can measure areas quickly. They are well suited to many polished glass and precision substrate applications. Transparent glass, however, can create measurement challenges because light may reflect from more than one interface. The setup must define which surface is being measured and how the instrument handles low contrast, slope, coatings and missing points.

NIST has reported that optical methods are increasingly used for surface texture and that discrepancies can occur between white-light interferometry and stylus measurement for rough surfaces in the 50 nm to 300 nm Ra range. This is a useful reminder that glass roughness numbers are method-dependent, not just material-dependent. (nist.gov)

Atomic force microscopy

Atomic force microscopy, or AFM, is valuable for nanoscale glass roughness, thin films and research samples. It can reveal texture that ordinary optical equipment may not resolve. The trade-off is that AFM usually measures a small area and requires careful sample preparation and interpretation. A clean 10 micrometer by 10 micrometer scan may be valid for the measured spot, but it may not represent an entire display panel, lens blank or production batch.

Scattering and optical performance tests

For optical glass, scatter measurement can be used as a performance-oriented test. It does not replace surface topography measurement because scatter is affected by more than roughness. Even so, it can be important when the final requirement is optical transmission, imaging contrast or laser performance. For critical optics, roughness data, scratch-dig inspection, coating inspection and scatter testing may all be needed.

Typical ranges and why a single target is risky

There is no universal acceptable surface roughness of glass. A finish that is suitable for a frosted panel would be unacceptable for a precision optical window. A finish required for a laser optic may be unnecessarily expensive for a protective cover.

Published research shows how wide the range can be. In a study of International Simple Glass, polished samples were reported with AFM RMS roughness of 10 ± 3 nm, while unpolished samples were reported at 440 ± 180 nm. A separate review of optical polishing methods reported ultrasmooth glass surfaces below 1 nm RMS in selected studies, including K9 glass processed to less than 1 nm RMS. These values should be read as study results, not as universal purchasing specifications. (nature.com)

For a practical specification, start from the function. If the part is a transparent window for imaging, roughness should be tied to scatter, clarity and coating needs. If it is a bonding surface, define the adhesive system, cleaning process and acceptable contact behavior. If it is decorative, specify appearance, gloss, haze, touch and cleaning performance in addition to roughness. See also: Buying Guides.

Cost also rises as the required roughness becomes lower and the inspection becomes more demanding. Sub-nanometer or single-digit nanometer roughness may require fine polishing, strict cleaning, stable metrology and controlled handling. Calling for an extremely smooth surface without a functional reason can increase cost and rejection risk without improving the final product.

How to write a practical roughness specification

A clear glass roughness specification should be written so that two qualified labs can understand and reproduce the requirement. The following items are usually more useful than a single number:

• Function: optical transmission, coating, sealing, bonding, decoration, tactile feel or cleanliness.
• Parameter: Ra, Rq, Rz, Sa, Sq, Sz or another named parameter.
• Unit: nanometers or micrometers, stated clearly.
• Standard or convention: for example ISO 10110-8 for optical texture notation or ISO 25178-2 for areal parameters.
• Measurement method: stylus, optical profilometer, AFM or another defined method.
• Spatial bandwidth: cutoff, filter, scan length, scan area and pixel or sampling spacing.
• Locations: number of measurement points and whether edges, center, coated areas or functional zones are included.
• Acceptance rule: maximum value, average value, lot rule and treatment of obvious defects.
• Surface condition: cleaned, coated, etched, strengthened, assembled or as-received.

For example, roughness below 10 nm is incomplete as a stand-alone requirement. A better requirement would state the measured surface, parameter, instrument class, scan area, filtering and pass/fail rule. The exact wording depends on the application, but the key is traceability. Without that detail, a supplier, customer and third-party lab may each be correct under their own measurement setup and still disagree.

Process choices that change glass roughness

Different finishing processes create different surface textures. Grinding removes material efficiently but leaves directional marks and subsurface damage that may need later polishing. Lapping can improve flatness and reduce height variation, but abrasive size and pad condition matter. Polishing reduces fine texture and can create very smooth optical surfaces when the process is controlled. Etching, sandblasting and acid frosting intentionally increase texture to create a matte appearance or functional wetting effect.

Coatings can either smooth or amplify the underlying texture depending on deposition method, thickness and material behavior. A thin coating may reproduce substrate roughness, while a thicker or poorly controlled coating may introduce its own morphology. Chemical strengthening and thermal processing can also alter surface condition indirectly through handling, cleaning or subsequent coating steps.

Because roughness can change after finishing, inspection should be linked to the stage that matters. If a glass part is polished and then coated, the final coated surface may be the functional surface. If a part is roughened for bonding and then cleaned aggressively, the cleaning procedure may change the actual texture available to the adhesive.

Common mistakes when comparing glass roughness data

Many disputes over glass roughness come from comparing unlike measurements. The first mistake is mixing Ra, Rq and Sa in one discussion without explaining the difference. The second is ignoring measurement area. A roughness value measured over a tiny AFM window can be valid but not representative of a larger panel. The third is failing to separate roughness from defects. A surface can have low average roughness and still fail because of scratches, digs, chips or stains.

Another common issue is over-specification. Smoother is not always better. Some adhesives, coatings and decorative finishes need controlled texture rather than the lowest possible number. The best specification defines the finish that supports the part function with a realistic measurement method and a stable acceptance rule.

For English-language content on related finishing terms, the Surface Finishes archive collects practical explanations for material and inspection topics.

Frequently asked questions

Is lower glass surface roughness always better?

No. Lower roughness can reduce optical scatter and improve some precision applications, but controlled roughness may help bonding, anti-glare appearance or tactile performance. The correct target depends on the function of the surface.

Can Ra be used for glass?

Yes. Ra can be used when a 2D profile measurement is appropriate, especially for comparative inspection. For optical or highly uniform surfaces, Rq, RMS roughness, Sa, Sq or spatial-frequency methods may provide more useful information.

Why do two labs report different roughness values for the same glass?

Different instruments, scan sizes, filters, surface preparation and data-processing rules can produce different values. This is especially common with transparent, coated or very smooth glass. A complete specification reduces this risk.

What is the best method for measuring very smooth glass?

There is no single best method for every case. Optical profilometry is fast and non-contact, AFM is strong for nanoscale local texture, and scatter testing may be useful when optical performance is the final concern. Critical parts often need more than one method.

Should roughness be specified before or after coating?

Specify the surface that controls performance. If the coating is the functional surface, post-coating roughness may matter more. If adhesion depends on the substrate texture before coating or bonding, the pre-coating surface should also be defined.