Watanabe & Co. Quartz & Silica Consulting

Material fundamentals

Fused natural quartz and synthetic fused silica.

Both are amorphous silicon dioxide. Chemically they are the same substance. In a fab they behave like two different materials, because everything that matters was decided by how the glass was made.

First, the vocabulary

The names are used inconsistently across the industry.

Fused quartz generally means glass melted from mined quartz — crystal or high-purity sand. Fused silica generally means glass built from a chemical precursor. But suppliers, datasheets and purchase orders use the two terms loosely and sometimes interchangeably, and "quartz glass" is often used as an umbrella for both.

This is not a pedantic distinction. A specification that says "fused quartz" without naming a grade, a metals limit and an OH range is not a specification — it is a hope. A large share of the disputes we are asked to unpick begin with two parties who each thought the word meant the thing they had in mind.

Throughout this site we use fused natural quartz for melt-from-ore material and synthetic fused silica for chemically synthesised material, and we recommend clients write specifications the same way — by route and by measured property, never by trade name alone.

Family 01

Fused natural quartz

Made by melting naturally occurring quartz. Its purity ceiling is set by the ore body, and the number of ore bodies good enough for semiconductor use is very small.

Manufacturing routes

Electric fusion. Quartz sand is melted in a refractory or graphite crucible under vacuum or an inert atmosphere. Because there is no hydrogen flame, the resulting glass has very low hydroxyl content — typically in the low tens of ppm or below — which is what gives electrically fused material its high-temperature creep resistance. Bubble content is higher than flame-fused material.

Flame fusion. Quartz grain is fed through an oxy-hydrogen flame and deposited onto a rotating target. The glass is more transparent and has far fewer bubbles, but the flame introduces water: OH content typically lands in the 150–250 ppm range, which lowers the viscosity at temperature and reduces sag resistance.

Plasma fusion. A hydrogen-free plasma is used instead of a flame, giving low OH with better clarity than electric fusion. It is a more expensive route and is used where both properties are genuinely required.

What you inherit from the ore

Natural quartz carries the trace elements of the deposit it came from. Aluminium is the dominant one, typically well over 10 ppm, with titanium, sodium, potassium, iron, calcium and lithium present at lower levels. Alkali metals matter disproportionately: sodium and potassium are mobile at process temperature, they accelerate devitrification, and in a device context they are classic yield killers.

It is also why feedstock provenance is a supply-chain question, not just a quality question. Commercially viable high-purity quartz deposits are concentrated in a handful of locations — most notably in the United States, Norway and Australia — and the number of companies able to refine to semiconductor grade is smaller still. A single-sourced feedstock two tiers below your direct supplier is a real exposure, and one most buyers have never mapped.

Where natural quartz remains the right answer: hot-zone structures, furnace and process tubes, boats and carriers, jigs and fixtures, sight glasses, and any part where mechanical robustness, thermal capability and cost per kilogram matter more than parts-per-billion metals. It is harder at the surface than synthetic material (≈650 vs ≈570 KHN) and it is substantially cheaper.

Family 02

Synthetic fused silica

Built from a chemical precursor rather than mined rock. Purity becomes something you engineer rather than something you are given.

Manufacturing routes

Flame hydrolysis (direct process). Silicon tetrachloride is oxidised and hydrolysed in an oxy-hydrogen flame, and the resulting SiO2 particles are deposited and fused directly into glass in a single step. This is the route that produces the largest boules — up to roughly two metres in diameter — and it delivers exceptional purity and homogeneity. The trade-off is high OH content, typically on the order of 1,000 ppm, along with residual chlorine.

Soot and consolidation (indirect / vapour phase). A porous soot body is deposited first, then dried and consolidated into transparent glass in a separate furnace step. Separating deposition from consolidation gives control over the atmosphere during drying, so OH can be driven to low levels — including below 1 ppm — while retaining synthetic purity.

Sol-gel and alkoxide routes. Hydrolysis of a silicon alkoxide such as methyl or ethyl silicate, gelled and sintered. Used for particular purity, geometry and near-net-shape requirements.

What you gain, and what you pay for it

Metallic contamination in semiconductor-grade synthetic silica is quoted in fractions of a part per million — lithium and copper at the 0.01 ppm level, sodium around 0.1 ppm, potassium and magnesium around 0.05 ppm. Bubble and inclusion content reaches the highest optical classes, and index homogeneity is good enough for deep-UV lithography optics and photomask substrates.

You pay in three currencies. Cost — several times natural quartz per kilogram. Hydroxyl — high-OH material has lower viscosity at temperature and sags sooner, so a direct-process grade may be the wrong choice for a long horizontal tube even though it is the purer material. And hardness — the surface is softer and marks more easily in handling.

The OH question is the one most often missed. It also sets the infrared behaviour: the familiar absorption features near 1.4, 2.2 and 2.7 µm are hydroxyl, so an IR application and a DUV application can want opposite ends of the same product family.

Where synthetic silica is not optional: anything in contact with the silicon melt, deep-UV and 193 nm optical paths, photomask substrates, plasma-facing liners and windows, and leading-edge thermal processing where alkali contamination would be a device-level failure.

Side by side

The comparison that actually drives decisions.

Fused natural quartz vs synthetic fused silica — indicative characteristics
Characteristic Fused natural quartz Synthetic fused silica
Raw material Mined quartz crystal or high-purity sand SiCl4 or silicon alkoxide precursor
Process Electric, flame or plasma fusion Flame hydrolysis, soot & consolidation, sol-gel
SiO2 purity ≈ 99.99 – 99.995 % ≈ 99.9999 % and above
Total trace metals Tens of ppm — Al dominant, plus Ti, Na, K, Fe Sub-ppm to ppb class
Typical OH content < 10 ppm (electric) · 150–250 ppm (flame) < 1 ppm (dried) · ~1,000 ppm (direct)
Bubbles & inclusions Present; grade-dependent Very low to effectively zero in top grades
UV transmission Limited below ~200 nm Good to ~185 nm; excellent at 248 / 193 nm
Index homogeneity Moderate — striae common High; suitable for precision optics
Surface hardness ≈ 650 KHN ≈ 570 KHN
Sag / creep resistance Better in low-OH electrically fused grades Depends entirely on OH — high-OH grades sag sooner
Devitrification tendency Higher — alkali and Al act as nucleation sites Lower — fewer nucleating impurities
Relative cost Baseline Several times higher per kg
Supply risk Tied to a few HPQ deposits worldwide Tied to precursor supply and a few producers

Values are indicative ranges drawn from published supplier data and technical literature, intended to frame a decision — not to replace a supplier's certified datasheet for a specific grade. We work from certified data on live projects.

The composite answer. The most instructive example is the CZ crucible. A crucible made entirely of synthetic silica would be prohibitively expensive; one made entirely of natural quartz would leach alkali and aluminium into an eleven-nines silicon melt. So the industry builds both: a natural quartz outer body providing strength and thermal mass, with a synthetic inner layer at the melt interface. Bubbles also grow more slowly in the synthetic layer, which matters over a long pull. Getting the layer architecture right is a materials problem, an economics problem and a process problem at the same time — and it is a good illustration of why "which material" is usually the wrong question.

Supplied forms

Boules, ingots and tubes.

The form the material arrives in constrains what can be made from it, what it costs, and where the defects will be.

Form 01

Boules & ingots

The primary synthetic form. Glass is deposited from the synthesis flame onto a rotating, retracting target, building a cylindrical or rectangular body that is then annealed, cored, sliced and finished. Direct-process boules reach roughly two metres in diameter; semiconductor-grade ingots are supplied in large square formats sized for photomask and large-substrate work.

  • Index homogeneity and striae grading
  • Annealing schedule and residual stress
  • Coring, slicing and yield planning
  • Radial and axial property variation
Form 02

Tubes

Drawn continuously from a melt or redrawn from a hollow preform, then cut, flame worked and finished. For diffusion and oxidation service the controlling parameters are wall thickness uniformity, straightness, bubble population and OH — with standard diameter and wall tolerances around ±0.15 mm and precision grades near ±0.08 mm.

  • Dimensional and ovality control
  • OH targeting for sag resistance
  • Flame working, flanging and joining
  • Cleaning, etching and surface preparation
Form 03

Crucibles

Formed by arc fusion of quartz grain against a rotating mould, producing an opaque bubble-containing outer wall for thermal insulation and strength, and a transparent inner layer at the melt interface. Consumable by nature — often a single crystal growth campaign — so cost, campaign length and contamination trade directly against each other.

  • Inner-layer material and thickness
  • Bubble structure and growth behaviour
  • Devitrification and barrier coatings
  • Campaign economics
Form 04

Plates, rods & machined components

Sliced, ground, lapped and polished parts: windows, viewports, showerheads, liners, rings, boats and carriers. This is where fabrication practice starts to dominate material choice — subsurface damage from grinding, contamination from tooling, and inadequate final cleaning cause more field failures than the base glass ever does.

  • Subsurface damage and edge quality
  • Flatness, parallelism and surface finish
  • Fire polishing and flame finishing
  • Final clean and packaging integrity

Reference

Properties that set the operating envelope.

Thermal & mechanical — typical values
Softening point≈ 1,683 – 1,700 °C
Annealing point≈ 1,215 °C
Strain point≈ 1,120 °C
Continuous service≈ 1,150 °C
Short-term maximum≈ 1,400 °C
Thermal expansion≈ 5.5 × 10⁻⁷ /°C
Density≈ 2.20 g/cm³
Knoop hardness570 – 650 KHN by type

How to read this table

The number that governs most component failures is not the softening point — it is the annealing point, around 1,215 °C. Above the strain point the glass relaxes; a horizontal tube held near that region will sag over time regardless of how far it is from softening. Low-OH material buys you margin here; high-OH material costs you margin.

The thermal expansion coefficient — roughly 5.5 × 10⁻⁷ /°C, about one twentieth of common borosilicate — is why quartz tolerates the ramps that semiconductor processing demands. It is also why joints between quartz and anything else deserve careful attention.

Chemical resistance is excellent against acids, and poor against hydrofluoric acid, hot phosphoric acid and strong alkalis. In fluorine plasma the material is consumed steadily; that is a consumable-life calculation, not a defect.

Devitrification — conversion of the glass to cristobalite at the surface during prolonged high-temperature service — is the usual end of life for thermal parts. It is nucleated by surface contamination and alkali, which is precisely why cleaning practice, handling discipline and base-material purity are one problem rather than three.

In practice

How we work a selection decision.

STEP 01

Define the actual environment

Peak and continuous temperature, dwell time, cycle count, gas and plasma chemistry, mechanical loading, and orientation. Most specifications describe the part; few describe what the part has to live through.

STEP 02

Establish the contamination budget

Which elements matter at what level, and how they would reach the wafer. This is what decides natural versus synthetic — not a general preference for purity.

STEP 03

Set the OH target

Sag resistance, IR and UV behaviour and hydrogen-related effects all follow from hydroxyl content. It should be a specified range with a test method, not an inherited default.

STEP 04

Cost the part over its life

Purchase price, expected campaigns before replacement, downtime to change it, and the scrap risk if it fails early. A cheaper part changed twice as often is frequently the more expensive one.

STEP 05

Qualify the supplier, then the lot

Audit the fabrication route, agree certified data and test methods, define incoming inspection, and establish a second source before you need one.

Bring us a selection problem →

References

Published sources behind this page

Figures on this page are indicative ranges compiled from public supplier datasheets and technical literature. They are provided for orientation. Design and procurement decisions should be made against a supplier's certified data for the specific grade.

  1. Heraeus Covantics — Fused silica for semiconductor manufacturing
  2. Shin-Etsu Quartz Products — Quartz glass manufacturer
  3. Tosoh Quartz Corporation — Quartz glass products
  4. Goodfellow — Fused silica vs fused quartz: key differences and properties
  5. ScienceDirect — Manufacturing and characterization of fused quartz crucibles for monocrystalline silicon
  6. Fused quartz properties and usage guide
  7. Semiconductor-grade fused quartz tubing — general properties

Deciding between grades, or between suppliers?

Send us the environment the part has to survive and we will tell you what we would specify — and why.

Start a conversation → See our services