Between 70–75% of every float glass batch is silica sand. That single fact explains why sand quality determines whether a production run yields premium architectural glass or green-tinted scrap. Just 0.01% excess Fe₂O₃, one hundredth of one percent can shift an entire furnace campaign into the reject bin. Float glass chemistry cannot be corrected once molten. You either get the sand right at intake or absorb the cost at inspection.
This guide covers the purity specifications that actually matter, the contaminants most likely to cause defects, and the supplier red flags worth catching before a truck dumps its load on your batch floor.
Silica provides the molecular backbone of float glass, its mechanical strength, chemical durability, and thermal stability all trace back to SiO₂ purity. Without tight chemistry, viscosity curves shift. Without controlled viscosity, tin bath spread rates drift and annealing stresses compound.
The defect physics are unforgiving. Seeds, tiny trapped gas bubbles form when oversized grains melt too slowly. Stones, which are unmelted refractory inclusions like zircon or chromite, survive the furnace entirely and tear through the tin bath. Cords create optical distortion visible as waviness in the final lite. A 5% rejection rate on full output is a six-figure loss before you start accounting for furnace downtime.
For analytical methods governing these determinations, ASTM C146 covers chemical analysis of glass sand. The ISO 3262 series addresses extender mineral specifications.
The chemistry of incoming sand dictates the color, clarity, and melt behavior of the ribbon. Impurities in the batch are impurities in the glass. The table below reflects qualification targets used across multiple float lines.
|
Oxide Parameter |
Premium Clear Float |
Standard Clear Float |
Why This Limit Matters |
|
SiO₂ (Silica) |
≥ 99.8% |
≥ 99.5% |
Forms the glass network; any reduction directly reduces durability and alters viscosity curves. |
|
Fe₂O₃ (Iron Oxide) |
< 0.015% (150 ppm) |
< 0.030% (300 ppm) |
Primary colorant. Even 0.1% produces a visible green cast; premium low-iron glass requires strict control. |
|
Al₂O₃ (Alumina) |
< 0.8% |
< 1.0% |
Increases durability slightly but raises melting temperature. Excess slows batch-to-glass conversion. |
|
TiO₂ (Titania) |
< 0.05% |
< 0.10% |
Imparts yellow-to-brown tint when combined with iron. Very undesirable in ultra-clear applications. |
|
Cr₂O₃ (Chromium Oxide) |
< 0.001% (10 ppm) |
< 0.002% (20 ppm) |
Extremely potent green colorant at ppm levels. One bad shipment can spike color beyond spec for months. |
|
ZrO₂ (Zirconia) |
< 0.01% |
< 0.02% |
Indicates heavy mineral contamination; zircon grains cause refractory stones that survive the melter. |
|
LOI (Loss on Ignition) |
< 0.3% |
< 0.5% |
Measures carbonates and organics. High LOI means foaming, redox shifts, and energy waste. |
Chromium is the element most likely to blindside you. At just 10 ppm, Cr₂O₃ produces a measurable green component in a 6mm lite held against white. Many suppliers report “nil” on their COAs without running sufficiently sensitive instrumentation. XRF screening is not adequate for chromium at these thresholds. Require ICP-MS or GF-AAS data specifically for Cr, Co, and V before approving any new source.
The gap between 99.5% and 99.8% SiO₂ looks trivial on paper. It is not. That 0.3% is typically alumina and iron. More alumina means higher melting energy and slower batch conversion. More iron means decolorizing additions, selenium or cobalt that cost money and complicate the melt chemistry. Both effects compound across a campaign.
Grain size distribution is the most frequently neglected parameter in supplier audits, and one of the most consequential. Target 90–95% of material retained between 105 µm (140 mesh) and 500 µm (35 mesh).
Grains above 500 µm melt too slowly and pass through the furnace partially unreacted, generating silica scum and cords in the ribbon. Fines below 105 µm cause a different problem: they fluidize in silos, segregate toward the center, and cause batch composition to drift throughout the shift. An operator batching by weight ends up adding proportionally less dry sand than intended once segregation takes hold.
Research from the International Commission on Glass Technical Committee on Melting found that maintaining a narrow particle size distribution reduces melting energy by up to 5% compared to broadly graded sands.
Keep moisture below 5%. This is not a chemistry issue, it is an economics issue. Every percentage point above five consumes additional furnace energy just to drive off water before melting can begin. More critically, variable moisture undermines batch weighing accuracy. An operator targeting 1,000 kg of sand at 8% moisture is actually adding 926 kg of silica and 74 kg of water. Batching errors compound from that point forward.
Rounded grains from mature sedimentary deposits flow smoothly through chutes and mix more homogeneously with soda ash and limestone. Angular, freshly crushed sand bridges in silos and flows inconsistently. For high-throughput batch systems, flowability is not a minor consideration, poor flow creates variance that looks like a chemistry problem until you trace it back to the sand pile.
Silica sand is never pure quartz in practice. Every deposit carries heavy minerals, clay coatings, and organic matter from its geological history. Each contaminant class produces a characteristic defect pattern.
Zircon (ZrSiO₄), rutile (TiO₂), ilmenite (FeTiO₃), and chromite (FeCr₂O₄) have melting points well above float furnace operating temperatures. A single 200-micron zircon grain can travel the entire melt path, embed in the tin bath, and produce a stone defect that shatters ribbon during annealing. These minerals concentrate when a supplier’s gravity spirals or magnetic separators are not properly maintained. Heavy mineral counts in grams per metric ton should be a required field on every COA.
Clay coatings carry alumina and iron directly onto grain surfaces, bypassing bulk chemistry screening. Organic material roots, humic acids, plant debris raises LOI and creates localized reducing conditions that disrupt sulfate fining behavior. Foaming at the spring zone frequently traces back to organic-contaminated sand, often from a supplier stockpile that sat wet too long.
Loss on Drying (LoD) measures free moisture at 110°C. Loss on Ignition (LOI) measures total weight loss after heating to 1,000°C, a number that includes carbonates decomposing, organics burning off, and bound water escaping. An LOI above 0.5% indicates either carbonate minerals generating CO₂ late in melting or organic carbon interfering with redox control. Do not accept LOI and LOD data without the distinction clearly labeled.
Deposit geology determines baseline purity, and processing determines consistency. Not all high-SiO₂ deposits produce float-grade sand at commercial scale.
The Lommel deposits in Belgium yield exceptionally pure, well-rounded grains with consistently low iron, a function of prolonged fluvial sorting. The St. Peter Sandstone formation beneath Ottawa, Illinois, supplies most of North America’s float glass industry with material running 99.7%+ SiO₂ and stable physical properties across seasons. Cape Flattery in Queensland, Australia, produces sand with low iron and tight particle distribution that has made it a preferred source for Asian float manufacturers.
Knowing the source deposit helps in qualification. Sands from immature geological settings or heavily weathered tropical deposits tend to carry more clay contamination and heavier heavy-mineral loads regardless of how well the supplier processes them.
A one-page certificate of analysis is not sufficient for float glass qualification. Demand the full package:
Inconsistent LOI values across shipments from the same source indicate stockpile management problems or inadequate washing. Chromium reported as “nil” without a detection limit stated means the instrument may not be sensitive enough to find it at 10 ppm. Suppliers who offer only pass/fail summaries rather than raw analytical data are hiding the trends you need to track.
Start with one truck. Sample aggressively every fifth tonne at minimum and run the full specification against each sample. If results hold, scale to one railcar, sampling every compartment. Only after several successful commercial-scale deliveries should you commit to vessel-volume contracts. Run the first trial blind if you can arrange it: standard commercial product reflects the supplier’s normal operation. A specially selected sample does not.
Defect-free float glass comes from getting three things right consistently: specifications tight enough to exclude the chemistry that causes defects, supplier audits that verify process consistency rather than just accepting a piece of paper, and in-house lab discipline that catches variance before it reaches the furnace.
A useful starting point: pull the last six months of incoming sand COAs and look for trends. Is LOI creeping upward? Are fines increasing? Is iron showing more variability shipment to shipment? Those trends are the early warning system. Catching them in the data is cheaper than catching them in the ribbon.
Between 70–75% of every float glass batch i
READ FULL