Two lost wax castings compared, one with a smooth silica sol finish and one with a coarser water glass finish
Send a drawing to a Chinese lost wax foundry and the first process question back is usually the same: silica sol or water glass? Those names describe the binder that builds the ceramic shell, and the choice decides surface finish, tolerance, minimum wall thickness, alloy range and price per kilogram. Choose well and you pay only for the accuracy you use. Choose poorly and you either machine away a cheap casting or overpay for tolerances nobody needs. Ningbo Ruican runs both routes with in-house CNC machining, and this guide shows where each one wins. See the capability overview at https://www.cnsandcasting.com/investment-casting/.
1. Shell building: what actually differs
Both processes share one sequence: inject wax, assemble a tree, dip in ceramic slurry, stucco with refractory grain, dry, repeat, dewax, fire, pour. The difference is binder chemistry and how each coat hardens.
Silica sol shells
The binder is colloidal silica, sub-micron particles suspended in alkaline water. Nothing sets it chemically; each coat hardens only by controlled evaporation, so the shell room has monitored temperature, relative humidity and airflow. A stainless build starts with a zircon flour prime coat and fine zircon stucco, then transition coats, then mullite or fused silica backups: six to nine coats, hours of drying between dips. Build takes four to seven days, followed by autoclave dewax and firing near 900 to 1100 C.
Water glass shells
Here the binder is sodium silicate, hardened by dipping into ammonium chloride, aluminium sulphate or a modern ester hardener, with cheaper silica flour and quartz sand refractories. Hardening takes minutes rather than hours, so four to six coats compress into one to two days, with lower firing at 700 to 900 C. The trade-offs: coarser surface reproduction, less hot strength from residual alkali and wider dimensional scatter.
| Attribute | Silica sol | Water glass |
|---|---|---|
| Binder | colloidal silica | sodium silicate |
| Hardening | controlled drying | chemical hardener dip |
| Prime refractory | zircon flour | silica flour |
| Backup refractory | mullite, fused silica | quartz sand |
| Typical coats | 6 to 9 | 4 to 6 |
| Shell build time | 4 to 7 days | 1 to 2 days |
2. Surface finish: Ra 3.2-6.3 vs Ra 6.3-12.5
A fine prime coat does not bridge small features, so silica sol shells reproduce the wax faithfully at about Ra 3.2 to 6.3 um (125 to 250 uin). Water glass shells usually land at Ra 6.3 to 12.5 um (250 to 500 uin). Two caveats apply. Roughness degrades with heavier sections and higher pour temperature, so a 3 mm wall and a 25 mm boss on one part will not measure alike. And Ra does not fully describe a cast surface: where appearance or cleanability matters, agree a physical comparison sample.
3. Tolerance: ISO 8062 CT4-CT6 vs CT6-CT8
Silica sol production routinely reaches CT4 to CT6; water glass lands at CT6 to CT8, per ISO 8062 casting tolerance grades. CT4 is demanding and tooling-dependent, while CT6 is realistic for well-controlled medium parts.
| Nominal size | CT4 | CT5 | CT6 | CT7 | CT8 |
|---|---|---|---|---|---|
| 25 mm | 0.30 | 0.40 | 0.60 | 0.80 | 1.10 |
| 100 mm | 0.50 | 0.60 | 0.90 | 1.10 | 1.50 |
| 250 mm | 0.60 | 0.80 | 1.10 | 1.40 | 2.00 |
The practical consequence is machining stock. At CT6 you can leave 1.0 to 1.5 mm per face; at CT8 plan 2.0 to 3.0 mm, plus allowances for mismatch and draft taper. Every extra millimetre is CNC cycle time.
4. Wall thickness and part size
Silica sol holds thinner walls because the denser shell resists distortion and fills cleanly: around 1.5 to 2.0 mm locally on small parts, 2 to 3 mm as the everyday floor. Water glass commonly stops at 3 to 4 mm, more where the casting carries load. Both cover parts from about 50 grams upward; silica sol is usual to roughly 25 to 50 kg, while water glass is preferred above that.
5. Alloy suitability
Both routes pour stainless, carbon steel, alloy steel, duplex and tool steel from the same induction melt shop. Differences appear around pour temperature, shell reactivity and how much surface stays unmachined.
| Alloy family | Grades | Silica sol | Water glass |
|---|---|---|---|
| Austenitic stainless | 304, 316, 316L | Preferred for hygiene and thin walls | Fine where all faces get machined |
| Martensitic, PH stainless | 17-4PH (H900, H1025), 420 | Recommended; stable through heat treatment | Acceptable for simple machined blanks |
| Duplex stainless | 2205 | Recommended; needs a refractory shell | Risky above about 1500 C pour |
| Carbon steel | 1020, 1045 cast equivalents | Used when tolerance demands it | Cost-effective default |
| Alloy steel | 4140, 4340 | Preferred for fatigue-loaded parts | Acceptable with generous stock |
6. Cost per kg and tooling cost
The figures below are budgetary planning ranges for stainless near-net blanks, ex-works, excluding alloy surcharge, machining and freight. Treat them as a starting conversation, not a quotation: price moves with casting weight, yield, section thickness, inspection scope and quantity.
| Cost element | Silica sol | Water glass |
|---|---|---|
| Simple single-cavity wax die | USD 2,000 to 3,500 | USD 1,500 to 2,800 |
| Complex die, cores or slides | USD 3,500 to 8,000 | USD 2,800 to 6,500 |
| Austenitic stainless casting | about USD 7 to 12 per kg | about USD 4.5 to 7.5 per kg |
| Carbon, alloy steel casting | about USD 4.5 to 7 per kg | about USD 3 to 5 per kg |
Below roughly 0.3 kg the per-kilogram gap nearly disappears, because cost then sits in wax injection, assembly, cut-off and inspection.
7. Lead time
Silica sol carries five to seven days of shell drying before the pour; chemical hardening removes most of that. Tooling plus first-article samples run about four to six weeks for water glass and five to seven weeks for silica sol. Repeat production once tooling is proven falls between 25 and 40 days for both, before heat treatment, machining and inspection. Allow two to three extra weeks for a PPAP Level 3 package. Ningbo Ruican also delivers CNC machined parts so casting, heat treatment and machining share one schedule.
8. When to downgrade to water glass
Downgrade when accuracy has no purpose: every functional face gets machined anyway; walls are 4 mm or thicker; the part is large or heavy; the alloy is plain carbon or low alloy; there is no Ra callout; converted tolerances are looser than CT7; annual volume is high enough that a 20 to 40 percent casting saving matters.
9. When to upgrade to silica sol
Upgrade when surface, thin walls or tight tolerance change downstream cost: Ra finer than 6.3 um on an unmachined face; walls below 3 mm; internal passages cleaned in place and unreachable by a tool; leak-tight bodies where porosity would force weld repairs; dimensions of CT5 or better; duplex and tool steel grades; cosmetic parts.
10. Hybrid strategies
- Split the drawing. Keep one process route per part number so costing stays simple.
- Cast near net, machine to finish. Pair a water glass blank with a rigid fixture and clear datums so CT8 scatter is irrelevant.
- Prototype in silica sol. Validate geometry with the better process, then switch when tolerances prove open.
11. Decision table
| Requirement | Route | Reason |
|---|---|---|
| As-cast Ra finer than 6.3 um | Silica sol | Water glass cannot meet it |
| Wall below 3 mm | Silica sol | Filling and shell strength |
| Tolerance tighter than CT6 | Silica sol | CT4 to CT6 capability |
| All faces machined | Water glass | Casting accuracy is unused |
| Duplex 2205 or tool steel | Silica sol | Pour temperature, reactivity |
| Part above 50 kg | Water glass | Shell economics dominate |
| Cleanability critical | Silica sol | Unmachined passage finish |
| Below 200 pieces per year | Either | Tooling dominates; quote both |
12. Worked trade-off examples
Example A: 316L butterfly valve body, 2.4 kg, 500 per year, internal flow passage unmachined. Water glass: casting at 6.20 USD per kg = 14.88; machining 27 minutes at 45 USD per hour = 20.25; tooling 2,400 USD over 1,000 pieces = 2.40; subtotal 37.53 plus 6 percent scrap, 2.25. Total 39.78 USD. Silica sol: casting at 9.40 USD per kg = 22.56; machining 17 minutes = 12.75; tooling 3,600 USD over 1,000 pieces = 3.60; subtotal 38.91 plus 1.2 percent scrap, 0.47. Total 39.38 USD. Landed cost is effectively equal, so choose silica sol: it also meets Ra 3.2 in the hygiene-critical passage.
Example B: carbon steel agricultural bracket, 8.6 kg, 300 per year, every mounting face machined. Water glass: casting at 3.10 USD per kg = 26.66; machining 34 minutes = 25.50; tooling 2,600 USD over 600 pieces = 4.33; subtotal 56.49 plus 4 percent scrap, 2.26. Total 58.75 USD. Silica sol: near-net weight 8.0 kg at 5.30 USD per kg = 42.40; machining 30 minutes = 22.50; tooling 4,000 USD over 600 pieces = 6.67; subtotal 71.57 plus 1 percent scrap, 0.72. Total 72.29 USD. Silica sol costs 13.54 USD more per part, about 4,060 USD per year, for accuracy the drawing never uses. Choose water glass and spend part of the saving on magnetic particle inspection of the load-bearing fillets.
FAQ
What surface roughness should I specify on the drawing?
Specify Ra 6.3 um for water glass parts and Ra 3.2 um where silica sol is used and the surface matters functionally. Anything finer needs polishing or machining, cost without added function.
Is EN 10204 3.1 certification available from Ningbo Ruican?
Yes. Ningbo Ruican issues EN 10204 3.1 certificates with heat-linked chemical and mechanical results under an IATF 16949 and ISO 9001 quality system. PPAP packages and First Article Inspection Reports are available for documented programs.
Can I switch a part from water glass to silica sol later?
Yes, but plan for it. Expect a revised shrinkage allowance, possibly a new gating layout and a fresh first-article inspection. Because the wax die is common to both routes, tooling often survives the change, which keeps the switch affordable.
Does a lower cost per kilogram always mean a cheaper part?
No. Extra machining stock, extra set-ups, higher scrap and cosmetic rework can erase a casting price advantage, as Example A shows. Compare total landed cost including machining, scrap, inspection and freight.
Are there alloys that should avoid the water glass route?
Duplex 2205 and high-alloy tool steels fit poorly, because they pour hot and react with residual alkali in the shell. Grain grades such as 17-4PH can run either way, but specify silica sol when stability through heat treatment matters.
What information gives the most accurate quotation?
Send a 3D model plus a 2D drawing with material grade, heat treatment, critical dimensions, machining datum scheme, NDT level, annual volume and packaging expectations. That removes nearly all quoting assumptions.