How to Read a TiO2 Certificate of Analysis

Every TiO2 shipment includes a batch-specific Certificate of Analysis (CoA). The CoA is the formal quality document that certifies the specific batch meets specification. Knowing how to read it lets you verify quality at receipt and catch issues before they affect production. A CoA that cannot be properly interpreted is a compliance liability, not an asset.

CoA structure: header and chemistry section

Header section: - Product name and grade designation (e.g., "SEMITI 996, Universal Rutile Titanium Dioxide") - Batch / Lot number (unique identifier) - Production date and shipment date - Container / FIBC numbers (for traceability) - Customer reference / Order number

Chemistry section: - TiO2 content (typically ≥93.0% for rutile, ≥98% for anatase): the primary chemistry assay. Below spec indicates dilution or contamination. - Rutile content (for rutile grades, ≥98.5%): proportion of TiO2 in rutile crystal form. Anatase grades have ≥99% anatase content. - Crystal form: rutile or anatase (or specific specifications for nano). - Surface treatment composition: e.g., "Al2O3 + organic" or detailed percentages if specified.

Physical properties: - Specific gravity / density: 4.0 g/cm³ for rutile, 3.9 for anatase — should be consistent batch-to-batch. - Oil absorption: gram of oil per 100 g of TiO2. Range typically 12–25 depending on surface treatment. Higher = needs more dispersant. - Loss on ignition (LOI): heating to 800°C, percentage weight loss. Should be ≤0.50%. - pH (aqueous suspension): 6.5–8.5 typical. Acidic batches indicate surface treatment issue. - Water-soluble matter: ≤0.50% typical. Higher indicates residual processing chemicals.

Physical, color, and optical properties on the CoA

Color properties: - CIE L*: lightness (97.0–98.8 range) - CIE a*: red-green (-0.5 to +0.5) - CIE b*: blue-yellow (1.0–2.5) - Whiteness Index (WI) or Yellowness Index (YI): derived color metric

These three (L*, a*, b*) tell you the color match against your reference. Major deviation (more than ±0.5 on any axis) means batch will look different in finished product.

Optical properties: - Tinting strength (Reynolds method or % vs standard): typically 1700–1950+. The functional opacity metric. - Particle size (D10, D50, D90): microns. Standard for coating-grade rutile: D50 0.25–0.30 μm. - 45 μm sieve residue: ≤0.02% typical. Higher indicates coarse particles or contamination.

Heavy metals: - Pb (lead): < 10 ppm typical (lower for premium grades) - Cd (cadmium): < 2 ppm - As (arsenic): < 3 ppm - Hg (mercury): < 1 ppm

Some CoAs also report Sb (antimony) and Ba (barium).

Functional spec (application-specific): - Thermal stability (for plastics grades): typically expressed as Δb* at 280°C or 300°C, 5 min hold - QUV gloss retention (for outdoor coatings grades): typically % retention at 2000 or 5000 hr - Photocatalytic activity (for cosmetic nano grades): % of uncoated baseline

How to use the CoA at receipt and red flags to watch for

Step 1: Verify batch identity Check that the CoA batch number matches the actual product (bag tags, container documentation). Mismatched batches are rare but happen — flag immediately.

Step 2: Compare against your spec For each critical parameter (typically TiO2 content, tinting strength, color L*/b*), verify the batch falls within your accepted range. Specific automotive OEM specs may have tighter tolerances than the CoA's normal range.

Step 3: Compare against historical batches Look for batch-to-batch variation in critical parameters. Sudden shifts (e.g., b* jumping from 1.8 to 2.4) suggest process change at the supplier — worth investigating before scaling production.

Step 4: Archive CoA Keep the CoA on file for 2+ years. If a quality issue surfaces in finished product, you'll need the CoA to investigate.

Common CoA red flags:

1. Missing batch number or production date: indicates poor traceability; refuse acceptance 2. TiO2 content right at minimum spec (e.g., exactly 93.0%): probably means the batch was at the edge, may have actual content lower; request retest or reject 3. b* significantly higher than historical range: indicates impurity carry-through; investigate before scaling 4. Heavy metals above spec: rare but critical; reject and request replacement 5. Test methods not specified: ask supplier to clarify which methods used (GB, ISO, ASTM)

Independent verification: For critical applications, periodic third-party testing (SGS, Bureau Veritas, Intertek) of samples confirms supplier CoA accuracy. Typical cost $500–1500 per analysis. Not necessary for every shipment, but worth periodic verification.

SEMITI CoA practice: - CoA prepared for every batch - PDF emailed at shipment date - Hard copy in document pouch with B/L - All test methods explicitly listed - Historical batch data available on request - Third-party retest available on customer request

Test methods referenced on CoAs: GB, ISO, ASTM comparison

Chinese TiO2 CoAs typically cite GB (Guobiao — national standard) test methods. International buyers may need to cross-reference these against ISO or ASTM equivalents to confirm comparability.

Key method equivalencies:

The GB and ISO methods for most parameters are technically equivalent and produce comparable results. The most meaningful difference is in tinting strength measurement: GB/T 5211.17 uses a Chinese national white reflectance standard, while ISO 787-24 uses the Reynolds number method referenced to a BCRA/NPL absolute standard. SEMITI CoAs report Reynolds number to facilitate direct comparison with Ti-Pure, Tronox, and Kronos CoA data.

Using CoA data for statistical process control (SPC)

Experienced TiO2 buyers maintain SPC charts of incoming CoA data to detect supplier process drift before it causes production problems.

Recommended SPC parameters for paint manufacturers: Track these parameters for every incoming TiO2 batch: - Tinting strength (Reynolds) — control limits: process average ± 3σ (typically ± 30–50 Reynolds units) - CIE b* — control limits: ± 0.3 units - Oil absorption — control limits: ± 2 g/100 g - D50 particle size — control limits: ± 0.03 μm

Plot each parameter on an X-bar control chart by batch/lot number. Trend detection (7 consecutive points moving in one direction) indicates supplier process drift before the parameter leaves spec — giving you time to investigate and adjust formulation or push back on the supplier before a production incident occurs.

Western Electric rules for CoA SPC: Apply the standard Western Electric detection rules: - Any single point beyond 3σ: immediate hold and investigation - 2 of 3 consecutive points beyond 2σ: warning flag, investigate - 7 consecutive points trending in one direction: process shift, contact supplier - 4 of 5 points beyond 1σ: watch signal, schedule supplier review

This level of rigor is standard at automotive OEM coatings suppliers and large industrial paint manufacturers who rely on TiO2 consistency as a critical raw material for color-matched production.

Reconciling CoA spec vs. in-plant performance discrepancies

Occasionally buyers find that an incoming TiO2 batch meets CoA spec but underperforms in their formulation. This discrepancy has identifiable causes.

Surface treatment variation within spec range: Surface treatment composition can vary within the specification tolerance without changing the CoA values. A batch at the lower bound of Al2O3 treatment may disperse more slowly, show higher flocculation tendency, and reduce in-formula hiding — all while meeting every CoA parameter. Solution: request tighter surface treatment tolerances in your supply agreement, or run in-formula screening on each incoming batch.

Particle size distribution width: D50 may be within spec, but a batch with D90 at the upper tolerance edge (e.g., 0.48 μm vs typical 0.40 μm) will show lower hiding power and more grind-gauge residue in your ink or coating. Always check D90 alongside D50 when a batch underperforms despite normal D50.

Lot blending at the warehouse: Some suppliers blend different production lots to hit average CoA targets. A blended lot can show a bimodal PSD not visible in standard D50/D90 reporting. Bimodal PSD causes dispersion problems and hiding loss. Request a full PSD curve (volume-weighted, not just D50/D90 points) if you suspect blending.