Bath Control | Management | Process governance
Statistical Process Control for Baths: Chart Setup & Rules
A bath can remain inside its chemical specification while its behavior changes enough to threaten the next load. Statistical process control, or SPC, gives management a structured way to detect that change, assign an investigation and document recovery. It does not authorize automatic chemical additions or replace product inspection.
This guide uses a conventional, additive-free Watts nickel bath as the worked example. Zinc-alloy bag hardware must enter it through an approved pretreatment and barrier-coating route; do not plate directly onto exposed zinc. Bright nickel, barrel plating and proprietary packages require separate operating instructions and baselines.
Evidence boundary: Published base-salt ranges are referenced below. Narrow operating bands, intervention thresholds, worked calculations and coupon results are illustrative commissioning examples, not QLQ customer test reports. Validated SPC baselines and measured batch comparisons are not available in the reviewed knowledge-base notes. Replace these examples with signed plant records before production use.
1. Decide what each chart must control
Write a chart charter before collecting numbers: tank identity, chemistry revision, product family, sampling point, measurement method, sampling interval, alarm owner and disposition authority. The laboratory owns analytical accuracy; the shift supervisor owns containment; the process engineer owns corrective action; quality owns product release.
Use individual measurements for periodic pH or chemical assays. Use rational subgroups for thickness only when several parts represent the same short production interval and comparable rack positions. Repeated readings of one coupon estimate measurement repeatability, not load-to-load variation. Do not pool rack and barrel data, different finish systems or unlike geometries.
Record actual timestamps, ampere-hours, exposed area, additions, dilution, filter changes and downtime alongside results. Missing readings stay missing. A dose followed immediately by another sample is a documented intervention, not an interchangeable routine observation. Check for serial correlation: densely sampled sensor data may exaggerate apparent stability and create misleading run-rule alarms.
2. Working Data & Formula Notes
The base composition below follows the Nickel Institute Nickel Plating Handbook. Typical values are selected example setpoints. Concentrations refer to the named hydrated salts, not elemental nickel.
| Component | Range | Typical | Function | Replenishment |
|---|---|---|---|---|
| Nickel sulfate, NiSO4·6H2O | 240–300 g/L | 270 g/L | Main nickel source | Assay and inventory balance; account for soluble anodes. |
| Nickel chloride, NiCl2·6H2O | 30–90 g/L | 60 g/L | Anode dissolution and conductivity | Chloride assay; avoid correcting total nickel blindly with chloride salt. |
| Boric acid, H3BO3 | 30–45 g/L | 40 g/L | Cathode-interface buffering | Assay; dissolve separately and verify after mixing. |
| Deionized water | Balance to working volume | Final volume mark | Solvent | Restore measured evaporation loss before interpreting concentration. |
| Approved pH adjustment reagent | As required by validated instruction | No fixed concentration | pH correction | Bench-test dose; controlled addition and remeasurement. |
This is not a bright-nickel recipe: no proprietary brightener, carrier or wetting-agent dose is implied. Their supplier-specific concentrations require confirmation. Use an approved makeup sequence and trained chemical-handling personnel.
For a dissolved component, the planning calculation is addition, kg = deficit, g/L × actual volume, L ÷ (1000 × assay fraction). At 1,000 L, a confirmed 2 g/L boric-acid deficit requires 2.02 kg of 99% material. Verify volume first, dissolve through the approved method, circulate and re-assay. A chart signal alone does not establish a deficit.
3. Operating bands are not statistical limits
Use three distinct layers: an engineered working band, a statistical alarm derived from stable history, and a stop threshold approved during process qualification. The following provisional guardrails illustrate management decisions for one rack-nickel recipe. They are not universal defect boundaries or permission to operate outside the working band.
| Parameter | Normal working band | Danger / stop threshold | Risk and response |
|---|---|---|---|
| pH, defined 25°C sample method | 4.0–4.5; target 4.2 | <3.8 or >4.8 | Hold loading; verify electrode and assay before adjustment. |
| Bath temperature | 53–57°C; target 55°C | <50°C or >60°C | Outside this example qualified thermal envelope; check circulation and heater control. |
| Nickel sulfate hexahydrate equivalent | 260–280 g/L | <240 or >300 g/L | Check total nickel, chloride and volume; do not infer salt independently from nickel alone. |
| Nickel chloride hexahydrate equivalent | 55–65 g/L | <30 or >90 g/L | Investigate chloride inventory and anode behavior. |
| Boric acid | 38–42 g/L | <30 or >45 g/L | Verify analysis and dissolution; investigate buffering or crystallization risk. |
| Average cathode current density | 2.8–3.2 A/dm² | <2.5 or >3.5 A/dm² | Outside the example load recipe; verify area, contacts and rectifier current. |
| Nickel thickness at critical location | 10–12 µm example target | <8 µm example drawing minimum | Quarantine affected loads; drawing requirements override this example. |
Values between the working band and stop threshold require verification and engineering review, not silent acceptance. Map sensor temperature and laboratory sample temperature separately. Automatic temperature compensation does not make pH readings taken at different sample temperatures chemically equivalent.
4. Set up the baseline and chart calculation
As a plant commissioning policy, collect at least 30 representative individual assays across shifts, additions and normal workload. This is a starting point, not proof of stability. Preserve abnormal points, investigate them and document any justified exclusion. Freeze approved Phase II limits with a revision date; do not recalculate after every unfavorable reading.
For an individuals chart, MR = |current result − previous result|, estimated sigma = mean MR / 1.128, and limits = mean ± 3 × estimated sigma. This method is described in the NIST individuals-chart guidance. Plot the companion moving-range chart to expose abrupt variability changes.
Suppose a stable example pH history has mean 4.20 and mean moving range 0.040. Estimated sigma is 0.0355; individual limits are approximately 4.094 and 4.306. A reading of 4.34 triggers investigation even though it remains inside the 4.0–4.5 working band. Conversely, a wide statistical limit cannot authorize an out-of-specification bath.
Validate measurement resolution before adopting these limits. Duplicate assays, a reference solution and cross-shift method comparisons help separate laboratory drift from bath drift. If measurement error consumes most of the observed spread, improve the measurement system before tightening process alarms. For subgroup charts, document subgroup size and use its corresponding constants; do not reuse individuals-chart limits.
5. Alarm rules and a closed-loop response
The NIST variables-chart guidance distinguishes specification limits from control limits and describes Western Electric rules. The core signals are one point beyond three sigma, two of three beyond two sigma on the same side, four of five beyond one sigma on the same side, or eight consecutive points on one side of the centerline.
Preselect the rule set; enabling every available rule increases false alarms. Apply zone rules to the individuals or subgroup-mean chart, not indiscriminately to moving ranges. Use a six-point monotonic trend only as a separately documented supplementary rule.
- Contain: identify production since the last verified acceptable check. Stop new loading for a confirmed danger-threshold breach; segregate potentially affected loads.
- Verify: retain the first result, resample at the defined point and check instrument standards. Never overwrite an alarm with its passing repeat.
- Diagnose: review additions, dilution, anode inventory, electrical contacts and upstream carryover. Nominate one investigation owner.
- Correct: execute one approved intervention, record quantity and mixing interval, then verify chemistry and representative product.
- Release: obtain engineering and quality approval against the drawing and reaction plan. Record the disposition of every held batch.
6. Coupon and batch comparison: keep the evidence chain
The table is a synthetic training dataset, not an actual experiment. It demonstrates the record format required for a controlled trial. Use five identically prepared copper coupons per condition, the same rack positions, a 20-minute nickel cycle, identical agitation and calibrated XRF readings at fixed locations. Test real hardware separately because geometry and pretreatment affect transferability.
| Trial / batch | Test condition | Nickel thickness, mean / minimum | Visible pit count per 10 cm², mean | Disposition example |
|---|---|---|---|---|
| A: reference | pH 4.20; 55°C; 3.0 A/dm²; 20 min; five coupons | 11.4 / 10.9 µm | 0 | Reference only; confirm hardware meets drawing. |
| B: pH excursion | pH 4.90; other conditions as A; five coupons | 11.2 / 10.5 µm | 4 | Hold: chemistry exceeds example stop threshold. |
| C: under-current challenge | 2.4 A/dm²; pH 4.20; 55°C; 20 min; five coupons | 9.1 / 8.5 µm | 0 | Investigate recipe deviation despite apparent thickness pass. |
| D: recovery check | Reference settings restored; five fresh coupons after verified mixing | 11.3 / 10.8 µm | 0 | Candidate recovery evidence, not automatic release. |
Archive individual readings, not just averages. A coupon minimum cannot stand in for recessed hardware locations. No salt-spray hours, hardness values or service-life claims are inferred here; those tests need their own specified method, coating stack, conditioning and acceptance criteria.
Two investigation scenarios
Scenario A: drifting pH after a dosing-system change. An eight-point same-side signal appears before a working-band breach. Compare pump delivery against a measured collection test and check sample preparation. If delivery has changed, repair and verify it rather than widening limits. Retain the pre-change baseline until a qualified process change justifies a new revision.
Scenario B: thinning deposits with stable chemistry. Hardware from one rack position becomes thin while pH and assays remain stable. Compare contact resistance, actual current and exposed area by rack. Repair the electrical path and verify critical-location thickness. Chemical additions would obscure the electrical fault. These are engineering training scenarios, not claimed customer outcomes.
7. Engineer Notes
- Mark the sampling point and use a fixed mixing delay; never sample directly beside a dosing outlet.
- Check the pH electrode with bracketing buffers before the first shift assay; log slope and offset against instrument instructions.
- Normalize solution volume before calculating additions; distinguish evaporation from drag-out loss.
- Verify cathode area when product mix changes; compare calculated density with actual rectifier current.
- Link each chemical addition to an assay, authorization, lot number and follow-up result.
- Retain coupon photographs and location-coded thickness readings with the production batch identifier.
- Review recurring signals weekly; fund measurement, filtration or contact repairs instead of accepting repeated manual correction.
8. Shift Checks
| Shift stage | Check item | Frequency | Acceptance standard | Record |
|---|---|---|---|---|
| Before shift | Volume, pH reference check, bath temperature | Before first load | Approved volume; valid instrument check; working bands met | Opening checklist, instrument ID, operator |
| Before shift | Nickel, chloride, boric acid; circulation and anodes | Opening assay and equipment check | Recipe bands; approved flow and anode condition | Laboratory sheet and maintenance status |
| During shift | Temperature, pH and chart signals | Temperature continuously; pH every two hours and after additions | No unaddressed alarm; working bands met | Timestamped chart and event log |
| During shift | Current, area and critical-location thickness | Every load; first load after intervention | Qualified load recipe and drawing limits | Load card, rack ID, XRF results |
| After shift | Closing assays, additions, held batches | Each handover | Inventory reconciled; no unexplained signal or unassigned hold | Signed handover and disposition register |
Management should track time to containment, repeat alarms, incomplete records and overdue corrective actions. Do not reward low alarm counts alone: suppressed alarms make that metric look better while increasing risk.
9. Questions technicians ask AI
These are practical technician-style prompts, not quotations from verified customer correspondence.
“My nickel bath pH is still in specification, but the chart says out of control. Should I add acid?”
Not from the chart alone. Verify the measurement and inspect recent additions, volume and sampling conditions. The signal indicates changed behavior, not a proven acid requirement. Keep affected loads traceable and follow the signed reaction plan.
“Can I use the same control limits for every nickel tank and both rack and barrel work?”
No. Separate tanks, chemistries and loading methods need representative baselines. Even identical recipes may have different sampling and operating variability. Share the investigation procedure, not automatically the numerical limits.
“After fixing the bath, how many good readings do I need before restarting production?”
There is no universal count. Establish restart evidence during qualification: verified chemistry after mixing, equipment checks and representative product tests. Quality must disposition held batches separately. Passing repeat assays do not erase the original excursion.
10. Source boundary and implementation support
QLQ knowledge-base context: the surface-treatment notes titled Continuous Plating and Zinc-Alloy Plating Defect Analysis and Treatment support checking bath composition, pH, temperature, current, filtration and electrical contacts. Original source-file identifiers are provided in the HTML metadata below. They do not supply validated SPC histories or the simulated trial results above.
Use the linked Nickel Institute source for base chemistry and the linked NIST pages for chart mathematics and signal rules. Plant-specific limits, alarm priorities and release criteria remain subject to engineering approval and the customer's specification.
What Would You Like to Solve?
Share your bath type, tank volume, sampling method, recent chart history and affected coating requirement with [email protected]. For coordinated equipment, chemistry and operating support, review turnkey surface-treatment line support. The useful starting point is a traceable problem statement, not an unsupported promise of capability.