Learn
OES Sample Preparation: How to Get Accurate Spark Spectrometer Readings
Sample preparation is the single largest source of error in spark OES analysis. This guide walks through surface preparation, electrode maintenance, argon purity, and the five most common mistakes that throw OES readings out of tolerance — with practical fixes that take minutes, not hours.
The 80/20 rule of OES accuracy
In ten years of installing OES instruments across foundries in 40+ countries, JIEBO engineers see the same pattern: instrument hardware is the cause of inaccurate readings perhaps 20 % of the time. The other 80 % is sample preparation, argon supply, or electrode wear. The instrument is rarely the problem.
This guide is a checklist. Run through it before you call us about a drift, an out-of-tolerance reading, or a "broken" instrument. Most calls we receive resolve at one of the five mistakes below.
1. Surface preparation — flat, fresh, dry
The spark must strike a flat, conductive, oxide-free surface. The standard preparation is to grind or mill the sample so the analysis area is a smooth disk of bare metal at least 20 mm in diameter. Three rules: flat (no rocking on the spark stand), fresh (analyzed within 5 minutes of grinding so surface oxide does not regrow), dry (no coolant residue, no grinding oil).
Recommended grinding media: 60-grit aluminum oxide belt for iron-base samples, 80-grit silicon carbide for aluminum and copper alloys. Do not use the same belt across base matrices — silicon carbide cross-contaminates aluminum readings with embedded Si grit.
2. Argon purity — the silent killer
OES requires argon at 99.999 % (5N) purity. Lower-grade argon contains oxygen and nitrogen that quench the plasma and shift element intensities — especially for low-emission elements (C, S, P, B). A 100 ppm oxygen contamination can shift the carbon reading by 10–30 % at production-grade levels.
Symptoms of bad argon: low elements suddenly drift low, calibration drift accelerates, the spark sounds different (more crackle, less clean snap). Fix: replace the cylinder, verify with a known reference sample, and check the argon regulator and lines for leaks. Cheap fitting tape on argon lines is a frequent culprit.
3. Electrode wear and gap
The counter-electrode (the small tungsten or silver pin opposite the sample) wears down with every burn. As it wears, the spark gap widens, the discharge characteristics shift, and intensities drift. JIEBO standard is to clean the electrode every 50–100 burns with a wire brush and re-tip every 1,000–2,000 burns depending on matrix.
For iron-base sparking, tungsten electrodes typically last 1,500 burns. For aluminum and zinc (which deposit material on the electrode), expect 800–1,200 burns before re-tipping. Keep a log: burns since last re-tip is one of the four numbers an OES technician should know without looking.
4. The sample stand and clamping pressure
A flat surface that does not clamp tightly to the spark stand is no better than a rough surface. The sample must seal the argon flow path; an uneven seal leaks air into the plasma and produces the same symptoms as bad argon. Test by listening: a clean seal produces a sharp single-pitch spark sound; an air-leaking seal sounds raspy.
For irregular samples (chip drillings, broken cast iron pieces), use the sample fixture supplied with the spectrometer. JIEBO's Innovate T5 and Noble T7 ship with adjustable holders that accommodate samples down to 15 mm diameter. The Surpass F1 mobile OES has a specialized fixture for irregular field samples like rebar and large castings.
5. Calibration drift vs sample-prep drift
Calibration drift is slow — a few percent over months. Sample-prep drift is fast — readings change between consecutive burns of nominally the same sample. If your sample-to-sample variation exceeds 1 % RSD for a stable matrix like Mn in steel, it is sample preparation, not calibration. Recalibrating does not fix a sample-prep problem; it just hides it for a few shifts.
Daily routine: run a control sample (a piece of certified reference material) as the first burn of every shift. Log the result. A trend chart of the control-sample carbon over 30 days will tell you instantly whether you have argon problems, electrode wear, or a real calibration drift.
Quick-fix table
When a reading is out of tolerance, run through this table top-to-bottom. Most problems resolve in the first three rows.
| Symptom | Likely cause | Quick fix | Time |
|---|---|---|---|
| All elements low simultaneously | Argon contamination or leak | Swap argon cylinder, check fittings for leaks | 10 min |
| Light elements (C, S, P, B) drift, metals stable | Argon humidity / oxygen contamination | Replace argon line dryer; check 5N grade | 15 min |
| Intensity drops gradually over a shift | Electrode wear | Clean electrode with wire brush; re-tip if needed | 5 min |
| Single-sample-to-sample variation > 1 % RSD | Surface preparation inconsistent | Re-grind with fresh belt; analyze within 5 min | 10 min |
| Calibration shifts over 1–4 weeks | True calibration drift | Recalibrate with CRMs | 30–60 min |
| Crackling spark sound | Sample-stand seal leaking air | Clean sample stand, check clamp pressure | 5 min |
| Aluminum readings high on Si only | Cross-contamination from SiC belt | Switch to dedicated Al-grade abrasive | 5 min |
Frequently asked questions
How fresh does "fresh" really mean?
Analyze within 5 minutes of grinding for iron and copper alloys. For aluminum, less — surface oxide forms in 60 seconds at room humidity, so analyze immediately after grinding for aluminum work.
Do I need a dedicated grinding belt per matrix?
Yes for aluminum and copper sample matrices. Silicon carbide grit embeds in aluminum and copper surfaces and shows up as a false Si peak; keep a separate aluminum-oxide belt for those matrices.
What argon flow rate should I run?
Approximately 3 L/min during sparking is the JIEBO Innovate T5 / Noble T7 specification. The instrument controls flow automatically; your job is to ensure the cylinder regulator outputs 0.3 MPa and the line is leak-free.
Can I detect bad argon without buying expensive gas analyzers?
Yes — burn a certified reference sample twice, swap to a fresh argon cylinder, burn again. If carbon, sulfur or phosphorus shifts more than 5 % between cylinders, your "old" cylinder was contaminated.
How often should I clean the spark stand?
Daily wipe-down at shift start, full cleaning weekly. The spark stand accumulates argon-fired material from previous samples; over a few weeks this becomes a cross-contamination source.