A glove-box spin coater is sold as one machine, but it is really two instruments joined together: an inert-atmosphere enclosure and a spin process head. When a perovskite film comes out the wrong thickness, or shows a pinhole pattern you have never seen before, the first question is which half of the machine caused it. That question can only be answered if the numbers were recorded while the equipment was being tested. The list below is the data set I would want in hand before signing an acceptance report.
Start with the atmosphere, because it sets the ceiling on everything else
Water and oxygen are the variables that quietly ruin thin-film results, so they need readings taken at several moments rather than one screenshot of a green display. Record the parts-per-million values for H2O and O2 at three points: right after the initial purge, immediately after an antechamber transfer, and halfway through a shift with gloves actually being used. A box that reaches its target only while nobody is working inside it is a different machine from one that holds that level under load. The published specification for the PowerDe glove-box series calls for water and oxygen below 1 ppm, and the water probe supplied with the system covers a 0-500 ppm range, which is wide enough to catch a bad transfer before it reaches the film.
Alongside the steady-state readings, capture the recovery time: how long the box takes to return to spec after the antechamber has been opened and a substrate has been passed in. Time the worst realistic cycle, not the best one, and note the analyzer model and its calibration date next to every number. A reading without a calibration date is an opinion.
- Overall leakage rate, measured over a defined hold period with the box sealed and idle. The stated figure for this series is below 0.05 vol% per hour; ask the vendor which instrument and which procedure produces that number, and repeat it yourself.
- Chamber pressure trace during real work. The automatic control band is specified at +15 mbar and -15 mbar. A pressure swing that pulls outside air through a glove port will defeat the purification loop long before the analyzer alarms.
- Room temperature and relative humidity during the test. The equipment is specified for a 15-30 ℃ working environment, so any result obtained outside that envelope is not comparable to the datasheet.
Then the spin process itself, measured on your substrate
Datasheet spin uniformity is quoted at ±3%, but that figure belongs to the vendor's test coupon and their ink. Verify it on the substrate and formulation you actually run. Measure thickness at a fixed grid, center, mid-radius and edge, with the same instrument and the same protocol on every run, and keep the measurement sheet with the run record. A single average thickness tells you almost nothing; a radial profile tells you whether the problem is dispense volume, ramp rate, or wetting behaviour.
Record the speed program as it actually ran, not as it was typed. The published range is 300-8500 rpm for silicon wafers up to 4 inches and 300-6000 rpm above that, and the shape of the curve matters more than the peak: ramp rate, dwell time, and the moment the dispense ends relative to spin start. Note the dispensed volume, the dispense position, and any delay between dispense and spin, because an unrecorded pause of a few seconds is enough to change film thickness on many formulations.
- Repeatability: run one recipe at least ten times back to back, then run it once more after the box has been opened and re-purified. Report the spread, not the best result.
- Chuck behaviour: confirm the vacuum hold, and watch for visible wobble or misting at the edge during ramp-up. Note the substrate size and whether the chuck was repositioned between runs.
- Ergonomics under real conditions: the box measures 1950 × 780 × 900 mm, the spin system occupies a 400 × 400 mm envelope, and the glove ports are 220 mm. Write down whether an operator can load a substrate and reach the chuck without fighting the gloves; an awkward station produces process drift that no sensor will explain.
- Defect photographs taken under the same lighting and magnification every time, so that a change in the film is a change in the film and not a change in the camera.
Control system, alarms and the record itself
The machine is operated through a touch screen and PLC, which makes it easy to assume the logging is already adequate. Check it before you rely on it. Recipes should be stored with a version number and a timestamp. Editing a speed program should leave a trace. Alarms for low purifier capacity, high humidity, door interlock, and loss of chuck vacuum should be written to a log with timestamps and exported in a format you can open without the vendor's software. Ask what happens when power drops mid-spin on a 220 V / 50 Hz supply: whether the run aborts cleanly, whether the recipe survives, and whether the event is logged. A machine that fails silently will cost you a week of unexplained film data.
Get the calibration certificates for the water and oxygen sensors, the recommended recalibration interval, and the replacement cost of the purifier media. Those three items decide the real cost of ownership far more than the purchase price.
Collect the transfer data while you are still in the lab
Most glove-box spin coaters are bought for development and then asked to hand a process to a line. The handover fails when the lab record stops at "film looked good". Write down the ink batch, solids content or molar concentration, filter size, dispense volume, substrate temperature, atmosphere readings at the moment of coating, and the anneal temperature and time. Those are exactly the fields someone will need when the same formulation is compared against slit coating equipment, where the coating window is set by web speed and head pressure instead of rpm. If the plan is a move to perovskite production line equipment, the lab record is the only baseline that makes a scale-up comparison honest.
What the acceptance pack should contain
| Data item | How to record it | What to compare it against |
|---|---|---|
| H2O and O2 levels | After purge, after transfer, mid-shift | The specified sub-1 ppm target, under load |
| Atmosphere recovery time | Timed on the worst realistic cycle | Your throughput plan, not the demo |
| Leakage rate | Sealed hold over a fixed period | The stated 0.05 vol%/h figure |
| Pressure trace | Continuous, while gloves are worked | The ±15 mbar control band |
| Speed program | Actual rpm trace with ramp and dwell | The recipe you intended to run |
| Thickness map | Fixed grid on your substrate and ink | Your process window, then the quoted ±3% |
| Run-to-run spread | Ten consecutive runs plus one after re-purge | The spread your process can tolerate |
| Alarm and event log | Exported file with timestamps | Every fault you induced during the test |
| Sensor calibration certificates | Copies in the acceptance pack | The recalibration interval you will have to meet |
Treat population figures with caution
Specifications also carry numbers that describe a fleet rather than your process, such as an equipment failure rate of 1% or a product pass rate above 99%. Neither is wrong, but neither predicts your Tuesday morning. Ask what conditions produced them, over how many machines and how many hours. Then trust the readings you took yourself, on your substrate, with your ink, recorded in a file you can open next year.
Before you sign
Atmosphere data, process data, and evidence that the machine keeps its own record: those three columns turn a film defect from a mystery into a diagnosable event. If you are evaluating a glove box spin coating equipment for perovskite, OLED, quantum-dot, OPV or lithium-battery work, send us your substrate size, your speed program and your target film, and we will run that test protocol against the machine before it ships.
