A roll-to-roll magnetron sputtering machine is usually bought on the strength of a specification sheet and a short demonstration run. Both are useful. Neither one describes how the machine behaves in the fourth hour of continuous coating, on a web that is not the supplier's test strip. The risks that decide whether the investment pays back are the ones that appear only while the line keeps running, and each of them can be closed with evidence you ask for before the order is placed.
What changes when the web never stops
In a batch chamber, the load goes in, the coating runs, and the door opens. If something drifts, the operator sees it at the next door opening, on one small sample. In a roll-to-roll line the substrate moves continuously through a deposition zone while fresh substrate keeps arriving, so a slow fall in sputter rate, a web that warms degree by degree, or a tension setting that changes as the supply roll shrinks in diameter all accumulate over hundreds of metres of material before anyone measures anything. The failure modes are not exotic. They are the ordinary ones, made expensive by the fact that the machine does not stop to show them to you.
Risk one: throughput that only exists at the test point
The product page for PowerDe's roll to roll magnetron sputtering coating machine describes the series and the applications it serves, but it does not publish a line speed. That is normal for this class of equipment, and it makes the speed the first number to request in writing, together with the conditions under which it holds.
A web speed measured on a bare test strip with a thin sacrificial coating is not the same number as a production speed. Ask for the speed with the target material, the power, the working pressure, the gas, and the film thickness attached. Then convert it. Speed multiplied by the length of the deposition zone gives the dwell time the web actually gets. The thickness you require, divided by the rate your power produces, gives the dwell time you need. If the second number is larger than the first, the line is slower than the brochure says, and the only fixes are more sources, more zone length, or a lower line speed.
Then calculate per shift rather than per hour. Roll changes, load-lock pumping, target changes, cleaning and threading all sit inside the day. A line that is fast between stoppages can still deliver a modest output, and the shift calculation is what tells you whether the machine meets the volume in your business case.
Risk two: uniformity across a web that is wider than a coupon
A batch chamber coats a small, flat, rigid sample. A moving web has edges, it flutters, its temperature changes along the run, and the erosion profile of the target moves as the target is consumed. Thickness variation across the width is the parameter that most often survives commissioning and then quietly destroys yield, because it is invisible in a single measurement.
Ask how uniformity is measured: across how many positions, at what points along the run, with what instrument, and whether the figures are taken hot or after cooling. Ask for a cross-web profile produced on your own material rather than a generic coupon, and ask for one at the start of a target's life and one near the end. Then ask what the correction method is, whether it is movable shutters, source-to-web distance, or a shaped magnetic field, and whether the operator can adjust it or whether it is fixed inside the machine.
Risk three: web handling, tension, and the cost of one break
On a coated roll, one web break costs the material inside the chamber, the run, and usually a clean-down before coating can restart. Thin polymer webs are easily damaged at the edges, and every splice that passes through the vacuum chamber is a chance for a flap to lift or tear.
Most of the practical difficulty on a line like this lives outside the sputter source, in the unwind and rewind. Ask how tension is set and how it is held as the roll diameters change, whether edge position control is included, what the threading procedure is after a break, and how long a restart takes in practice. Ask for the minimum substrate thickness the line can run reliably and the maximum, the splice type the machine tolerates, and the web width available after the guides are fitted. These answers belong in the quotation, not in a conversation.
Risk four: the heat budget of the substrate
Sputtering puts energy into the web as well as into the film. A polymer substrate that survives a two-minute test may discolor, shrink, or relax once it has been under the source for a long run at the power your thickness requires. The product page does not publish a substrate temperature limit, so the heat budget should be treated as an open item rather than an assumption.
Ask what the web temperature reaches at the speed you intend to run, and how that figure was obtained, since a contact reading, an infrared reading and an inferred calculation will not agree with each other. Ask what is available to control it: chilled rollers, backing gas, reduced duty cycle, or a change of source-to-web distance. Then verify it during the trial with a sensor on the web itself, not on the chamber wall, and with the run long enough to reach a steady state.
Risk five: repeatability, target life, and particles
A rate that is correct on Monday and drifting by Wednesday is a process you do not own. Ask what the expected target utilisation is, how the rate is monitored during a run, and what the end-of-life criterion for a target is. Ask how often the chamber must be cleaned, what generates particles between cleans, and what keeps them out of the film.
The product page states that the sputtering chamber is made of SUS 304 stainless steel with the inner surface, including the shielding plates, mirror polished to obtain a clean working environment and a good vacuum level. That construction is exactly what particle control depends on, because a smooth, low-outgassing interior pumps down faster and gives loose material fewer places to hide. It is still worth asking how the shielding is removed for cleaning, how long a clean-down takes, and how many runs are expected between cleans.
Ask for run-to-run data rather than one successful run: two or three consecutive rolls on the same recipe, with the measurements from each one.
Risk six: spares, consumables, and support distance
A vacuum system is a collection of consumables, and the response time of the nearest engineer matters more than any line in a brochure. Ask which items are expected to be replaced, at what intervals, at what price, and who holds the stock: feedthrough seals, shutter drive parts, pump service kits, power supply fuses, gauges and sensors. Ask whether they are standard parts available locally or proprietary parts that ship from the factory.
Ask who commissions the machine, in what language support is given, and what the response commitment is. Ask what the warranty covers and what voids it, including whether your own operators are allowed to open the chamber. Then ask for the interface data your building needs before the machine is built: installed electrical load, cooling water flow and temperature, compressed air, floor space, door width, and lifting access. None of these appear on the product page, and all of them are civil and utility decisions with long lead times.
Risk seven: scope, and what the description actually promises
The same product page describes the series as serving research and development work in perovskite solar cells, OLED, lithium battery, quantum-dot LED and organic photovoltaic films, states that it is capable of producing nanoscale oxide, nitride, metal and other thin film materials, and also mentions the evaporation of large-sized perovskite devices and batch evaporation. That wording deserves attention before money changes hands. Sputtering and evaporation are different processes, and a series description that covers both is not a specification for the machine you would receive.
Close this risk with a configuration sheet rather than a discussion: which process the delivered machine performs, the usable web width, the number and type of sources, whether load-lock, unwind and rewind are included, and which items are quoted as options. A risk that is only visible in a description disappears when the same scope is written on a numbered quotation.
A trial on your own web, with the acceptance numbers agreed first
Almost every risk above is closed by one action: a run on your substrate, at your thickness, for a full working day, with the acceptance figures agreed before the run starts and not after it. Send your own sample rolls and your target film requirement in writing, and ask the supplier to confirm the parameters they will run. Ask the trial record to include the cross-web profile at the start and at the end, the web temperature during the run, the number of breaks and stoppages, the time taken to restart, and the measurements from each roll.
Send the record itself to the supplier rather than a summary of it. A summary invites agreement; a record invites corrections, and the corrections are where the useful information is.
A one-page risk register before the purchase order
Put the whole review on one page: the risk, who owns it, the evidence that closes it, and the date that evidence is due. Throughput, uniformity, web handling, substrate temperature, repeatability, serviceability and scope fit on a single sheet, and that sheet changes the conversation from a brochure to a list of documents.
If a supplier can hand over run data on your material, a stated measurement method, a consumables list with prices, and a configuration sheet, most of the investment risk has already moved to their side of the table. If they cannot, the gap in the answers tells you what you are being asked to buy on trust.
Where the coating step is part of a wider line, review the automated production line solutions so that handling, drying and inspection are planned together with the deposition step instead of after it. To start the conversation with the specific numbers listed above, send your substrate sample and film requirement through the contact page.
