Setting a lead-free reflow profile from the paste datasheet

The first lead-free batch comes out of the oven and something is off. A row of 0402 resistors stands on end, a QFN has a dull, grainy fillet on one side, and the boards that went through last look different from the boards that went through first. The oven was set to the profile printed on the paste jar, so the profile “should be right.”

The profile on the jar is not the profile the board saw. It is the manufacturer’s guideline for the paste, measured on their test vehicle, and it says nothing about your board’s copper, your connector’s thermal mass, or how your oven’s zones actually behave with two boards in it instead of a full belt. Getting from that datasheet to a working profile is a small piece of engineering, and most of the defects a small line sees trace back to one of a handful of places on that curve.

What the paste datasheet is actually telling you

A lead-free paste sheet gives you four or five numbers, and each one guards against a different failure. Reading them as a shopping list, rather than as one shape to copy, is what makes the sheet useful.

Ramp rate and maximum slope are two different limits. Indium’s process note on reading a paste spec draws the line clearly: the maximum slope is the steepest instantaneous heating rate, usually in the first zone, and its purpose is to protect components from thermal shock. The average ramp rate from room temperature to peak is a separate number that protects the paste itself. It sets how fast the flux volatiles leave, and a ramp that is too fast is the usual cause of solder balling, beading and tombstoning. The note gives a commonly recommended upper slope of 3 °C per second and an average ramp in the 1 to 2 °C per second range (Indium, max slope vs. ramp rate). An oven that hits the average but spikes at the entrance can pass the first number and fail the second.

Soak is optional, and the sheet usually says so. Most modern pastes are written for a ramp-to-peak profile, a near-linear climb to the reflow zone, and offer a soak as the fallback. Indium’s own guidance is that the soak exists to let the largest thermal masses catch up with the smallest before anything melts, and that it should be used only when thermal uniformity cannot be achieved otherwise (Indium, ramp-to-peak). A soak that a board does not need keeps the flux active longer than it has to be, and Circuitnet’s process experts add the other cost: the laminate spends longer above its glass-transition temperature (Circuitnet, ramp vs. soak). Soak profiles were the default when infrared ovens produced large gradients across a board. Forced convection changed that, which is why the paste vendors now lead with ramp-to-peak.

Time above liquidus and peak are a window, not a point. SAC305 melts near 217 °C. The sheet gives a band for how long the joint should stay above that and a band for the peak. Both bands have to hold at the coldest joint on the board, and the upper edge of the peak band has to hold at the hottest component. On a board with one big shielded inductor and a field of 0201s, those two constraints pull in opposite directions. That is the whole problem, and it is why you cannot set a profile without measuring the board.

Cooling rate matters too. A fast, controlled cool gives a finer grain structure and brighter fillet. Most sheets specify a maximum, and a small oven with a short cooling zone can fail it by letting the board sit in the last hot zone while the belt stops.

Profiling a board you have two of

Profiling means attaching thermocouples to the board, running it through the oven, and reading what each joint saw. A small line skips this because it feels like a factory step. It is the one step that turns the datasheet into a real profile, and it needs one sacrificial board, a profiler or a logging thermometer that survives the oven, and an hour.

Where the thermocouples go decides whether the data means anything. The standard practice is a minimum of four: one in the air in front of the board, one at the hottest point, one at the coldest point, and one on any temperature-sensitive part. Hottest points are bare copper pads and small passives. Coldest points are under plastic-bodied packages and on the leads of coils, transformers and connectors (I-Connect007, practical thermal profiling). If there is a BGA, the joint that matters is under the center of the package, and the honest way to measure it is to drill from the back to a pad and attach there.

How they are attached matters almost as much. High-temperature solder is the preferred attachment. A thermocouple taped down with polyimide tape reads the tape, not the joint, because the tape is an insulator, and Indium’s profiling guidance says plainly that it is not recommended for that reason. Thermally conductive epoxy or aluminum tape is the fallback when solder is not possible, using as little material as possible so the attachment itself does not add thermal mass (Indium, attaching thermocouples). Keep the profiler body well behind the board so its own mass does not drag the readings.

The number to look at first is the spread between the hottest and coldest thermocouple at peak. The profiling guide above treats a spread of about 10 °C or less at peak as the goal on a large board. If the spread is bigger than the gap between the paste’s minimum peak and the most sensitive component’s maximum, no profile exists that satisfies both, and the fix is on the layout or in the parts list, not in the oven. That is worth knowing before the boards are built, which is one reason we look at thermal mass distribution during layout review and not only at the assembly stage.

Reading defects back onto the curve

Each common lead-free defect maps to a region of the profile. The mapping is not perfect, but it tells you where to look first.

Defect	Look first at	Why
Tombstoning	Ramp through melting	Unequal wetting at the two ends; slower, more uniform heating through liquidus evens it out
Solder balls, beading	Early ramp	Volatiles driven off too fast spatter paste off the pad
Grainy or dull fillets, poor wetting	Peak and time above liquidus at the coldest joint	The joint did not get hot enough for long enough
Voids under QFN and power pads	Peak, time above liquidus, and when the flux activates	Flux outgassing trapped in the molten joint
Discolored laminate, popped caps	Peak at the hottest point, total time hot	The hottest component exceeded its limit

Voiding deserves one more sentence, because it is where the datasheet and the board most often disagree. KIC’s white paper on voiding attributes it to flux outgassing while the solder is molten, and names peak temperature, time above liquidus, ramp rate and soak as the profile levers. Its practical advice is narrower: use ramp-to-peak rather than a soak when chasing voids, and note that the greatest consensus favors a higher peak, with a longer time above liquidus helping wetting and reducing flux entrapment. Ramp rate matters because it sets when the flux activates and the volatiles leave, but the paper does not say a slower ramp reduces voiding on its own. Its example window moves a common lead-free window from 45 to 75 seconds above liquidus and a 235 to 255 °C peak to 60 to 90 seconds and 245 to 265 °C, and warns that this is close to some components’ maximum temperatures (KIC, reflow profiling to minimize voiding). Moving peak and time above liquidus up to chase voids is legitimate only if the hottest thermocouple says the most sensitive component can take it. Component datasheets cite a peak classification temperature from the JEDEC moisture and reflow sensitivity standard, and that number is the limit the part survived qualification at, not a recommended process temperature.

Two things that look like profile problems are not. Paste that was cold when it was printed, or that sat open on the stencil past its working life, prints badly and reflows worse, and no oven setting recovers it. AIM’s handling guidance is representative: let the container reach room temperature on its own before opening, with several hours preferred, and get the printed board into the oven promptly rather than letting the deposit dry (AIM, solder paste handling). Storage temperature and shelf life differ by paste, so read the sheet for the one on your bench. Moisture in the components is the other impostor. A part that popped or delaminated in the oven usually needed baking, which we covered in When SMT parts that sat open need baking before reflow.

Keeping the profile once you have it

A profile is tied to a board, an oven and a load. Change any of them and re-check. The things that quietly change it on a small line:

• Load. One board in a batch oven sees a different air mass than four. A profile set on the first board of the day may run hot on a full tray.
• Panel versus single. A panelized run has more copper and rails than the singles used to profile. Profile the panel.
• Paste lot or vendor change. Even the same alloy from another vendor carries a different flux with different ramp needs. Re-read the sheet.
• Layout revision. Adding a ground pour or a shielded inductor moves the coldest joint.

Keep the profile record with the board revision, not with the oven. When a board comes back for a second build a year later, the profile and the thermocouple positions are part of its build documentation, the same as the stencil file.

For boards we assemble, we profile against the actual paste datasheet and the components’ limits and keep that record with the revision; sending the paste you specify, the assembly drawing and any parts with a reflow limit below the usual is enough for us to start.