Thermally Assisted Removal: Where the Heat Goes and How Much on Threadlocked Micro Screws

The most common standoff in rework is a single M1.0 screw with one corner of its head already rounded off, where more torque means cam-out and a little more means a sheared head, while medium-high strength threadlocker holds it firm. The standard answer is heat. But a quick pass with a hot air gun and delivering heat precisely where it is needed, then removing within the correct temperature window, are separated by a full set of parameters.

First, why heat works. Anaerobic threadlockers are acrylic systems that cure into a crosslinked polymer network filling the thread clearance and locking it. Above the glass transition temperature, chain mobility increases and modulus and shear strength fall rapidly, dropping breakaway torque to a fraction of its original value. The softening range differs sharply by strength class. Medium strength, usually marked blue, softens noticeably between 150 and 200 degrees Celsius. Medium-high and high strength, marked red or green, typically need 230 to 250 degrees Celsius to release without damaging the parts. Treating these two classes alike is why so many attempts end with "we heated it and it still will not move" — the temperature was simply too low.

The real conflict is that heat does not act on the adhesive alone. The conduction path runs from the screw head through the shank to the engaged thread and into the clamped parts, so heat enters at the head and travels backwards. For the engaged thread to reach softening temperature, the head must be hotter still, and both the head and the clamped parts are usually more heat-sensitive than the threadlocker. This is the thermal budget: the usable ceiling is set by the most fragile component on the workpiece, not by the adhesive.

Working out that budget starts with listing the temperature limits on the workpiece. Lead-free solder on a PCB reflows near 260 degrees Celsius, but only for seconds; prolonged dwell above 200 degrees Celsius risks popcorning and delamination. Lithium cells should never exceed 60 to 80 degrees Celsius under any circumstances. ABS parts begin to soften around 90 degrees Celsius and PC around 130. Camera modules, FPC connectors and already-mounted electrolytic capacitors each carry their own ceiling. Around a single screw there are often three or four such limits at once, and the lowest one governs the station. This is why small screw removal equipment with thermal assistance normally uses a small, positionable heating unit rather than a broad heat source: the smaller the footprint, the easier the thermal budget is to hold.

Choosing a heating method is essentially a trade between footprint and controllability. Hot air is convective: wide coverage, steep gradients, heat spilling onto neighbouring components, but inexpensive and flexible, which suits a rework bench. A contact heater conducts heat into a concentrated, repeatable spot, at the cost of contact pressure and interface consistency — press it poorly and no heat goes in. Induction acts on metal only, non-contact and fast, concentrating heat in the screw and metal insert and staying comparatively kind to nearby plastics and PCB, yet it is equally sensitive to nearby shields, standoffs and heat sinks and needs a workpiece-specific coil shape. Infrared and laser are point heaters operating in milliseconds to seconds with a small heat-affected zone, at the cost of price and the need for precise aiming. A whole-plate oven suits only workpieces that can be baked as a unit and is the least flexible option.

Measuring temperature is where more mistakes happen than in choosing a setpoint. The reliable method is a thermocouple bonded or welded directly to the screw head, reading actual head temperature and closing the loop on heater power. Infrared sensing looks convenient but emissivity varies enormously with surface condition — zinc plated, black oxide, stainless, already oxidised — and readings can differ by tens of degrees, making it a trend indicator rather than a control input. Another frequent error is measuring the heater outlet instead of the workpiece, two values that diverge widely in contact heating.

Heating also produces a secondary effect: the head surface changes. Oxidation, discolouration and bluing alter reflective behaviour, and a vision template threshold calibrated beforehand may no longer hold, showing up as localisation failures or false rejects. This deserves particular attention on stations equipped with an automatic vision screw removal machine. The practical fix is to split the recipe into separate parameter sets for pre-heat localisation and post-heat verification rather than sharing one. Otherwise the symptom observed on the floor is that the machine cannot find the screw after heating, while the root cause is a change in imaging conditions.

The time window is a parameter too, not simply a matter of removing once it is hot. Ramp rate should not be too fast, since rapid heating creates local overshoot and thermal stress, especially on thin walls and multilayer boards. Soak time must be long enough for heat to travel from the head to the engaged thread, which varies widely with the heat capacity and conductivity of the clamped parts: seconds for a small M1.0 assembly, considerably longer on a heavy metal housing. Removal must happen before the adhesive re-stiffens; too early and it has not softened, too late and strength recovers and torque climbs again. Standard practice is a trial temperature-time-torque curve to locate the plateau where torque drops clearly, then placing the process window in the middle of that plateau rather than at its edge.

The cost of overheating usually appears after the screw is out. Grade 8.8 and 10.9 screws temper above 200 degrees Celsius, hardness falls, and the torque-angle curve of the subsequent re-tightening can no longer be trusted — the screw may yield before reaching nominal torque. Plating oxidises and blues, failing cosmetic criteria outright. Plastics soften and deform, changing assembly clearances. Local PCB delamination leaves electrical risks that only surface at test. Carbonised adhesive residue turns into hard particles that contaminate nozzles and bits and degrade pick-and-place accuracy downstream. And the step most often skipped: residual adhesive in the threaded hole must be cleaned, by nylon brush or dedicated cleaning, or the residue cushions the re-tightening and measured torque reads systematically low.

A complete thermal removal recipe therefore contains seven items: heating method, heater power or setpoint, target temperature and measurement point, soak time, removal torque ceiling, cooling method and residue cleaning method. Written as these seven, the process is reproducible. Written as "heat then remove", results change with the operator and the batch. In projects involving threadlocked fasteners, an M0.8-M2.0 screw removal solution normally fixes these items as recipe parameters in the machine program, called up by screw size and adhesive class. The point is not to complicate the process but to turn "can it be removed" from a matter of feel into a repeatable result.