Two plastic halves have to become one product. That decision — how they join — quietly sets your assembly cycle time, your sealing strategy, your repair policy, your tooling, and whether a returned unit can be opened without destroying it. Teams spend months on wall thickness and rib layout, then pick the joining method in an afternoon. That is backwards: the method constrains the geometry, so it belongs in the design while the parts are still soft. There are three families — chemical, thermal, and mechanical — and each has a natural home.
Chemical joining
Solvent bonding
A solvent dissolves the surface of both parts; the polymer chains interdiffuse and the joint becomes continuous material. Done right the bond is as strong as the parent plastic and optically invisible, which is why acrylic cases and polycarbonate light pipes are solvent bonded rather than glued. It works only on amorphous thermoplastics — acrylic, polycarbonate, ABS, PVC, polystyrene — and does nothing to polyethylene, polypropylene, acetal or nylon. Methylene chloride and MEK solvents carry real ventilation obligations, so plan the workstation, not just the joint.
Cyanoacrylate
Fast, cheap, tempting. It cures in seconds against surface moisture and bonds most rigid plastics, but it is brittle, poor in peel, weak above about 180 °F, and prone to blooming a white haze. Fine for small non-structural bonds and prototypes; bad for anything that flexes or takes impact.
Epoxy
Two-part epoxy gives the highest strength and the best gap filling, bonds dissimilar materials including plastic-to-metal, and resists heat and chemicals. The costs are cure time, metering equipment, and shelf life.
UV-cure adhesive
Cures in seconds under a lamp — epoxy-like strength with a controllable cure. Light must reach the bond line, so at least one part has to transmit UV. Excellent for clear windows, lenses, and medical fluidics; useless in an opaque blind joint.
Whatever the chemistry, surface energy decides everything: polyolefins and acetal are so low in surface energy that no adhesive holds without plasma, flame, or chemical etch pretreatment. If your material choice is still open, weigh the joining consequences while choosing the plastic.
Thermal joining — welding
Ultrasonic welding
The volume default. A horn vibrates at 20–40 kHz; friction at a molded energy director melts a thin layer and the parts fuse in 0.2–1 second. Tooling is a horn and a fixture, typically $3,000–$8,000. It needs a designed joint molded into the part, near-field access for the horn, and compatible polymers on both sides — see ultrasonic welding for joint design and failure modes.
Hot-plate welding
A heated platen melts both mating faces, retracts, and the parts are pressed together. Slow at 10–30 seconds, but it produces the strongest hermetic joints available in plastic, tolerates large irregular parts, and works on polypropylene and polyethylene where ultrasonics struggle.
Spin welding
Circular joints only: one part spins against the other until frictional melt occurs, then stops under pressure. Cheap tooling, fast cycle, excellent seals — the standard for round filter housings and caps. You cannot control final angular orientation without a stop feature.
Laser and vibration welding
Laser transmission welding passes a beam through a transmissive upper part into an absorbing lower part, melting only the interface — no vibration, no particulate, ideal for sealed electronics. Vibration welding is linear friction, used for parts too large for an ultrasonic horn.
Mechanical joining
Heat staking
A molded boss passes through a hole in the mating part and a heated or ultrasonic tip reforms its head into a rivet. Fast, cheap, no consumables, holds PCBs firmly — and permanent.
Screws into plastic
Thread-forming screws designed for plastic run directly into a molded boss. Cheap and serviceable, but each reassembly degrades the thread and three to five cycles is a realistic life. Boss geometry — hole diameter, wall thickness, root radius — matters more than the screw, and it is the most common source of cracked bosses in the field. The rules are in threads in plastic parts.
Threaded inserts
Brass inserts installed by heat, ultrasonics, or molded in place give a real metal thread that survives hundreds of cycles, for a few cents and an operation. They are the correct answer for any enclosure opened repeatedly in service, and installation method drives pull-out strength — see installation methods that hold.
Snap fits
Zero added parts, zero cycle time, zero consumable cost — the cheapest joint that exists and the hardest to design well. The failure mode is a beam that either creeps open over time or cracks on the third disassembly. Snap-fit design is a real engineering exercise, not a late addition.
Selection table
| Method | Works on | Relative strength | Cycle time | Serviceable | Sealed |
|---|---|---|---|---|---|
| Solvent bond | Amorphous only (PC, PMMA, ABS, PVC) | Very high | Minutes + fixture | No | Yes |
| Cyanoacrylate | Most rigid plastics | Low, brittle | Seconds | No | Partial |
| Epoxy | Nearly all, incl. dissimilar | High | Minutes to hours | No | Yes |
| UV-cure adhesive | One part must transmit UV | Medium-high | Seconds | No | Yes |
| Ultrasonic weld | Similar thermoplastics, rigid | Very high | Under 1 s | No | Yes, with shear joint |
| Hot plate | Most, incl. PP and PE | Highest | 10–30 s | No | Hermetic |
| Spin weld | Circular joints, most thermoplastics | Very high | 1–5 s | No | Hermetic |
| Laser weld | Transmissive + absorbing pair | High | 1–10 s | No | Hermetic, clean |
| Heat stake | Most thermoplastics | Medium | 1–3 s | No | No |
| Self-tapping screw | All | Medium | 2–5 s each | Limited (3–5 cycles) | With gasket |
| Threaded insert | All | High | 2–4 s + screw | Yes, hundreds of cycles | With gasket |
| Snap fit | All, ductile preferred | Low-medium | Instant | Yes, few cycles | No |
Choosing without regret
Work through it in this order. Does the joint have to be reopened in service or at end of life? If yes, inserts or snaps — read design for repairability before you argue otherwise. Does it have to seal? Then welding, or adhesive with a compressed elastomer, and the seal deserves its own gland design. Are both parts the same polymer family? If not, welding is out. What is the volume? Below a few thousand units, screws and adhesives avoid tooling; above that, welding and staking repay their fixtures in labor within months. And remember that every method except snaps and screws is permanent — which is a customer-service decision as much as an engineering one. Reducing the number of joints, the core of design for assembly, beats optimizing any of them.
Projects House designs plastic assemblies with the joint chosen up front: joint geometry molded in, process specified, fixtures scoped, and a validation plan for pull, peel, and leak testing. Send your geometry and volume through our contact form.