Short answer: take the foam tape off this joint and bed the lens on a two-part RTV silicone with a silane primer on both faces, in a 0.5-1 mm [0.02-0.04 in] glue line. That combination holds a polycarbonate lens on an anodized aluminium housing through -40 to 85 C, road vibration and a 1 m [39 in] dunk. If the lens carries structural load, use a flexible two-part epoxy at the same 1 mm line. If you want the assembly re-openable, run a butyl band in a groove, the way headlamps do it. The tape prototypes failed for two physical reasons: the sealed anodize they sat on, and the 45 x 10-6/K expansion mismatch between polycarbonate and aluminium. Both have a fix, and each fix is a number.
Why the tape leaked on the anodize
Anodizing grows a thin ceramic oxide on the aluminium, and the sealing bath that follows most anodizing closes the oxide's pores. To an adhesive that sealed layer is a smooth, low-energy ceramic that wets poorly, which is the classic failed-tape result: weak adhesion first, leaks right behind it. The oxide itself can work in your favour. Bonding studies on phosphoric-acid-anodized aluminium show the adhesive flowing into the micro-porous oxide and locking into it, which is why aircraft bond shops specify unsealed anodize on bond lands. If you control the anodizing specification, order the land unsealed or phosphoric-anodized, and leave the rest of the housing sealed against corrosion.
The tape also needs a geometry most lens flanges lack. Foam-tape design guidance calls for about 70 % contact on a flange flat within 0.1 mm [0.004 in], laid as a die-cut ring in one closed loop. Straight tape runs and stamped frames leak at the ends of the runs, where water wicks along the foam. Above about 90 C [194 F] continuous service the tape class is out of range as well. The same oxide questions come up in aluminium enclosure oxide and finishes.
The chemistries that hold this lens
The values below are typical data-sheet ranges for each chemistry class; a named product's own data sheet wins over any class range.
| Chemistry | Lap shear on prepared aluminium | Elongation at break | Cure | Comes apart | Where it wins |
|---|---|---|---|---|---|
| Two-part RTV silicone + silane primer | 1-2 MPa [150-290 psi] | 200-600 % | 4-24 h at 23 C, no air needed | cut and re-bead | the default for a sealed lens |
| Flexible two-part epoxy | 10-20 MPa [1450-2900 psi] | 100-150 % | hours at 23 C, post-cure helps | one-way | the lens carries load |
| Structural acrylic (MMA) | 15-25 MPa [2200-3600 psi] | 50-100 % | minutes to fixture | one-way | grips lightly oiled metal with minimal prep |
| Butyl band | seal only, no strength | stays putty | applied hot into a groove | re-softens near 65 C [150 F] | joints you re-open routinely |
| VHB-class acrylic foam tape | 0.7-1 MPa [100-150 psi] | several hundred % | immediate | peels | flat stiff flanges, sheltered duty |
Two more classes sit at the edges of this choice. A moisture-cure polyurethane of the marine lens class gives a strong, flexible, permanent bond after a denatured-alcohol wipe, and suits one-off shop work well. UV-curing adhesives cure only where light reaches: polycarbonate transmits UV and anodized aluminium stops it dead, so a UV adhesive suits a thin edge fillet cured through the plastic, in a 0.1-0.5 mm [0.004-0.02 in] line, and nothing wider under an opaque flange.
The mismatch that breaks stiff adhesives
Polycarbonate expands at about 65-70 millionths per kelvin and aluminium alloys at about 23, so the pair moves relative to itself by 45 x 10-6/K. A stiff, thin bond converts that motion into stress at the interface. A soft, thick line converts the same motion into harmless shear inside the glue. The working rule for cyclically bonded joints: shear strain = (relative expansion x span x temperature swing) divided by the glue-line thickness. Keep that strain under about a third of the adhesive's elongation at break, and the bond cycles for years. A silicone at 200 %+ elongation has room to spare. A rigid structural epoxy at a few percent elongation has none, whatever its lap-shear number says.
Sizing the glue line for a 100 mm lens from -40 to 85 C
The case: a 100 mm [4 in] polycarbonate lens bonded at 23 C [73 F] onto an anodized aluminium flange, cycled to an 85 C hot soak and a -40 C cold soak, the usual automotive exterior range.
Inputs:
- Bond span: L = 100 mm, the lens diameter.
- Relative expansion: 45 x 10-6/K, which is 68 for polycarbonate minus 23 for aluminium.
- Worst swing: 85 C - 23 C = 62 K on the hot side; the cold soak gives 63 K, the same magnitude.
Rule in words: differential movement = relative expansion x span x temperature swing, and shear strain = that movement divided by glue-line thickness.
Arithmetic:
- Movement: 45 x 10-6/K x 100 mm x 62 K = 0.28 mm.
- In a 0.2 mm [0.008 in] thin rigid line: 0.28 / 0.2 = 140 % strain.
- In a 0.5 mm line: 0.28 / 0.5 = 56 %.
- In a 1.0 mm line: 0.28 / 1.0 = 28 %.
Decision: silicone tolerates about 67 % working strain, a third of its 200 % elongation, so 0.5 mm and 1.0 mm both pass. A flexible epoxy tolerates about 50 %, a third of its 150 %, so it needs at least 0.28 / 0.5 = 0.56 mm; order 1.0 mm shims. The thin rigid line at 140 % cracks on the first thermal cycle. Set the glue line at 1.0 mm and every elastomeric chemistry on the table clears with margin.
Preparing the two surfaces
- Abrade the anodized land with a P320-P400 abrasive to open the sealed glaze; you are cutting through the seal, not removing the coating.
- Wipe the land with isopropyl or denatured alcohol on a lint-free cloth and let it dry.
- Wipe the polycarbonate mating face with isopropyl alcohol only; acetone and ketone solvents craze polycarbonate.
- Coat both faces with the adhesive maker's silane primer in a thin film, and give it its flash time, typically 15-30 minutes on silane-primer data sheets.
- Set 1 mm shims or a machined step on the land, lay the bead, and press the lens down onto them evenly.
- Cure 24 h at 23 C before handling. For an epoxy, add a post-cure 5-10 C above your part's maximum service temperature, 90-95 C here, to cut creep. Epoxy cure guidance puts glass transition 10-15 C above the cure temperature, and one widely used flexible aerospace epoxy cures at room temperature to a Tg near 21 C.
- In a closed ring joint, use a two-part addition-cure silicone; a moisture-cure RTV needs air and only cures about 2 mm [0.08 in] of depth per day from the exposed edge.
Wear nitrile gloves and keep air moving while the solvents are open. The same prep sequence applies when bonding anodized enclosure lids without losing the seal.
Designing the land the bead sits in
Give the adhesive a land 4-6 mm [0.16-0.24 in] wide, machined flat within 0.1 mm, with the glue line set by the shims or a machined step. A raised lip holds the bead in compression and makes water climb a step; the same trade-offs drive raised lip or flat face gasket choices. Lay the bead inside the fastener line, so screws carry clamp and adhesive carries seal, then tool a small fillet at the outer edge to remove the place where peeling starts. For a window in an enclosure door, the opening fundamentals live in the cutouts, glands and HMI windows guide, and a bonded display bezel follows the same land-and-line logic in bonded bezel display mounting. A taped unit you inherit is a different job, covered in removing an adhesive-mounted HMI screen from its cutout.
Which chemistry by what happens at teardown
Three choices, three futures:
- Sealed for life: two-part epoxy or structural acrylic. Maximum strength, zero service path.
- Opened once or twice, ever: two-part RTV silicone. Cut the line, re-bead, re-prime.
- Opened routinely: butyl band in a groove; it re-softens near 65 C [150 F] in an oven, which is how every lamp shop opens a headlamp.
Decision for the automotive lens: primer-treated two-part silicone at a 1.0 mm line, because vibration and water reward the softest durable line. Move to the flexible epoxy when the lens itself is structural, say when it carries a boss or a bracket. Keep butyl when the part has to come apart in service, and accept that it seals rather than holds.
What to specify on the drawing
| Spec line | Write this on the drawing |
|---|---|
| Joint | PC lens on machined anodized flange, land 5 mm [0.2 in] wide, flat within 0.1 mm |
| Surface prep | P320-P400 abrade, alcohol wipe, silane primer both faces, 15-30 min flash |
| Adhesive | two-part addition-cure RTV silicone, Shore A 25-40, elongation over 200 % |
| Glue line | 1.0 mm [0.04 in], set by shims |
| Cure | 24 h at 23 C [73 F] |
| Proof test | thermal cycle -40 to 85 C, then 1 m [39 in] of water for 30 min, no water inside |
That last line is the IP67 water test: 1 m of water for 30 min, on top of a dust-tight first digit. Ratings on a finished enclosure are the manufacturer's ratings, and NEMA itself tests nothing. Our enclosures are rated NEMA 4X / IP66 by the manufacturer and not UL listed, so a lens bond earns its rating when your proof test passes.
If the lens is a window in a stainless door instead, draw the opening and the gasket land in the cutout designer or send a drawing through custom stainless enclosures; stainless takes the same abrade-and-prime prep and bonds more readily than sealed anodize. Custom work is delivered in 10 business days from your approval of the drawing. For displays, the HMI enclosures we build come with the screen fitted and an IP65-type front seal with the display installed, manufacturer rated, from the 10.4 in panel PC housing to the 15 in housing.
Enclosures in this article
| Enclosure | Material | Outside H x W x D / display | Rating |
|---|---|---|---|
| 10.4" HMI enclosure | 304 stainless | 10.2 x 12.44 x 3.6 in | IP65-type front seal with the display installed, manufacturer rated; not UL listed |
| 15" HMI enclosure | 304 stainless | 12.91 x 16.06 x 3.88 in | IP65-type front seal with the display installed, manufacturer rated; not UL listed |
Questions
Why did VHB tape leak on polycarbonate lens prototypes?
Two stacked reasons. Most anodizing is sealed in a hot bath that closes the oxide pores, leaving a low-energy ceramic that the tape wets poorly. And foam tape needs about 70 % contact on a flange flat within 0.1 mm, laid as a die-cut closed ring; straight runs leak at their ends. An optically clear foam tape does hold PC on anodized aluminium in non-automotive duty when the flange is truly planar and the ring is die-cut.
Can a UV-curing adhesive bond the lens instead?
Only where light reaches the whole line. Polycarbonate transmits UV and anodized aluminium blocks it, so a UV adhesive suits a thin edge fillet cured through the plastic, around a 0.1-0.5 mm [0.004-0.02 in] line, and joints where at least one part is translucent. A full land hidden under an opaque flange stays soft in the middle.
Will the silicone fog the inside of the lens?
It can, if you pick the wrong cure chemistry. Acetic-cure RTV releases acetic acid while it cures, which hazes optics and attacks nearby electronics. Use a neutral-cure or two-part addition-cure silicone, the low-outgassing type used on aerospace optics, and give a moisture-cure bead its 2 mm [0.08 in] per day of cure-through before you close the joint.
How hot can the bonded joint run?
Silicone elastomers serve from about -60 to 200 C [-76 to 392 F], so an 85 C under-bonnet soak sits far inside that range. A room-temperature-cured epoxy is the one to watch: one flexible aerospace grade cures to a glass transition near 21 C [70 F], so post-cure it 5-10 C above your maximum service temperature before the part ships.
Do I need to strip the anodize before bonding?
Open it, keep it. A light P320-P400 abrasion plus an alcohol wipe cuts through the sealed glaze and exposes the porous oxide the adhesive keys into. If you control the anodizing specification, the strongest surface is an unsealed or phosphoric-acid-anodized land, with the rest of the housing sealed for corrosion. Bond a scrap piece both ways and pull it apart: failure inside the adhesive is the result you want.
How do I get the lens off again for repair?
It depends on the chemistry you chose. Butyl re-softens near 65 C [150 F] in an oven, which is how lamp shops re-open headlamps. A silicone line cuts with a knife and re-beads after re-priming. An epoxy or acrylic joint is a one-way assembly, so specify a replaceable gasket design there if service access matters.





