Crystal LabFerrara · UniFe & INFN

Research › X-ray and γ-ray optics

Crystal bonding

Glue-free assemblies that keep their curvature

Bonded to a substrate figured to the right shape, a crystal takes that shape and keeps it, with no glue and no holder.

The physics.

Anodic bonding of a crystal to a figured substrateLeft: a flat crystal plate above a glass substrate whose surface has been figured to the wanted curvature. Right: heated and under a voltage, with the crystal positive, sodium ions in the glass drift away from the interface, the surfaces are pulled together and bond, and the crystal takes the shape of the substrate. 1 crystal and substrate2 bonded, with its shape crystal plateglass figured to shape heatvoltage + crystal (anode)− cathodeNa⁺Na⁺Na⁺ Si–O–Si bonds at the interface,no glue layer
A flat crystal bonded anodically to a figured glass substrate takes its curvature. Curvature and thicknesses are exaggerated.

Bent crys­tals are usually curved by me­chan­i­cal holders that press or clamp them. The result is never exactly the in­tend­ed shape: bending a plate one way also curves it the other way (an­ti­clas­tic cur­va­ture), clamps can twist it, and the holder adds mass and bulk, a burden for the go­ni­om­e­ters that align crys­tals to a beam. Glue avoids the holder but leaves a layer that creeps and changes with tem­per­a­ture, so the shape drifts.

Bonding joins the crystal di­rect­ly to a sub­strate figured to the wanted shape with optical preci­sion, and the crystal takes that shape. In anodic bonding silicon and an alkali-contain­ing glass are heated to a few hundred degrees and a voltage of a few hundred volts is applied, with the crystal pos­i­tive. Sodium ions in the glass drift away from the inter­face, the electro­static attrac­tion pulls the sur­faces into full contact, and oxygen from the glass bonds to the silicon.

In thermo-compres­sion bonding two metal layers are pressed to­geth­er while hot until they inter­diffuse; a gold–ger­ma­nium eu­tec­tic joins ger­ma­nium crys­tals. For quartz or lithium niobate we look for sub­strates whose thermal expan­sion matches the crystal, since any mis­match leaves stress and extra cur­va­ture when the as­sem­bly cools.

The same plat­form serves three fam­i­lies of devices: silicon crys­tals for TeV beams at the LHC, bonded to sub­strates with radii of about 10 m for deflec­tions of several mil­li­ra­di­ans; spher­i­cal and cy­lin­dri­cal crys­tals for X-ray spec­tros­co­py and imaging; and rows of aligned tiles for Laue lenses.

What we have shown.

A silicon crystal anodi­cally bonded on a cy­lin­dri­cal lens de­flect­ed high-energy hadrons as well as a con­ven­tion­al bender, with almost no torsion. CRYSTAL BOND, the INFN na­tion­al project we lead with Legnaro, Trieste and Frascati, extends anodic and thermo-compres­sion bonding on pre-shaped sub­strates to new LHC de­flec­tors and to X-ray lenses, from syn­chro­tron en­er­gies to the hard X-rays of Laue lenses. With INAF Brera we adapted sil­i­cate bonding to curved X-ray tele­scope optics, dis­pensed by a col­lab­o­ra­tive robot.

180 GeV/c

The first crystal bent by anodic bonding, tested with hadrons at the CERN SPS for TWOCRYST: it deflected the beam like a conventionally bent crystal, with almost no torsion.

European Physical Journal C 85 (2025)

Si on glass

Silicon wafers bonded to glass, and silicon crystals bonded onto convex cylindrical glass lenses, the first steps towards Laue-lens tiles.

Proc. SPIE 12576 (2023)

2026

A SUSS SB8 Gen2 bonder, upgraded with Electron MEC, for anodic and thermo-compression bonding of crystals on curved substrates.

Read first.

  1. Performance of short and long bent crystals for the TWOCRYST experiment at the Large Hadron ColliderBandiera L., Cai R., Carsi S. … Guidi V., Malagutti L., Mazzolari A. et al. · European Physical Journal C 85 (2025)
  2. Anodic bonding to manufacture Laue lenses for high-energy astrophysicsMazzolari A., Frontera F., Romagnoni M., Guidi V., et al. · Proc. SPIE 12576, 125760F (2023)

All 2 papers on this line →