Ceramics and glasses · Workhorse
Sapphire
Savalues describe: single crystal
Why it wins
Single-crystal alumina: hard, broad transmission, brazeable into titanium sleeves for sealed miniature quantum packages, bakes hard.
Why not the alternative
Fused silica is cheaper but cannot be metal-brazed as robustly.
Watch out
Birefringent, orientation matters; expensive in large diameters.
Properties
The values below are candidate: compiled from the sources named, not yet individually validated. Provisional provenance: CoorsTek, CeramTec, Kyocera technical ceramic data; Kurt J. Lesker materials notes.
Wet cleaning
| Recipe | solvent then piranha or RCA-type clean for optical surfaces, ultrapure rinse, dry N₂ |
|---|---|
| Forbidden | abrasive contact: it scratches the seat, not the sapphire |
| Limit | optical surface quality, measured by scatter |
Vacuum and outgassing
| Outgassing, unbaked (10 h) | 5e-10 mbar·L/s/cm² |
|---|---|
| Outgassing, baked | 5e-13 mbar·L/s/cm² |
| Vapour pressure | none |
Temperature
| Bake, assembled | 450 °C |
|---|---|
| Vacuum degas | 1400 °C |
| Braze / H2 firing | 1600 °C |
| Metallurgical limit | service to 1800 °C |
Thermal
| CTE | 5.3 ppm/K |
|---|---|
| Thermal conductivity | 35 W/m·K |
| Thermal conductivity (cryo) | very high below 30 K W/m·K |
| Specific heat | 750 J/kg·K |
| Emissivity | 0.3–0.9 by wavelength |
| Melting / softening | 2040 °C |
Mechanical
| Strength | 400–700 flex MPa |
|---|---|
| Elongation | nil |
| Young's modulus | 400 GPa |
| Hardness | 2000+ HV |
| Density | 3.98 g/cm3 |
Electrical and magnetic
| Relative permeability | 1.0 |
|---|---|
| Resistivity | >1e16 Ω·cm |
| Dielectric strength | 17 kV/mm; tan δ 1e-5 |
Engineering
| Corrosion | inert |
|---|---|
| Joining | brazed to Kovar or Ti after metallisation |
| Process notes | no porosity, no grain boundaries: the cleanest oxide surface available |
| Availability and cost | standard viewport sizes in stock |
Brazing
Sourced pairings
| With | Filler | Atmosphere | Expansion gap at set |
|---|---|---|---|
| Niobium RRR 300 | BCu-1 | vacuum | expansion of side A absent: 1084.62 °C is outside the range of sapphire (20 to 1017 °C), never extrapolated |
| Ti-6Al-4V (off the list of 19) | named only in prose: Ag-Cu-Ti structured foil, liquidus 790 C | vacuum | no filler identified, nothing to derive |
| Kovar (FeNiCo) | Incusil-ABA | vacuum furnace, oxygen-free | 0.04 % at 605 °C Kovar (FeNiCo) on the outside matched |
| Austenitic stainless 304L / 316L / 316LN ESR
one built object
pressure window of the Gas Cherenkov Detector 3, a gamma diagnostic for inertial confinement fusion used at the Omega Laser Facility and the National Ignition Facility: a sapphire-to-stainless brazement bought complete and already proof-tested, then machined to final dimensions, welded to a tube section and a flange, and installed in the pressure cell | named only in prose: not named by the source. LANL defined the features required on the stainless flange and purchased a complete, fully tested sapphire brazement from EnvirOptics Inc. | not stated by the source | no filler identified, nothing to derive |
| Niobium RRR 300
one built object
taper insulator seal of the Thermionic Fuel Element: a metallized single-crystal alumina ring brazed to niobium skirts, the upper one thinned to 0.38 mm and formed into a convolution, welded between the emitter and the trilayer, built in lots and irradiated in the FFTF and EBR-II reactors | 35Nb-65V | not stated by the source | set temperature absent: no solidus recorded for nb-v-35-65 |
| Niobium RRR 300
one built object
higher-order-mode damper prototype for the 56 MHz superconducting cavity of RHIC, built at Jefferson Lab for Brookhaven: a sapphire window brazed onto niobium cuffs forming the vacuum barrier of the coaxial line, installed in the RHIC cavity in April 2014 | Incusil-ABA | not stated by the source | 0.03 % at 605 °C Niobium RRR 300 on the outside matched |
| Niobium RRR 300
one built object
RF feedthrough of the higher-order-mode couplers of the High Gradient and Low Loss cavities of the CEBAF 12 GeV upgrade, 32 units built: a single-crystal sapphire dielectric brazed to a niobium probe on one side and to a copper sleeve captured in the stainless steel mounting flange on the other | named only in prose: not named in this paper, which says only that the concept was to directly braze the niobium probe to a single-crystal sapphire dielectric and the sapphire to a substantial copper sleeve captured in the stainless steel mounting flange | not stated by the source | no filler identified, nothing to derive |
| Niobium RRR 300
one built object
the DECA, Diamond Electrical Contact Assembly, of the diamond amplified photocathode capsule: a brazed stack of diamond, niobium washer, sapphire washer and niobium washer, assembled and then mounted in the assembly chamber and tested | Ticusil | not stated by the source | 0.02 % at 780 °C Niobium RRR 300 on the outside matched |
| Molybdenum and TZM
one built object
support insulators of the electrostatic extraction deflectors of the K500 superconducting cyclotron of the NSCL, the planar end design that replaced the failing earlier one and was in routine use at the time of writing | named only in prose: not named by the source, which gives the atmosphere and the metallizing but not the filler | hydrogen furnace | no filler identified, nothing to derive |
| Titanium Gr 2 / Ti-6Al-4V
one built object
weld-in window cell for the nonmagnetic UHV chambers of atomic fountain clocks: an optical blank brazed into a thin weld collar which is then welded into the chamber. The cells are fabricated commercially and mounted, one titanium chamber carrying several windows, three welded directly into the chamber and one into a titanium ConFlat flange. | named only in prose: the source names a family and not an alloy: sapphire and titanium can be brazed with copper-silver-titanium (or closely related) brazes, and this type of braze uses a small addition of titanium to a copper-silver eutectic to enhance wetting to ceramics and metals other than copper and silver. No trade name and no composition are printed. | not stated by the source | no filler identified, nothing to derive |
| copper, grade not stated by the source
one built object
sealed microwave window of a high power gyrotron: a single crystal Al2O3 wafer 80 mm across and 1 mm thick, C crystal face, index (0001), supplied by Taizhou Chenzhiyi Optical Materials, vacuum brazed onto copper and then assembled into a window | named only in prose: a ternary active filler metal Ag-Cu-Ti in the mass ratio 69:28:3, laid about 0.15 mm thick on the copper surface over a layer of commercial flux, with a small weight for contact. That ratio is not any alloy of this catalogue and it is NOT fitted to the nearest one: Ticusil is 68.8Ag-26.7Cu-4.5Ti and the titanium fraction is not the same number. | vacuum | no filler identified, nothing to derive |
Preparation
- none
- high temperature metallizing, with a vapour deposited tungsten coating between the metallizer and the braze to protect the unmetallized areas from reacting with the braze alloy
- 50 micrometre Ticusil foil punched into washers of 8 mm inside and 8.5 mm outside diameter
- a standard Mo-Mn metallizing procedure on each end of the sapphire cylinder
- none stated. The filler is laid straight on the copper over a flux layer and the sapphire is not metallized.
Crystal anisotropy
This material is a single crystal, and its expansion is not the same along the two axes. The two are given side by side above 1026.85 °C and neither is retained: averaging them would fabricate a number no one measured, and choosing one would settle an orientation that the source of this card does not declare. The curve of this card does not use them.
| °C | // c-axis | // a-axis | gap |
|---|---|---|---|
| 1026.85 | 0.892 | 0.830 | 0.062 |
| 1126.85 | 0.998 | 0.928 | 0.070 |
| 1226.85 | 1.105 | 1.027 | 0.078 |
| 1326.85 | 1.215 | 1.128 | 0.087 |
| 1426.85 | 1.325 | 1.230 | 0.095 |
| 1526.85 | 1.440 | 1.333 | 0.107 |
| 1626.85 | 1.555 | 1.440 | 0.115 |
Why neither axis is retained
Sapphire is single-crystal Al2O3 and its expansion is not the same along the two axes of the crystal. The recommended-values table of the TPRC volume prints both, and above the end of this card's own curve at 1017 C it is the only reading the corpus has. The two axes are served SIDE BY SIDE AND NEITHER IS RETAINED, for two reasons that pull the same way. Averaging them would fabricate a number no one measured. Choosing one would settle an orientation that THE SOURCE OF THIS CARD DOES NOT DECLARE: the card reads grade single crystal, orientation not specified, and no document of this dossier says which way the crystal of a viewport is cut. The gap between the axes is 0.062 percent of length at 1300 K and 0.115 at 1900 K, which is seven to eight percent of the value, and a reader who needs it to a percent has to know his orientation before he can use either column. The polycrystalline column of the same table is NOT carried here: this card is a monocrystal.
Y. S. Touloukian, R. K. Kirby, R. E. Taylor and T. Y. R. Lee, Thermophysical Properties of Matter, the TPRC Data Series, volume 13, Thermal Expansion of Nonmetallic Solids, figure and table number 35R, page 176, DTIC ADA129116
Recommended values for aluminum oxide, polycrystalline column, tabulated in kelvin against percent linear expansion referenced to 293 K, which is the 20 C reference of this series. Values copied row by row and never derived, temperatures converted from kelvin by subtracting 273.15. The report states the axial recommended values are based on data for pure samples, that the polycrystalline values are calculated from the axial values, and that both are considered accurate to within 3 percent. Melting point given as about 2319 K. The data pages of this volume carry no text layer, so the table was read from a 300 dpi rendering of page 176.
Chemical affinity
What documents state about a BINARY SYSTEM, one element of a filler metal faced with one element of this material. An affinity is not a verdict on a joint: a joint brings many binaries together at once, some helpful and some harmful, and what follows is the list of those the corpus can name. The count below is not written anywhere. It falls out of the crossing of this material's elements with the composition of every filler the corpus holds, so it grows on its own the day a binary enters the corpus.
Elements of this material: Al O, from the chemical name.
interfacial reaction that wets9 fillersTi against O1 source+1 statement
The two elements react at the interface and the product of that reaction is what the rest of the filler wets. This is the mechanism that makes active metal brazing work, and it is the only one of the six that a joint wants.
9 fillers: 35Au-62Cu-2Ti-1Ni, 48Ti-48Zr-4Be, 49Ti-49Cu-2Be, 56Zr-28V-16Ti, 71.5Ti-28.5Ni, Cusil-ABA, Cusin-1 ABA, Incusil-ABA, Ticusil
Conditions and verbatim
Printed page 857, section on active metal brazing. The article states that titanium is the most extensively studied and widely used active element addition to filler metals formulated to directly braze high-melting oxide ceramics, citing Ref 37 to 41, and that the critical interfacial reaction product in the case of oxide ceramics brazed with titanium-containing filler metals is either TiO or Ti2O3, with appreciably higher adhesion in systems that result in the formation of TiO, citing Ref 42. The article gives no temperature for this reaction and none is written here.
Source[A] Joining, in Engineered Materials Handbook Desk Edition, M. M. Gauthier editor, ASM International, 1995, pages 846 to 864, DOI 10.31399/asm.hb.emde.a0003056
Source conditionsHandbook article, consulted as a PDF whose printed page numbers run 846 to 864. The article is itself a digest of the Ceramics and Glasses volume 4 of the Engineered Materials Handbook, so its statements are second-hand summaries of the papers it cites by Ref number. The Ref numbers are kept in the conditions of each entry, because a reader who wants the primary measurement needs them.
Grades
| c-plane | windows and qubit substrates, no birefringence on axis |
|---|---|
| a-plane / r-plane | when polarization or epitaxy dictates |
| HEM or Kyropoulos optical grade | low scatter, large aperture viewports |
Used by
Tools
- Brazing Abacussapphire
- Brazing Route Enginesapphire