1. Bonding Experiments for Cryogenic Detectors R. Douglas 1, K. Haughian 1, A. A. van Veggel 1, L. Cunningham 1, G. Hammond 1, G. Hofmann 2, J. Hough 1, A. Khalaidovski 3, J. Komma 2, I. Martin 1, P. Murray 1, R. Nawrodt 2, S. Rowan 1, Y. Sakakibara 3, T. Suzuki 3, K. Yamamoto 3 Thursday 5 th December 2013 ELiTES Meeting, Tokyo 1 University of Glasgow; 2 Friedrich-Schiller-Universität Jena; 3 University of Tokyo

2. Talk outline Hydroxide catalysis bonding Sapphire bonding at the IGR Sapphire bonding at the ICRR Indium bonding Summary

3. Hydroxide catalysis bonding Thermal noise of the test masses and suspensions in GW detectors is an important limit to detector sensitivity. Hydroxide catalysis bonding can be used (GEO, aLIGO, aVirgo) to form quasi- monolithic suspensions of fused silica One possible way to reduce thermal noise is to cool test masses and suspensions (of appropriate materials): Sapphire (KAGRA) and Silicon (ET) are candidate materials for cooled use. Studies of the properties of hydroxide catalysis bonds formed between sapphire and silicon (breaking strength, thickness, thermal conductivity etc) are required to assess feasibility for use in cooled crystalline suspensions Steel wires Penultimate mass Silica fibres End/input test mass Ear Attachment or ‘ear’ Steel wire break-off prism Ear Weld horns Fibres

4. The hydroxide catalysis bonding procedure Chemistry of bonding of fused silica: Aqueous alkaline hydroxide solution placed between surfaces to be jointed Hydration and etching Polymerisation Dehydration Similar chemistry for jointing other oxide surfaces Small volumes of solution used (~0.4  l/cm 2 ) Several stringent requirements which must be fulfilled if bonds are to be successful Surface must be very clean Surfaces must have good match of global figure (typically flat to /10) Once made bonds are allowed to cure for four weeks to reach their maximum strength

5. Both tensile and shear strength of interest A 4-point-bending test is applied to break the bonds. This allows a value of tensile strength to be calculated. Experiments to test shear strength designed both at the ICRR and Glasgow Tensile strength Shear strength

6. Sapphire bonding Sapphire has a hexagonal crystal structure: We are interested in the a-axis, the c-axis and the m-axis

7. Sapphire axes in KAGRA The mirrors will have an orientation such that the sapphire c- plane in the mirror will be perpendicular to the c-plane of the fibre Thus investigating the properties of bonds between c-plane and a- plane sapphire and c- plane to m-plane sapphire is of interest

8. Hydroxide catalysis bonding of sapphire Areas for investigation: How does bond strength depend on type of solution used? Strength of bonds at cryogenic temperatures Effect of thermal cycling on bond strength Can we re-bond if something breaks? How strong will bonds be if different crystalline axes are bonded together? Can we bond if surfaces jointed have flatness poorer than /10? Will increasing the volume of bonding solution used help in bonding surfaces with flatness poorer than /10? Red = Results obtainedBlue = Ongoing

9. Bonding surface C-plane Aqueous solutions studied: – sodium silicate (as used in aLIGO) contains pre-dissolved SiO 2 ; aids formation of polymer-like chains in bond – sodium hydroxide and potassium hydroxide absence of SiO 2 may allow thinner bonds – sodium aluminate of interest for optical purposes – better index match to sapphire? All solutions were prepared with de-ionised water to have a pH of 12 – (to match the pH obtained when using a ratio of commercial sodium silicate solution to water of 1:6 as used in aLIGO) The samples were bonded, cured and broken at room temperature to assess their tensile strength Studies of bond tensile strength vs. chemical composition of solution

10. For comparison: GEO and aLIGO: K. Haughian Thesis (2012): Silica jointed using sodium silicate solution (Tensile strengths ~ 16 MPa) Suzuki et al (2006): Sapphire-sapphire jointed using potassium hydroxide solution (Shear strengths of ~ 7 MPa) Dari et al (2010): Sapphire-sapphire bonds jointed using potassium hydroxide solution (Shear strengths of ~ 1.5 MPa) For KAGRA strengths of ~ 10 MPa will be sufficient. Bond tensile strength at room T as a function of solution used Tensile strength (MPa)

11. Bonds tensile strength at cryogenic temperature, and after thermal cycling Two further sets of sodium silicate bonds created: One set were broken in a liquid nitrogen bath One set were thermally cycled – cooled to 10 K and then increased back to room T three times Tensile strength (MPa)

12. Re-bonding sapphire Almost all sapphire pieces survive strength testing intact – of order 15% are damaged when the bonds are broken Re-bonding the pieces could provide information about ability to repair breakages in bonded sections of a suspension Samples already used once in bonding experiments were cleaned and re-bonded using sodium silicate solution Example of undamaged surfaces after strength testing Example of damaged surfaces after strength testing

13. Tensile strength of bonded sapphire after re-bonding Slightly lower than those made with pristine samples; however the strengths are still good (and are strong enough for use in typical detector suspensions) Tensile strength (MPa)

14. Strength of bonds between different crystalline axes Bonds for KAGRA will be C-to-A or C-to-M due to the crystal orientation of the mirror and the fibres A new set of bonds were created to measure the effects of bonding different crystal axes together

15. Are we able to bond at all if parts aren’t as flat as hoped? Generally samples are used which have surfaces to be jointed with flatness of <λ/10 (λ=633 nm) Unfortunately, sapphire is difficult to polish and most of the samples procured by ICRR for the study of the effect of crystal axis/plane on bond strength had poorer flatness than λ/10 For the initial set of experiments the best samples were chosen, Allowing 9 bonds each of C-to-A, C-to-M and C-to- C type samples to be produced. These each had a flatness of <λ/4 Bonding surface and plane of interest

16. Shear strength of bonds between different crystalline axes tested at KEK at 10 K (Preliminary) C-to-M appears to be the most promising of the available options Both the C-to-A and C-to-M bonds are strong enough for KAGRA The red point in the C-to-C set could not be broken, and may have a strength even greater than 119 MPa

17. Are we able to bond at all if parts aren’t as flat as hoped? A new set of experiments was devised, using the remaining samples which had flatnesses ranging from ~ λ/4 to ~ λ 1. C-to-A with samples with flatnesses of ~ λ/4 bonded to samples with flatnesses of >λ/4. To determine the effect on bond strength if one surface is significantly flatter than the other 2. A-to-A with samples with flatnesses of >λ/4; half using the 0.4 μl/cm 2 of solution and half using 0.8 μl/cm 2 of solution. 3. M-to-M the same way as experiment 2 These bonds have been created and their strengths will be measured at by somebody at the ICRR in the New Year, when they will be fully cured To determine the effect on bond strength if one surface is significantly flatter than the other To determine whether increasing the volume of solution will improve strength when bonding with surfaces of poor flatness

18. Thermal conductivity of bonds Sapphire was jointed using sodium silicate solution to allow studies of bond thermal conductivity (Glasgow) Typical bond brought to the ICRR where an experiment was set up to test its thermal conductivity – Work ongoing Thermal conductivity set-up at IGR Thermal conductivity set-up at ICRR

19. Summary - Sapphire Sodium silicate solution produces the strongest bonds Breaking bonds at liquid nitrogen temperatures appears to have no detrimental effect on the measured strength compared to breaking them at room temperature Thermally cycling bonds appears to reduce the average bond strength; after 3 cycles average strength still ~40MPa Re-bonding samples reduces average bond strength; however re-bonded samples still have average strength >40MPa C-to-M bonds initially appear to be stronger than C-to-A at 10 K.

20. Alternative bonding techniques: Work in Glasgow in mid-90s considered range of techniques for jointing fused silica to use in GEO600 suspensions at room T: – Optical contacting (rejected due to poor reliability and possibility for substrate damage) – Direct welding of fibres to silica test masses (rejected due to likelihood of damaging test masses due to relaxation of thermal stresses) – Indium bonding (well known as a technique to joint glasses) In addition to: – Hydroxide catalysis bonding Reported in PhD thesis of S. Twyford.

21. Fused quartz test mass suspended by silica fibres jointed using different techniques Comparison of loss measurements were made of internal modes of test mass when: – slung by wires – suspended via an indium bond: – suspended via a hydroxide catalysis bond To form indium joint: the samples were heated to 140 °C (just below the melting temperature of indium) An ultrasonic soldering iron was used to deposit indium on surfaces to be jointed - breaks the thin oxide later that forms on the surface of the indium Bond ~50 µm thick formed (estimated from measuring the mass of the indium used) fused quartz T-piece, stub and cylinder post jointed to mass using Indium bond 15 mm wide flat polished along the mass 64 mm diameter by 70 mm long fused quartz test mass Fused silica fibres welded to T-piece

22. Measured Mechanical Losses of Test mass modes: Bond Type 39.7 kHz49.5 kHz50.6 kHz60.1 kHz Weld(5.5 ± 0.2) × 10 -7 (6.2 ± 0.1) × 10 -7 (1.9 ± 0.1) × 10 -6 (7.1 ± 1.1) × 10 -7 Indium(1.4 ± 0.2) × 10 -5 (2.1 ± 0.2) × 10 -6 (4.3 ± 0.3) × 10 -5 (3.0 ± 0.2) × 10 -5 Hydroxide catalysis (1.8 ± 0.1) × 10 -5 (7.7 ± 1.8) × 10 -7 (2.9 ± 1.2) × 10 -7 (4.8 ± 0.6) × 10 -6 Wire-slung7.3 × 10 -7 ---

23. Initial Results Concluded – mechanical losses associated with using indium jointing were tolerable when scaled to use in full- size suspensions – low melting point meant it was not practical for use in Room T suspensions (e.g: suspensions are heat treated for cleaning purposes) – of interest for cryogenic suspensions?

24. Recent studies Clamped cantilever with indium layer applied Kelvin Nanotechnology Ltd in (Glasgow) fabricated silicon cantilevers: 35 mm long, 5 mm wide and 54.5 µm thick Thin layer of indium (t Indium = 530 ± 30 nm) evaporated onto one face of the cantilever at Univ. of Jena Mechanical losses measured between 10 K and 80 K with and without indium applied Mechanical loss of indium calculated Nb : – exposed surface layer of indium oxidised – Indium coating rather than a joint or bond

25. Loss of indium film at cryogenic temperature Loss of approximately 5 x 10 -4 at a few 10’s of K Broadly consistent with values from Liu et al (1999) FEA models of thermal noise from indium bonds used in an Advanced detector-like suspension geometry at 40K suggest noise would be ~x10 lower than sensitivity of ET. Mode 4 at f = 2222 Hz

26. Summary – indium bonding Not ideal for use at room temperature when suspensions may be heated for e.g. cleaning purposes Loss at low (10’s of K) temperatures makes it of potential interest for use in construction of cryogenic suspensions Further work: construction of small prototype sapphire suspensions using indium bonding.

27. Thank you for your attention References: [1] A. Dari et al. Breaking strength tests on silicon and sapphire bonding for gravitational wave detectors. Classical and Quantum Gravity, 27(4): 045010, 2010 [2] T. Suzuki et al. Application of sapphire bonding for suspension of cryogenic mirrors. Journal of Physics: Conference series, 32, pp 309, 2006 [3] K. Haughian. PhD Thesis. University of Glasgow, 2013 [4] A. A. van Veggel et al. Strength testing and SEM imaging of hydroxide-catalysis bonds between silicon. Classical and Quantum Gravity, 26(17): 175007, 2009 [5] X. Liu et al. Low-temperature internal friction in metal films and in plastically deformed bulk aluminium. Physical Review B, 59(18), 1999