Bonding as an assembly
technique in upgrades to
aLIGO and detectors like ET
and KAGRA
R. Douglas1, A. A. van Veggel1, K. Haughian1, D. Chen2, L. Cunningham1,
R. Glaser3, G. Hammond1, G. Hofmann3, J. Hough1, M. Masso Reid1, P.
Murray1, R. Nawrodt3, M. Phelps1, S. Reid4, S. Rowan1, Y. Sakakibara2,
T. Suzuki5, C. Schwarz3, K. Yamamoto2
1
SUPA, Institute for Gravitational Research, University of Glasgow, Glasgow, UK
ICRR, University of Tokyo, Tokyo, Japan
3 Friedrich Schiller University, Jena, Germany
4 SUPA, University of the West of Scotland, Paisley, UK
5 High Energy Accelerator Research Organisation, Tokyo, Japan
2
Amaldi Gwangju, June 2015
Introduction
• GEO600, aLIGO, and advanced VIRGO use quasi-monolithic fused silica test
mass suspensions with superior thermal noise performance at room
temperature
• Hydroxide catalysis bonding (HCB) is used in all to attach interface pieces to
the mirrors, allowing attachment of the fibres (which are welded)
Steel wires
Penultimate mass
Attachment or ‘ear’
Steel wire break-off
prism
Silica fibres
Fibres
Weld horns
Ear
End/input test mass
Ear
• We are considering upgrades to detectors at room temperature (A+ and ETHF) and various cryogenic temperatures (Voyager, ET-LF, KAGRA)
LIGO-G1500838
2
Chemistry
This method can create strong, durable bonds.
Chemistry of bonding between silica surfaces:
Hydration and Etching
Dehydration
Polymerisation
-
OH ions from the
bonding solution
OH ions form weak fill any open bonds
bonds with surface Si on the surface
atoms
Bonding
Bonding
solution
solution with
OH OH OH OH
lots
with- of
anfree
OH
ions
excess of
SiO
Si
Si
free OH
ions
Si
O
Bonds between the Si Bulk Silica
atoms and the bulk weaken
due to the extra OH ions
bonding to the Si atom
O
O
Silicate molecule
breaking
away from
being
released
from
the bulk into
and into
the the
Bonding solution with
bonding solution
lots of free OH ions Si(OH)-5
H2O
H2O
H2O
OH
OH
OH
OH
OH
SiO2 Si OH H O
SiO2
OH
Si OH H O Si OH H O Si OH
H
Si O OHOH Si
i
OH
OH
OH
OH H
Si
O
O
O
Bulk Silica
LIGO-G1500838
H
H
Bond
3
Feasibility study: how?
• Can we bond sapphire and silicon using HCB?
 Yes, we can
 Silicon surfaces must be oxidised to facilitate reliable bonding
 Chemistry is different for sapphire: aluminate chains
• Are there other jointing techniques that may be suitable?

Indium bonding but only under compression and if suspension is not
baked out
• What is the strength of bonds in conditions consistent with
those of cryogenic suspension, assembly and operation?
LIGO-G1500838
4
Bond Strength
Two types of strength tests carried out: 4 point bend tester to
measure tensile strength and a torsional shear strength tester.
Load, F
4 contact
Bonded sample
points
d
L
Bond
l
3F ( L  l )
 tensile 
2
2bd
Where b is the breadth of the sample
 max_shear
T

2
0.208ab
Where T is the torque, a is the length and b is
the width of the cross-section of the sample
LIGO-G1500838
5
Cryogenic strength of HCBs between sapphire
Load,
F
•
•
•
40x10x5 mm samples,
10x5 mm M-to-M-plane bond in the centre
Results published Douglas et al., CQG, 2014
4 line
contacts
Bonde
d
sampl
e d
L
[MPa]
strength [MPa]
Shearstrength
Tensile
100
120
Room temperature
pH 12
Pristine samples
M-to-M
80
60
40
20
0
Sodium
aluminate
Bon
d
l
100
[MPa]
strength [MPa]
Shearstrength
Tensile
120
Sodium
hydroxide
Potassium
hydroxide
Sodium
silicate
Bonding solution LIGO-G1500838
80
60
40
20
0
R.T.
77 K
Temperature
6
Bond strength: sapphire, curing time
Load,
F
•Curing time for bonds
between silica has been
found to be ~4 weeks.
4 line
contacts
Bonded
sample
d
Bonded sapphire sample
5x5x40 mm. C-C plane. 5x5 mm
bonding surface
L
Bon
d
l
•All C-C samples broke clean
across the bond.
•Strength within error the
same for the 3 different
curing times.
•Curing time of sapphire
doesn’t appear to be longer
than 4 weeks. Investigations
are ongoing.
•Strength for A-A, M-M and CC within error the same.
LIGO-G1500838
7
Bond strength: silicon, thermal cycling
•Check the effect of thermal cycling on the strength of
line
bonds as suspended silicon masses are likely to go 4contacts
through multiple thermal cycles in their lifetime.
•40 bonds tested in set 1, 28 bonds tested in set 2.
•Thermal cycling done in Jena
~4 K – ~293 K, 2 hours per cycle.
Load,
F
Bonde
d
sampl
e d
L
l
Bon
d
Oxidised silicon sample
10x5x20 mm
Boron doped P-type <100>
~150 nm dry thermal oxide
Observations:
• Overall average strength
doesn’t change
• Two strength groups
(thought to be linked to
bond quality).
LIGO-G1500838
8
Bond strength: questions remaining?
Silicon
• What is the minimum oxide layer thickness for ion-beam sputtered coatings?
• What is causing difference in strength between different sets of bonds: bond
quality?
• How can we best to assess this?
• Can strength/mechanical loss be improved through heat/vacuum
treatment?
• What is the curing time?
Sapphire
• Influence of thermal expansion differences between different crystal
orientations on bond strength when bonding e.g. c-a plane or c-m plane?
• Influence of polishing process/surface damage on strength.
• Literature study suggests surface reactivity higher for mechanically polished
surfaces as opposed to pristine surfaces
• Can strength/mechanical loss be improved through heat/vacuum
treatment?
• Bond strength of sapphire notLIGO-G1500838
polished or cut on specific crystal plane?
9
Thermal conductivity
• Thermal conductivity has now
been measured of:
Heater and
temperature
sensor used
to thermally
stabilise the
clamp
1. Silicon-silicon HCB bonded
cylinders (25 mm x 76 mm)
down to 30 K (Lorenzini, ET
Clamped
silicon sample
T2 ΔL T1
Heater
used to
induce
heat flow
meeting Jena, 2010)
2. Silicon-silicon HCB bonded
samples (5x5x40 mm) down to
8K
3. Sapphire-sapphire HCB bonded
samples (Poster C. Schwarz, D.
Chen, ET meeting 2014, Lyon).
4. Sapphire-sapphire indium
bonded sample
Results suggest drop in thermal
conductivity is very small.
LIGO-G1500838
10
Summary
• HCB between silicon and sapphire have been extensively demonstrated
to be strong enough for typical gravitational wave detector suspensions
• Thermal cycling of HCB bonded silicon does not negatively impact
strength of bonds but more work is needed to be able to assess bond
quality
• Thermal conductivity measurements of HCB and indium bonds appear
to be high enough for suspension attachments.
• Aim: Understand better the influence of cross-sectional size on
thermal conductivity
• For more information see Marielle van Veggel’s recent GWADW talk
(LIGO-G15008) which includes details on loss measurements.
LIGO-G1500838
11
Thank you for your attention
Thank you for your attention
Any questions?
LIGO-G1500838
12
LIGO-G1500838
Bond strength: what do we know?
Silicon
• In Glasgow we bond with sodium silicate solution. (Potassium
hydroxide is also possible, see Dari et al., CQG, 2010)
• Using a wet thermal oxide layer, we need 50 nm oxide layer to get
reliably strong bonds (Beveridge et al., CQG, 2011)
• Strengths tested at R.T. and at 77 K
• Ion beam sputtered oxide layer is an excellent and probably more
practical alternative (Beveridge et al., CQG, 2013)
• Average strength values are 20–40 MPa (depending on type of
material, crystal orientation, temperature of test.
• Strength at 77 K higher than at R.T.
• Recent thermal cycling tests (3,10 and 20 times down to ~8 K) of
<100>-<100> silicon (2 sets tested), suggest that the average stays
constant. We are developing techniques to assess the quality of
bonds during the bonding procedure to ensure only high quality bonds
are accepted.
LIGO-G1500838
14
Silica
Silicon
0
Cumulative dL/L [-]
1.0E-04
5.0E-05
0.0E+00
-5.0E-05
-1.0E-04
-1.5E-04
-2.0E-04
-2.5E-04
-3.0E-04
-3.5E-04
100
200
Temperature [K]
300
Silica
Silicon
dL/L difference
300
Perpendicular to C-axis
7.0E-06
6.0E-06
5.0E-06
4.0E-06
3.0E-06
2.0E-06
1.0E-06
0.0E+00
Coefficient of linear
thermal expansion [1/K]
3.0E-06
2.5E-06
2.0E-06
1.5E-06
1.0E-06
5.0E-07
0.0E+00
-5.0E-07
-1.0E-06
-1.5E-06
200
100
Temperature [K]
Parallel to C-axis
0
Cumulative dL/L [-]
Coefficient of linear
thermal expansion [1/K]
Stress due to thermal expansion differences
1.0E-04
-5.0E-19
-1.0E-04
-2.0E-04
-3.0E-04
-4.0E-04
-5.0E-04
-6.0E-04
-7.0E-04
-8.0E-04
0
LIGO-G1500838
100
200
Temperature [K]
300
parallel to c-axis
perpendicular to c-axis
Difference between planes
300
200
100
Temperature(K)
0
15
Bond mechanical loss, overview
• Silica-silica HCB bond made with sodium silicate solution @ RT
Bond loss: 0.11 ± 0.02 (Cunningham, CQG, 2010)
• Silica-silica HCB bond (same as above) after 3 years and after heat treatment
150 C for 48 hours. Bond loss: 0.06 ± 0.01 (Haughian, thesis, 2012)
• Silicon-silicon HCB bond made with sodium silicate solution @ RT. Not an
ideal bond (offset). Bond loss: 0.27-0.52 (Haughian, thesis, 2012)
• Sapphire-sapphire HCB made with sodium silicate solution @ RT temperature
•
Bond loss @ R.T. 0.03± 0.01
•
Bond loss @ 20 K (3 - 7)x10-4 ± 1x10-4
• Sapphire-sapphire indium bond
•
Bond loss @ 20 K (2 – 3)x10-3
(Talk K. Yamamoto, ET meeting 2014, Lyon).
LIGO-G1500838
16
Thermal noise associated with HCBs in aLIGO
• In aLIGO thermal noise arising from
bonds was to be 10% of the total
thermal noise budget @ 100 Hz
• Thermal noise budget @ 100 Hz is
2.510-24 /Hz
 Bond thermal noise budget per
test mass 710−22 m/√Hz, (T01007501,2009)
• Calculated: 5.4 10−22 m/√Hz )
assuming bond loss of 0.11 at 293 K
Hild, S., CQG. 29 (2012) 124006
Assumptions for calculations following:
• bond thermal noise budget < 10% of overall thermal noise budget @ 100 Hz
• Ear size (where unknown): same ratios and average shear stress (0.17 MPa)
as aLIGO.
LIGO-G1500838
17
Thermal noise from bonds for cryogenic detectors
Name detector
A+2
ET-LF1 (option 1)
ET-LF1 (option 2)
TM material
Silica
silicon
sapphire
Temperature TM [K]
293
10
10
Bond area 1 ear [mm x mm]
39x118
45x136
45x136
Assumed bond thermal noise
requirement [m/Hz]
310-22
210-22
210-22
Bond loss [-]
*1.110-1
**1.110-1
*710-4
Calculated expected thermal noise
(±15%)
[m/Hz]
5.310-22
1.410-22
1.210-23
* Measured value ** Assumed value (bond loss) silica-silica bond @ R.T
LIGO-G1500838
18