Transcript Oxidation of silicon - uni

Silicate bonding on silicon and silica
S. Reid, J. Hough, I. Martin, P. Murray, S. Rowan,
J. Scott, M.v. Veggel
University of Glasgow
Introduction: Current applications



Originally developed for NASA’s
Gravity Probe B mission, launched
April 2004. (Gwo et al., patent)
GEO600 currently operates with
quasi-monolithic fused silica
suspensions and mirrors. This
technology allows improved
thermal noise in the suspension
systems.
Picture of a GEO600 sized silica test
mass in Glasgow with silica ears
jointed using hydroxy-catalysis
bonding
Construction of the ultra-rigid,
ultra-stable optical benches for
the LISA Pathfinder mission.
Silica fibres are welded to the ears in
the completion of the lower-stage of
the GEO600 mirror suspension. 2
Introduction: Planned applications


The upgrades for Advanced
LIGO plan to incorporate the
GEO600 technology for
significantly improved thermal
noise performance – and under
consideration for Advanced
VIRGO (in addition to other
improvements, e.g higher
power lasers).
Construction of the ultra-rigid,
ultra-stable optical benches for
LISA.
eg. LIGO
AdvLIGO
Wire loop
arm length: 4 km
Quadruple stage
silica ribbons/fibres
10-21
10-22
h
10-23
10-24
Design sensitivity curves for the LIGO
3
and AdvLIGO detectors.
Introduction: Future applications (silicon)

The construction of a 3rd generation
gravitational wave observatory within
Europe E.T. (Einstein Telescope) and
under consideration for the construction
of the DUAL resonant mass detector.
Virgo+ 2008
LIGO 2005


Bars 2005

Virgo Design

GEO-HF
2009

DUAL Mo

(Quantum Limit)

Advanced LIGO/Virgo (2014)

Credit: M.Punturo
Einstein GW Telescope

M. Punturo et al.
4
Studies on silicate bonding of silica in relation
to the Advanced generation of GW detectors

Settling time



Bond structural properties


investigate details of the underlying chemistry e.g. Through the Arrenhius equation
time available for bond adjustment and alignment
Close inspection of bonds through electron microsopy can reveal properties such
as: bond thickness, molecular structures in addition to
imperfections/inhomogeneities at the microscopy-scale.
Bond mechanical properties


Mechanical strength (studying the factors responsible for strength and reliability)
Mechanical loss in addition to bond thickness will allow the level of mechanical
dissipation associated with the bond layer to be calculated – thus allowing precise
modelling of the level of thermal noise expected from silicate bonds in future
gravitational wave detectors.
5
Studies on silicate bonding of silica in relation
to the Advanced generation of GW detectors
Settling time
experiments
Bond structural
properties
Bond mechanical
properties
SEM
SEM
Above plot showing two bonded
silica cylinders, studied before
and after silicate bonding.
Above plot showing settling time
as a function of temperature for
silica-silica bonds
TEM
Activation energy:
Ea = 0.545 eV per
molecule of OH−
AFM
S. Reid et al.,
PLA 363 341-345 (2006)
TEM
7.9 GPa
(81±4) nm
Experiments suggest that the level
of loss associated with silicate
bonding may lie:
fbond ~ (0.3→1.2)×10-1
(across the different
measured modes).
6
Required studies on silicate bonding
of silicon in relation to future detectors

Settling time

Bond structural properties

Bond mechanical properties



Surface preparation (oxidisation
techniques)
Thermomechanical properties
Temperature cycling
effects/failures
In addition to characterising
these properties in relation
to silicon-silicon bonds,
it is also necessary to
understand the required
surface preparation and the
cooling performance
available for bonded silicon
components.
7
Wealth of literature on oxidation techniques

For example: Deal, B.E., Grove, A.S. General relationship for the
thermal oxidation of silicon. Journal of Applied Physics, vol. 36,
no. 12, pp. 3770 – 3778, 1965

Found quantitative relations of the rate of growth of thermal oxide
8
Relevant literature knowledge

Qualitative statements B.E. Deal, A.S. Grove





Wet oxidised surfaces give a less dense silicon oxide then dry oxidised
surfaces – possibility of having an effect on bond strength and
thermal noise.
Carrier gas for wet oxidation doesn’t make a difference in oxidation
speed (nitrogen or oxygen). Thus undissociated H2O is the oxidising
agent.
In dry oxidation molecular oxygen is oxidising agent.
Flow speed in wet oxidation doesn’t influence speed of oxidation.
Higher oxidation temperature gives higher density.
9
Visable appearance of thermal oxides on silicon

Color of oxide as a function of thickness

Note, this is also dependent on viewing angle.
10
Oxidation results (oxide colors) in Glasgow



Shown layer
thicknesses are
expected layer
thicknesses
Colours don’t match
with corresponding
layer thicknesses on
graph in previous slide
Colours don’t match
between wet and dry
oxidation of the same
prospected thickness
11
Oxidation results (change in flatness)
PV flatness change as a function of oxidation regime
Spikes on
edge of
surface
Delta PV flatness
good side
300
PV flatness change [nm]
Localised
dip in
surface
350
Delta PV flatness
bad side
250
200
150
100
50
0
-50
50 nm 100 nm 200 nm 100 nm 50 nm 100 nm
wet
dry
wet
dry
wet
dry
1000C 1000C 1000C 920C 1000C 1000C
200 nm
dry
1000C
oxidation regime
12
Oxidation results (change in flatness)
300
PV flatness [nm]
250
Before oxidisation
After oxidisation
200
150
100
50
0
Batch no. 1
Batch no. 2
13
Bond thickness

Comparison of silica-silicon SEM images
40 nm
14
Thermal conductivity of silicate bonds
in collaboration with Firenze




The first set of silicate bonded silicon-silicon have been fabricated
with varying volumes of 1:6 sodium silicate solution at Glasgow.
Samples sent to Florence for thermal conductivity measurements.
Volumes of bonding solution: 0.4 ml cm-2, 0.2 ml cm-2 and
0.1 ml cm-2.
(Advanced LIGO specification)
See following talk by Enrico Campagna.
diameter = 25.4 mm
12.7 mm
15
Mechanical strength of bonds

Initial tests showed that a pair of silicate bonded 1” silicon disks
supported 40Kg for 2 week (~1 MPa).
wire loop
rubber ring
40Kg
lightly
clamped
40Kg load suspended
Si-Si sample under load
16
Test setups for mechanical strength
testing of silicon-silicon

New strength testing setups have been designed and ready for use.
(M. v. Veggel)
pure shear
strength test
Four-point bending test
(peeling test)
17
Temperature cycling


The ability of silicate bonds to withstand repeated temperature
cycles must be verified, in addition to withstanding the thermal
stresses that may be induced during cooling.
Repeated cycles from room temperature to 77K were performed on
bonded samples of silicon with no bond failures (in addition to this
various samples of different materials including SiO2-ZnSe, SiO2-Ge,
SiO2-ULE, SiO2‐Al2O3, all of whom have different coeff. of thermal
expansion)
18
Conclusion



Silicate bonding appears to be a highly promising technique for the
construction of cryogenic and ultra-low loss monolithic suspensions
Current estimates suggest that the thermal noise associated with
silicate bonding will have a negligible contribution to the overall
thermal noise in Advanced LIGO and likewise Advanced VIRGO.
Future work: extend the studies of bond thermal noise
19