Development of Fused S ilica S uspension Fibres for Advanced Gravitational Wave Detectors TAUP S endai 11 th S eptember 2007 Alastair Heptonstall Institute for Gravitational Research University of Glasgow Hept onst all, Bart on, Cagnoli, Cumming, Faller, Hough, Jones, Mart in, Rowan, S t rain, Veggel, Zech
Monolithic suspensions for advanced detectors Development of monolithic suspensions � is based on experience from the GEO600 suspensions This talk will cover aspects of � production and testing of suspension elements suitable for Adv. LIGO and upgrades to Virgo The criteria that must be met by ribbon � fibres for Adv. LIGO: � S trength (x3 safety margin) � Thermal noise performance To meet these criteria we require � � Breaking stress > 2.4 GPa � Intrinsic loss <3 x 10 -11 / t, where t is the thickness of the ribbon Hept onst all, Bart on, Cagnoli, Cumming, Faller, Hough, Jones, Mart in, Rowan, S t rain, Veggel, Zech
Improving fibre pulling technology Advanced LIGO suspensions require ±1.9% � tolerance on fibre dimensions. This is a slight increase on the ±2.1% � achieved in GEO600. Repeatability and tolerance in flame � pulling machines is limited by gas regulation and slack in mechanical parts. A new machine was developed in � Glasgow using a CO 2 laser and high precision drive systems Designed for both ribbon and cylindrical � fibre production to be suitable for both LIGO and Virgo upgrades. The machine is also capable of welding � fibres. Hept onst all, Bart on, Cagnoli, Cumming, Faller, Hough, Jones, Mart in, Rowan, S t rain, Veggel, Zech
Pulling fibres using the CO 2 laser Hept onst all, Bart on, Cagnoli, Cumming, Faller, Hough, Jones, Mart in, Rowan, S t rain, Veggel, Zech
Virgo laser pulling machine installation Hept onst all, Bart on, Cagnoli, Cumming, Faller, Hough, Jones, Mart in, Rowan, S t rain, Veggel, Zech
Controlled shaping of the neck 1400 Diameter ( µ m) 1000 600 200 0 1 2 3 4 Length (mm) Diameter of CO2 pulled fibre in region of taper 2000 2000 Diameter ( µ m) 1500 1500 Diameter (micron) 1000 1000 500 500 0 0 1 2 3 4 5 6 7 8 9 10 1 3 0 2 5 4 Travel /mm Length (mm) Hept onst all, Bart on, Cagnoli, Cumming, Faller, Hough, Jones, Mart in, Rowan, S t rain, Veggel, Zech
Mechanical loss in CO 2 laser pulled fibres 1.E-06 1.E-06 8.E-07 8.E-07 6.E-07 Loss 6.E-07 Loss 4.E-07 4.E-07 2.E-07 2.E-07 0.E+00 0.E+00 1 10 100 1000 10000 1 10 100 1000 10000 Frequency (Hz) Frequency (Hz) uprasil 300 fibres of diameter ~470 µ m were measured Four S � Initial analysis of losses shows a surface loss consistent with: � h φ = × − 12 4 . 7 10 m surface From Penn et al we can calculate values: � h φ = × − 12 6 . 05 10 m for suprasil 2 surface h φ = × -12 3.25 10 m for suprasil 312 surface S uprasil 300 is not necessarily expect ed to be similar to 312 or 311 as it � has a different manufacturing process and a lower OH content Hept onst all, Bart on, Cagnoli, Cumming, Faller, Hough, Jones, Mart in, Rowan, S t rain, Veggel, Zech
Where does dissipation arise in our material? In order to reduce thermal noise we need to reduce � dissipation. To do this we must first understand where it arises. � Loss in fused silica is normally split into two categories � Bulk A very low level dissipation in the body of the � material recently shown to be due to the residual effects of dissipation due to a two level system S urface A much higher level of dissipation in the � damaged surface layer The dominant loss mechanism depends on surface to volume � ratio. This can now be controlled to a level acceptable for next � generation detectors However a better understanding of the physics of these loss � mechanisms is needed to reduce thermal noise for future detectors Hept onst all, Bart on, Cagnoli, Cumming, Faller, Hough, Jones, Mart in, Rowan, S t rain, Veggel, Zech
Recent measurements at Glasgow (1) 3.5E-07 3.0E-07 Measured loss minus Loss measurements made on laser � 2.5E-07 thermoelastic pulled fused silica fibres have shown 2.0E-07 a length dependence to dissipation 1.5E-07 1.0E-07 This is consistent with a source of � 5.0E-08 loss close to the top of the fibre 0.0E+00 0 0.05 0.1 0.15 1/ Length This has been shown analytically and � using finite element modelling S ource of loss thought to be due to � welding This is a previously unknown source � of loss – highly relevant for development of detector suspensions Hept onst all, Bart on, Cagnoli, Cumming, Faller, Hough, Jones, Mart in, Rowan, S t rain, Veggel, Zech
Recent measurements at Glasgow (2) 1.E-06 Each weld gives different � 470um welded directly on fibre value for loss 8.E-07 470um welded When viewed under a � residual loss directly on fibre 6.E-07 microscope possible loss 470um welded 4.E-07 directly on fibre mechanisms can be seen 345um welded 2.E-07 Fibre attached using thick � with a few mm welded with a neck above the neck neck shows lowest loss as 0.E+00 0 0.02 0.04 0.06 0.08 0.1 0.12 less energy stored in weld 1/ length Hept onst all, Bart on, Cagnoli, Cumming, Faller, Hough, Jones, Mart in, Rowan, S t rain, Veggel, Zech
Recent measurements at Glasgow (3) Analysis of dissipation in fibres has shown evidence of a � frequency dependent bulk loss seen at a higher than expected level Approximately 10 times that seen in bulk samples � At higher frequencies this contributes as much as 25% of loss � 2.40E-07 Measured loss minus thermoelastic 2.20E-07 2.00E-07 1.80E-07 1.60E-07 1.40E-07 1.20E-07 1.00E-07 0 500 1000 1500 2000 2500 3000 3500 4000 Frequency (Hz) Hept onst all, Bart on, Cagnoli, Cumming, Faller, Hough, Jones, Mart in, Rowan, S t rain, Veggel, Zech
Ribbon fibre development Hept onst all, Bart on, Cagnoli, Cumming, Faller, Hough, Jones, Mart in, Rowan, S t rain, Veggel, Zech
Ribbon cross-sectional shape development Initial cross-section First ribbon fibres pulled had a non- � rectangular cross-section due to heat loss from edges. Laser was run at close to maximum power � due to heat loss. Polished aluminium heat shield was � developed to reflect heat back at edges. Using heat shield Further improvements to the symmetry of the � fibre neck and cross section were achieved by using slides on either side to reduce the edge effects. Laser stabilisation has been significantly � improved Using heat shield & side slides � Fast sensor � Wedged Brewster window for pick-off Profile of pull has been investigated to create � good shapes for the neck regions Hept onst all, Bart on, Cagnoli, Cumming, Faller, Hough, Jones, Mart in, Rowan, S t rain, Veggel, Zech
Profiling of ribbon dimensions Hept onst all, Bart on, Cagnoli, Cumming, Faller, Hough, Jones, Mart in, Rowan, S t rain, Veggel, Zech
S trength and bounce frequency testing Hept onst all, Bart on, Cagnoli, Cumming, Faller, Hough, Jones, Mart in, Rowan, S t rain, Veggel, Zech
Welding technology Hept onst all, Bart on, Cagnoli, Cumming, Faller, Hough, Jones, Mart in, Rowan, S t rain, Veggel, Zech
Bonding test mass ears at LAS TI (1) Hept onst all, Bart on, Cagnoli, Cumming, Faller, Hough, Jones, Mart in, Rowan, S t rain, Veggel, Zech
Bonding test mass ears at LAS TI (2) Hept onst all, Bart on, Cagnoli, Cumming, Faller, Hough, Jones, Mart in, Rowan, S t rain, Veggel, Zech
Conclusions Based on the experience of the flame pulling machines used for the GEO600 suspensions � we have designed and built new fibre pulling machines using CO 2 lasers Laser pulled cylindrical fibres have a surface loss at a similar level to flame pulled fibres � Data shows evidence of length dependent loss which appears to be related to the quality � of weld There is strong evidence of frequency dependence in residual loss of fibres st udied � This appears to arise due t o dissipation in the bulk of the fibre material but at a higher � level of loss than is seen for larger ‘ bulk’ samples Both the above effects need included in any model of suspension thermal noise in � monolithic silica suspensions Further studies in progress � The construction of the monolithic pendulum stage for LAS TI has begun, with successful � bonding of the ears to both the penultimate and test masses Hept onst all, Bart on, Cagnoli, Cumming, Faller, Hough, Jones, Mart in, Rowan, S t rain, Veggel, Zech
Recommend
More recommend
