Current status of coating research in Glasgow

1. Current status of coating research in Glasgow I. Martin1, M. Abernathy1, H. Armandula2, R. Bassiri1, N. Beveridge1, E. Chalkley1, C. Comtet3, M. M. Fejer4, A. Gretarsson5, G. Harry6, D. Heinert7, J. Hough1, I. MacLaren1, C. Michel3, J-L. Montorio3, N. Morgado3, R. Nawrodt7, S. Penn8, S. Reid1, R. Route4, S. Rowan1, C. Schwarz7, P. Seidel7, W. Vodel7, A. L.Woodcraft1 1SUPA, University of Glasgow. 2LIGO Laboratory, California Institute of Technology. 3LMA, Lyon. 4Stanford University. 5Embry-Riddle Aeronautical University.6LIGO Laboratory, Massachusetts Institute of Technology. 7University of Jena. 8Hobart and William Smith Colleges.

2. Overview • Temperature dependence of coating dissipation • Effect of doping on low T dissipation in Ta2O5 • Heat treatment of Ta2O5 coatings • Ion beam sputtered silica • Comparison to multilayer coating loss results • Comparison of silica deposited by ion beam sputtering, e-beam evaporation and thermal oxidation • Other possible coating materials – hafnia • Studies of coating mircoscopic structure and short range order • Coating material properties • Density • Young’s modulus

3. Thermal noise associated with the dielectric mirror coatings predicted to be a significant limit to the sensitivity of second generation detectors Coating thermal noise will limit sensitivity between ~ 40 and 200 Hz 10-22 Strain(Hz-1/2) 10-23 10-24 101 102 103 Frequency(Hz) Coating thermal noise • Level of thermal noise related to mechanical dissipation of coating material, f(w) • Current coatings made of alternating layers of SiO2 (low refractive index) and Ta2O5 (high index) • Experiments have shown • Ta2O5layers are the dominant source of mechanical dissipation • Doping theTa2O5withTiO2can reduce the dissipation Projected Advanced LIGO sensitivity curve

4. 34mm 48mm thick Coating applied here 500mm Cryogenic loss measurements of single layer coatings • Studying the temperature dependence of the dissipation is of interest to: • Understand dissipation mechanisms in the coatings • Evaluate coatings for possible use at cryogenic temperatures • Single layers of coating materials (0.5 mm thick) applied to silicon cantilever substrates, fabricated at Stanford • Coatings supplied by LMA and CSIRO Titania doped tantala (LMA) coatedsilicon cantilever in clamp Silica (LMA) coatedsilicon cantilever in clamp

5. Measuring coating loss • Bending mode of cantilever excited electrostatically • Loss f(w) obtained from exponential fit of amplitude decay • Coating loss calculated from the difference in loss of a coated sample and an identical uncoated sample: Ratio of energy stored in cantilever to energy stored in coating Difference in loss between coated and un-coated cantilevers

6. Dissipation peak in tantala coatings • Initial measurements below 77 K carried out in collaboration with Jena University • Dissipation peak at ~ 18 – 20 K found in Ta2O5 coatings • Loss of silicon cantilever substrate increases at low temperature due to thermal oxide layer - necessary for coating adhesion. (c) (b) (a) Loss of (a) uncoated silicon cantilever with thermal oxide layer, (b) cantilever coated with 500 nm of TiO2-doped Ta2O5 (14.5 % Ti) and (c) the calculated loss of the coating layer

7. Effect of doping on loss of Ta2O5 coatings Comparison of dissipation in doped (14.5 % TiO2) andun-dopedTa2O5 for 3rd (left) and 4th (right) bending modes. • Doping reduces the height of the peak and slightly increases the width of the peak • Doping reduces loss of Ta2O5 throughout temperature range, with the exception of the wings of the peak

8. Analysis of coating loss peak (TiO2 doped coating) Most dissipation mechanisms described as Debye relaxations between equilibrium states, with relaxation timet: Tpeak increases with mode frequency – this indicates a thermally activated dissipation mechanism, wheret is given by the Arrhenius equation: - t0is the relaxation constant - Ea is the activation energy for the dissipation process

9. Analysis of coating loss peak • Activation energy associated with the dissipation peak calculated from Arrhenius law • (40 ± 3) meV for doped Ta2O5 • (29 ± 2) meV for undoped Ta2O5 • The low temperature dissipation peak in fused silica has a similar activation energy (44 meV) • Oxygen atoms undergo thermally activated transitions between two stable bond orientations • Width of peak thought to be related to the distribution of Si-O bond angles • Similar dissipation mechanisms may be responsible for loss peak in Ta2O5 Martin et al, Class. Quant. Grav. 25, 055005 (2008)

10. Asymmetric double-well potential model of dissipation in silica • Stable Si-O bond orientations can be represented by an asymmetric double-well potential (right) • Amorphous structure results in a distribution of both the potential barrier g(V) and the asymmetry f(D) • Activation energy calculated from Arrhenius law corresponds to the average barrier height in this distribution • Mechanical loss can be shown to be related to the distribution of barrier heights g(V)1,2: 1K. S. Gilroy and W. A. Phillips, Philosophical Magazine B 43 (1981) 735. 2K. A. Topp and D. G. Cahill, Z. Phys. B: Condens. Matter 101 (1996) 235. • grepresents the coupling between strain and the dissipation mechanism • Ciiis the elastic constant of the material • f(D) ~ const = f01

11. ADWP model – barrier height distribution • The barrier height distribution function can be calculated from temperature dependent loss data • TiO2 doping • shifts the peak in the barrier distribution to a higher barrier height • appears to increase the width of the distribution. • Other methods of altering the bond angle distribution of interest – perhaps heat treatment? • A study of the effect of heat treatment on low temperature dissipation is underway

12. Effect of heat treatment temperature (i) (ii) • Large loss peak at ~ 80 to 90 K in un-doped Ta2O5 coating (CSIRO) annealed at 800 °C, perhaps due to (expected) onset of polycrystalline structure • Smaller, broader peak at ~35 K in un-doped Ta2O5 annealed at 300 °C – activation energy 138 meV. (c) (a) Above: Electron diffraction measurement of Ta2O5 annealed at (i) 600 °C, showing amorphous structure and (ii) 800°C, showing crystalline structure. Left: Loss at 1.9 kHz of 0.5 mm Ta2O5 coatings annealed at (a) 600,(b) 300 and (c) 800°C (b)

13. Comparison of SiO2 and Ta2O5 • Ta2O5 has higher loss throughout temperature range, but loss of SiO2 will also be significant below 100 K Scatter at higher temperatures, possibly due to loss into clamp. Repeated room temperature measurement of silica coating loss.

14. Comparison to multilayer results of Yamamoto et al • Loss of single SiO2 and Ta2O5 layers used to calculate loss in a 31 layer multilayer coating, as measured by Yamamoto et al* • Yamamoto’s results: • No large peak at 20 K – evidence of a peak at higher temperature • Difference in results could possibly be explained by: • Differences in annealing temperature and / or coating layer thickness? • Further study of multilayer coatings of interest (a) (b) (a) Calculated multilayer coating loss at 1 kHz compared to (b) Yamamoto’s measured multilayer loss at 1.1 kHz *Yamamoto et al., Phys. Rev. D 74, 022002 (2006)

15. Silica loss at low temperature – comparison of deposition methods • Bulk silica (a) & thermal oxide (b) grown on siliconhave a dissipation peak at ~ 35 K1,2 • e-beam SiO2 (5.5 kHz)1 (c) • 109 nm thick e-beam film shows no peak at 35 K • Higher loss than bulk and thermal oxide above 40 k • Ion beam sputtered silica (d) • Broad loss peak at ~ 23 K (c) (a) (b) (d) • Significantly lower loss and different temperature dependence to similar thickness of e-beam SiO2 • Plan to continue investigation of effect of deposition technique / material microstructure. Currently studying thermal SiO2 on Si cantilevers in collaboration with Jena and Legnaro 1White and Pohl, Phys. Rev. Lett. 75 (1995) 4437, 2Cahill and Van Cleve Rev. Sci. Inst. 60 (1989) 2706.

16. Alternative high index materials – hafnia? • Investigation of possible alternative high index coating materials • Initial studies of 0.5 mm thick HfO2 coating, heat treated at 300°C show: • a broad dissipation peak at ~ 50 K • a plateau, or possibly a second peak, at ~200 K. • Study of the effect of heat treatment on loss of hafnia planned 500nm HfO2 coating on a silicon cantilever Coating loss for four bending modes of cantilever between 54 and 3200 Hz uncoated silicon cantilever

17. Comparison of loss in HfO2 and Ta2O5 Initial absorption studies of a multilayer silica-hafnia coating by Markosyan et al. (Stanford) lie in the range 60-80ppm, considerably higher than required. Further investigation required. • Both coatings heat treated at 300 °C • Similar loss below 50 K • Dissipation peak at higher T (50 K) in hafnia • Material properties for thin-film hafnia are not well studied – both coating loss and thermal noise depend on Young’s modulus

18. Ta2O5 layer SiO2 layer Other coatings research (1) • Studies of microscopic properties of coatings, using electron microscopy. Collaboration with colleagues in Oxford to carry out structural modelling using Reduced Density Function analysis of electron diffraction results High resolution TEM image of silica/tantala multilayer coating and a convergent beam diffraction measurement of a doped tantala layer. • Ultracryogenic loss measurements in collaboration with Legnaro and Jena • Preliminary measurements of thermal oxide carried out – planned measurements of Ta2O5 may yield more information on dissipation processes (e.g. dissipation by tunneling through potential barrier at T < 4 K)

19. Other coatings research (2) • Density – measure mass of coated cantilever before and after removal of coating with HF etch • Initial density measurement of 7675 ± 461 kg m-3 for LMA undoped ion beam sputtered tantala Slightly higher than recent measurement of 7150 kg m-3 from LMA Rutherford back scattering data (Nazario Morgado’s talk, 1st ET general meeting) • Measurements of Young’s modulus in progress • Measurements by nano-indentation with AFM • Measure change in radius of curvature when a coated cantilever is heated, using Zygo interferometer Zygo interferometer image of the curvature of a cantilever

20. Conclusions • TiO2 doping (14.5 %) increases the activation energy of the dissipation peak in Ta2O5 by ~ 30 % • Possibly due to a shift in the distribution of ADWP barrier heights to higher energy • A full understanding of the dissipation mechanisms may allow coating loss to be further reduced, improving the sensitivity of future detectors • Preliminary results of the mechanical loss of hafnia show similar levels of mechanical loss to tantala below ~50 K • Optical properties and Young’s modulus require further investigation • Coating material properties being investigated by ellipsometry, AFM and electron microscopy techniques • Low operating temperatures (~ 10 K), new and / or improved coating materials or a reduction of the total coating thickness may be attractive to gain a significant reduction in coating thermal noise for ET