SRF Surface Studies and the High Field Q-slope Mystery Alexander Romanenko Cornell University CLASSE

Outline High field Q-slope (HFQS) in niobium cavities Sharp decrease in Q-factor with field Empirically found “cure” – baking at 100-120C for 28-48 hours But explanation missing Interesting physics Understanding will significantly improve the current cavity treatment procedure Ways to find the explanation Cavity tests Surface studies Models studied Oxygen diffusion Magnetic field enhancement Interface tunnel exchange Newer ideas and experiments 11-Jan-19 A.Romanenko

Q-slope and baking HFQS is eliminated (EP) or improved (BCP) by vacuum Real Nb cavities Ideally HFQS is eliminated (EP) or improved (BCP) by vacuum baking @100-120C for 24-48 hours Adapted from B. Visentin (Saclay) 11-Jan-19 A.Romanenko

High Field Q-slope - History Initially HFQS was called “European headache” since cavities chemically polished in Europe had it and electropolished and tested cavities at KEK (Japan) did not have it Difference was due to the process KEK used to improve vacuum after high pressure rinsing – empirical HFQS cure was found by accident 100-1400C high vacuum annealing for 24-48 hours Removes HFQS in EP cavities Improves HFQS in BCP cavities 11-Jan-19 A.Romanenko

Temperature Mapping Thermometry board T-map Nb cavity on the test stand Thermometry boards mounted to the cavity 11-Jan-19 A.Romanenko

Non-uniformity of heating Some regions (hot) have higher losses [Why?] Temperature Mapping dT, mK log(T) Hot “Cold” log(Epk) Epeak = 50 MV/m, Hpeak = 123 mT Non-uniformity of heating Some regions (hot) have higher losses [Why?] 11-Jan-19 A.Romanenko

Heating non-uniformity Two main directions Baking effect Study Nb samples treated similarly to cavities Heating non-uniformity Cut tested Nb cavities and analyze hot/cold regions 11-Jan-19 A.Romanenko

Cavity Measurements vs. Surface Studies Cavity measurements – macro characterization Averaged over the whole surface characteristics Not possible to pinpoint the physical entity responsible for the HFQS Surface studies – micro characterization Local probing of the surface Depth resolution of a few nm is needed 11-Jan-19 A.Romanenko

Nb2O5 2-5 nm Nb ~40 nm London penetration depth @ 2K Nb Cavity Surface Hydrocarbons ~A few monolayers Nb2O5 2-5 nm NbOx, x<2.5 Nb ~40 nm London penetration depth @ 2K 11-Jan-19 A.Romanenko

Most popular model: O Adapted from G.Ciovati SRF’07 talk 11-Jan-19 A.Romanenko

Secondary Ion Mass Spectrometry Very sensitive – ppb detection possible Destructive depth profiling BUT Instrumental effects – preferential sputtering of oxygen, roughness effect on signal, chemical information not reliable due to sputtering-induced ion production 11-Jan-19 A.Romanenko

5 nm 100 sec = 1 nm ToF-SIMS results G. Eremeev and J. Francis 11-Jan-19 A.Romanenko

Arguments against O model No evidence of oxygen-enriched layer underneath oxide No evidence of oxygen diffusion at 100-120C baking temperature Oxygen depth profile does not change after baking 11-Jan-19 A.Romanenko

Magnetic field enhancement (MFE) Chemically polished surface Electropolished surface Field enhancement region with enhancement factor , if  H > Hc the region becomes normal conducting 11-Jan-19 A.Romanenko

BCP and EP cavities behave similarly before baking Arguments against BCP and EP cavities behave similarly before baking But roughness is different No evidence of MFE at grain boundary steps from T-maps Possibly overestimated the field enhancement – might be a negligible effect No difference in roughness between hot/cold spots 11-Jan-19 A.Romanenko

One of the possible oxide-related loss mechanisms Oxide losses Adapted from G.Ciovati SRF’07 talk One of the possible oxide-related loss mechanisms 11-Jan-19 A.Romanenko

X-Ray Photoelectron Spectroscopy Conduction Band Valence Band L2,L3 L1 K Fermi Level Free Electron Incident X-ray Ejected Photoelectron 1s 2s 2p Elemental composition within first few nm (except for H and He) Chemical state information Sensitivity limited to about 0.1-1 at.% 11-Jan-19 A.Romanenko

No change in oxide structure H.Tian, JLab, FNAL Material Workshop, 2007 11-Jan-19 A.Romanenko

XPS: Oxide structure Is non-uniformity of heating caused by oxide structure differences? – No! Nb5+ Hot Nb0 Valence band Hot Cold Cold 11-Jan-19 A.Romanenko

Arguments against “oxide models” Oxide structure change after baking is reversible Namely cancelled by air exposure Whereas cavity Q vs. E improvement is preserved No difference in the oxide between “hot” and “cold” spots 11-Jan-19 A.Romanenko

Other aspects to explore Crystalline orientation of individual niobium grains Deformation (stress) Vacancies and dislocations – defects of the crystalline lattice … 11-Jan-19 A.Romanenko

Electron Backscattered Diffraction A tool to study crystalline microstructure Based on diffraction of backscattered electrons Information depth – 20-100 nm Crystallographic orientation mapping Information on crystal defects distribution 11-Jan-19 A.Romanenko

EBSD: Is grain orientation responsible? (Small grain BCP cavity) Hot Cold 11-Jan-19 A.Romanenko

Recent ideas and methods Scattering off magnetic impurities (i.e. Nb12O39) – [T.Proslier et al., App.Phys.Let. – in print] Scanning Tunneling Microscopy (STM) Role of dislocations, niobium vacancies and Vac-H complexes Positron Annihilation Spectroscopy (PAS) Electron Backscattered Diffraction (EBSD) Detailed studies of the Nb/oxide interface Transmission Electron Microscopy (TEM) 11-Jan-19 A.Romanenko

High field Q-slope is not yet understood Conclusion High field Q-slope is not yet understood Set of possible causes is narrowing down Several proposed mechanisms (roughness, grain orientation, oxide losses) have been eliminated 11-Jan-19 A.Romanenko