In-vacuum realisation of the new definition of the kilogram Stuart Davidson, National Physical Laboratory, UK JRP Coordinator - NewKILO 1
The SI unit of Mass The kilogram is the unit of mass; it is equal to the mass of the International Prototype of the Kilogram Dates from the first meeting of the CGPM (Conference General des Poids et Mesures) held in 1889. Currently realised via The International Prototype of the Kilogram (IPK), a cylinder of platinum-iridium alloy, stored at the Bureau International des poids et Mesures (BIPM) in Sèvres near Paris. There are about 100 copies of the Kilogram which are used to provide national mass standards around the world 2
Traceability Problems with the current definition For mutual recognition and equivalence all measurements must be traceable back to the IPK No laboratory can establish an independent mass scale (compare this with e.g. the meter or the kelvin) Stability The stability of the mass scale is limited by the (unknown) stability of the artefact Copies of the kilogram are known to drift 3
Problems with the current definition Only three periodic verifications have ever taken place All copies cleaned before comparison with IPK Some copies (national standards) have changed by up to 50 µg Changes in the values of kilogram copies with relation to IPK, at the 2 nd and 3 rd periodic verifications 4
Traceability Problems with the current definition For mutual recognition and equivalence all measurements must be traceable back to the IPK No laboratory can establish an independent mass scale (compared with e.g. the meter or the kelvin) Stability The stability of the mass scale is limited by the (unknown) stability of the artefact Copies of the kilogram are known to drift Since there is no ultimate (stable) point of reference the entire mass scale could be experiencing a much greater drift 5
6
Perhaps these weights are stable and IPK is losing mass 7
Perhaps all Pt-Ir weights are gaining mass (e.g. by mercury accretion) 8
Why do the kilograms change value - the surface of the weights The kilogram (platinum-iridium) Pt and Ir oxide Carbonaceous contamination (H-C) Water Water layer acts as a buffer but gradually hydrocarbon contamination increases Storage in vacuum removes the (loosely bonded) water and (therefore) doesn t necessarily improve stability 9
What are the alternatives? the routes to Re-definition Requirements The definition must be traceable to fundamental physical constants The reproducibility of the definition must be better than 2 parts in 10 8 (20 micrograms on 1 kilogram) Techniques International Avogadro project measures/fixes Avogadro constant N A Watt balance measures/fixes the Planck constant h 10
Avogadro based kilogram Avogadro constant (N A ) number of elementary particles in a Mole N A = 6.022 140 x 10 23 Mole Amount of a substance containing as many elementary particles as there are atoms of carbon on 12 g of 12 C So if we collect a certain number of particles we have a defined mass 11
Avogadro kilogram Practical realisation Realisation via a 93.6 mm diameter sphere pure single-crystal silicon (X-ray crystal density experiment) Measure: Sphere volume and mass Out of roundness of sphere is about 50 nm N n. M. a Isotopic composition Difficult, so we use enriched 28 Si Lattice spacing X-ray/optical interferometry Surface chemistry To determine oxide thickness and contamination A m 3 12
Electrical kilogram watt balance Balance electrical and mechanical power; M.g = B.i.l BUT field strength (B) and coil length (l) are difficult to measure to the required accuracy SO M.g I.l.B 13
Electrical kilogram watt balance Use a 2 part experiment 1. Static phase - coil force balances the weight (M.g = B.I s.l) 2. Moving phase - coil goes through the field at a constant velocity, u to generate a voltage,v m (V m = B.l.u) M.g = V m I s /u Electrical units Josephson voltage and quantum Hall resistance are realised via the Planck constant (h) B I F g F e L v 14
The new definition "The kilogram, kg, is the unit of mass; its magnitude is set by fixing the numerical value of Planck s constant (h) to be equal to exactly 6.626 06X 10-34 when it is expressed in the unit s 1 m 2 kg, which is equal to J s. " Note: Either the XRCD or watt balance experiment can be used to derive a value of h, and therefore realise the kilogram, since the Avogadro and Planck constants are linked, with an uncertainty of less than 1 in 10 9, by the von Klitzing and fine structure constants. 15
Implementing the new realisation To implement the redefinition there is a requirement to develop a practical means to link the realisation experiments to the SI and to allow maintenance and traceability of the mass scale following redefinition. 3 key aspects: Provide a means of accurately fixing the Planck (and Avogadro) constant with reference to the International Prototype Kilogram (IPK) Allow dissemination of the new realisation at the level of the NMIs which should be achieved with uncertainty contributions smaller than the required (relative) uncertainty of the realisation (2 in 10 8 ) Provide a means of maintaining the standard between realisations (WB and Avogadro Key Comparison) 16
Current Traceability International Prototype Kilogram Effect of redefinition on uncertainties Possible Future Traceability Primary realisation (in vacuum) Vacuum/vacuum transfer Vacuum/air transfer Vacuum/inert gas transfer U =? µg U = 0 µg National primary Standard (Pt/Ir) U = 5 µg National secondary standards (SS) Mass comparison U 2 µg SS-Pt/Ir mass comparison U 10 µg (air buoyancy) Air/vacuum transfer U =? µg Mass comparison U 10 µg Target U = 20 µg National primary standards (AuCu, Si?) U = 20 +? µg National secondary standards (SS) Group of artefacts (Pt/Ir, SS, Si, AuCu) for medium-term maintenance of unit U = 10 +? µg U = 12 µg End user primary standard (SS) U = 15 µg SS-SS comparison U 2 µg SS-SS comparison U 2 µg U 20 +? µg End user primary standard (SS) Target U = 25 µg 17
Implementation - key areas of research Next generation mass standards (watt balance and vacuum compatible, optimised for stability) Procedures for air/vacuum transfer to optimise the repeatability of the process Characterisation of the surface of mass standards to understand dynamic sorption mechanisms and effects of cleaning and storage in various media Develop apparatus and optimise procedures for the storage, cleaning and transportation of primary mass standards Understanding behaviour of weights in vacuum is vital to all these areas 18
Materials for standards and storage New mass standards and materials Platinum-iridium was a good choice for 1889 Stainless steel is at least as stable and is widely used so its surface properties need to be well understood New materials such as nickel-superalloys and (single crystal) tungsten offer some improvements with regard to magnetic properties (for use in watt balance experiments) but finishing techniques need to be optimised. 19
Materials properties for mass standards Material Density (kg.m -3 ) Hardness (Hv) v ( 10-5 ) PtIr 21530 175 +24 Ir 22500 380 +5.9 Au-Ag9-Cu12-Pt4* 15870 250-2.8 W 19300 200 +5.5 Stainless Steel 8000 175 +300 Si 2300 1000-0.3 Udimet 720 8100 500 +44 Au electroplated Cu 8937 200-230 -10(bulk Cu) Rh electroplated Cu 8937 600-900 -10(bulk Cu) 20
Typical surface roughness of mass standards Rms height (nm) Au #01 16.4 ± 0.7 Au #02 16.5 ± 0.4 Au #03 17.5 ± 0.9 Au #04 17.7 ± 0.8 Rh #01 10.9 ± 0.6 Rh #02 9.5 ± 0.3 Rh #03 9.5 ± 0.5 Rh #04 11.2 ± 0.5 Ir #16 4.0 ± 0.3 Ir #18 4.7± 0.1 Typical R A values: SS (E1) 50 nm Pt-Ir 10 nm Silicon 5 nm Ir #20 5.0 ± 0.2 Ud #17 2.4 ± 0.4 Ud #18 2.4 ± 0.3 Si W0 1.6 ± 0.1 Si W5 1.7 ± 0.1 Si W10 1.8 ± 0.1 21
Comparison of materials for new standards Material Advantages Disadvantages Platinum-iridium Stainless steel Silicon Gold alloy Iridium Nickel super alloy Single crystal tungsten Plated copper Well characterised material Easy to machine Well characterised material Used for the majority of current weights Excellent surface finish achievable Natural silicon readily available Very low magnetic permeability Good magnetic properties Easy to machine Dense Hard Good magnetic properties Dense Similar density to stainless steel Hard Good magnetic properties Excellent magnetic properties Density similar to that of Pt-Ir Excellent magnetic properties Easy to manufacture Similar density to stainless steel Expensive Relatively high magnetic permeability High magnetic permeability Complex alloys - Surfaces difficult to analyse Low density so must be weighed in vacuum Potential static issues Low relative hardness Samples analysed showed anomalous sorption characteristics Difficult to machine Expensive Relatively expensive and difficult to obtain Difficult to manufacture artefacts High quality crystals of suitable size expensive and difficult to obtain Quality of artefacts relies on good coating process and material Au soft, Rh show inclusions form polishing 22
Materials for standards and storage New mass standards and materials Platinum-iridium was a good choice for 1889 Stainless steel is at least as stable and is widely used so its surface properties need to be well understood New materials such as nickel-superalloys and (single crystal) tungsten offer some improvements with regard to magnetic properties (for use in watt balance experiments) but finishing techniques need to be optimised. Materials for support and storage PEEK, PTFE and Torlon evaluated for the contact of weight during storage Contact tests (10,000 cycles) on stainless steel surface using pegs (r = 5 mm) of the material under test Thermal Desorption Spectroscopy tests for outgassing Aluminium and titanium tested for weigh support (balance pan) 23
Materials for weight support and storage PEEK PTFE 24
Materials for weight support and storage Mass spectrometry confirms presence of PTFE 25
TDS on storage container materials
Materials for weight support and storage Aluminium Titanium 27
Air-vacuum transfer procedures Air vacuum transfer procedures Measure the change on mass due to air vacuum transfer Optimisation of the transfer process to minimise variability of mass (due to surface sorption of water) Understand the long term surface changes which lead to instability in mass standards Investigate the possible use of nitrogen to passivate surfaces and improve repeatability 28
Sorption/Pressure correlation 7.2000 7.1950 7.1900 Mass difference / mg 7.1850 7.1800 7.1750 7.1700 7.1650 7.1600 Cycle1 Cycle2 Cycle3 Cycle4 Cycle5 7.1550 1.0E-03 1.0E-02 1.0E-01 1.0E+00 1.0E+01 1.0E+02 1.0E+03 1.0E+04 1.0E+05 1.0E+06 Pressure / Pa 29
Water sorption with pressure (METAS) measured with a Quartz Crystal Microbalance 30
Overlayer depth / nm Mass change / µg XPS / gravimetric correlation 17 16 15 14 13 XPS overlayer depth measurement (continuous line) and gravimetric data (points) 26 24 22 20 18 12 16 11 10 9 8 7 6 5 4 3 14 12 10 8 Overlayer depth (Air) Overlayer depth (Vacuum) 6 Overlayer depth (Transfer) 4 Mass gain (Air) 2 Mass gain (Vacuum) 0 Mass gain (Transfer) -2 0 5 10 15 20 25 30 35 40 45 50 31 EURAMET Time / months NewKILO JRP
Surface Analysis - XPS XPS has been used to characterise the surfaces of SS and Pt-Ir artefacts Fitting of carbon (C1s) peak Growth in overlayer is (mainly) due to accretion of hydrocarbons Angle resolved XPS used to determine surface stoichiometry 32
XPS - Growth of contamination 33
Effect of venting cycles on surface contamination 34
Thermal desorption spectroscopy used to evaluate venting cycles (CNAM) Nitrogen cycles show less rapid recontamination of samplers But a steady-state of contamination will still eventually be reached Using an intermediate nitrogen stage on the air-vacuum cycling does not inhibit recontamination and adds complexity to the process 35
Storage and Cleaning of mass standards Storage of mass standards In order to improve the stability of mass standards in future alternative storage media are investigated Nitrogen, Argon, Vacuum, (air) The effect of transition between these media and air (and vacuum) also needs to be considered Cleaning The traditional BIPM nettoyage-lavage involves rubbing with alcohol soaked chamois and spraying with steam, it is not repeatable UV/Ozone and Plasma techniques are viable alternatives to nettoyage-lavage Storage media after cleaning have an effect on the rate of recontamination 36
Mass storage inert gas vs. air (NPL) Silicon and SS weights stored in air or inert gas. Mass measurements made in vacuum. Artefacts stored in air show increase in mass due to surface contamination. Artefacts in argon show slight decrease in mass Care with handling and transfer of artefacts (to balance) is critical in maintaining stability. 37
New cleaning techniques for mass standards Traditional cleaning is generally by nettoyagelavage at the BIPM or solvent (manual or ultrasonic) at NMIs New techniques using UV activated ozone and low pressure (H 2 and O 2 ) plasma have been developed These techniques are non-contact therefore less user dependent and more controllable than manual methods Parallel evaluation of 3 cleaning techniques UV/Ozone, Hydrogen plasma and BIPM method has been undertaken (gravimetrically and using surface analysis) Subsequent storage and cycling conditions (air, vacuum and nitrogen) have been evaluated to characterise recontamination 38
Comparison of cleaning methods for different materials (METAS) Vacuum cycling has (mild) cleaning effect on new samples Plasma more effective than UV/Ozone (at METAS) BIPM method (as applied by METAS) not effective 39
Cleaning and storage Silicon (NPL) Overlayer depth measured by XPS UV better than plasma (NPL) both more effective than BIPM method (as executed by NPL) Vacuum stored samples gained most contamination after cleaning 40
Thermal desorption cleaning effect (CNAM) As received BIPM Vac-Air-Vac Vac-N 2 -Vac Thermal desorption (to 350 º C) has increased cleaning effect (over BIPM method) Nitrogen venting cycle retains the cleanest surface condition 41
Recontamination of samples after hydrogen plasma cleaning (METAS) 42
Brief summary and conclusions New mass standards and materials Platinum-iridium and stainless steel still perform well but there are possible problems with magnetic properties. Harder materials will also be more resistant to surface wear. Tungsten (poly or single crystal) has good properties but finishing techniques need optimising PEEK is a good material for storage containers. Care to avoid thirdbody wear Titanium is good for balance pans but surface wear is also a function of surface finish Air-Vacuum transfer Desorption of water approximately linear down to 0.1 Pa then minimal Good repeatability but there is a hysteresis Surface finish (R A < 50 nm) has low correlation with surface sorption 43
Brief summary and conclusions Cleaning of mass standards Current methods are user dependant and therefore have variable effects UV/Ozone and plasma techniques are more controllable/effective but need careful implementation to optimise results Media for storage (and transfer) Inert gas (argon or nitrogen) improves long term stability Vacuum storage (after cleaning) generally increases the recontamination rate Nitrogen (vacuum) venting cycles reduce re-contamination but a steady-state will still eventually be reached Adding a nitrogen stage to air-vacuum cycling does not improve re-contamination 44
Timescale CCM roadmap to redefinition 45
Impact of the kilogram redefinition Realisation can be via watt balance or Avogadro experiments thus the unit of mass can theoretically be established by any NMI Initially the best uncertainty of realisation will be 2 in 10 8 (compared with zero currently) Procedures must be in place initially to link the realisation experiments with IPK and subsequently to disseminate the unit to end users Additional uncertainty in this dissemination (from vacuum to air) needs to be small to minimise the increase in CMCs available to users - currently 15 µg for 1 kg (k = 1) Result minimal impact for end users 46
A collaborative research project between 11 European NMIs and including external partners NPL, CMI, CNAM, DFM, EJPD, LNE, MGRT, MIKES, PTB, SMU, TUBITAK INRIM, NRC, BIPM, KRISS, Häfner, Mettler-Toledo, Sartorius, Troemner TU-Ilmenau, IPQ, IMBiH www.newkilo.eu 47
Thank you for your attention 48