FINAL DETERMINATION OF THE
BOLTZMANN CONSTANT k BY
DIELECTRIC-CONSTANT GAS THERMOMETRY
(DCGT)
C. Gaiser, B. Fellmuth et al.
CCT/17-30
Bernd Fellmuth ”28th CCT Meeting” Page2
Improvements & Final Result
First Results & Difficulties
Dielectric-Constant Gas Thermometry
History & Motivation
Uncertainty estimates & correlations
Outline
Bernd Fellmuth ”28th CCT Meeting” Page3
Motivation and History
Situation in 2003: Fact: Only one measurement of the Boltzmann constant on the level of
2 ppm (NIST acoustic gas thermometry). Demand: A sound new definition of the kelvin should not only be based on
speed of sound measurements. Reaction: First internal study at PTB in 2003 for a determination of k with DCGT
based on the experience gained at low temperatures (T < 30 K)
Component State of the art (2003) Goal c (p = 0) determination 3 ppm 1 ppm
pressure measurement 4 ppm 1 ppm
Compressibility keff 13 ppm 0.5 ppm
Polarizability 0 2 ppm 0.5 ppm
Impurities 5 ppm 0.5 ppm
Adsorption 0.5 ppm 0.2 ppm
T measurement 2 ppm 1 ppm
Type B combined 15 ppm 2 ppm
Bernd Fellmuth ”28th CCT Meeting” Page4
pC
CpCeffrr 1
)0(
)0()(k
c
Temperature Pressure Capacitance ratio
Ice
Water
vapour
Water
Dielectric-Constant Gas Thermometry (DCGT)
Measuring quantity :
r dielectric constant
0 electric constant
0 atomic polarizability
keff effective compressibility
c electric susceptibility
p pressure
T temperature
Clausius-Mossotti equation
combined with the ideal-gas law:
0r
r 0
1
2 3
p
kT
Bernd Fellmuth ”28th CCT Meeting” Page5
Main Challenges
Temperature
Pressure
Change of capacitance ratio
Goal:
ur(k) ≈ 2 ppm
Bernd Fellmuth ”28th CCT Meeting” Page6
Determination of the Boltzmann constant with DCGT
Isotherm measurement ( T ≈ TTPW )
First series coefficient:
1
eff
TPW0
01
33
k
kTA
...
333
3
3
2
21
cccAAApLinear regression:
3
13 eff
1
TPW0
0
k
AT
kResult:
c
Bernd Fellmuth ”28th CCT Meeting” Page7
21 K to 27 K kDCGT = 1.380657(22) •10-23 J/K
u(k)/k = 15.9 ppm 2
TPW kDCGT = 1.380654(13) •10-23 J/K
u(k)/k = 9.2 ppm 1
Determination of the Boltzmann constant with DCGT 2011
1 B. Fellmuth et al., Metrologia 48, 382-390 (2011) 2 C. Gaiser and B. Fellmuth, Metrologia, 49, L4-L7 (2012)
Bernd Fellmuth ”28th CCT Meeting” Page8
Reducing the uncertainty of the effective compressibility
Improvements in RUS measurements (e.g. temperature-dependent measurements)
Refinement of evaluation models (FEM, Monte-Carlo simulation)
Test samples also for the insulation materials (Al2O3)
Determination of the thermal expansion coefficient and the molar specific heat capacity
adiabatic isothermal
Resonant ultrasound spectroscopy
(RUS):
Bernd Fellmuth ”28th CCT Meeting” Page9
Reducing the uncertainty of the effective compressibility
FEM calculations of effective compressibility for specific capacitor geometry
optimization of capacitor design
Switch from stainless steel to tungsten carbide as capacitor material
Result (2013) Composite isothermal compressibility:
-9.370 x 10-13 Pa-1 (urel = 0.17%)
Stainless steel
keff ≈ -2.0 ∙ 10-12 Pa-1
Tungsten carbide
keff ≈ -0.9 ∙ 10-12 Pa-1
Bernd Fellmuth ”28th CCT Meeting” Page10
Reducing the uncertainty of the susceptibility
Careful analysis of the capacitance measuring network (chokes) and switch to low noise cables
Only one cable for the zero detector (increase of the sensitivity by nearly a factor two)
Measurement of the unbalanced signal of the null detector with a bandwidth of 0.01 Hz
0 2 4 6 8 10 12 14 16
-8
-6
-4
-2
0
2
4
6
8
2011
2013
Lo
ck-in
vo
lta
ge
(n
V)
time (min)
Bernd Fellmuth ”28th CCT Meeting” Page11
Purity of the measuring gas
99.99999% Helium (Linde AG)
Gas purifier (adsorber) (Micro Torr SP70, SAES Pure Gase, Inc.)
Helium purifier (getter) (HP2, Valco Instruments, Co. Inc.)
Component Certificate
Gas (ppb)
Specification
Getter (ppb)
Specification
Adsorber (ppb)
H2 < 30 < 10 < 0.1
H2O < 50 < 10 < 0.1
O2 < 30 < 10 < 0.1
CO < 30 < 10 < 0.1
CO2 < 30 < 10 < 0.1
N2 < 10
Hydro-carbons < 1 < 10 < 0.1
Noble gases
Bernd Fellmuth ”28th CCT Meeting” Page12
Purity of the measuring gas
Mass-spectrometry measurements
Experiment (UHP gas tubing)
Mass-spectrometry measurements
Component Detection limit (ppb)
H2 < 300
H2O < 20
O2 < 10
CO < 100
CO2 < 50
N2 < 100
CH4 < 20
Ne < 10
Ar < 10
Kr < 10
Xe < 10
Bernd Fellmuth ”28th CCT Meeting” Page13
Purity of the measuring gas
99.99999% Helium (Linde AG)
Gas purifier (adsorber) (Micro Torr SP70, SAES Pure Gase, Inc.)
Helium purifier (getter) (HP2, Valco Instruments, Co. Inc.)
Mass-spectrometry measurements
Experiment (UHP gas tubing)
Mass-spectrometry measurements
Component Mass-Spec and
Getter Spec. (ppb)
Sensitivity
in He
Uncertainty
(ppm)*
H2 < 10 4 0.02
H2O < 10 160 0.9
O2 < 10 10 0.06
CO2 < 50 10 0.3
N2 & CO < 100 8 0.4
Ne < 10 2 0.01
Ar < 10 8 0.05
Kr < 10 10 0.06
Xe < 10 16 0.09
Combined uncertainty 1.0
*(asymmetric rectangular distribution)
Bernd Fellmuth ”28th CCT Meeting” Page14
Results 2013 (10 isotherms)
Bernd Fellmuth ”28th CCT Meeting” Page15
Component u(k)/k ∙106
Overall weighted mean of
A1 values 2.6 (6.3)
Component u(k)/k ∙106
Susceptibility measurement
(capacitance change) 1.0 (1.0)
Pressure measurement 1.9 (1.9)
Temperature 0.3 (0.3)
Determination of the effective
compressibility 2.4 (5.8)
Head correction 0.2 (0.2)
Impurities (measuring gas) 1.0 (2.4)
Surface layers (impurities) 0.5 (1.0)
Polarizability ab initio
calculation (theory) 0.2 (0.2)
Typ A Typ B
Uncertainty budget DCGT 2013 (TPW)
Combined standard uncertainty: 4.3 ppm (9.2 ppm) B. Fellmuth et al., Metrologia 48, 382-390 (2011),
C. Gaiser et al., Metrologia, 50, L7-L11 (2013)
Bernd Fellmuth ”28th CCT Meeting” Page16
kDCGT = 1.380654 •10-23 J/K
u(k)/k = 9.2 ppm 1
kDCGT = 1.3806509 •10-23 J/K
u(k)/k = 4.3 ppm 2
Determination of k with DCGT (TPW)
1 B. Fellmuth et al., Metrologia 48, 382-390 (2011), 2 C. Gaiser et al., Metrologia, 50, L7-L11 (2013)
The 2013 value supersedes
the 2011 one
Bernd Fellmuth ”28th CCT Meeting” Page17
Improved pressure standard
2 cm2 – Systems
7 MPa
20 cm2 – Systems
0.7 MPa
1163 1159
1162
1342
1343 1341
Bernd Fellmuth ”28th CCT Meeting” Page18
159171
294
231
294
177
315294294
252
168
210
456
609618
1159
/116
2
1162
/116
3
1163
/115
9
1341
/115
9
1159
/134
2
1343
/115
9
1162
/134
1
1342
/116
2
1162
/134
3
1163
/134
1
1163
/134
2
1343
/116
3
1341
/134
2
1343
/134
1
1342
/134
3
A
Improved pressure standard
u2013(p)=1.9 ppm u2015(p)=1.0 ppm
T. Zandt et al., Metrologia, Special-issue on k (2015)
Cross-float measurements
Bernd Fellmuth ”28th CCT Meeting” Page19
kDCGT = 1.380654 •10-23 J/K
u(k)/k = 9.2 ppm 1
kDCGT = 1.3806509 •10-23 J/K
u(k)/k = 4.3 ppm 2
Determination of k with DCGT (TPW)
1 B. Fellmuth et al., Metrologia 48, 382-390 (2011), 2 C. Gaiser et al., Metrologia, 50, L7-L11 (2013) 3 C. Gaiser et al., Metrologia, Special-issue on k (2015)
kDCGT = 1.3806509 •10-23 J/K
u(k)/k = 4.0 ppm 3
Bernd Fellmuth ”28th CCT Meeting” Page20
Component u(k)/k ∙106
Overall weighted mean of
A1 values 2.6
Component u(k)/k ∙106
Susceptibility measurement
(capacitance change) 1.0
Pressure measurement 1.0
Temperature 0.3
Determination of the effective
compressibility 2.4
Head correction 0.2
Impurities (measuring gas) 1.0
Surface layers (impurities) 0.5
Polarizability ab initio
calculation (theory) 0.2
Type A Type B
Remaining potential of uncertainty reduction
What is next: DCGT measurements with two new
different tungsten carbide cylindrical
capacitors
Hope: Extremely stable results
reduction in the Type A uncertainty.
Consistent results
reduction of u(keff)
Bernd Fellmuth ”28th CCT Meeting” Page21
New massive shielding of the capacitor electrodes and more stable insulating discs
Parallel measurement of 2 different capacitors against two different types of reference capacitors
Improvements since 2014
bottom: top:
Bernd Fellmuth ”28th CCT Meeting” Page22
Improved stability due to switches inside the measurement system
Reduction of Type A uncertainty by a factor 2
Improvements since 2014: Type A
Bernd Fellmuth ”28th CCT Meeting” Page23
Use of two different types of tungsten carbide Dkeff ≈ 6 %
Refinement of evaluation models (FEM, analytic approximation)
RUS measurements on many samples > consideration of kvol()
Consistent results for the two capacitors led to reduction of u(keff) by a factor of 2
Resonant-ultrasound spectroscopy
(RUS):
Improvements since 2014: compressibility
Bernd Fellmuth ”28th CCT Meeting” Page24
Results 2017 (8 isotherms, 2 capacitors)
11 22 33 44 55 66 77 881 2 3 4 5 6 7 8
4.39713x109
4.39716x109
4.39719x109
4.39722x109
A1TC2-single-isotherm
A1TC2-weighted-mean
u(A1TC2-weighted-mean
)
A1TC3-single-isotherm
A1TC3-weighted-mean
u(A1TC3-weighted-mean
)
A1
TC
3(P
a)
Isotherm
1 2 3 4 5 6 7 8
4.39677x109
4.39680x109
4.39683x109
4.39686x109
Isotherm
A1
TC
2(P
a)
Bernd Fellmuth ”28th CCT Meeting” Page25
Uncertainty Budgets (ppm)
Component TC1 (2013) TC2 (2017) TC3 (2017)
Type A estimate 2.62 1.60 1.86
Type B estimates
Susceptibility measurement (DC/C) 1.00 0.40 0.40
Determination of the compressibility keff 2.35 0.65 1.53
Temperature TTPW (traceability to the TPW) 0.30 0.30 0.30
Pressure measurement (7 MPa) 1.00 1.00 1.00
Head correction (pressure of gas column) 0.20 0.20 0.20
Impurities (measuring gas) 1.00 1.00 1.00
Surface layers (impurities) 0.50 0.50 0.50
Polarizability from ab inio calculations (theory) 0.20 0.10 0.10
Combined standard uncertainty 3.97 2.36 2.89
DCGT > second primary thermometry method with ur(k) < 3 ppm
Second condition of the CCT for the new definition of the kelvin fulfilled!
Bernd Fellmuth ”28th CCT Meeting” Page26
Weighted mean considering fully correlations
Component Correlation
Type A estimate Partial
Susceptibility measurement (DC/C) Complete
Determination of the compressibility keff Partial
Temperature TTPW (traceability to the TPW) Complete
Pressure measurement (7 MPa) Complete
Head correction (pressure of gas column) Complete
Impurities (measuring gas) No (independent)
Surface layers (impurities) No (independent)
Polarizability from ab inio calculations (theory) Complete
kWM = 1.3806482 with ur(k) = 1.94 ppm
( 0.2 ppm smaller than CODATA 2014)
(Relative standard uncertainty without consideration of correlations:
1.66 ppm, i.e. smaller by about 20%)
Bernd Fellmuth ”28th CCT Meeting” Page27
Comparison of the three kTCk values
TC1 TC2 TC3
1.380644
1.380648
1.380652
1.380656
k x
10
23(J
/K)
Capacitor
kTCk
kWM
u(kWM
)
Bernd Fellmuth ”28th CCT Meeting” Page28
Results at low temperatures
10 100-10
-8
-6
-4
-2
0
2
(T-T90
)DCGT 2014 (He)
(T-T90
)DCGT 2014 (mean C1-C2-He-Ne)
(T-T90
)DCGT 2010 corrected
(T-T90
)AGT NPL 2016
(T-T90
)AGT LNE/NIST 2006
(T-T90
)AGT INRIM 2016
(T-T90
)CCT 2010
(T-T
90)
(mK
)
T(K)
Gaiser, Fellmuth: Phys. Stat. Sol. B 253, 1549 (2016): Method for extrapolating
compressibility data of solids from room to lower temperatures
Gaiser, Fellmuth, Haft: Metrologia 54, 141 (2017): Primary thermometry from
2.5 K to 140 K applying dielectric-constant gas thermometry