Methane Standards Made in Whole and Synthetic Air Compared by

Subscriber access provided by McMaster University Library

Article

Methane standards made in whole and synthetic air compared by CRDS and GC-FID for atmospheric monitoring applications Edgar Flores, George C. Rhoderick, Joele Viallon, Philippe Moussay, Tiphaine Choteau, Lyn Gameson, Franklin R Guenther, and Robert Ian Wielgosz Anal. Chem., Just Accepted Manuscript • DOI: 10.1021/ac5043076 • Publication Date (Web): 13 Feb 2015 Downloaded from http://pubs.acs.org on February 18, 2015

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a free service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are accessible to all readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

Analytical Chemistry is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 25

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Analytical Chemistry

Methane standards made in whole and synthetic air compared by CRDS and GC-FID for atmospheric monitoring applications Edgar Flores*1, George C. Rhoderick2, Joële Viallon1, Philippe Moussay1, Tiphaine Choteau1, Lyn Gameson 2 , Franklin R. Guenther 2and Robert Ian Wielgosz1. 1

Bureau International des Poids et Mesures (BIPM), Pavillon de Breteuil, F-92312 Sèvres Cedex, (33) 1 45 07 70 92, [email protected]

2

Chemical Sciences Division, Material Measurement Laboratory, National Institute of Standards and Technology, 100 Bureau Drive, MS-8393

Gaithersburg, Maryland 20899-8393, United States.

KEYWORDS: WS-CRDS, GC-FID, Methane (CH4), standards, measurement.

Abstract There is evidence that the use of whole air versus synthetic air can bias measurement results when analyzing atmospheric samples for methane (CH4) and carbon dioxide (CO2). Gas chromatography with flame ionization detection (GC-FID) and Wavelength Scanned - Cavity Ring Down Spectroscopy (WS-CRDS) were used to compare CH4 standards produced with whole air or synthetic air as the matrix over the mole fraction range 1600 nmol mol-1 to 2100 nmol mol-1. GC-FID measurements were performed by including ratios to a stable control cylinder, to obtain a typical relative standard measurement uncertainty of 0.025 %. CRDS measurements were performed using the same protocol, and also with no interruption for a limited time period without use of a control cylinder, to obtain relative standard uncertainties of 0.031 % and 0.015 % respectively. This measurement procedure was subsequently used for an international comparison, in which three pairs of whole air standards were compared with five pairs of synthetic air standards (two each from eight different laboratories). The variation from the reference value for the whole air standards was determined to be 2.07 nmol mol-1 (average standard deviation) and that of synthetic air standards 1.37 nmol mol-1 (average standard deviation). All but one standard agreed with the reference value within the stated uncertainty. No significant difference in performance was observed between standards made from synthetic air or whole air and the accuracy of both types of standards was limited only by the ability to measure trace CH4 levels in the matrix gases used to produce the standards.

1

ACS Paragon Plus Environment


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 2 of 25

INTRODUCTION A comparison study of the differences in primary gas standards for methane (CH4) in air, performed in 20031,2, found the standard deviation of results around the reference value to be 30 nmol mol−1, and 10 nmol mol−1 for a more limited set of standards. These results can be compared to the level of agreement that is currently required from field laboratories routinely measuring atmospheric CH4 levels, set by Data Quality Objectives (DQO) established by the World Meteorological Organization (WMO)3,4 to meet the scientifically desirable level of compatibility for CH4 measurements on a global scale, which is currently set at 2 nmol mol−1 (1 sigma). In 2005, a new scale for CH4 measurements was established by the National Oceanic and Atmospheric Administration (NOAA), acting as the Central Calibration Laboratory (CCL) for CH4 in the WMO Global Atmosphere Watch (GAW) programme5. The scale was designated NOAA046. This scale comprises sixteen gravimetric standards covering the CH4 mole fraction range 300 nmol mol-1 to 2600 nmol mol-1. In 2012 Rhoderick et al.7 published a series of new CH4 standards with improved accuracy developed by the National Institute of Standards and Technology (NIST) achieving comparability against standards from the Korea Research Institute of Standards and Science (KRISS) of +2.3 nmol mol-1 (KRISS being higher) and of −2.54 nmol mol-1 against NOAA (NOAA being lower). In 2013, Brewer et al.8 described a new series of highly accurate CH4 and CO2 standards made of a whole air matrix at the National Physical Laboratory (NPL) with an isotopic distribution matching the atmospheric distribution for CO2 and that had excellent agreement against NIST standards for CH4 mole fraction values. The use of whole air versus synthetic matrix is still under debate, in particular because of the potential biases that can be observed when using spectroscopic techniques for analysis such as Wavelength Scanned - Cavity Ring Down Spectroscopy (WS-CRDS)9 that has been rapidly deployed for atmospheric measurements, with good reported linearity (as described by Crosson et al.10, Chen et al.11 Winderlich et al.12) and stability (Karion et al.13). Potential sources of possible bias have been reported, including pressure broadening effects (PBEs), first described by Chen et al.11, Rella et al.14 and finally by Nara et al.9, who quantified the bias introduced by deviations in the main component fractions within synthetic air matrices compared to ambient air composition. Another issue with CRDS instruments is their specificity to only one isotopologue of the molecule. However, whilst this is a source of bias which needs to be corrected for when measuring CO2

15

, the impact on CH4 measurement is considered less

2


Page 3 of 25

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


important because the isotopic distribution in standards produced from fossil fuel sources of CH4 and in atmospheric air are sufficiently close, and the resulting in biases9 are not greater than the measurement precision of the CRDS instrument. In addition, several publications describe efforts to characterize the uncertainty of these instruments. Rella et al.14, for example, performed an intensive side by side comparisons. Likewise, Allan variance analysis has been used by some authors (Winderlich et al.12, Stowasser et al.16) to determine the frequency at which a CRDS analyzer should be calibrated. Finally, Rella et al.14 proposed the correction factors which should be applied needed to compensate for the effects arising from changes in the amount fraction of water vapour present in the atmospheric mixtures. This last effect is not considered since the focus of this work is on measurement of ultra-dry gas mixtures. In this work we describe the preparation of the CH4 standards by NIST, in whole scrubbed real air (CH4 free real air) and in synthetic air. The analytical instruments maintained in the BIPM laboratories and the methods developed to compare the standards with minimum uncertainty are discussed. Two measurement methods have been developed for CRDS: the first uses a control standard to correct the drift of the CRDS (as for gas chromatography with flame ionization detection, GC-FID) and the other is based on the raw instrument response to all the cylinders analyzed over a limited period of time. These methods are presented here together with the detailed uncertainty budget and consider contributions such as the short-term repeatability, the intermediate precision, instrument drift and potential systematic biases due to the difference in the matrix gas composition, as well as differences in isotopic distribution of CH4 in cylinders coming from different sources. The methods were then used to compare two sets of cylinders produced by NIST, and to quantify any differences between the mixtures made of synthetic and whole air matrices. Subsequently, the BIPM piloted the international comparison CCQM-K8217 (the CCQM is the Consultative Committee for Amount of Substance: Metrology in Chemistry and Biology ) in which the preparative capabilities for gravimetric primary reference mixtures of CH4 in air from eight laboratories was compared, in the range (18002200) nmol mol-1. This comparison is the most recent in a series of comparisons on CH4 at ambient levels organized by the CCQM1,2,18. The results obtained during the CCQM-K82 comparison are reevaluated here to compare the groups of standards prepared in whole air versus synthetic air. Six of the eighteen mixtures submitted by participants in the comparison consisted of whole scrubbed real air, while the rest were prepared from synthetic air following strict limits on their composition.

3



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 4 of 25

NIST CH4 STANDARDS Ten compressed gas mixtures containing CH4 in a balance of air were prepared gravimetrically by the NIST Gas Sensing Metrology Group (GSMG) according to the GSMG Quality System (QMIII-839.03) Technical Procedure 839.03.07. The composition of the standards was selected based on recent measurements of dry air composition in monitoring sites used for the WMO GAW programme 19. Five were prepared using a balance gas of dry, whole real air (scrubbed of water vapour and CH4) and five were prepared using a synthetic balance gas of oxygen (O2), argon (Ar), CO2 and nitrogen (N2). The standards were developed from level 5 (standards with nominal methane mole fractions of 5 µmol mol-1 to 25 µmol mol-1,) Primary Standard gas Mixtures (PSMs) of the 2010 CH4 in air complete suite, and were prepared using the same methods and techniques as described elsewhere7. These level 5 PSMs were used to prepare the current NIST atmospheric concentration level PSMs. The PSMs were prepared at nominal mole fractions of 1700 nmol mol-1, 1800 nmol mol-1, 1900 nmol mol-1, 2000 nmol mol-1, and 2100 nmol mol-1. Two PSMs were prepared at each nominal mole fraction; one in the dry, whole, CH4 whole scrubbed real air and one in synthetic air. The pure CH4 (Advanced Gas Technologies, Palm, PA, USA), cylinder # 4991582, used to prepare the original full suite as previously described

7

was analyzed for impurities including N2, O2, Ar, CO2 and hydrocarbons. Gas

chromatography was used for the analysis using several types of detectors based on thermal conductivity (TCD), flame-ionization (FID) and helium-ionization (HID). CH4 was determined to be (99.9993 ± 0.00006) % pure. The manufacturer of the pure CH4 gas confirmed that it had been produced from liquefied natural gas, which meant that the isotopic composition of the gas would be in the range of –(43±7) ‰ for δ13C (VPDB) and –(185±20) ‰ for δD (VSMOW) with (±1 SD). The synthetic air from Air Liquide America Specialty Gases (ALASG), Plumsteadville, PA, and CH4 whole scrubbed real air from Scott-Marrin, Riverside, CA, were analyzed for impurities and constituents of air using the following instruments: O2 (Siemens Oxymat 6, paramagnetic technical principle which uses alternating pressure methods); Ar (Agilent GCTCD); CO2 (Agilent GC/TCD); CH4 (Agilent GC/FID and Picarro CRDS); and the N2 composition was determined by subtraction of all other constituents. The GC conditions for the Ar analyses include the use of a 4.8 m x 0.32 cm stainless steel column packed with Molesieve 5A , 30 ml/mim He column flow, temperature programmed from -25 OC for 5 min., 60 OC /min to 200 OC, hold for 1 min. then to - 25 OC at 60 °C/min

4


Page 5 of 25

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


and hold for 1 min, TCD @ 150 OC, reference flow at 45 ml/min He, and a 0.25 ml sample loop. Each instrument was calibrated with the appropriate primary standard mixtures prepared gravimetrically. The mole fractions and associated uncertainties were calculated as previously described. Table 1 lists the cylinder number for each PSM together with their mole fraction as prepared by the gravimetry procedure, and relative uncertainty expressed as a level of confidence of approximately 68 %, where k =1. After preparation, the PSMs were analyzed to verify the gravimetric preparation. Analyses were performed on a GC-FID operated at 250 °C. A 3.66 m by 6.35 mm stainless steel column packed with Poropak Q (80-100 mesh) was used at a temperature of 40 °C for the analysis. The helium column carrier, purged through a heated VICI Model HP2 (Valco Insturments Co. Inc.) gas purifier, was used at a flow rate of 60 mL min-1. The FID was supplied with 320 ml min-1 of pure O2 and 40 mL min-1 of hydrogen. A 10 mL stainless steel sample loop was used to introduce the CH4/air samples onto the column. A whole air mixture contained in a 29.5 L high-pressure aluminum cylinder, CC324321, and previously verified against the NIST’s GSMG atmospheric suite of PSMs, was used as a control. A minimum of six ratios of each PSM to the control were calculated using peak area data. A generalized least squares linear program, GenLine20, was used to plot the average ratio and standard error of the ratio (x-axis) against the mole fraction prepared by gravimetry and uncertainty (y-axis). The standard error was calculated as: s/√n where s is the standard deviation. All CH4/air PSMs (real and synthetic matrix) were in total agreement and pass the test. The mole fraction for the control whole air sample determined from this regression line was (1887.56 ± 0.62) nmol mol-1, where the relative uncertainty is expressed at k = 1. In addition, each subset of five PSMs in whole air and five in synthetic air were plotted separately and the mole fraction of the control is predicted from the regressions.

NIST MEASUREMENTS OF TRACE CH4 LEVELS IN THE MATRIX GASES Synthetic air from Air Liquide America Specialty Gases (ALASG), Plumsteadville, PA, and CH4 whole scrubbed real air from Scott-Marrin, Riverside, CA, were analyzed for trace CH4. A GC-FID was used to separate the components of air using a 3.66 m by 3.18 mm stainless steel column packed with Porapak Q (80-100 mesh) at a temperature of 40 °C. The helium column carrier, purged through a heated gas purifier, was used at a flow rate of 30 mL min-1. The FID (250 °C) was supplied with 320 mL min-1 of O2 and 40 mL min-1 of H2. A 6-port stainless steel

5



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 6 of 25

gas sampling/injection valve was used to introduce the matrix air samples onto the column via a 10 mL stainless steel sample loop. A nominal 12 nmol mol-1 CH4 in air gravimetric standard was used to calibrate the GC-FID. Additionally, CH4 determinations were made on the balance air using both Tiger Optics (Laser Trace Model) and Picarro CRDS (G2301 Model) systems. Results from each of these instrument methods were in agreement within the stated uncertainties. The amount of trace CH4 in the dry whole air cylinders ranged from 1.0 ± 0.1 nmol mol-1 to 2.8 ± 0.1 nmol mol-1, and < 1.0 nmol mol-1 in the synthetic air cylinders.

6


Page 7 of 25

1 2 3 4 5 6 7Number of 8 9Cylinder 10 11 12 13 14 15 16 17 CAL018193 18 19 20 CAL018196 21 22 FF4234 23 24 25 CAL018215 26 27 28 CAL018226 29 30 31 FF4190 32 33 34 CAL018216 35 36 37 38 39 40 41 42 43 44 45 46 47 48


CH4

mole

fraction

Standard uncertainty

Assigned

Assigned

Assigned

Assigned

Assigned

Assigned

Assigned

Assigned

CO2

standard

Ar

standard

O2

standard

N2

standard

mole fraction

uncertainty

mole fraction

uncertainty

mole fraction

uncertainty

mole fraction

uncertainty

Air matri x

k =1

k =1

k =1

k=1

xCO2

u(xCO2)

xAr

u(xAr)

xO2

u(xO2)

xN2

u(xN2)

(µmol mol-1)

(µmol mol-1)

(µmol mol-1)

(µmol mol-1)

(cmol mol-1)

( cmol mol-1)

( cmol mol-1)

( cmol mol-1)

(nmol mol-1)

(nmol mol-1)

1637.42

0.56

WA

387.311

0.035

9.356

0.009

20.901

0.003

78.145

0.008

1704.63

0.59

SA

390.939

0.042

9.380

0.012

20.926

0.002

78.182

0.011

1815.72

0.66

SA

376.226

0.046

9.290

0.025

20.889

0.004

78.144

0.001

1819.22

0.66

WA

388.571

0.037

9.371

0.006

20.908

0.003

78.116

0.008

1906.34

0.66

WA

387.000

2.600

9.380

0.012

20.912

0.003

78.112

0.009

1929.63

0.64

SA

376.075

0.189

9.285

0.093

20.887

0.004

78.147

0.009

1969.34

0.75

SA

376.329

0.177

9.312

0.025

20.840

0.003

78.190

0.009

7



1 2 3CAL018191 1970.90 4 5 6FF4259 2069.00 7 8 2082.14 9FF4222 10 11 12 13 14 15 16 17 Table 1. 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48

Page 8 of 25

0.74

WA

388.832

0.430

9.366

0.010

20.910

0.004

78.114

0.010

0.71

SA

374.500

0.200

9.328

0.027

20.839

0.004

78.191

0.012

0.81

WA

388.891

0.047

9.372

0.012

20.919

0.004

78.105

0.010

Mole

fraction

and

uncertainty

for

each

CH4/air

PSM.

SA:

Synthetic

air.

WA:

whole

air.

8


Page 9 of 25

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


ANALYTICAL INSTRUMENTS, METHODS AND UNCERTAINTIES The BIPM-CH4 gas facility includes a GC-FID (Agilent, series 7890A), a CRDS (Picarro Inc., CA, USA, model G1202) and an auto-sampler (a dedicated ensemble using a VICI 16 port valve). All cylinders were handled according to the recommendations described in ISO 1666421. The cylinders were sequentially analyzed using a 16inlet automatic gas sampler connected to the cylinders, GC-FID and CRDS analyzer.

Measurement by GC-FID

Setup The system is a Agilent series 7890A GC-FID, equipped with a stainless steel column packed with Poropak Q (80100 mesh). The FID was supplied with 320 ml min-1 of O2 with a purity of 99.9995 %, from a high-pressure cylinder, and 40 ml min-1 of H2 with a purity of 99.9999 % produced on site by a commercial generator (Claind, model CG2200) by water (demineralized) electrolysis. The GC-FID operated at 250 °C. A 6 m by 6.35 mm column was used at a temperature of 35 °C for the analysis. The helium column carrier with a purity of 99.9999 % passed through a patented getter-based purifier for extra purification (SAES getter®, model PS2-GC50-R-2), at a flow rate of 70 ml min-1. A 5 ml stainless steel sample loop was used to introduce the CH4/air sample onto the column. The pressure in the sample loop was measured by a calibrated pressure sensor with a resolution of 0.2 hPa and recorded at each injection. Finally the peak areas were quantified online by the software ProChem (SRA, version 2.5.7) and processed using an in-house developed LabView program CH4-GC Computation, version 1.0.1.3.

Analytical method In this method the measured quantity was the drift corrected ratio to the control cylinder. In order to calculate this number, two CH4/air gas standard mixtures were analyzed between one CH4/air gas mixture used as control standard (A). The measurement sequence started by measuring the CH4 peak area response of three replicate analyses of the control cylinder (A1, A2, A3) and then three replicate analyses of two CH4/air standards, PSM1 and PSM2, for finalizing again with three replicate analyses of the control cylinder. Since the analysis of each CH4/air gas mixture takes 15 minutes, this measurment sequence takes about 180 minutes to complete. The average of the three replicate


9


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 10 of 25

peak areas of the control before and after was then used to correct for drift, and the drift corrected value was used as the denominator in the ratio of the standard and control peak areas. After that a new sequence started following the same order but using two new CH4/air standards. These measurements were repeated until all CH4/air standards were measured. This entire sequence was repeated seven times to constitute one series of measurements, and each series was itself repeated three times, so that each cylinder was analyzed twenty one times in total. For each cylinder in one series, the drift corrected average ratio to the control standard of seven repeated

RGC

measurements

was first calculated. Results of the three series were then combined with the calculation of the

weighted mean R wGC , defined as: 3

(1)

R wGC = ∑ l =1 wl R GC _ l

where the weight wl for the series l is defined as

wl =

(

1/ u RGC _ l

)

2

3

∑1/ u ( R

GC _ l

l =1

(

.

)

(2)

2

)

and u R GC _ l is the standard deviation of the mean of the seven repeats.

Uncertainty The uncertainty of the weighted mean

(

u R wGC

)

2

=

R wGC is simply given by the equation:

1

(3)

3

∑1/ u ( R l =1

GC _ l

)

2

In addition to this component, the intermediate precision, or stability of the instrument response during the measuring period, of the GC-FID was evaluated by the standard deviation of the drift corrected ratio to the control cylinder of one standard exclusively used for this purpose, giving u Int Pr e = 0.0002. This component was further


10

Page 11 of 25

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


combined with the weighted mean uncertainty to provide the combined standard uncertainty associated with measurements performed with GC-FID, resulting in a typical relative standard uncertainty of 0.025 %.

Measurement by CRDS A CRDS Picarro, G1202 model, was used as another independent measurement technique. Two methods, one based on the instrument response (Method 1) and one on the ratio to a control cylinder (Method 2), were used for the measurements. In Method 1 the measurand was the CRDS response averaged over five minutes. The measurements were done without interruption in a minimized period of time (see section 0, Type A uncertainty:). In Method 2, the measurand was the ratio of the CRDS response to a control standard (the same standard as for the GC-FID Method). In both methods, the CRDS was used as a comparator, so that the main sources of uncertainty are statistical effects, as described further in detail. In addition pressure broadening and isotope ratio effects are potentially different for each analyzed gas mixture, therefore constituting a source of uncertainty that can affect both methods.

CRDS Method 1

Analytical method In this method the measured quantity was the CRDS response defined as the average of the value over the last five minutes of measurement, recorded after three minutes of flush. This optimum measurement time was determined from an Allan variance analysis, as is reported later. All the cylinders included in one comparison were analyzed sequentially. This measurement sequence was then repeated three times, and the average of the three results calculated. One important point to note in this method is that the analysis has to be performed within a limited time to benefit from the low noise and low drift rate of the instrument. As an example the comparison CCQM-K82 included as many as 16 cylinders, yet the total analysis was still limited to 18 hours.

Uncertainty Four effects are considered as important uncertainty contributors for the CRDS measurements when comparing standards produced from the same source of CH4, these are: the short term repeatability, the response drift, the potential PBEs and an additional experimental variance. The influence of these contributors is separated in Type A and Type B uncertainties using the recommendations in the Guide to the expression of uncertainty in measurement, GUM22.


11


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 12 of 25

Type A uncertainty: The first uncertainty component to consider is the short-term repeatability of the instrument, defined as the Allan deviation23. This was evaluated during time series measurements of two CH4/air gas mixtures in the mole fraction range of 1.8 µmol mol-1 to 2.2 µmol mol-1. For each mixture, the Allan deviation was calculated on the CRDS raw data (measurement time of 1 s). White noise behaviour was observed with a measurement averaging time up to top = 300 s. The measurand was then defined as the average response of the CRDS value over five minutes of measurements and the measurement response of the instrument was then associated with an uncertainty equal to the Allan deviation at top: u Alla = 0.1 ppb. This value was found to be constant over the CH4 mole fraction range of 1.8 µmol mol-1 to 2.9 µmol mol-1. Stowasser et al.16, Crosson et al.10, Chen et al.11, Winderlich et al12, and Rella et al.14 have also used Allan variance analysis to determine the noise in the CRDS response and obtained comparable values to those reported here. A second uncertainty component to consider is the possible instrument response drift during the time of one entire measurement sequence, including one analysis of all cylinders. This component, udrift, was considered to be equal to the standard deviation of the instrument response to a control cylinder over a period of 6 hours, the average duration of one measurement sequence, and was found to be equal to 0.11 ppb. The two components described above are combined in the Type A uncertainty using the following equation, in which the Allan deviation is divided by three due to the further average of three measurements:

2 u Allan 2 uA = + u Drift =0.124 ppb 3

(4)

Type B uncertainty: The potential biases introduced into the CRDS measurement results due to variation in the composition of the air matrix, as quantified by Nara et al.9 on a very similar instrument, Picarro model G-1301, allows limits on the composition of the air matrix established for the main components of synthetic air standards to be compared. As strict limits are imposed on the matrix composition of all analyzed mixtures, both from NIST and in the CCQM-K82 comparison, this bias was considered to be equal to zero, with an associated non-zero uncertainty of Type B,


12

Page 13 of 25

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


u B = u Broad . If we assume that there is equal probability of a bias due to the PBEs within the calculated limits, the uncertainty follows a rectangular distribution. For example the lowest amount of N2 in a CCQM-K82 standard was 77.9814 % and the highest 78.25379 %, resulting in ∆xCH4_N2 = 0.0479 ppb and −0.09659 ppb respectively. The uncertainty expressed in ppb is then equal to

u CRDS ( x N 2 ) = 0 .04168

.

Using the same methodology, uncertainties associated with deviations in O2 and Ar were calculated to be uCRDS(xO2) = 0.0373 ppb and uCRDS(xAr) = 0.0187 ppb. Combining the three components for N2, O2 and Ar, the uncertainty associated with PBEs is uBroad = 0.059 ppb. This uncertainty was applied to the values of all cylinder analyzed by CRDS. Uncertainty from additional experimental variance: As each measurement sequence was repeated three times, the average of the three results for each cylinder was calculated and its associated uncertainty defined as the standard deviation of the mean σ. This was a conservative approach to cover any additional variance not covered by other terms already considered, and was generally not a major contributor to the overall uncertainty of the measurement. Combined standard uncertainty: Finally, the combined uncertainty of all these contributors is given by:

( )

u yA =

u A2 + u B2 + σ 2

(5)

( ) is the standard uncertainty of the CRDS responses for 6 hours of measurements. This corresponds to

where u y A

a relative standard uncertainty of 0.01 %.

If the method is used to compare standards produced with different CH4 sources, an additional uncertainty component to cover variance arising in CRDS response due to possible isotopic variation in the CH4 in different standards should be added. Previous publications9 have concluded that the isotopic bias for CH4 measurements with CRDS are not significant as the potential bias is of the same magnitude as their reported analytical precision of their


13


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 14 of 25

CRDS instrument of ±0.3 nmol mol-1. This conclusion is reexamined and described as follows. Assuming that pure CH4 used in the preparation of standards originates from natural gas, the reported24 mean isotopic composition and (±1 SD) range around the mean that could be expected is –(43±7) ‰ for δ13C (VPDB) and –(185±20) ‰ for δD (VSMOW). Gas samples at the extremes of this range would lead9 to biases in the CRDS measured CH4 mole fraction values of +0.34 nmol mol-1 and −0.38 nmol mol-1. Considering a rectangular probability distribution between these limits, allows a type B standard uncertainty to be calculated to cover potential variations in CRDS measurements occurring due to potentially different isotopic mixtures in the gas measured, uδ =0.21 ppb, which

( )

would result in a value of u y A of 0.015 % when expressed as a relative standard uncertainty.

CRDS Method 2

Analytical method In this method the measured quantity was the drift corrected ratio to the control cylinder, matching GC-FID measurements. The same measurement sequence was also used, with two CH4/air gas mixtures analyzed against a CH4/air gas mixture used as a control standard. One measurement sequence was also repeated three times to

ࡾ૛ . constitute one series of measurements, and the drift corrected ratios R2 further averaged in തതതത Uncertainty The calculation of the combined uncertainty in the method CRDS 2 was carried out in two steps. First, the uncertainty associated with the drift corrected ratio to the control cylinder, u (R 2 ) was calculated. Then this contribution was combined with another two contributions that are the intermediate precision term, u Intpre , and an additional experimental variance,

σ

.

Drift corrected ratio uncertainty: Although ratios to a control cylinder were calculated in method 2, this did not prevent potential biases due to PBEs. On the contrary, the treatment of this uncertainty component is more complex, because different biases can occur for the analyzed cylinder and the control cylinder used to calculate the drift corrected ratio R2. The uncertainty was therefore quantified using the uncertainty calculation software GUM22. The uncertainty associated with the CRDS


14

Page 15 of 25

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


response for each standard cylinder was considered to be the combination of two contributions, the short term repeatability uAllan and the PBEs uBroad, as determined in method 1. Since one control cylinder was used for all the measurements, only the short term repeatability uAllan was considered to contribute to the uncertainty in this case. The drift correction therefore is different for cylinders analyzed just after the control, and for cylinders analyzed just after this (or just before the next control analysis). To account for this, two slightly different treatments were performed that resulted in different uncertainties according to the position of the cylinder in the measurement sequence. Typically, the relative standard uncertainties of 0.0100 % and 0.0125 % respectively were obtained for cylinders analyzed just after and just before the control. Intermediate precision: An additional term describing the uncertainty due to the intermediate precision of the CRDS ratio measurements ഥ ૛ ). This uncertainty term was calculated by the standard deviation of the ratio between a specific was added to ࢛(ࡾ cylinder and the control cylinder during the comparison period. This term was equivalent to u Intpre = 0.000075.

Additional experimental variance: Again, as each measurement sequence was repeated three times, the standard deviation of the mean was calculated and further combined with other uncertainty components. Combined standard uncertainty: The final expression to determine the global uncertainty for each cylinder using CRDS method 2 is given by the equation:

( )

2 2 u R 2 = u (R 2 ) + u Intpre +σ .

(6)

This corresponds to a relative standard uncertainty of 0.025 %, using the largest values obtained by a conservative approach. If the method is used to compare standards produced with different CH4 sources, an additional uncertainty component to cover variance arising in the CRDS response due to possible isotopic variation in the CH4 in different


15


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 16 of 25

standards should be added. This standard uncertainty component, uδ, was previously computed to be 0.21 ppb, ഥ ૛ ) of 0.031 % when expressed as a relative standard uncertainty. which would result in a value of ࢛(ࡾ

COMPARISON OF NIST WHOLE AIR VERSUS SYNTHETIC AIR STANDARDS

Linearity of analytical methods Six standards (CAL018193, CAL018226, CAL018191, FF4234, FF4190 and CAL018216) from a lot of ten were used to demonstrate the comparability between the standards produced by NIST, which consisted of different gas matrices, and the others were used during method development and validation for the comparison. In order to first verify the linearity of each method, a calibration line was evaluated using the Generalized Least Squares (GLS) approach described in ISO 6143:200125,. The calculations were performed using B_LEAST, a computer program that estimates a calibration function by taking into consideration uncertainties on both axes for regression analysis, and in which ISO 6143:2001 methodology is implemented. The results are listed in Table 2. The Goodness-of-Fit (GOF), defined as the maximum value of the weighted residuals25, in all cases is less than 2 meaning that all standards constitute a consistent calibration set within the analytical uncertainties as determined by the BIPM and the preparation uncertainties assigned by NIST. All three methods show good linearity with respects to the scale made up of the six selected standards. The values of the slope b1 were equivalent, for CRDS 2 and GC-FID methods, taking into account u(b1), since the same control standard was used for both. Table 2. Ouput from GLS Algorithm in this Analysis Modea Parameter

GC-FID

CRDS-2

CRDS-1

b0

−3.9335 (ppb)

−7.8621

−8.002

b1

1.906 × 103

1.910 × 103

1.0033

u(b0)

4.8472 (ppb)

5.1799

4.0798

u(b1)

4.9547

5.2933

2.1962 × 10-3

cov (b0, b1)

−23.96 (ppb)

−27.35

-8.9375 × 10-3


16

Page 17 of 25

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


SSD rem

1.5428

1.1067

1.6919

GOF

0.6392

0.673

1.0173

b0, b1, u(b0), u(b1), and cov u(b0, b1) are the parameters of a straight-line model calibration function for the GC-FID, CRDS Method 1 and CRDS Method 2 analyses responses against xCH4.

a

Comparability between whole and synthetic air standards In order to verify the comparability between whole and synthetic air standards, calibration curves were calculated for each method using only the standards made from whole air as the calibration set (CAL018193, CAL018226 and CAL018191), this allows predicted values for the synthetic air standards to be calculated and compared to their certified gravimetric values. The difference (Diff) between the gravimetric mole fractions assigned by NIST xNIST and the calculated predicted value xPV was determined for each synthetic air standard and for each method. The combined standard uncertainty associated with Diff is expressed as:

2 2 u ( Diff Method ) = uNIST + uPV

(7)

The calculated set of nine differences is plotted in Figure 1 for each of the methods, CRDS 1 (black dots), CRDS 2 (red dots) and GC-FID (green dots). As can be observed all differences agree with zero within their expanded uncertainty (expressed at 95 % level of confidence). This means that none of the methodologies are biased with respect to each other and that for both measurement techniques the two gas matrices used by NIST to produce standards gave equivalent results within their stated measurement uncertainties.


17


5

CRDS 1 CRDS 2 GC-FID

4 3 -1

Diff / (nmol mol )

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 18 of 25

2 1 0 -1 -2 -3 -4 -5

FF4234

FF4190

CAL018216

Figure 1. Difference between CH4 gravimetric mole fractions and predicted values using CRDS Method 1 (black dots), CRDS Method 2 (red dots) and GC-FID (green dots) of the NIST synthetic mixtures at 1815.72 nmol mol-1 (FF4234), 1929.63 nmol mol-1 (FF4190) and 1969.34 nmol mol-1 (CAL018216). The error bar represents the expanded uncertainty at a 95 % level of confidence.

COMPARISON OF WHOLE AIR VERSUS SYNTHETIC AIR STANDARDS FROM DIFFERENT SOURCES A target standard uncertainty for CH4 in air standards of 1 nmol mol-1 has been recommended5 in order to achieve the desired uncertainties for atmospheric monitoring laboratories, which are expected to have results with standard uncertainties equal to or better than 2.2 nmol mol-1. However, a more stringent treatment that will result in negligible impact of the standard on the uncertainty of the measurement result, would require the standard uncertainty of primary standards to be better or equal to 0.5 nmol mol-1 and the agreement within a set of primary standards (described by a standard deviation of their dispersion from a reference value) to be also equal to or better than 0.5 nmol mol-1. Uncertainties on standards at these levels would result in the standard contributing to less than 5 % of a measurement uncertainty of a laboratory claiming a standard uncertainty of 2 nmol mol-1. These required values can be meaningfully compared with the levels of agreement between standards obtained and described in this paper, and more widely with international comparisons performed within the last ten years.


18

Page 19 of 25

During comparisons performed in 2012, prior to the CCQM-K82 comparison, the measured differences between laboratories ranged from less than 0.1 nmol mol-1 (between NIST and NPL)16 to 4.8 nmol mol-1 (between KRISS and NOAA) 15. The CCQM-K82 comparison with eight participants was carried out in 2013. Participants were asked to produce CH4 in air mixtures using their usual process, either in whole scrubbed real air or in synthetic air but under strict specifications for the gas matrix composition. Six of sixteen standards were produced in whole air. Figure 2 reproduces the degrees of equivalence of the key comparison CCQM-K82 as reported elsewhere17. The key comparison reference value (KCRV) was calculated using only the measurement results of the CRDS method 2 and the self-consistent set of cylinders comprised all cylinders except one. The Goodness-of-Fit of the regression performed with this data set is equal to 1.72, demonstrating consistency of the ensemble of selected standards. In Figure 2 the standards made using whole air matrix (WA) are indicated with red dots and those made using synthetic air (SA) are shown in black dots.

10

WA SA

8 6 4 -1

D / (nmol mol )

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


2 0 -2 -4 -6 -8 -10 KRISSKRISS NIM NIM NIST NIST NMIJ NMIJ NOAANOAA NPL NPL VNIIM VNIIM VSL VSL

Laboratory

Figure 2. Graph of equivalence for the key comparison CCQM-K82. The error bar represents the expanded uncertainty at a 95 % level of confidence. For each pair of standards the degree of equivalence for the low mole fraction standard is plotted before the high mole fraction standard. The improvement in the comparability of CH4 in air standards during the period 2003 to 2013 is evidenced using the results of the CCQM-K82 comparisons and those of CCQM-P41 part 1 and part 2 organized in 20031,2. Figure 3 plots the degrees of equivalence obtained during the three comparisons. As can be observed the results have substantially improved. Using the standard deviation of all degrees of equivalence for each comparison as a global indicator of agreement, a reduction by a factor of more than fifteen is observed, from 30 nmol mol-1 in 2003 to 1.7 nmol mol-1 in 2013.


19


150

CCQM-P41 part 1, 2003 coordinated by VSL

CCQM-P41 part 2, 2003, coordinated by VSL

100 -1

Di / (nmol mol )

CCQM-K82, 2013 coordinated by BIPM/NIST

50

KRISS KRISS NIM NIM NIST NIST NMIJ NMIJ NOAA NOAA NPL NPL VNIIM VNIIM VSL VSL

0

KRISS GUM NIST NMi VSL CEM NMIA NPL NMIJ BAM

-50 IMGC CENAM CEM NOAA NIST KRISS CSIRO-AR NMIJ NMi VSL NMIA NPL

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 20 of 25

Figure 3. Degree of equivalence between CH4 gravimetric mole fractions during the last ten years. The error bar represents the expanded uncertainty at a 95 % level of confidence.

The 2003 comparison1,2 of the differences in primary gas standards for CH4 resulted in a standard deviation of results of 30 nmol mol-1, and 10 nmol mol-1 for a more limited set of standards. For the six NIST standards studied in 2013 the variation from the reference value for the whole air standards was characterized by a standard deviation of 0.74 nmol mol-1 (0.04 % relative) and that of synthetic air standards by a standard deviation of 0.29 nmol mol-1 (0.015 % relative). These results are similar in magnitude to the stated standard gravimetric uncertainties of these standards which range from 0.56 nmol mol-1 to 0.81 nmol mol-1. For the totality of the standards compared in CCQM-K82, the variation from the reference value for the whole air standards was characterized by a standard deviation of 2.07 nmol mol-1 (0.1 % relative), and those of synthetic air standards by a standard deviation of 1.06 nmol mol-1 (0.055 % relative), and all but one standard agreed with the reference value within the stated uncertainty. Greater variability is observed for standards from different laboratories than from a single laboratory. The most probable cause for this is the uncertainty and difficulty in obtaining accurate measurements of trace CH4 levels in the matrix gases as described by Rhoderick et al.7. However, no significant difference in performance between standards made synthetically or using whole air was observed and the accuracy of both types of standard is now limited by the ability to measure trace CH4 levels in the matrix gases used to produce the standards.


20

Page 21 of 25

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


CONCLUSIONS The measurement methods described in this paper have allowed whole air and synthetic air standards to be compared with sufficient precision to be able to conclude that the choice of either matrix can lead to equivalent measurement results by either GC-FID or CRDS measurements, if certain limits in the mole fraction composition of the air matrix used to produce the standards (amount of N2, O2 and Ar) are respected. The comparability of CH4 in air standards has improved considerably within the last ten years, with the standard deviation of the values of standards from their reference values, analyzed in the most recent international comparison, determined to be 1.7 nmol mol-1. The major factor in this dispersion is most likely the uncertainty arising from the measurement of trace CH4 in the matrix gas, which requires measurements at similarly low mole fractions both in whole air and synthetic air component gases. Reduction in this uncertainty and improvements in the accuracy of measurement techniques at the 1 nmol mol-1 level of CH4 in air will contribute significantly to the further improvement in comparability of CH4 standards at the 2000 nmol mol-1 level. Ideally, future instrumentation that will allow the measurement of trace CH4 to (0.1 ± 0.05) nmol mol-1 would contribute significantly to the overall accuracy in CH4 measurement in air standards. At this stage the contribution to the uncertainty in CRDS measurements of CH4 mole fractions at atmospheric levels arising from isotopic variation in calibration standards will become more significant and require further characterization.

ACKNOWLEDGMENTS The authors would like to acknowledge the Gas Analysis Working Group (GAWG) of the Consultative Committee for Amount of Substance: Metrology in Chemistry and Biology (CCQM) for the useful discussions.

Disclaimer: Certain commercial equipment, instruments and materials are identified in order to specify experimental procedures as completely as possible. This does not imply a recommendation or endorsement by the Bureau International des


21


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 22 of 25

Poids et Mesures and the National Institute of Standards and Technology nor does it imply that any of the materials, instruments or equipment identified are necessarily the best available for the purpose.

References: (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14) (15) (16) (17) (18) (19) (20) (21) (22) (23) (24) (25)

Veen, A. M. H. v. d.; Brinkmann, F. N. C.; Arnautovic, M.; Besley, L.; Heine, H.-J.; Esteban, T. L.; Sega, M.; Kato, K.; Kim, J. S.; Castorena, A. P.; Rakowska, A.; Milton, M. J. T.; Guenther, F. R.; Francey, R.; Dlugokencky, E. Metrologia 2007, 44, 08002. Veen, A. M. H. v. d.; Brinkmann, F. N. C.; Arnautovic, M.; Besley, L.; Heine, H.-J.; Esteban, T. L.; Sega, M.; Kato, K.; Kim, J. S.; Castorena, A. P.; Rakowska, A.; Milton, M. J. T.; Guenther, F. R.; Francey, R.; Dlugokencky, E. Metrologia 2007, 44, 08003. GAW; Brand., W. A., Ed.: Jena, Germany,, 2009. GAW; WMO, 2009; pp 49. GAW, 2011; pp 67. Dlugokencky, E. J.; Myers, R. C.; Lang, P. M.; Masarie, K. A.; Crotwell, A. M.; Thoning, K. W.; Hall, B. D.; Elkins, J. W.; Steele, L. P. Journal of Geophysical Research: Atmospheres 2005, 110, D18306. Rhoderick, G. C.; Carney, J.; Guenther, F. R. Analytical Chemistry 2012, 84, 3802-3810. Brewer, P. J.; Brown, R. J. C.; Miller, M. N.; Miñarro, M. D.; Murugan, A.; Milton, M. J. T.; Rhoderick, G. C. Analytical Chemistry 2014, 86, 1887-1893. Nara, H.; Tanimoto, H.; Tohjima, Y.; Mukai, H.; Nojiri, Y.; Katsumata, K.; Rella, C. W. Atmos. Meas. Tech. 2012, 5, 2689-2701. Crosson, E. R. Applied Physics B 2008, 92, 403-408. Chen, H.; Winderlich, J.; Gerbig, C.; Hoefer, A.; Rella, C. W.; Crosson, E. R.; Van Pelt, A. D.; Steinbach, J.; Kolle, O.; Beck, V.; Daube, B. C.; Gottlieb, E. W.; Chow, V. Y.; Santoni, G. W.; Wofsy, S. C. Atmos. Meas. Tech. 2010, 3, 375-386. Winderlich, J.; Chen, H.; Gerbig, C.; Seifert, T.; Kolle, O.; Lavric, J. V.; Kaiser, C.; Hofer, A.; Heimann, M. Atmos. Meas. Tech. 2010, 3, 1113-1128. Karion, A.; Sweeney, C.; Wolter, S.; Newberger, T.; Chen, H.; Andrews, A.; Kofler, J.; Neff, D.; Tans, P. Atmos. Meas. Tech. Discuss. 2012, 5, 7341-7382. Rella, C. W.; Chen, H.; Andrews, A. E.; Filges, A.; Gerbig, C.; Hatakka, J.; Karion, A.; Miles, N. L.; Richardson, S. J.; Steinbacher, M.; Sweeney, C.; Wastine, B.; Zellweger, C. Atmos. Meas. Tech. 2012, 6, 837-860. Lee, J.-Y.; Yoo, H.-S.; Marti, K.; Moon, D. M.; Lee, J. B.; Kim, J. S. Journal of Geophysical Research: Atmospheres 2006, 111, D05302. Stowasser, C.; Buizert, C.; Gkinis, V.; Chappellaz, J.; Schüpbach, S.; Bigler, M.; Faïn, X.; Sperlich, P.; Baumgartner, M.; Schilt, A.; Blunier, T. Atmos. Meas. Tech. 2012, 5, 999-1013. Flores, E.; Viallon, J.; Choteau, T.; Moussay, P.; Wielgosz, R. I.; Kang, N.; Kim, B. M.; Zalewska, E.; Veen, A. A. M. H. v. d.; Konopelko, L.; Wu, H.; Qiao, H.; Rhoderick, G.; Guenther, F. R.; Watanabe, T.; Shimosaka, T.; Kato, K.; Hall, B.; Brewer, P. Metrologia (Tech.Suppl.) 2014 (submitted for publication). Alink, A. Metrologia 2000, 37, 35. Park, S. Y.; Kim, J. S.; Lee, J. B.; Esler, M. B.; Davis, R. S.; Wielgosz, R. I. Metrologia 2004, 41, 387. Milton, M. J. T.; Harris, P. M.; Smith, I. M.; Brown, A. S.; Goody, B. A. Metrologia 2006, 43. ISO In 16664:2004. BIPM; IEC; IFCC; ISO; IUPAC; IUPAP; OIML., 2008. Allan, D. W. In IEEE T. Instrum. Meas., 1987; Vol. IM-36; pp 646-653. Quay, P.; Stutsman, J.; Wilbur, D.; Snover, A.; Dlugokencky, E.; Brown, T. Global Biogeochemical Cycles 1999, 13, 445-461. ISO In 6143:2001, 2001.


22

Page 23 of 25

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Difference between CH4 gravimetric mole fractions and predicted values using CRDS method 1 (black dots), CRDS method 2 (blue dots) and GC-FID (red dots) of the NIST synthetic mixtures at 1815.72 nmol mol-1 (FF4234), 1929.63 nmol mol-1 (FF4190) and 1969.34 nmol mol-1 (CAL018216). The error bar represents the expanded uncertainty at a 95 % level of confidence. 205x157mm (300 x 300 DPI)



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Graph of equivalence for the key comparison CCQM-K82. The error bar represents the expanded uncertainty at a 95 % level of confidence. For each pair of standards the degree of equivalence for the low mole fraction standard is plotted before the high mole fraction standard. 205x157mm (300 x 300 DPI)


Page 24 of 25

Page 25 of 25

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Degree of equivalence between CH4 gravimetric mole fractions during last twenty years. The error bar represents the expanded uncertainty at a 95 % level of confidence. 209x148mm (300 x 300 DPI)


Methane Standards Made in Whole and Synthetic Air Compared by

Recommend Documents