Progress on sampling and analysis of carbonaceous aerosols: The EUSAAR strategy Jean-Philippe Putaud, Fabrizia CAVALLI, EC-JRC-IES

▪ Positive artifact mitigation ▪ Negative artifact assessment 1 Progress in the validation of the sampling train for particulate carbonaceous matter: ▪ Positive artifact mitigation ▪ Negative artifact assessment 2 Progress in defining a reference method for thermal-optical analysis of OC+ EC: ▪ CEN WG on OC and EC in PM 3 EUSAAR intercomparisons for OC / EC / TC measurements: ▪ Results of the 3rd and 4th intercomparisons ▪ Next intercomparison in 2010 Briefly explain the content of this presentation

Positive artifacts 1- POC / EC sampling train validation ▪ Positive OC artifacts contribute 14 – 70 % of particulate OC on plain QFF (site and season dependent) ▪ Denuder mean performance ranges from 43 to 93% (site and season dependent) ▪ Positive OC artifacts are reduced to 5 – 21 % The issue of positive artifact has been successfully addressed

Negative artifacts: new tests 1- POC / EC sampling train validation Negative artifacts: new tests Sampling head ▪ NO news from supplier on special production of suitable sorbents ▪ NO alternative in the very recent literature + blank problems ▪ Tried again quartz fiber filter as back-up filter in two different experiment types Denuder Carbon honeycomb monolith Definition of negative artifact: volatilisation of SVOC from filter-bound aerosol particles. the assessment of the sampling negative artifact is performed using sorbent substrate back to the front QFF. C-impregnated glass fiber filters has been usuallproduced by a company disapperaed in the mean time. The production has therefore stopped and no other suppliers were willing of setting-up special production of such substrates. Further in the very recent literature no alternative has been proposed and tested and alternative sorbent were chacarteised by strong contaminations and fast aging. We have decided therefore to test again QFF as sorbents or back up filters in two different exeroments: Front quartz filter Back-up filter

Two back-up quartz filters 1- POC / EC sampling train validation Negative artifacts: new tests Sampling head 24-hr sampling 20 cm/s face velocity 14 < T < 28°C Denuder Carbon honeycomb monolith The first experiment consists of annealing two QFF back to the QFF Characteristic of the sampling Front quartz filter Two back-up quartz filters

1- POC / EC sampling train validation Negative artifacts: new tests Describe figure. Conclusions: ▪ efficiency 80% comparing the amount of OC on the two back-up; ▪ negative artifact = only 3% compared to particulate OC comparing the amount of OC found on the back-up filters with that obtained on the front OC ▪ very limited set of data Quartz fiber filter as back-up absorbent: ▪ efficiency 80% ▪ negative artifact = only 3% compared to particulate OC ▪ very limited set of data

C D A B 1- POC / EC sampling train validation Day i Day i + 1 Negative artifacts: new tests additionally C D A B Teflon filters Sampling head 24-hr sampling 20 cm/s face velocity Denuder Alternatively in a more complex a test: …describe test. At Day i we performed to parallel aerosol sampling with these identical configuration. In such configuartion, OC on that filters consists of equations The filter B sampled at day I is further sampled for 24h the following day with in front a teflon filter to remove particulate OC. We will have that …equation We can now compare OC A and C and the difference represents the residual positive artifact and the negative artifact on day i+1 In addiation on day 1+I we determined sampling through the sampling train D the residual positive artifact. We can now estimate the remianing negative artifact. Front quartz filter Day i Day i + 1

1- POC / EC sampling train validation Negative artifacts: new tests in Ispra DAY i A & B OC = OCparticulate i + OCresidual positive art i+ OCnegative art i DAY i+1 C OC = OCparticulate i + OCresidual positive art. i+ OCnegative art i + + OCresidual positive art i+1+ OCnegative art i+1 C-A ΔOC = OCresidual positive art i+1+ OCnegative art i+1 Additionally, DAY i+1 D OC = OCresidual positive art i+1 This slide was inserted 20 Apr. 2010

1- POC / EC sampling train validation Negative artifacts: preliminary results C-A ΔOC = OCresidual positive art i+1+ OCnegative art i+1 D OC = OCresidual positive art i+1 ▪ The OCresidual positive art is 0.37  0.17 gC/m3 (0.54  0.21 gC/cm2) in winter and 0.57  0.22 gC/m3 (0.77  0.32 gC/cm2) in summer. Range confirmed in 2010 tests, 0.42  0.22 gC/cm2. The negative artifact would represent up to 20% of the OCparticulate Conclusion

2- Towards a reference method for OC / EC thermal–optical analysis The EUSAAR_2 protocol Fraction_Carrier gas Temp C Time s OC1_He 200 120 OC2_He 300 150 OC3_He 450 180 OC4_He 650 EC1_He/O2 500 EC2_He/O2 550 EC3_He/O2 700 70 EC4_He/O2 850 80 Analysis: Thermal optical method developed within the EUSSAR community in which major biases affecting the existing protocol NIOSH and IMPROVE has been mitigated. F. Cavalli, M. Viana, K. E. Yttri, J. Genberg, and J.-P. Putaud (2010). Toward a standardised thermal-optical protocol for measuring atmospheric organic and elemental carbon: the EUSAAR protocol. Atmos. Meas. Tech., 3, 79–89.

2- Towards a reference method for OC / EC thermal–optical analysis CEN working group on OC /EC 36 resolutions so far Resolution 3 • Analysis: EC and OC data should allow/support mass closure of PM along with other PM constituent determined e. g. within the AQD framework. Therefore a mass based technique should be used. • Links to other standards and network protocols The European Standard should not be developed in isolation from relevant work in other communities especially EMEP, EUSAAR and the US/Canada. Resolution 4 WG 35 agrees that the candidate standard method will be based on the analysis of samples collected on filters using the thermal method with optical correction for charring. The protocol together other is under the evalution of the the CEN working group on OC EC to eentually recieive the status of european standard methods. The activity of this working group started one year ago and up to now 36 resolution has been taken.

CEN working group on OC /EC 2- Towards a reference method for OC / EC thermal–optical analysis CEN working group on OC /EC Resolution 5 WG 35 agrees that a literature review and data analysis needs to be undertaken including: • OC and EC levels in blank filters, • Artifacts (i.e. filter types), • Round robin tests, • Role of carbonates. These results will feed into an uncertainty assessment. Resolution 6 WG 35 agrees that a field study is needed to set up a well validated standard. The field study shall allow the evaluation of different protocols of the thermal method with optical correction for charring. Both optical parameters transmission and reflectance shall be recorded to allow an evaluation of charring correction. This will include both off-line and on-line analysis. Remind how these figures are obtained

CEN working group on OC /EC 2- Towards a reference method for OC / EC thermal–optical analysis CEN working group on OC /EC Resolution 21 To extend and clarify Resolution 7, WG 35 agrees that the choice of analytical protocol for EC and OC (and possibly TC and IC) in the standard shall be based on: - comparability with one of the three widely-used existing protocols EMEP (EUSAAR2), NIOSH, and IMPROVE, and reproducibility (ie relative insensitivity to small variations in the actual method used) - lack of systematic bias or interference (eg consideration of the effect of IC, results of analysis of “pure” OC or EC) practicality (ie time and cost implications). Remind how these figures are obtained

CEN working group on OC /EC 2- Towards a reference method for OC / EC thermal–optical analysis CEN working group on OC /EC Resolution 35 Given -the urgent need for Member States to have a standardised method for EC/OC, in the absence of a mandate from the Commission (DG ENV), the WG agrees that they will work on a CEN Technical Report for EC/OC as a priority. This TR will describe several methods (e.g. thermal protocols) that will give different results for EC and OC, because validation data is needed both to specify one standard method and to properly characterise that method. The standard method will be proposed within 3 years from the mandate from the DG ENV. Remind how these figures are obtained

3- OC / EC / TC intercalibration - 2008 ▪ Ambient aerosol samples collected by 4 partners at their sites (BIR, MSY, KPO, ISP) ▪ Samples distributed to 12 EUSAAR-NA2 partners + 12 associates ▪ Use of the EUSAAR_2 thermal-optical protocol was advised Remind how these figures are obtained

TC relative to avrage TC from EUSAAR partners 3- OC / EC / TC intercalibration - 2008 TC relative to avrage TC from EUSAAR partners On average, On average, TC carbon can be determined within ±30% error wrt average by - 10 among 10 EUSAAR partners - 7 among 9 EUSAAR associates

EC / TC relative to EUSAAR _2 3- OC / EC / TC intercalibration - 2008 EC / TC relative to EUSAAR _2 To quantify the differences between methods and among laboratories, EC/TC ratios (Fig. 3) are considered. Such ratios are independent of the possible spatial heterogeneities in filter loadings. These data confirm that EC values obtained with NIOSH-like methods are on average 30% lower than the values obtained with the EUSAAR-2 protocol. On average, EC / TC ratio can be determined within ±30% error wrt average by ▪ 7 among 8 EUSAAR partners ▪ 4 among 9 EUSAAR associates

3- OC / EC / TC intercalibration - 2009 ▪ NIST reference material 8785 (urban dust on quartz fibre filter) ▪ Samples distributed to 12 EUSAAR NA2 partners The last intercomparison (2008) showed that the standard deviation among EUSAAR partners (average 17%) could reach 24% for a single sample collected in Norway (NOR-1). It was first suspected that a fraction of the standard deviation was due to filter loading heterogeneities, but further tests performed later at JRC demonstrated it was not the case. The marginally acceptable standard deviation in TC determination should therefore come from inaccurate measurements by some of the participants. As the true value for ambient samples is unknown, it was impossible to determine which participant was delivering inaccurate values. That is why it was decided to have an additional intercomparison exercise for carbonaceous species determination with respect to the workplan. This intercomparison (2009) was based on NIST (USA – National Institute of Standards and Technology) reference materials 8785, for which reference values for TC amounts are available. For cost reasons, the intercomparison ws limited to EUSAAR partners.

3- OC / EC / TC intercalibration – NIST reference material TC amounts (µg / cm²)

TC relative to NIST reference values 3- OC / EC / TC intercalibration TC relative to NIST reference values Figure 2 shows that the ratio between the TC amount determined by each participant and the reference value ranges 0.87 – 1.55. It should be noted that the discrepancies observed in 2009 with respect to reference values were not observed in 2008 when comparing the result of each partner to the average values. To explain these inconsistencies, we first tried to check the calibration of the instrument at the time the reference filters were analyzed. Data were provided by 5 partners only. No obvious problem was detected. To further check the external standard used to calibrate the instrument, it was advised to double check the calibration by injecting known volumes of pure C-containing gases. Results of this test were reported by 2 partners only so far. It is top important to pursue this investigation to determine whether it is possible that the NIST reference material are getting contaminated while aging, e.g. by trapping volatile OC during storage and transport.

EC/TC relative to NIST reference values 3- OC / EC / TC intercalibration EC/TC relative to NIST reference values 2008 This intercomparison suggests that EUSAAR partners using the EUSAAR_2 protocol obtain EC/TC ratios for the same reference material differing by a factor 1.7. This is not acceptable. The errors in EC/TC seem to be quite systematic: some partners generally report higher values than others.

3- OC / EC / TC intercalibration- NIST reference material TC : conclusions ▪ most partners report TC values that clearly larger than the ref. value ▪ 2009’ biases cannot explain the ±30% deviation observed in 2008 ▪ the overall uncertainties are large => this intercomp. is not stringent EC /TC : conclusions ▪ EC / TC ratios range from 0.26 to 0.55 ▪ differences in line with previous intercomparison Remind how these figures are obtained

fabrizia.cavalli@jrc.ec.europa.eu 3- OC / EC / TC intercalibration Next intercomparison in 2010 ▪ Ambient aerosol samples ▪ Soot loaded filters In June 2010 punches will be distributed among partners. EMEP stations reporting OC/EC data are invited to participate please send an e-mail to: fabrizia.cavalli@jrc.ec.europa.eu Remind how these figures are obtained

Positive artifacts 1- POC / EC sampling train validation Definition of positive artifact Characterise and mitigate positive artifact

Positive artifact mitigation 1- POC / EC sampling train validation Positive artifact mitigation Tests to assess the contribution of positive artifact to the particulate organic carbon has been performed at nine Eusaar stations during different seasons Assessment of the positive artifact contribution of OC particulate collected on plain QFF filters i.e. without the use of any denuder. Range 15 -70% depending on site and season with generally lower contribution in winter. To mitigate this bias we recommended the use of a honeycomb denuder in front of the QFF. The efficiency of the denuder has been determined at each background site Range and lower efficiency in summer. Positive artifact without denuder: teflon-QFF versus palin QFF Denuder efficiency: Teflon-QFF versus Teflon-denuder-QFF

Positive artifact determination 1- POC / EC sampling train validation Positive artifact determination If denuder with such efficiencies are employed the positive artifact is considerably mitigated contributing 5 to 20% of the OC particulate. If we compare the extent of the positive artifact to the TC aerosol content, we can notice, thathigher contribution of positive artifacts occur when the TC are lower rendering more problematic this bias.

3- OC / EC / TC intercalibration- 2008 EC / TC ratio