Experience in the Use of the LBMA Reference Materials Mike Hinds Royal Canadian Mint LBMA Assayer and Refiner March 2011 1 LBMA RM Project 2007-2010 2 Gold Reference Materials AuRM1 and AuRM2 Available since June 2009 2 Silver Reference Materials AgRM1 and AgRM2 Available since June 2010 Next Reference Materials need to decided Input from Refiners/Assayers required 7-8 March 2011 1
Use of Reference Materials Verify the accuracy and precision of existing analytical methods for the determination of elements in the matrix Solid sample: Spark AES and Spark Ablation ICP-AES Solution: ICP-AES and ICP-MS Use for the calibration of spectrometers Validate the analysis of in-house reference materials by using as quality control or validation standard 7 Laboratories Participating Laboratories 5 Spark OES (solid sample) 2 ICP-AES (solution) THANK YOU For Sharing Your Results! 7-8 March 2011 2
Comparing Lab Results to CRM Values Statistical Test: Student s t-test t = (x crm x lab )/s( (1/n crm + 1/n lab )) if t < t tab, then no significant difference Use only if s crm s lab If not then the calculation is more complicated CRM standard deviation (s crm ) CI = ± 2s crm Inspection of Overlap between CRM values and Lab values Lab Measurement Overlap with LBMA RM Concentration Values Lab values = CRM values = Excellent Good Acceptable Review precision 7-8 March 2011 3
Overlap with CRM concentration values Slight Overlap Bias Problem Significant Bias Example of Lab Result Comparison Te in AgRM2 AgRM2 Certified LAB LAB LAB LAB LAB LAB LAB Te Concentration A B C D E F G Concentration, ppm 38.1 40.0 56.4 31.7 37.4 40.8 44.6 39.2 ± s 1.2 2.5 1.6 3.9 2.1 0.8 0.8 n 8 4 2 5 13 6 6 ± 2s 3.5 2.3 5.0 3.1 7.9 4.2 1.6 1.6 Lower 34.6 37.6 51.4 28.6 29.6 36.6 43.0 37.6 Upper 41.6 42.3 61.4 34.8 45.3 45.0 46.2 40.8 Overlap Good Bias Slight Overlap Wide Confidence Interval Good Bias Good G B SO Wide CI G B G 7-8 March 2011 4
Ag in AuRM1 Examination of Confidence Intervals AuRM1 Certified LAB LAB LAB LAB LAB LAB LAB Ag Concentration A B C D E F G Concentration, ppm 20.0 19.4 23.9 20.3 22.2 33.3 20.6 ± s 0.1 1.9 0.4 0.3 1.3 0.3 N 4 5 5 7 2 3 ± 2s 0.8 0.2 3.7 0.8 0.5 2.5 0.6 Lower 19.2 19.2 20.2 19.5 21.7 30.8 20.0 Upper 20.8 19.6 27.7 21.1 22.7 35.8 21.2 Overlap Good Sl.Overlap Wide CI Good Bias Bias Good G Wide CI G B B G Lab Results for AgRM1 A B C D E F G Method SpOES SpOES SpOES ICP SpOES ICP SpOES Al Wide CI G G B As G G G G G Au Wide CI G B G WideCI WideCI WideCI Bi G G G G G SO G Cd G G G G SO B Cr G B G B Cu G G G G G G G Fe SO G G G G G G Mg G SO G B Mn G G G SO Ni G G SO G G G G Pb G G SO B G B G Pd G G G G SO B B Pt B G G G G G Rh G SO SO G Sb G G G G SO SO Se G G G G G B G Si G G B Sn SO G G G G G Te G B G Wide CI G SO SO Zn G G G SO G G G 7-8 March 2011 5
Lab Results for AgRM2 A B C D E F G Method SpOES SpOES SpOES ICP SpOES ICP SpOES Al G wide CI B B As G G G G B G Au G G SO G G B G Bi G G G G G SO G Cd G SO B G SO B Cr G G G G B Cu G B G G SO G G Fe SO B SO SO B B B Mg B G G B Mn G B SO SO Ni G G G SO SO SO G Pb G G G B G B B Pd G SO B G G G G Pt G G G G G G Rh G G G SO Sb G G G G SO G Se G G G wide CI SO G G Si wide CI G B Sn wide CI B G SO G G Te G B SO wide CI G B G Zn G G wide CI G G G SO Lab Results for AuRM1 A B C D E F G Method SpOES SpOES ICP SpOES ICP SpOES Ag G Wide CI G B B G Al G G G As G G G B Bi G G G SO G Ca G G Cr Wide CI G G SO Cu B G G G G G Fe G G SO G G Mg G G G G Mn G G G G Ni G G G G G Pb G G G G G G Pd B G G G G G Pt G G G G Wide CI G Rh B G G G Sb G G G Se G G G Si G G G SO Sn G G G G Te G G B B Ti G G G G Zn G G G G G 7-8 March 2011 6
Lab Results for AuRM2 A B C D E F G Method SpOES SpOES ICP SpOES ICP SpOES Ag G G G SO SO B Al Wide CI G G As Wide CI G G B Bi G Wide CI B SO G Ca G G Cr G G G G Cu B SO G G G G Fe G G SO B G Mg SO G G B Mn SO G G G Ni SO G SO B G Pb G SO SO G B G Pd B Wide CI G G SO G Pt B G G G G G Rh B G G G Sb Wide CI G G Se B B SO Si B G G B Sn B G G G Te SO G Wide CI G Ti G G G G Zn G G G G SO % Determined Elements in RM Overlap Categories AgRM1 AgRM2 AuRM1 AuRM2 Good Overlap Overlap with Wide CI Slight Overlap Bias No Overlap 70.5% 60.7% 83.5% 61.9% 4.9% 4.9% 3.1% 6.2% 13.1% 15.6% 5.2% 15.5% 11.5% 18.9% 8.2% 16.5% 7-8 March 2011 7
Observations High % of agreement with RM values More agreement with AgRM1 and AuRM1 Suggests a need for higher element concentration RMs Every laboratory has issues to be addressed OK, this was expected There has never been such collaboration before High quality RM s covered many elements were not available before Expect more agreement in 2013 Some laboratories reported n=2 replicates Standard Deviation calculated but not valid Observations Fe in AgRM2 may need to be re-examined examined by the RM Project Steering Committee Elements that are challenging*: AgRM1 AgRM2 AuRM1 AuRM2 Cd, Cr, Mg, Pb, Pd, Rh, Te Au, Cd, Cu, Mg, Mn, Ni, Pb, Pd, Sn, Te Ag, Te Ag, Bi, Cu, Fe, Mg, Ni, Pb, Pd, Se, Si * 2 or more indicators of lab bias (red or grey) 7-8 March 2011 8
Next Steps Collect RM data from participating labs in 2013 Expect to see better agreement Hope to have more labs sharing data and experiences Continue work on more Reference Materials Survey in 2010 inconclusive No one potential RM was wanted more than others Poor response Next Reference Materials Options 99.5 99.9 Au and Ag Reference Materials Higher concentrations of all elements Much higher concentration levels Silver: Au, Cu, Pd, Pt Gold: Ag, Cu, Pd, Pt Current Ag & Au RM indicate bias at higher levels Ag & Au RM s with elements less commonly analysed elements (current RM s cover 70% elements) Impure Ag & Au with % level impurities - RF 7-8 March 2011 9
Acknowledgements 7 Participating Laboratories LBMA RM Steering Committee Members LBMA Royal Canadian Mint Slides for Discussion Definition of Purity Uncertainty 7-8 March 2011 10
Variation in the Number of Elements Determined in Gold and Silver Lab A Lab B Lab C Lab D Lab E Lab F Lab G Silver 19 15 21 + 21 + 12 14 Gold 5 21 22 + 16 11 22 + Other elements not in the RM s may be determined by each lab Definition of 99.99% Au and Ag In 2008, LBMA published a list of elements that are determined by the Referees. Good Delivery Rules Annex M Core Elements Gold: 26 Silver: 23 Additional Elements Gold: 16 Silver: 9 7-8 March 2011 11
LBMA CORE Elements Gold Silver Ag Cu Pt Pd Rh Ni Fe Pb Zn As Se Sb Bi Cr Mn Co Al Ti Si Ca Cd Mg Sn Te Ir Ru Au In LBMA Additional Elements Gold Silver B In Mo K Na Os Tl Zr Be Li W P Hg Ge Ga V Rh Ti 7-8 March 2011 12
Uncertainty Determining Too Few Elements Risk of not measuring impurities present Risk of over reporting purity (High Bias) Determining Too Many Elements Increase in uncertainty If LOD s included in calculation of purity Risk of under reporting purity (Low Bias) Conservative Normal Distribution of Results 7-8 March 2011 13
Extreme Case Sum of Impurites = 100 ppm 100 ppm impurities Au < 99.99% Au > 99.99% Establish a Maximum Limit Based on Uncertainties and Bias 100 ppm impurities Max Limit Au > 99.99% 7-8 March 2011 14
Calculation of Uncertainty Many Different Ways of Calculation Sum (Limit of Detection) Sum (1/2 Limit of Detection) Square Root (Sum of Variances) LOD = 3σ Can validly add variances then take square root U = 2 x (sqrt(σ(lod/3)(lod/3) 2 )) Example Calculations Consider: 26 elements determined in Gold 6 elements determined above LOD s: : Sum = 60 ppm Sum of Biases = + 5 ppm 20 elements at or below LOD of 1 ppm each 7-8 March 2011 15
Summary Account for uncertainty in estimating a maximum limit for sum of impurities (< 100 ppm) Include bias in the calculations for uncertainties Compare to LBMA RM s Different methods of uncertainty calculations yield different results 7-8 March 2011 16