Evaluation of sample preparation methods for elemental profiling of wine by

ICP-MS: comparison of direct dilution, microwave digestion, and filtration

Joshua D. Godshaw1, Helene Hopfer3, Jenny Nelson2, and Susan E. Ebeler1
1Department of Viticulture and Enology, University of California, Davis, One Shields Ave, Davis, CA 95616 EWCPS 2017
2Agilent Technologies, Inc., 5301 Stevens Creek Blvd, Santa Clara, CA 95051 Poster number: 11
3Department of Food Science, The Pennsylvania State University, University Park, PA 16802

Introduction Experimental Results and Discussion

Wine elemental composition is known to vary with Principal Component Analysis
respect to origin, grape variety, environment, and
winemaking practices. Elemental analysis of wines is Direct Dilution Acidification Filtration prior to Microwave
Acidification
usually performed employing inductively coupled plasma (DD) Prior to
(FA)
Digestion (MW)
Filtration (AF)
mass spectrometry (ICP-MS). ICP-MS analysis of wine is
challenging as the variable level of organic matrix
components requires special operating conditions and
matrix-matching. Additionally, organic matrix components Samples diluted 1:3 with Samples diluted 1:3 with
Sample filtered using
Agilent Captiva Premium
2 x 1mL concentrated HNO3

and suspended particulates can build up in the sample 5% HNO3 (v/v) 5% HNO3 (v/v) added to 2 mL sample in
syringe filter (PTFE, 15 mm, PTFE microwave tube
0.45 µm)
introduction system. Sample preparation prior to analysis
offers the opportunity to eliminate or minimize these
interferences. However, the absence of a universal
pretreatment for wine ICP-MS analysis has contributed to Diluted sample filtered using
Agilent Captiva Premium Filtered samples diluted
Capped sample tubes
digested using Milestone
conflicting recommendations of best practices. syringe filter (PTFE, 15 mm,
0.45 µm)
1:3 with 5% HNO3 (v/v) UltraWAVE according to
manufacturer’s
recommendations (Table 2) Figure 1. PCA biplot of scores of wine means and loadings of significant elements by
ANOVA in first two principal components ( 54.42 % total variance).

2mL aliquot of digested

Purpose
sample diluted 1:5 with
Ultrapure water

To compare sample preparation methods in the

elemental analysis of wine by ICP-MS:
Milestone UltraWAVE
• Direct analysis after dilution (DD) Parameter
Power:
Value
1500 W
Pressure: 150 bar

• Acidification prior to filtration (AF)

Temperature: 240°C
Temperature ramp time: 20 minutes
Hold time: 10 minutes
Figure 2. PCA biplot of scores of wine means and loadings of significant elements by
Table 2. Milestone UltrWAVE microwave digestion settings. ANOVA in first and third dimensions ( 54.42 % total variance).
• Filtration prior to acidification (FA)
Spike Recovery
• Microwave assisted acid digestion (MW)
Results and Discussion
Limits of Detection (LOD) and Method Blank
Experimental Concentrations

Experimental Design Instrumental LOD Mean Method Blank Concentrations

 Samples: prepared in triplicate by each method (μg/L) (μg/L)

 Unfiltered nor fined wine samples of Chardonnay (C), Pinot

Noir (PN), Syrah (S), and Tempranillo (T) with 12-15%
Isotope Mode AF-FA-DD MW AF DD FA MW
7
Li NG 0.096 0.030 1.23 1.77 1.50 0.962
ethanol content (UC-Davis winery) 27
Al He 0.748 1.035 <LOD 1.84 <LOD 13.72
Figure 3. Mean 65Cu spike recovery (%) of different preparation methods and sample types (n=3). Means not
sharing a letter are statistically significantly different by Tukey’s HSD.
 Method blanks (BL) were a 12% ethanol solution prepared 47
Ti He 0.220 0.182 0.611 <LOD <LOD 2.27
via all four methods 51
V He 0.009 0.009 0.013 0.020 0.015 0.029
52
 All wines were centrifuged at 4°C and 5000 x g for 10 55
Cr He 0.205 0.052 <LOD 1.04 <LOD 5.07

minutes prior to preparation to reduce suspended

Mn He 0.063 0.042 <LOD 0.772 <LOD 20.08
59
Co 0.004 0.003 <LOD 0.026 <LOD 1.36
particulates in unrefined wine samples
He
60
Ni He 0.080 0.044 <LOD 0.600 <LOD 5.22
 Calibration 63

65
Cu He 0.046 0.044 <LOD <LOD <LOD 0.248

 43 isotopes were quantified via a 6 point external calibration 0.017 0.011 <LOD <LOD <LOD 0.045
Cu He
66
Zn 0.261 0.192 <LOD <LOD <LOD 1.16
ranging from 0.1 to 500 μg/L
He
71
Ga He 0.004 0.004 0.007 0.006 0.010 <LOD
 Cu in wine was also quantified using isotope dilution after 75
As HEHe 0.012 0.012 0.056 0.154 0.116 0.106
spiking with 100 μg/L 65Cu 78
Se HEHe 0.074 0.019 <LOD <LOD <LOD 0.280

 Spikes
85
Rb He 0.040 0.068 0.467 0.558 0.279 0.130
88
Sr He 0.019 0.021 0.065 <LOD <LOD 0.050
 100 μg/L 65Cu and 5 μg/L 206Pb stable isotope standards 93
Nb He 0.007 0.007 <LOD 0.018 <LOD 0.525
were chosen to represent typical wine levels 98
Mo NG 0.070 0.018 <LOD 0.072 0.099 1.95 Figure 4. Mean 206Pb spike recovery (%) of different preparation methods and sample types (n=3). Means not
sharing a letter are statistically significantly different by Tukey’s HSD.
 Quality Control
101
Ru He 0.008 0.010 0.009 <LOD <LOD <LOD
Cu Isotope Dilution
103
Rh He 0.002 0.001 <LOD <LOD <LOD 0.034
 Instrument performance was continuously monitored using 107
Ag He 0.014 0.002 <LOD <LOD <LOD 0.138
an internal standard (ISTD) mix containing 1 µg/L 6Li, 45Sc, 111
Cd He 0.007 0.009 0.040 <LOD 0.040 <LOD
72Ge, 89Y, 115In, 159Tb, 209Bi in 1% HNO 123
Sb 0.040 0.013 <LOD 0.060 <LOD 0.732
3 He
 Repeat analysis of 10 μg/L calibration standards of each
125
Te NG 0.003 0.003 <LOD 0.007 0.004 0.005
Figure 5. Comparison of isotope
calibration series approximately every 15 samples accurate
133
Cs He 0.017 0.015 0.029 0.033 0.074 0.033
137 dilution and external calibration results
and precise within 20% of value and <20% RSD per
Ba He 0.038 0.043 <LOD <LOD <LOD 0.124
140
for 63Cu shown averaged by sample and
Ce 0.002 0.004 0.003 0.002 <LOD <LOD method. Black line represents identical
analytical run
He
141
Pr He 0.000 0.003 <LOD 0.003 <LOD <LOD results (slope=1), blue line is the best
 Statistical Analysis (P ≤ 0.05) 146

147
Nd He 0.002 0.003 0.005 0.002 0.002 <LOD
fit linear approximation of the data.

Analysis of Variance (ANOVA) 0.002 0.002 0.002 0.004 <LOD <LOD

Sm He
 153
Eu He 0.000 0.003 <LOD 0.001 <LOD <LOD
 Multivariate Analysis of Variance (MANOVA) 157
Gd He 0.001 0.004 <LOD 0.004 <LOD <LOD
 Principal Component Analysis (PCA) 163
Dy He 0.001 0.005 <LOD <LOD <LOD <LOD
165
 Tukey’s Honestly Significant Differences (HSD) Ho He 0.000 0.003 0.001 0.000 <LOD <LOD
166
Er He 0.001 0.003 <LOD 0.003 <LOD <LOD • R: instrument response ratio of 63Cu:65Cu
169
Tm 0.001 0.002 <LOD 0.001 <LOD <LOD • mspike: mass of spike solution added to sample
He
172 • msample: mass of sample
Yb He 0.001 0.003 0.008 0.004 0.005 0.005 • Wspike: atomic weight of Cu in spike
Wsample: atomic weight of Cu in sample
Instrumental Parameters
181
Ta He 0.020 0.035 <LOD 0.032 <LOD 1.94 •
182 • A63 and A65: abundances of 63Cu and 65Cu in spike solution
W He 0.109 0.013 <LOD 0.189 0.122 1.27
205
• B63 and B65: natural abundances of 63Cu and 65Cu in sample
Tl He 0.002 0.003 0.005 0.007 0.008 <LOD
206
Pb He 0.009 0.003 <LOD 0.140 <LOD 0.030

Conclusions
208
Pb He 0.006 0.004 0.017 0.130 <LOD 0.043
Agilent Technologies 8800x ICP-MS 238
U He 0.004 0.001 0.009 <LOD 0.009 <LOD
Parameter Value
RF power:
RF matching:
1550 W
1.8 V
• Significant effect of preparation method observed
Nebulizer pump speed: 0.1 rps • 37 isotopes in wine significantly differed by method
Replicates: 3 • MW is most variable preparation treatment in wine analysis
Sweeps per replicate: 100 Table 3. Limits of detection (LOD) and average method blank (n=3) concentration of 43 isotopes
monitored by ICP-MS without a collision gas (NG), in helium mode (He), and high energy helium
and significantly differed from all other methods for 21
isotopes measured when averaged over the wines
Carrier Gas (Ar): 1.05 L/min
Collision Gas (He): 5 mL/min (He mode) mode (HEHe). Limits of detection are expressed as 3.14 times the standard deviation (n=6 for direct
10 mL/min (HEHe mode)
methods or n=8 for MW) of matrix matched calibration blanks per analytical run. LODs for direct
methods shown are the average of two analytical runs.
• Number of steps present risk for contamination
• Statistical significance may not mean scientific significance
• All methods tested adequately separated the different wine
Table 1. ICP-MS operating conditions. samples (Figure 2), although extreme care must be taken if
using MW for absolute quantitation