1 Economically Motivated Adulteration of Lemon Juice:  Cavity Ring Down Spectroscopy in 
2 Comparison to Isotope Ratio Mass Spectrometry – Round-Robin Study 
3  
4 Madhavi Manthaa*, Kevin M. Kubachkaa, John R. Urban a, Anatoly Chernyshevb, William A. Markc, 
5 Christine Franced, Michelle Chartrande,Jonathan Hachef, Sylvain Decoeurg and Haiping Qih 
6  
7 aUS FDA Forensic Chemistry Center, Cincinnati, OH 45237, USA 
8 bAnalytica Laboratories Ltd, New Zealand 
9 cEnvironmental Isotope Laboratory, University of Waterloo, Waterloo, Ontario, Canada 
10 d Smithsonian Museum Conservation Institute, Suitland, MD, U.S.A 
11 e National Research Council, Ottawa, Ontario, Canada 
12 f Canadian Food Inspection Agency, Ottawa, Ontario, Canada 
13 g Canada Border Services Agency, Ottawa, Ontario, Canada 
14 h USGS/Reston Stable Isotope Laboratory, Reston, VA, U.S.A 
15  
16  
17  
18 * Corresponding author. 
19  
20 E-mail address: madhavi.mantha@fda.hhs.gov(M.Mantha)    
21                            kevin.kubachka@fda.hhs.gov(K.M.Kubachka) 
22                           John.Urban@fda.hhs.gov(J.R.Urban) 
23      anatoly.chernyshev@analytica.co.nz(A.Chernyshev)      
24      wamark@uwaterloo.ca(W.A.Mark) 
25      FranceC@si.edu(C. France) 
26      Michelle.Chartrand@nrc-cnrc.gc.ca(M. Chartrand) 
27      Jonathan.Hache@inspection.gc.ca(J. Hache) 
28      Sylvain.Decoeur@cbsa-asfc.gc.ca(S. Decoeur) 
29      haipingq@usgs.gov(H. Qi) 
30  
31        
32  
33  
34  
35  
36  
37  
38  
39  
40  
41 Reference to any commercial materials, equipment, or process does not, in any way, constitute 
42 approval, endorsement, or recommendation by the US Food and Drug Administration. 
43  
44 All views and opinions expressed throughout the presentation are those of the presenter and do not 
45 necessarily represent views or official position of US Food and Drug Administration. 
46  
47  
48  
49  
50  
51  
52 Economically Motivated Adulteration of Lemon Juice:  Cavity Ring Down Spectroscopy in 
53 Comparison to Isotope Ratio Mass Spectrometry – Round-Robin Study 
54  
55 ABSTRACT 
56 Economically motivated adulteration (EMA) of foods is an increasing concern in the recent years. In 
57 this work, an optimized sample preparation procedure for the determination of lemon juice 
58 adulteration was validated using Elemental Analyzer – Isotope Ratio Mass Spectrometry (EA-IRMS); 
59 additionally, 69 imported lemon juice samples were evaluated using Combustion Module – Cavity 
60 Ring Down Spectrometry (CM-CRDS) and compared  to the well-established EA-IRMS.  Equivalency 
61 of CM-CRDS to EA-IRMS was further demonstrated by conducting a round-robin study involving eight 
62 laboratories throughout the United States, Canada, and New Zealand. Overall, the results obtained 
63 for CM-CRDS were statistically indistinguishable from the results obtained using EA-IRMS for EMA 
64 lemon juice analysis.  
65 1 Introduction 
66 Lemon juice is a one of the common targets of Economically Motivated Adulteration (EMA). Other 
67 commonly reported targets for EMA of foods include honey, olive oil, dairy products, spices and other 
68 citrus juices 1. Available data shows that the US imported roughly 37% of the world’s crop of lemons 
69 and limes 2, while Argentina and Italy are the world’s largest suppliers of lemon juice 3. This study 
70 focuses on carbon Stable Isotope Ratio Analysis (SIRA) measuring 13CVPDB (‰) values of 69 
71 imported lemon juice samples for detecting adulteration. 13CVPDB (‰) is 13C values expressed in 
72 (units of %), defined as parts per thousand differences in the l3C/12C ratio of a sample from that in 
73 standard Vienna PeeDee Belemnite (VPDB).4-5 
74 Lemon juice consists of approximately 9% dissolved solids. The solids are approximately 60% citric 
75 acid, 20% sugars 4, minor amounts of malic acid and other components. The quality and price of 
76 lemon juice is directly related to the relative quantities of citric acid and sugars present in the juice.6 
77 There are two primary ways of lemon juice adulteration for economic gain, addition of inexpensive 
78 sweeteners, like cane sugar and/or high fructose corn syrup (HFCS), and/or the addition of 
79 exogenous citric acid. Lemon trees follow a C3 photosynthetic pathway 4 with 13CVPDB values 
80 typically ranging from -23.5 to -26.6‰ 6; therefore both sugars and citric acid naturally produced by 
81 lemons exhibit similar 13CVPDB values.  In contrast, cane sugar and HFCS, are produced in a C4 
82 photosynthetic pathway 4. Commercial citric acid is commonly manufactured by fermenting some 
83 sugars that follow C4 photosynthetic pathway  by certain Aspergillus niger strains; correspondingly, 
84 its carbon isotopic composition is close to that of C4 sugars.   Sugars and citric acid derived from C4 
85 plants have less negative (or more positive) 13CVPDB values, ranging from -13.1‰ to -9.8‰.  Addition 
86 of C4 derived sugar or citric acid to a C3 lemon juice, enriches the 13C content and increases the 
87 natural 13C  VPDB of the lemon juice in the more positive direction, which is indicative of adulteration 4. 
88 Note all subsequent reported 13C values imply 13CVPDB. 
89  Applications of carbon SIRA using an elemental analyzer interfaced with an isotope ratio mass 
90 spectrometer (EA-IRMS) to determine the 13C values of various components of  lemon juice 6-9 and 
91 other fruit juices 7, 9-12 related to EMA are reported in the scientific literature.  In various reports, the 
92 mean and standard deviation of 13C values of citric acid isolated from lemon juice were -25.5 ± 1.2‰ 
93 (n=84) 6; -24.8 ± 1.1‰ (n=10) 7; -24.1‰, -25.1‰ and -24.2‰ (fresh squeezed lemons, n=3) 4; and -
94 25.8 ± 0.8‰ (fresh squeezed lemons, n=7) and -25.9 ± 0.7‰ (lemon juice concentrates, n=7) 9.  The 
95 reported averages ranged from 25.8‰ to -24.1‰, which corresponds well to the expected values for 
96 C3 plants.  To our knowledge, no official regulations for adulteration of lemon juice with exogenous 
97 citric acid exist, therefore the data presented above was used to support the criterion for this study 
98 that a citrate value greater than -23.0‰ is indicative of adulteration. Commercial citric acid, potentially 
99 used for adulteration, is derived from fermentation of a variety of common carbohydrate sources. 
100 When produced by fermentation of corn syrup, 13C of -9.7 and -10.1‰ 4 have been reported, 
101 corresponding to the expected values for C4 plants. Other fermentation sources of citric acid include 
102 paraffin, petroleum, or beet molasses with reported 13C values ranging from of  -27.2‰ to -25.2‰ 4, 
103 consisted with C3 plants.  Citric acid from these other sources is not detectable as adulterant using 
104 the current conventional methods, including the method which is presented here. 
105  A citric acid isolation method commonly utilized in the literature is based on work by Doner 4, in 
106 which organic acids are precipitated by adding excess calcium hydroxide.  However, CO2 may 
107 become trapped during the industrial production of Ca(OH)  62  and potentially bias 13C values of the 
108 calcium citrate. Therefore, the precipitation procedure was optimized and validated within our 
109 laboratory using EA-IRMS for analysis.  The standard method for carbon SIRA is EA-IRMS, a 
110 technique requiring an elevated degree of technical knowledge for operation as well as high cost of 
111 purchase and maintenance.  To make this method more versatile, the same isolates were analyzed 
112 using another carbon SIRA technique, combustion module – cavity ring down spectrometry (CM-
113 CRDS), which has lower operation costs, simpler analysis, increased robustness (further discussed 
114 elsewhere 13).  CM-CRDS has been used previously for detecting EMA in honey 13-16, but to our 
115 knowledge detection of EMA in lemon juice by CM-CRDS has not been reported. In this study, we 
116 report comparative analysis of calcium citrate precipitates from 69 lemon juice samples by EA-IRMS 
117 and CM-CRDS as part of a single laboratory validation for CM-CRDS applied to EMA of lemon juice. 
118 As a follow up, data from a round-robin study in which with eight participating laboratories analyzed 
119 citric acid isolated from six lemon juice samples by EA-IRMS with three laboratories also utilizing CM-
120 CRDS. 
121  
122 2 Materials and Method 
123 2.1 Reagents and Standards 
124 Water used throughout the experiments was ultrapure deionized water (DIW) with resistivity of at 
125 least 18 MΩ·cm obtained from a Milli-Q system (Bedford, MA, USA) unless otherwise noted. Citrate 
126 was precipitated using various combinations of ammonium hydroxide (Fisher OPTIMA, Fair Lawn, 
127 NJ, USA), sodium hydroxide (Fisher Scientific), calcium chloride dihydrate (Sigma Aldrich, St. Louis, 
128 MO, USA), calcium hydroxide (Fisher and Acros Organics), calcium nitrate (Fisher Scientific). 
129 Validation experiments were carried out using two commercially available citric acids, citric acid 
130 monosodium salt (C6H7NaO7), from Aldrich Chemical (St. Louis, MO, USA) (13C -12.16‰, 
131 designated Source A) and citric acid anhydrous (Acros Organics Fair Lawn, NJ, USA) (13C -24.15‰, 
132 designated Source B). Three lemon juice samples (Brands 1, 2, and 3) purchased from local markets 
133 in Cincinnati, OH, along with a freshly squeezed lemon juice (referred to herein as “fresh lemon 
134 juice”) from 20 locally purchased lemons (Cincinnati, OH) were also analyzed as part of the validation 
135 experiments as in-house controls. For both EA-IRMS and CM-CRDS analysis, citrate samples were 
136 placed in 5 x 9 mm tin foil capsules from Costech Analytical Technologies, Inc. (Valencia, CA, USA). 
137 Acetanilide (Costech Analytical Technologies, Inc) was used to condition the reactors, verify proper 
138 sample combustion, and as a quality control check for 13C values. Standards used for 13C 
139 normalization to the international Vienna Pee Dee Belemnite scale were purchased from NIST 
140 (Gaithersburg, MD, USA): NIST Reference Material (RM) 8542 (IAEA-CH-6, Sucrose, 13CVPDB = -
141 10.45 ± 0.07‰), 8573 (USGS40, L-glutamic acid, 13CVPDB = -26.39 ± 0.09‰), 8543 (NBS 18, calcite,  
142 13CVPDB = -5.01 ± 0.07‰), and 8574 (USGS41, L-glutamic acid, 13CVPDB = +37.63 ± 0.10‰). The 
143 linearity, sensitivity, and precision of CM-CRDS were determined by using citric acid (Fisher) 13. All 
144 subsequent reported 13C values infer 13CVPDB. 
145  
146 2.2  Samples 
147 Sixty-nine lemon juice samples were selected from a 2013 FDA import assignment for analysis by 
148 EA-IRMS and CM-CRDS. Due to the lack of certified lemon juice reference materials, freshly 
149 squeezed lemon juice from 20 locally purchased lemons was used as an in-house control, which was 
150 prepared and analyzed along with each analytical batch to verify consistent method performance.  
151 Two adulterant solutions of 6% (w/w) citric acid (from Source A or B) were prepared and mixed at 
152 proportions of 0%, 5%, 10%, 20%, 50% and 80% (w/w) to the locally purchased lemon juice, Brand 
153 #2 to serve as adulterated samples. Samples selected for the round-robin study were chosen from 
154 the 69 above, two adulterated, two inconclusive (< 0.4‰ from the cut off value of -23‰), and two 
155 unadulterated.  In order to provide sufficient, well-homogenized material for analysis by eight 
156 laboratories, three individual preparations of each selected sample were combined, mixed with 
157 spatula thoroughly, and split into eight portions for distribution.  Each portion was homogenized and 
158 prior to distribution, at least 3 portions of each sample were analyzed by the US FDA laboratory using 
159 EA-IRMS to ensure adequate homogenization.  
160 2.3  Instrumentation and Operating Principles 
161 Lemon juice samples were analyzed using both EA-IRMS and CM-CRDS. The elemental analyzer for 
162 the EA-IRMS system was the Costech Elemental Combustion System (ECS) model 4010 from 
163 Costech Analytical Technologies, Inc. (Valencia, CA, USA) interfaced to a Thermo Delta V Advantage 
164 (Thermo-Scientific, Waltham, MA, USA) with a Conflo IV gas flow controller (Thermo Fisher, Bremen, 
165 Germany).  For CM-CRDS, the samples were combusted in the Combustion Module, Model 02 by 
166 Costech with the Liaison interface module, and CO2 Cavity Ring Down Spectrometer analyzer, model 
167 G2121-i, both from Picarro Inc. (Santa Clara, CA, USA). Nitrogen (99.9998%) was used as a carrier 
168 gas for the CM-CRDS. The principle, operation and comparison of CM-CRDS with EA-IRMS are 
169 discussed in detail in Mantha et al. 13, Balsley-Clausen et al.17, and Crosson et al.18.  
170  
171 2.5 Sample Preparation Procedure 
172 The method for precipitation of citric acid from lemon juice was adapted from Doner et al. 4, 
173 AOAC Official Method 981.0919, and AOAC Official Method 982.2120.  The procedure was optimized 
174 as detailed in the method validation section and the finalized conditions are listed as follows.  
175 Approximately 10 mL of lemon juice was poured into a 50 mL centrifuge tube and centrifuged for 10 
176 minutes at 3000 rpm, discarding the precipitated pellet (consisting of pulp and extraneous material in 
177 the lemon juice).  The pH of the supernatant was adjusted to 8.5 or above using concentrated 
178 ammonium hydroxide.  Approximately 2 mL of 3M calcium chloride (CaCl2•2H2O) was mixed with the 
179 supernatant and heated in an oven at 60°C for at least two hours to precipitate the citrate (as calcium 
180 citrate. The precipitate along the supernatant was centrifuged for 10 minutes at 3000 rpm and 
181 vacuum filtered. The precipitate was washed twice with 5 mL of DIW, once with 5 mL of acetonitrile, 
182 and finally with 5 mL of DIW. The precipitate was transferred into a petri dish and dried in an oven at 
183 60°C for over two hours. The dried calcium citrate was gently pulverized and thoroughly 
184 homogenized. Triplicate portions of the precipitated calcium citrate (0.3– 2.0 mg for EA-IRMS and 
185 0.7– 6.0 mg for CM-CRDS) were weighed into tin capsules for determination of 13C values. 
186 2.6 Multi-Laboratory Round-Robin Study Parameters  
187 Eight laboratories, four from Canada, three from the US and one from New Zealand volunteered to 
188 participate in the round-robin study conducted to evaluate the equivalency of CM-CRDS to EA-IRMS 
189 applied to EMA of lemon juice. The sample set provided to these laboratories (distributed by the US 
190 FDA laboratory) included calcium citrate samples isolated from two adulterated, two not adulterated, 
191 and two inconclusive juices out of the set of 69 previously described. Three of the eight laboratories 
192 conducted the study using both EA-IRMS and CM-CRDS; five laboratories used EA-IRMS only.  
193 Each laboratory used at least two standards for normalization of 13C values and at least one 
194 verification standard to check the stability of the run during the analysis sequence. Table 1 specifies 
195 the standards utilized by each laboratory for quality control.  Laboratories were left to their own quality 
196 control guidelines to ensure their reported values were appropriate.  All laboratories used 
197 normalization standards with 13C values which bracketed that of the sample range, and check 
198 standards with the normalization range. For laboratories that used both EA-IRMS and CM-CRDS, the 
199 same standards were used for both techniques.  
200  
201  
202  
203  
204  
205 Table 1: The isotopic standards used in the study  
EA-IRMS/CM-CRDS 
Normalization Normalization Normalization Normalization Check Check Check Check 
Standard - 1 Standard - 2 Standard - 3 Standard - 4 Standard - 1 Standard - 2 Standard - 3 Standard - 4
Lab 1 A B - - B C D A
Lab 2 E F G H I K - -
Lab 3 H B - - R - - -
Lab 4 H B - - M - - -
Lab 5 A N - - O P - -
Lab 6 J L - - K - - -
Lab 7 A N K - S - - -
Lab 8 A B - - Q - - -
A NIST RM 8573 L-Glutamic Acid; -26.39 ‰ J USGS 61, Caffeine; -35.05 ‰
B NIST RM 8542 Sucrose; -10.45 ‰ K USGS 62, Caffeine; -14.79 ‰
C IAEA-CH3 Cellulose; -24.72 ‰ L USGS 63, Caffeine; -1.17 ‰
D EIL-72 Cellulose; -25.47 ‰ M Fructose -1: -21.1 ‰
E Fructose -ILS; -10.98 ‰ N NIST RM 8574 L-Glutamic Acid; +37.63 ‰
F Galactose ILS; -21.41 ‰ O Acetanilide; -26.3 ‰
G Sucrose ILS -26.02 ‰ P Urea_UIN3; -11.7 ‰
H NIST RM 8540 'Polyethelene Foil; -32.15 ‰ Q Acetanilide; -28.32 ‰
I Nicotinamide ILS; -22.95 ‰ R HP-V3 (In-Lab Honey Check); -25.66 ‰
206 ILS Internal Laboratory Standard S Cane Sugar; -11.83 ‰  
207  
208 3 Results and Discussion 
209 3.1 Sample Preparation Optimization 
210 The precipitation of citrate was initially carried out with calcium hydroxide (Ca(OH)2) as described in 
211 Doner et al. 4 To assess the 13C values of the isolated citrates, the two citric acid sources (A & B) 
212 were dissolved in degassed water (~6% w/w, to mimic approximate citric acid levels in lemon juice), 
213 isolated, and analyzed by EA-IRMS. The resulting 13C values of the isolated citrate were compared 
214 to that of the respective neat form.  When using the procedure from Doner and coworkers, the values 
215 were similar, however, the method blanks (water rather than lemon juice) used to test contribution 
216 from reagents exhibited an elevated CO2 signal, indicating the presence of a carbonaceous impurity 
217 in the commercial attributed to dissolved carbonates in the calcium hydroxide 6.  Alternate sources 
218 from various vendors of calcium hydroxide were tested and all contained detectable carbon 
219 impurities. The observed amount of the carbon from calcium hydroxide had a negligible impact on the 
220 13C values of the isolated citrate. However, due to an unpredictable quantity of such impurities, such 
221 interference should be avoided.    Reducing the amount of calcium hydroxide resulted in an 
222 unacceptably low yield of citrate. Similar experiments were carried out substituting calcium hydroxide 
223 with calcium chloride (CaCl2) or calcium nitrate (Ca(NO3)2), or using sodium or ammonium hydroxides 
224 to adjust the pH. Precipitation of the citrates with Ca(NO3)2, in combination with either NaOH or 
225 NH4OH freshly prepared solutions, resulted in a slight positive shift in 13C values relative to the neat 
226 citric acids.  Precipitation of citrates with CaCl2, in combination with either NaOH or NH4OH, produced 
227 results consistent with the neat citric acids and with no measurable carbon signal for method blanks.  
228 Use of NH4OH provided better control of the pH adjustment and produced more precipitate than was 
229 achieved with NaOH.  For this study, CaCl2 was used along with NH4OH to precipitate citrate from 
230 lemon juice, lemon juice concentrate and citric acid samples, using the finalized conditions described 
231 in Sample Preparation Procedure.  
232  
233 3.2 Single Laboratory Validation Utilizing Modified Sample Preparation Procedure 
234 The modified sample procedure was validated for both accuracy and precision by EA-IRMS analysis, 
235 using two commercially available citric acids (Source A and B), three locally purchased lemon juice 
236 (from concentrate) samples and freshly squeezed, locally purchased lemons. 
237 3.2.1 Accuracy 
238 Assessment of accuracy was based on a comparison of results obtained from testing the calcium 
239 citrate precipitated from solutions of two pure citric acid sources, to the results obtained from the neat 
240 citric acid by EA-IRMS.  The mean 13C obtained from calcium citrate isolated from a 6% (w/w) 
241 aqueous solution of Source A (apparently derived from a C4 plant source) was -12.23 ± 0.7‰ (n=3, 
242 ±2σ) and that from the 13C obtained from the neat citric acid, -12.16 ± 0.04‰ (n=2, ±2σ).  Similarly, 
243 the mean 13C obtained from calcium citrate isolated from a 6% (w/w) aqueous solution of Source B 
244 (apparently derived from a primarily C3 plant source) was -23.94 ± 0.02‰ (n=3, ±2σ) and that from 
245 the 13C obtained from the neat citric acid (-24.15 ± 0.02‰, n=3, ±2σ).  These results demonstrated 
246 that the precipitation process does not induce significant isotopic fractionation to citric acid.   
247 Accuracy of the procedure was further demonstrated by comparison of results from the multi-
248 laboratory round-robin study presented in Section 3.4 
249 3.2.2 Precision 
250 The short term and intermediate precision of the modified procedure was demonstrated by 
251 comparison of the results obtained for three bottles of each of three brands of locally purchased 
252 lemon juice and juice from freshly squeezed lemons.  Three different analysts performed triplicate 
253 citrate isolations on each of the three bottles and the fresh lemon juice; each analyst performed their 
254 analysis on a separate day. Each of the citrates were weighed in triplicate and analyzed by EA-IRMS.  
255 The results are summarized in Table 2.  The mean standard deviation for thirty-six sets of triplicate 
256 analysis (triplicates from four lemon juice sources analyzed by three analysts) was 0.06‰ (max = 
257 0.13‰). The mean standard deviation for 27 preparations from each brand (three preparations, three 
258 days, three analysts/brand) was 0.10‰ (max = 0.11‰). The results obtained for each brand by the 
259 three analysts agreed to within 0.2‰.  
260 Table 2: Precision  
Lemon Juice Type Analyst 1 Analyst 2 Analyst 3 
13C (n=9, ± 2σ) 13C (n=9, ± 2σ) 13C (n=9, ± 2σ) 
Brand 1 -26.86 ± 0.04 ‰ -26.95 ± 0.18 ‰ -26.99 ± 0.14 ‰ 
Brand 2 -26.79± 0.12 ‰ -26.95 ± 0.12 ‰ -26.98 ± 0.08 ‰ 
Brand 3 -24.44 ± 0.12 ‰ -24.51 ± 0.28 ‰ -24.60 ± 0.10 ‰ 
Fresh Lemon Juice -25.40 ± 0.10 ‰ -25.39 ± 0.16 ‰ -25.53 ± 0.10 ‰ 
261  
262 The precision of the procedure was further demonstrated by comparison of results obtained for three 
263 brands of lemon juice from concentrate (three bottles each of brand, analyzed by three different 
264 analysts on three different days) using the modified sample preparation to historical results (past 
265 results produced by US FDA laboratory) derived from the Ca(OH)2 based method.  The mean 13C 
266 values obtained for the validation trials were not statistically distinguishable from the historical 13C 
267 values as seen on Table 3. The historical values for the lemon juices and the fresh lemon juice was 
268 obtained by using lemon juice from the same bottle, for each brand and fresh lemon juice for a period 
269 of 2 years. 
270 Table 3: Precision comparison with the historical results 
Lemon Juice Type Average 13C Historical Average 13C 
(n=27, ± 2σ) (± 2σ) 
Brand 1 -26.96 ± 0.16 ‰ -26.88 ± 0.52 ‰ 
Brand 2 -26.91 ± 0.20 ‰ -27.03 ± 1.52 ‰ 
Brand 3 -24.52 ± 0.22 ‰ -24.42 ± 0.64 ‰ 
Fresh Lemon Juice -25.44 ± 0.18 ‰ -25.38 ± 0.26 ‰ 
271  
272 3.2.3. Verification of Adulteration Detection Threshold 
273    
274 Brand #2 lemon juice was adulterated with 6% of two commercially available citric acids (Source A 
275 (13C = -12.16‰) and Source B (13C = -24.15‰)) by 0%, 5%, 10%, 20%, 50% and 80% (w/w). The 
276 citrate was precipitated and analyzed by EA-IRMS.  The data is presented in Figure 1. In this study, 
277 lemon juice samples were considered adulterated when the 13C values of citrate were more positive 
278 than -23‰. The detection of adulteration is possible when citric acid derived from a C4 plant is added. 
279 The exact detection threshold is also dependent on the 13C of the original juice. In this particular 
280 example, an adulteration of lemon juice Brand #2 (13C = -26.91 ± 0.07‰) with the addition of a C4 
281 based citric acid (Brand A, 13C = -12.16 ± 0.22‰) would be interpreted as adulterated at 
282 approximately 25% w/w or greater (based on a citric acid 13C value of > -23‰).  Adulteration with a 
283 citric acid source from a C3 source (Source B) is practically undetectable using the given 
284 methodology. 
285  
286  
287  
288  
289 Figure 1: The change in 13C values of precipitated calcium citrate upon addition of solutions 
290 of 6% (w/w) citric acid from Source A (blue) and Source B (green). 
291  
292 3.2.4. Analyte Response Linearity and Sensitivity Determination 
293 The IRMS linearity criteria used within our laboratory, based on manufacturer’s recommendations, 
294 was a slope less than 0.066‰/V 5; this criteria was routinely confirmed over a signal range from 500 
295 mV to 10,000 mV for m/z 44 y 13. The CM-CRDS was demonstrated to have a linear response over a 
296 12CO2 concentration range from 1,000 to 9,000 ppm (which corresponds to 0.25 mg to 2.25 mg of 
297 carbon) 13. 
298  
299 3.2.5. Analytical Working Range 
300 The approximate amount of calcium citrate needed to produce EA-IRMS signals in the range of 1,000 
301 mV to 10,000 mV for m/z 44 is 0.08 mg (at 0% sample dilution) to 10 mg (at 95% sample dilution). 
302 The typical analytical portion of calcium citrate used for EA-IRMS, in this study, was 0.3 mg to 2 mg.  
303 Similarly, for CM-CRDS, the amount of calcium citrate needed to produce signals in the range of 
304 1000 ppm to 9,000 ppm 12CO2 was 0.7 mg to 6 mg 13, which was the typical analytical portion of 
305 calcium citrate used in this study, expanding upon the manufacturer’s recommended linearity range of 
306 2000 to 5000 ppm 
307 3.2.6 Comparability of Accuracy 
308 The comparison of 13C values of calcium citrate isolated from the 69 imported lemon juice samples 
309 is shown in Table 4 and Fig. 2. The results are in a good agreement. The average difference between 
310 the measured 13C values was -0.14‰ with a range of -0.30‰ to 0.13‰.  This represents a general 
311 bias of CM-CRDS values being slightly negative compared to EA-IRMS values. The average bias of -
312 0.14‰ is less than the generally acceptable standard deviation of 0.2‰ for EA-IRMS and 0.3‰ for 
313 CM-CRDS, therefore it was deemed insignificant.  
314  
315 3.2.7 Comparability of Adulteration Classification 
316 For the purpose of this study, lemon juice samples were classified as adulterated when the 13C 
317 values of citrate were greater than -23‰ and not adulterated otherwise. The results were considered 
318 inconclusive when the 2σ range around the mean overlapped the classification threshold of -23‰ (σ 
319 only includes analysis variability among replicates).  The classification (not adulterated, inconclusive, 
320 or adulterated) based on the CM-CRDS results were in good agreement with the classification based 
321 on EA-IRMS results.  Fifty-seven samples were classified as not adulterated, nine samples were 
322 classified as adulterated, and one sample classified as inconclusive by both techniques.  Although the 
323 overall replicate variability has been shown to be smaller for EA-IRMS, the two samples (#36 and 38) 
324 classified as inconclusive based on EA-IRMS results, were classified as not adulterated based on 
325 CM-CRDS results, due a smaller replicate variation (±2σ).   
326 3.2.8 Comparability of Precision 
327 The mean standard deviation for 69 sets of triplicate analysis by CM-CRDS (0.06‰) compared well 
328 with the mean standard deviation (0.07‰) obtained for the same samples by EA-IRMS.  The pooled 
329 standard deviations for samples classified as adulterated (0.05‰) and for samples classified as not 
330 adulterated (0.05‰) by CM-CRDS were similar to the pooled standard deviations obtained by EA-
331 IRMS (0.06‰) and (0.04‰), respectively.  
332 Sixty-five of the sixty-nine samples were analyzed in triplicates, each from three separate isolations of 
333 citrate from lemon juice. Four other samples were analyzed in triplicate, but from single isolation of 
334 citrate from the lemon juice. Standard deviations of the 13C values from both analysis methods were 
335 similar, which demonstrates the reproducibility of the precipitation method. 
336  
337 Figure 2: Comparison of 69 calcium citrate isolates as determined by IRMS and CRDS. 
338  
339  
340  
341  
342  
343  
344  
345  
346  
347 Table 4: 13C for 69 calcium citrate isolates as determined by IRMS and CRDS. 
Sample 
IRMS CRDS No. IRMS CRDS 
Sample  
No. δ13C ± 2σ, ‰ Results1 δ13C ± 2σ, ‰ Results1 δ13C ± 2σ, ‰ Results1 δ13C ± 2σ, ‰ Results1 
1 -26.80±0.04 NA -26.86±0.02 NA 36 -23.18±0.22 I -23.42±0.17 NA 
2 -26.84±0.04 NA -26.88±0.06 NA 37 -23.46±0.02 NA -23.56±0.06 NA 
3 -16.98±0.06 A -17.13±0.06 A 38 -23.28±0.28 I -23.49±0.08 NA 
4 -19.73±0.12 A -20.00±0.16 A 39 -23.70±0.14 NA -23.97±0.30 NA 
25 -27.23±0.02 NA -27.30±0.14 NA 40 -23.31±0.02 NA -23.40±0.04 NA 
26 -24.17±0.14 NA -24.22±0.14 NA 41 -23.49±0.32 NA -23.47±0.14 NA 
7 -12.72±0.12 A -13.00±0.12 A 42 -23.47±0.24 NA -23.57±0.10 NA 
28 -11.00±0.12 A -11.14±0.08 A 43 -23.54±0.26 NA -23.53±0.10 NA 
9 -14.43±0.04 A -14.64±0.12 A 44 -23.64±0.18 NA -23.63±0.10 NA 
10 -24.06±0.02 NA -24.12±0.10 NA 45 -23.27±0.08 NA -23.41±0.10 NA 
211 -24.20±0.24 NA -24.22±0.12 NA 46 -25.71±0.38 NA -25.70±0.08 NA 
12 -25.06±0.04 NA -25.34±0.12 NA 47 -27.02±0.08 NA -27.13±0.20 NA 
13 -23.52±0.06 NA -23.69±0.18 NA 48 -27.07±0.22 NA -27.34±0.20 NA 
14 -23.21±0.18 NA -23.35±0.12 NA 49 -27.10±0.26 NA -27.09±0.04 NA 
15 -23.16±0.22 I -23.28±0.28 I 50 -26.98±0.22 NA -27.16±0.26 NA 
16 -23.31±0.04 NA -23.35±0.12 NA 51 -27.12±0.04 NA -27.24±0.16 NA 
17 -23.41±0.123 NA -23.59±0.14 NA 52 -26.95±0.26 NA -27.01±0.10 NA 
18 -23.61±0.30 NA -23.76±0.22 NA 53 -27.11±0.16 NA -27.22±0.30 NA 
19 -24.43±0.06 NA -24.71±0.18 NA 54 -27.03±0.04 NA -27.07±0.12 NA 
20 -26.82±0.04 NA -27.11±0.26 NA 55 -27.01±0.14 NA -27.10±0.22 NA 
21 -23.37±0.04 NA -23.67±0.08 NA 56 -27.23±0.163 NA -27.43±0.18 NA 
22 -12.36±0.22 A -12.59±0.06 A 57 -27.08±0.02 NA -27.26±0.16 NA 
23 -12.27±0.08 A -12.57±0.08 A 58 -27.07±0.10 NA -27.24±0.12 NA 
24 -14.12±0.08 A -14.19±0.04 A 59 -27.04±0.06 NA -27.21±0.02 NA 
25 -25.12±0.12 NA -25.19±0.20 NA 60 -27.17±0.08 NA -27.04±0.08 NA 
26 -19.96±0.14 A -20.12±0.04 A 61 -27.12±0.04 NA -27.11±0.08 NA 
27 -23.94±0.20 NA -24.12±0.16 NA 62 -26.97±0.04 NA -27.23±0.10 NA 
28 -23.51±0.043 NA -23.57±0.06 NA 63 -27.00±0.10 NA -27.25±0.10 NA 
29 -23.46±0.06 NA -23.54±0.05 NA 64 -27.08±0.06 NA -27.14±0.08 NA 
30 -23.62±0.08 NA -23.67±0.12 NA 65 -26.93±0.20 NA -27.12±0.12 NA 
31 -23.41±0.04 NA -23.68±0.12 NA 66 -26.99±0.06 NA -27.26±0.22 NA 
32 -27.17±0.04 NA -27.29±0.10 NA 67 -26.90±0.04 NA -27.15±0.28 NA 
33 -23.33±0.06 NA -23.62±0.16 NA 68 -26.52±0.08 NA -26.66±0.30 NA 
34 -23.43±0.10 NA -23.45±0.12 NA 69 -27.03±0.44 NA -27.11±0.10 NA 
35 -23.16±0.10 NA -23.45±0.04 NA      
Unless otherwise noted, 2σ is based on analysis of one weighing for each of three individual isolation preparations  
1NA = Not Adulterated, A = Adulterated, I = Inconclusive    2 Triplicate analyses of a single preparation.   3n=2 
348  
349  
350  
351 3.3 Analytical Results from the Round-Robin Study 
352 Reported averages (±2σ, ~95% confidence) provided by eight laboratories for the six citrate samples 
353 analyzed by EA-IRMS and CM-CRDS are presented in Table 5a and 5b, respectively. The overall 
354 average of the reported 13C values for each of the six citrate samples are not statistically 
355 distinguishable when comparing those from each technique. The largest spread between reported 
356 average values was 0.48‰ for EA-IRMS and 0.37‰ for CM-CRDS.  Additionally, the largest 
357 difference between a reported value from a laboratory and the overall average (using the same 
358 technique), was 0.36‰ (sample 23 by laboratory 4 using EA-IRMS), however, this was not a 
359 statistical outlier using the Grubb’s test. For a given sample, the overall average of the reported 
360 citrate 13C values differed by <0.15‰ between CM-CRDS and EA-IRMS, which is within generally 
361 accepted analysis deviations of 0.2‰ and 0.3‰ for EA-IRMS and CM-CRDS, respectively.  
362 Repeatability (within laboratory, r) and reproducibility (among laboratories, R), were estimated using 
363 the AOAC International Interlaboratory Study Workbook for Blind (Unpaired) Replicates 21.  The 
364 average r and R for all six samples were 0.17‰ and 0.30‰, respectively, for IRMS, and 0.18‰ and 
365 0.37‰, respectively, for CRDS. A comparable inter-comparison study was performed by Guillou et. 
366 al. 12, involving seventeen laboratories examining acids isolated from juices and analyzed by EA-
367 IRMS. The resulting r and R values were 0.58‰ and 1.75‰, respectively. These relatively large 
368 values were attributed to the fact that each of the seventeen laboratories isolated the acids prior to 
369 analysis, whereas in this study, the samples were prepared in one laboratory, homogenized, and 
370 distributed for analysis.   Perhaps a more comparable criterion derives from ten of the participants in 
371 the Guillou study that analyzed the reference material NBS 22 (13CVPDB = -29.73 ± 0.09‰) with an 
372 average 13C value of -29.8‰, with repeatability and reproducibility of 0.20‰ and 0.27‰, 
373 respectively, however this only represents the analysis of one sample rather than six in the presented 
374 study.   
375 In our study, the results from each of the participating laboratories had allowed to correctly classify 
376 the previously determined non-adulterated and adulterated samples using both techniques.  Only one 
377 result would have been reported as inconclusive (sample number 15, by laboratory 8 using CM-
378 CRDS), the remainder of the “inconclusive” samples would be classified as not adulterated.                                                          
379  
380 Table 5a: 13C values of Calcium Citrate determined by EA-IRMS. The overall average 
381 represents an unweighted average of the reported averages for each sample. 
EA-IRMS (δ13C ± 2σ (‰ ) )
Samples Adulterated Inconclusive Not Adulterated
Lab # 7 23 15 16 64 69
Lab 1 -12.76 ± 0.18 -12.42 ± 0.08 -23.26 ± 0.06 -23.25 ± 0.08 -26.88 ± 0.08 -26.90 ± 0.04
Lab 2 -12.93 ± 0.04 -12.55 ± 0.08 -23.35 ± 0.02 -23.36 ± 0.02 -27.07 ± 0.08 -27.01 ± 0.08
Lab 3 -12.90 ± 0.06 -12.52 ± 0.06 -23.40 ± 0.10 -23.30 ± 0.04 -27.06 ± 0.04 -27.08 ± 0.16
Lab 4 -12.71 ± 0.16 -12.09 ± 0.12 -23.33 ± 0.06 -23.31 ± 0.06 -26.99 ± 0.02 -26.75 ± 0.16
Lab 5 -12.82 ± 0.08 -12.51 ± 0.06 -23.33 ± 0.02 -23.35 ± 0.14 -27.02 ± 0.44 -26.94 ± 0.14
Lab 6 -12.81 ± 0.44 -12.57 ± 0.30 -23.34 ± 0.08 -23.43 ± 0.32 -26.93 ± 0.32 -26.96 ± 0.18
Lab 7 -12.78 ± 0.04 -12.41 ± 0.06 -23.29 ± 0.04 -23.29 ± 0.04 -26.99 ± 0.04 -26.98 ± 0.04
Lab 8 -12.69 ± 0.12 -12.27 ± 0.08 -23.16 ± 0.22 -23.31 ± 0.04 -27.08 ± 0.06 -27.03 ± 0.44
382 Average -12.80 ± 0.16 -12.42 ± 0.33 -23.31 ± 0.15 -23.33 ± 0.11 -27.00 ± 0.14 -26.96 ± 0.20  
383 Table 5b: 13C values of Calcium Citrate determined by CM-CRDS.  The overall average 
384 represents an unweighted average of the reported averages for each sample. 
CM-CRDS (δ13C ± 2σ (‰ ) )
Samples Adulterated Inconclusive Not Adulterated
Lab # 7 23 15 16 64 69
Lab 1 - - - - - -
Lab 2 - - - - - -
Lab 3 -12.74 ± 0.18 -12.57 ± 0.18 -23.63 ± 0.20 -23.40 ± 0.20 -27.20 ± 0.12 -27.22 ± 0.34
Lab 4 - - - - - -
Lab 5 - - - - - -
Lab 6 - - - - - -
Lab 7 -12.80 ± 0.24 -12.47 ± 0.16 -23.26 ± 0.20 -23.34 ± 0.26 -26.99 ± 0.24 -26.94 ± 0.16
Lab 8 -13.00 ± 0.12 -12.57 ± 0.08 -23.28 ± 0.29 -23.35 ± 0.12 -27.15 ± 0.08 -27.11 ± 0.10
385 Average -12.85 ± 0.27 -12.54 ± 0.12 -23.39 ± 0.42 -23.36 ± 0.06 -27.11 ± 0.22 -27.09 ± 0.28  
386  
387 4. Conclusion 
388 While readying our laboratory to determine lemon juice adulteration via exogenous citric acid addition, 
389 we improved a previously presented citrate isolation procedure to remove possible carbon 
390 contamination.  After extensive testing of this procedure using EA-IRMS analysis, several imported 
391 samples were also analyzed by CM-CRDS.  Both techniques showed excellent agreement in the 
392 determination of 13C values of the calcium citrate precipitated from the 69 lemon juices using the 
393 improved methodology. A round-robin study involving eight laboratories was successful in assessing 
394 the accuracy of the CM-CRDS compared to EA-IRMS. Given that the data produced by CM-CRDS is 
395 statistically indistinguishable (<0.15‰ difference) from EA-IRMS, CM-CRDS could be implemented 
396 as an alternative analysis technique for the determination of adulteration in lemon juice. It should be 
397 noted that the overall standard deviations associated with replicate variability of CM-CRDS were 
398 slightly higher than those of EA-IRMS, which could potentially lead to more inconclusive results 
399 compared to EA-IRMS.  Additionally, more citric acid is needed for analysis via CM-CRDS than for 
400 EA-IRMS.  Although these issues are worth considering, they are minor and do not preclude 
401 CM-CRDS from this application.  Furthermore, this study adds to the growing body of literature that 
402 supports CM-CRDS as a comparable technique for multiple matrices.  
403 References 
404 1. Everstine, K., Spink, J.,Kennedy, S., Journal of Food Protection (2013), 76 (4), 723-735. 
405 2. USDA, (2018).U.S. Department of Agriculture: 2018;  
406 3. FAO, (2003).Food and Agricultural Organizations of the United Nations: Rome, Italy, 2003;  
407 4. Doner, L. W., J. Agric. Food Chem. (1985), 33, 770-772. 
408 5. Carter, J. F.,Barwick, V. J., Good Practice Guide for Isotope Ratio Mass Spectrometry. 2011. 
409 6. Gonzalez, J., Jamin, E., Remaud, G., Martin, Y.-L., Martin, G. G.,Martin, M. L., Journal of 
410 Agricultural and Food Chemistry (1998), 46 (6), 2200-2205. 
411 7. Jamin, E., Martin, F., Santamaria-Fernandez, R.,Lees, M., Journal of Agricultural and Food 
412 Chemistry (2005), 53, 5130-3. 
413 8. Jamin, E., Gonzalez, J., Bengoechfa, I., Kerneur, G., Remaud, G., Naulet, N.,Martin, G. G., 
414 Journal of AOAC International (1998), 81 (3), 604-609. 
415 9. Jamin, E., González, J., Bengoechea, I., Kerneur, G., Remaud, G., Iriondo, C.,Martin, G. G., 
416 Journal of Agricultural and Food Chemistry (1998), 46 (12), 5118-5123. 
417 10. Jamin, E., Gonzalez, J., Remaud, G., Naulet, N.,Martin, G. G., Journal of Agricultural and Food 
418 Chemistry (1997), 45 (10), 3961-3967. 
419 11. Antolovich, M., Li, X.,Robards, K., Journal of Agricultural and Food Chemistry (2001), 49 (5), 
420 2623-2626. 
421 12. Guillou, C., Koziet, J., Rossmann, A.,Martin, G. J., Analytica Chimica Acta (1999), 388 (1), 
422 137-143. 
423 13. Mantha, M., Urban, J. R., Mark, W. A., Chernyshev, A.,Kubachka, K. M., Journal of AOAC 
424 International (2018). 
425 14. Picarro. Application Note AN 023  Picarro’s Combustion Module-CRDS Provides Excellent 
426 Data Using the Approved AOAC Internal Standard Isotope Ratio Analysis (ISCIRA) Method for Honey 
427 (AOAC 998.121). [Online], 2010. www.picarro.com. 
428 15. Picarro. Application Note  AN 022 Combustion Module-CRDS for δ13C analysis of imported 
429 honey and detection of adulteration. [Online], 2010. www.picarro.com. 
430 16. Kivrak, S., Kivrak, I.,Karababa, E., Food Science and Technology (Campinas) (2016). 
431 17. Balslev-Clausen, D., Dahl, T. W., Saad, N.,Rosing, M. T., J. Anal. At. Spectrom. (2013), 28 (4), 
432 516-523. 
433 18. Crosson, E. R., Ricci, K. N., Richman, B. A., Chilese, F. C., Owano, T. G., Provencal, R. A., 
434 Todd, M. W., Glasser, J., Kachanov, A. A., Paldus, B. A., Spence, T. G.,Zare, R. N., Analytical 
435 Chemistry (2002), 74 (9), 2003-2007. 
436 19. AOAC. In Official Methods of Analysis, AOAC INTERNATIONAL: Gaithersburg, MD, 1981; Vol. 
437 Method 981.09. 
438 20. AOAC. In Official Methods of Analysis, AOAC INTERNATIONAL: Gaithersburg, MD, 1982; Vol. 
439 Method 982.21. 
440 21. AOAC INTERNATIONAL (2014) Interlaboratory Study Workbook for Blind (Unpaired) 
441 Replicates version 2.1. 
442