This document describes a new method for rapidly determining nitrate levels in vegetables using gas chromatography-mass spectrometry (GC-MS). Nitrate is extracted from frozen vegetable samples in hot water and converted to volatile nitrate ethyl ester (EtONO2) using simple aqueous chemistry. The EtONO2 derivative is then separated from the vegetable matrix as a gas and sampled in the headspace for analysis by GC-MS. The method allows for clean chromatograms and analysis completion within 1.8 minutes. Isotope dilution with 15N-labeled potassium nitrate as an internal standard provides high-precision quantitation results within 1%. The method provides a simple, rapid and accurate means for nitrate analysis in vegetables

1. Accepted Manuscript
Rapid determination of nitrate in vegetables by gas chromatography mass
spectrometry
Beatrice Campanella, Massimo Onor, Enea Pagliano
PII: S0003-2670(17)30579-2
DOI: 10.1016/j.aca.2017.04.053
Reference: ACA 235212
To appear in: Analytica Chimica Acta
Received Date: 7 February 2017
Revised Date: 18 April 2017
Accepted Date: 22 April 2017
Please cite this article as: B. Campanella, M. Onor, E. Pagliano, Rapid determination of nitrate in
vegetables by gas chromatography mass spectrometry, Analytica Chimica Acta (2017), doi: 10.1016/
j.aca.2017.04.053.
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to
our customers we are providing this early version of the manuscript. The manuscript will undergo
copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please
note that during the production process errors may be discovered which could affect the content, and all
legal disclaimers that apply to the journal pertain.

2. M
ANUSCRIPT
ACCEPTED
ACCEPTED MANUSCRIPT
Rapid determination of nitrate in vegetables by gas chromatography mass
spectrometry
Beatrice Campanellaa
, Massimo Onora
, Enea Paglianob,∗
aConsiglio Nazionale delle Ricerche (CNR), Istituto di Chimica dei Compisti Organometallici, UOS di Pisa, Via Moruzzi 1,
56124 Pisa, Italy
bNational Research Council of Canada, 1200 Montreal Road, K1A0R6 Ottawa, Ontario, Canada
Abstract
In this study we present a novel isotope dilution gas chromatography method for the determination of
nitrate in vegetables. The analyte was extracted in water at 70 ◦
C and mixed with isotopically enriched
15
NO−
3 internal standard. The sample was centrifuged and the supernatant reacted with sulfamic acid for
removal of nitrite, and with triethyloxonium tetrafluoroborate for converting NO−
3 into volatile EtONO2.
This simple aqueous chemistry allowed for separation of analyte from sample matrix in the form of a gaseous
derivative which could be sampled in the headspace before GC–MS analysis. This key-feature of the method
made possible the collection of clean chromatograms within an elution time of only 1.8 min. Detection of
EtONO2 could be performed using electron impact ionization with a standard GC–MS setup. The method
was optimized and validated for the analysis of nitrate in fresh vegetables in the 10-10,000 μg/g range with
a detection limit of 2 μg/g. Due to the use of primary isotope dilution quantitation, traceable results of
high-precision were attained.
Keywords: Nitrate, Vegetable, Isotope Dilution, Gas chromatography Mass Spectrometry
1. Introduction1
Inorganic nitrates are ubiquitous in the environment and can occur in foodstuff as additives (E252) or2
contaminants. The major contribution of nitrate to human diet is due to vegetables, where NO−
3 can easily3
reach the part-per-thousand level [1].4
Despite nitrate being relatively non-toxic for humans, attention should be given to nitrite which is its main5
metabolic byproduct [1]. In this regard, NO−
3 is reduced to NO−
2 by enzymatic reactions occurring in saliva,6
and the resulting nitrite may react with secondary and tertiary amino compounds to form highly carcinogenic7
N -nitroso compounds [2]. Furthermore, NO−
2 inactivates hemoglobin by converting the oxyHb(Fe2+
) into8
metHg(Fe3+
). This last effect can have severe consequences for infants and may lead to a deadly condition9
∗Corresponding author
Email address: <enea.pagliano@nrc-cnrc.gc.ca> and <enea.pagliano@outlook.com> (Enea Pagliano)
Preprint submitted to Analytica Chimica Acta May 15, 2017

3. M
ANUSCRIPT
ACCEPTED
ACCEPTED MANUSCRIPT
known as methemoglobinemia [3]. Fatal nitrate intoxication has also been reported in cattle fed with10
contaminated fennels [4].11
For these reasons, an Acceptable Daily Intake (ADI) of 3.7 mg/kg NO−
3 b.w./day was established by the12
European Food Safety Authority (EFSA) and endorsed by the Joint FAO/WHO Expert Committee on Food13
Additives (JECFA) in early 2000 [5]. Furthermore, nitrate levels in spinach and lettuce have been regulated14
in the EU with the Commission Regulation (EC) No 1881/2006. The ESFA reported that the overall human15
exposure to nitrate due to vegetables is unlikely a health risk, but 2.5% of the population of some Member16
States may exceed the ADI and a portion of only 47 g of arugula with median level of nitrate, may already17
exceed the ADI threshold [5].18
Most of the studies and regulation written on this theme consider only the risks associated with nitrate19
intake and not its potential benefits. In this regard, recent investigations have shown that dietary nitrates20
are part of the so called nitrate-nitrite-NO pathway and can be beneficial for the health of cardiovascular21
system [6, 7, 8, 9]. In particular, Weitzberg and Lundberg have invited researchers and health authorities22
to consider if the concerns over dietary nitrate are indeed leading to health benefits or if they are limiting23
the consumption of an essential nutrient [8].24
Assessment of pros and cons related to dietary nitrate will need further investigations and availability of25
rapid and precise methods for measuring nitrate could play an important role. Several methodologies for26
nitrate in foodstuff have already been described in literature [10]. The spectrophotometric determination of27
nitrate was achieved by reduction of the analyte to nitrite followed by derivatization with sulphanilamide28
and N -(1-naphthyl)-ethylenediamine to a pink azo-dye detectable at 540 nm (Griess method). NO−
3 to NO−
229
reduction has been attained both enzymatically [11] and with metallic powders based on Zn [12, 13] and30
Cd [14, 15] in batch or flow-injection. Colorimetric methods based on Griess chemistry found criticism due31
to poor nitrate recovery with bias higher than 60% [16]. Electrochemical methods have also been proposed32
for the direct detection of nitrate in vegetables [17, 18, 19]. Although such methods are rapid, interference33
due to matrix components may be encountered. The most popular approaches for NO−
3 determination in34
vegetables are the ones based on high pressure liquid chromatography (HPLC) [20, 21], ion chromatography35
(IC) [22, 23, 24] and capillary electrophoresis [25, 26] with non-specific conductivity or UV–vis detectors.36
These methods are accurate, but can be time consuming and they need the vegetable sample extracts to37
be pre-cleaned by filtration or solid phase extraction (SPE) for avoiding contamination and deterioration of38
the analytical column.39
In this study a rapid headspace gas chromatography mass spectrometry (GC–MS) method for the deter-40
mination of nitrate in vegetables is presented. The novel method could be applied directly on vegetables41
extracts without the need for filtration or SPE. The use of MS allowed for high-precision isotope dilution42
quantitation avoiding shortcomings associated with non-specific detectors which are often used in combi-43
nation with liquid chromatography. Furthermore, headspace sampling allowed for clean chromatography44
2

4. M
ANUSCRIPT
ACCEPTED
ACCEPTED MANUSCRIPT
with no column degradation. NO−
3 was extracted from the frozen matrix in hot-water at 70 ◦
C [16] and45
converted into volatile EtONO2 using simple aqueous chemistry at room temperature. EtONO2 was readily46
separated from the matrix under gaseous form and could be sampled, without further purification, in the47
vial headspace [27]. On a DB-5.625 column, the GC–MS analysis was complete in only 1.8 min, reagents48
cost per analysis were better then 0.5 USD, and with the use of 15
NO−
3 internal standard results within 1%49
precision could be obtained.50
2. Materials and methods51
2.1. Reagents and standards52
Potassium nitrate 15
N-labeled was purchased from Cambridge Isotope Laboratories (K15
NO3, 99% of 15
N53
enrichment) and used as internal standard. The Fluka Nitrate Standard for IC (TraceCERT R
, 1004 ± 254
μg/g of nitrate in water) was used as primary standard of natural isotopic composition. Triethyloxonium55
tetrafluoroborate (chemical purity ≥ 97%), sulfamic acid (chemical purity ≥ 97%), acetone and acetonitrile56
of HPLC grade were purchased from Sigma-Aldrich. All preparations and dilutions of samples and standards57
were performed gravimetrically using ultrapure water (18.2 MΩ·cm at 25 ◦
C).58
2.2. Vegetable samples59
Samples of fresh raw lettuce (Lactuca sativa), spinach (Spinacia oleracea), carrot (Daucus carota subsp.60
Sativus), kale (Brassica oleracea var. Sabellica), arugula (Eruca vesicaria) and endive (Cichorium endivia)61
were purchased in a local Italian supermarket in September 2016. Such vegetables were selected because62
they are the main source of nitrate in human diet [1]. The samples were carried to the laboratory and their63
outside leafs, stems and stalks were removed. The samples were then rinsed in water, drained on tissue64
paper, transfered into PE bags and kept frozen at −18 ◦
C until analysis. Such preliminary operation were65
carried out within the day of purchase. Sample homogenization were performed on the frozen materials using66
a blade food processor (model Tritty-76006; Termozeta, Bareggio, MI, Italy) in action for four intervals of67
15 s. Within each griding, the sample was stirred with a spoon, scraping the material on the walls of the68
container. In some of the experiments for method validation, a freeze-dried sample of spinach was analyzed.69
A 300 g amount of fresh pre-washed spinach was purchase in a local supermarket in Canada (October 2016),70
frozen at −80 ◦
C and placed in a Thermo Scientific 5L ModulyoD Freeze Dryer for 48 hours. The material71
was then crushed in a plastic ball mill and the resultant power was frozed at −80 ◦
C following a second72
freeze dry cycle for 48 hours. The residual 30 g of spinach powder was kept at −20 ◦
C.73
3

5. M
ANUSCRIPT
ACCEPTED
ACCEPTED MANUSCRIPT
2.3. Gas chromatography mass spectrometry method74
Nitrate was measured by GC–MS after derivatization to volatile nitrate ethyl ester (EtONO2). Some of75
the measurements were performed on an Agilent 6850 gas chromatograph equipped with an Agilent 5975c76
mass spectrometer detector and a CTC CombiPAL autosampler. The sample was prepared in a 10 mL77
headspace vial (Agilent, P/N 8010-0139) and incubated at 60 ◦
C for 5 min following 500 μL of headspace78
sampling (syringe hold at 95 ◦
C and flushed with He for 5 min to avoid carry over effects). The inlet was79
held at 150 ◦
C and the headspace was injected in 8:1 split with a helium flow rate of 1.1 mL/min. The80
oven was held at 85 ◦
C for 6.5 min and the elution was carried out on a high-polarity column (DB-FFAP81
nitroterephthalic acid modified polyethylene glycol column: 60 m length × 0.250 mm ID × 0.50 μm film).82
The transfer line was set at 230 ◦
C. The measurements of nitrate in vegetables were also performed on a83
Hewlett-Packard 5973 GC–MS. In this case, the sample was prepared in 4 mL vials with a screw cap with84
PTFE/silicone septum and the sampling of 500 μL of headspace was manually performed with a gas tight85
syringe at room temperature. The inlet was held at 100 ◦
C and the injection was performed in 8:1 pulse86
split mode (30 psi for 2 min) with a helium flow rate of 1 mL/min. Isothermal elution was performed at87
30 ◦
C for 1.8 min on a non-polar column (DB-5.625 (5%-phenyl)-methylpolysiloxane column: 30 m length88
× 0.250 mm ID × 0.25 μm film). The transfer line was set at 230 ◦
C.89
EtONO2 detection was attained in positive EI at 70 eV using the standard autotune procedure for mass90
calibration. Acquisition was performed in Total Ion Chromatography (TIC) for identification and in Selected91
Ion Monitoring (SIM) for quantitation purpose monitoring m/z signals at 46 and 47 Da with a dwell time92
of 50 ms for each signal.93
2.4. Ion chromatography UV–vis method94
A Dionex DX-500 ion chromatograph equipped with a GP40 gradient pump and a Rheodyne 7125 injector95
(25 μL sample loop) was employed. The chromatographic separation was carried out in 25 min with a flow96
rate of 1 mL/min of an isocratic carbonate mobile phase prepared to reach a concentration of 2.3 mmol/L97
Na2CO3 and 0.8 mmol/L NaHCO3. A Dionex IonPac AS12 column (4 × 250 mm, 9 μm particle size) with98
a Dionex IonPac AG12 guard column (2 × 50 mm, 13 μm particle diameter) were employed for separation.99
Ion suppression was achieved using a Dionex ASRS 300 (4 mm) self-regenerating suppressor in recycle mode.100
Detection of nitrate was obtained with a AD-20 variable wavelength UV-vis detector at 205 nm. A Dionex101
PeakNet chromatography data system (version 4.5) was used for data acquisition and processing. For the102
calibration of the instrumental response, standard solutions of nitrate were prepared in water and used for103
the construction of the calibration plot (1, 2.5, 5, 7.5, and 10 μg/g NO−
3 ). The nitrate was extracted from the104
sample in water at 70 ◦
C for 15 min with the same procedure proposed for the GC–MS method (Paragraph105
3.3). Before injection, the extracts were filtered through a 0.45 μm Agilent Captiva premium syringe filter106
following SPE on a Dionex On-Guard II RP. The SPE cartidge was conditioned with 5 mL of methanol and107
4

6. M
ANUSCRIPT
ACCEPTED
ACCEPTED MANUSCRIPT
10 mL of water. A 2 mL volume of the sample extract was loaded on the column discarding the first eluate;108
an additional 2 mL were passed through, collecting the eluate in a plastic tube. The SPE cartridge did not109
retain nitrate which was fully recovered in the eluate fraction. The eluate was finally diluted to fit in the110
linear range of the method (0 – 10 μg/g NO−
3 , R2
= 0.9987) and injected in the ion chromatography system.111
2.5. Safety considerations112
Triethyloxonium tetrafluoroborate is a crystalline salt able to perform ethylations in both aqueous and113
non-aquous medium [28, 29]. Et3O+
[BF4]−
should be handled in a vented fumehood with adequate PPEs114
and stored in a freezer at −20 ◦
C. The potential risk handling Et3O+
[BF4]−
is softened by its non-volatile115
nature. In an aqueous medium Et3O+
[BF4]−
undergoes complete acid hydrolysis within 80 min at 18 ◦
C116
[30]:117
Et3O+
+ H2O Et2O + EtOH + H+
118
In order to avoid exposure to volatile byproducts, the vials hosting the derivatization should be prepared119
and cleaned in a vented fumehood.120
3. Results and discussion121
3.1. Derivatization and reagent preparation122
Triethyloxonium tetrafluoroborate salt has recently been used as derivatizing agent for the determination of123
several inorganic anions – including nitrate [31, 32] – by gas chromatography mass spectrometry [33, 34]. The124
ability of this reagent to perform ethylation directly in aqueous sample solution [28] is a great advantage125
that Et3O+
chemistry owns over other classic alkylation methods commonly employed in GC–MS [35].126
Et3O+
[BF4]−
converts NO−
3 to the corresponding ethyl ester accordingly to the following reaction scheme:127
Et3O+
(aq) + NO−
3 (aq) EtONO2(g) + Et2O(g)128
The EtONO2 derivative is volatile and can be sample from the headspace of the vial hosting the derivati-129
zation. In other words, the derivatization is responsible for a first gross separation of the analyte from the130
matrix which is not injected into the gas chromatograph. The headspace injection results in baseline clean131
chromatography and prevents the deterioration of the GC column. In our laboratory the same column has132
been in use for over five years without any changes in its performance.133
Et3O+
[BF4]−
is commercially available as a solid salt. In previous work [32] the preparation of the reagent134
was attained just before derivatization by dissolving Et3O+
[BF4]−
in water following its prompt addition135
to the samples. This modus operandi did not permit storage of the reagent solution, therefore the leftovers136
after each derivatization were disposed of.137
5

7. M
ANUSCRIPT
ACCEPTED
ACCEPTED MANUSCRIPT
In this study, an alternative method for preparing the reagent solution was implemented. The solubil-138
ity/stability of Et3O+
[BF4]−
was tested in acetonitrile and acetone. Reagent solutions were prepared by139
dissolving 1 g of Et3O+
[BF4]−
in 1 mL of organic solvent kept at −20 ◦
C. No difference in the alkylation140
power were observed when Et3O+
[BF4]−
was prepared in water or in these two organic solvents. Such solu-141
tions were stored at −20 ◦
C overnight. Even at such low temperature a reaction occurred between Et3O+
142
and acetone: the medium color turned orange showing possible polymerization reactions. On the other hand,143
the acetonitrile solution appeared unchanged. Such solution was then stored at −20 ◦
C and tested along a144
period of two months without noticing variations in its alkylation power (see Supporting Information). In145
this study, acetonitrile was selected for the preparation of Et3O+
[BF4]−
solution for derivatization.146
Table 1: Nitrate analytical response (in terms of isotope ratios, rA*B) for the calibration blends measured
at different times by GC–MS
NO−
3
a
15-Sep 16-Sep 26-Sep 28-Sep 3-Oct 6-Oct 14-Oct 25-Oct Average
0.00 0.0117 0.0123 0.0115 0.0127 0.0148 0.0137 0.0147 0.0159 0.0134 ± 0.0016 (12%)
1.03 0.0787 0.0802 0.0786 0.0803 0.0797 0.0826 0.0803 0.0825 0.0804 ± 0.0015 (1.9%)
10.1 0.6768 0.6711 0.6752 0.6753 0.6762 0.6781 0.6806 0.6773 0.6763 ± 0.0027 (0.4%)
25.2 1.6762 1.6694 1.6812 1.6722 1.6811 1.6695 1.6717 1.6734 1.6743 ± 0.0047 (0.3%)
50.2 3.2804 3.3011 3.2694 3.2543 3.2791 3.2817 3.2728 3.2712 3.2762 ± 0.0133 (0.4%)
73.7 4.7671 4.7923 4.7903 4.7989 4.7849 4.7841 4.8081 4.7911 4.7896 ± 0.0119 (0.2%)
98.4 6.3307 6.3372 6.3556 6.3229 6.3855 6.3502 6.3330 6.3608 6.3470 ± 0.0203 (0.3%)
a
Six calibration blends and a blank were prepared by mixing varying amounts of primary standard
(NO−
3 of natural isotopic composition) with a set amount of 15
NO−
3 internal standard (15 μg/g for
all blends). In the first column the mass fraction of NO−
3 standard of natural isotopic composition is
reported in μg/g.
3.2. Preparation of isotope dilution calibration plot147
Isotope dilution (ID) was employed for quantitation with a fully 15
N-enriched nitrate as internal standard.148
ID is regarded as a high-end method for generating precise analytical data traceable to SI units [36]. The149
use of such nearly perfect internal standardization compensates for analyte losses during sample preparation150
and matrix effects [37]. Several mathematical approaches have been proposed for the treatment of isotope151
dilution data [38] and in this study a calibration plot with gravimetric preparation was adopted [39]. Six152
calibration blends and a blank were prepared by mixing varying amounts of primary standard (0.1 - 1.0 - 2.5153
- 5.0 - 7.5 - 10 mL of 1004 μg/g NO−
3 of natural isotopic composition) with a set amount of internal standard154
(1.5 mL of 991 μg/g NO−
3
15
N-enriched) and diluting the mixtures with water to 100 mL. Such solutions155
were employed across the experiments to verify the instrumental response. A 2 mL aliquot of each solution156
was transfered to a vial with 50 μL of Et3O+
[BF4]−
acetonitrile solution prepared as described in section157
6

8. M
ANUSCRIPT
ACCEPTED
ACCEPTED MANUSCRIPT
Figure 1: Calibration plot for the determination of nitrate.
3.1. The reaction was held at room temperature for 30 min following GC–MS analysis (section 2.3). The158
mass spectrometer was set to monitor m/z value of 46 Da and 47 Da. The first signal corresponds to the159
ion 14
NO+
2 mostly abundant in the analyte of natural isotopic composition, whereas the second corresponds160
to the ion 15
NO+
2 mostly abundant in the internal standard [40]. The gravimetric calibration curve was161
prepared by plotting the quantity wA* · mA*/mB (wA* is the mass fraction of the nitrate primary standard,162
in this case 1004 μg/g, while mA*/mB is the ratio between the mass of primary standard (mA*) and internal163
standard (mB) in each calibration blend) versus rA*B (isotope ratio obtained by dividing the peak area of164
the GC–MS signal at 46 Da by the one at 47 Da). A complete description of the quantitation method165
is reported in the Supplemental Information and the teoretical aspects can be found elsewhere [39]. The166
fitting of the calibration function was preformed with the LINEST function of Microsoft Excel using both167
a linear curve and a Pad´e[1,1] rational function [39]. The second fitting strategy allows accounting for any168
deviations from linearity that are always present in an isotope dilution plot [41]. However, in this case such169
deviations were very modest and a linear fit was satisfactory for the purpose of this application.170
The calibration blends were remeasured for a period of over one month. As reported in Table 1, the nitrate171
analytical response (in terms of isotope ratios, rA*B) was very consistent. In this regard, verification of the172
calibration plot (Fig. 1) could be done on a monthly base or every time the MS detector was tuned.173
3.3. Sample preparation and derivatization174
The freezing of the sample at −18 ◦
C before homogenization is essential for this method because it assures175
cell-breakage and complete recovery of incurred nitrate [16]. Frozen sample were homogenized using a blade176
food processor following extraction of the analyte in water at 70 ◦
C, as recommended by several authors177
7

9. M
ANUSCRIPT
ACCEPTED
ACCEPTED MANUSCRIPT
Figure 2: Sample preparation for the determination of nitrate in vegetables. The first route (Prep. A)
describes an ideal isotope dilution experiment where the internal standard (IS) is mixed with the sample
at first. The second route (Prep. B) describes the sample preparation employed within the study. The
analytical results produced by the two methods are identical within the limit of the experimental error, but
Prep. B employs much less 15
NO−
3 internal standard resulting in a more economical approach.
[10, 16, 42, 24] and official methods [43]. GC–MS sample preparation is outlined in Fig. 2. In a formal178
isotope dilution method, the internal standard should be added to the sample at first (Fig. 2, Prep. A) [37].179
In this case, however, the same quantitative results were obtained when the 15
NO−
3 internal standard was180
added after the aqueous extraction (Fig. 2, Prep. B). This second route was preferred because it entailed a181
considerable saving on the amount of 15
NO−
3 used in each analysis, whereas the traditional isotope dilution182
approach could be considered for method validation.183
A 2 g aliquot of frozen homogenized sample (or 0.2 g of freeze dried material) was transfered to a glass vial184
and 20 mL of water were added. The solution was vortexed and maintained for 15 min in a heating block at185
70 ◦
C. After cooling to room temperature, such mixture was shaken and 2.5 mL of the sludge were transfered186
in a 15 mL centrifuge tube. At this point a certain aliquot of the same 15
NO−
3 internal standard solution187
employed for the preparation of the calibration plot was added. The amount of 15
NO−
3 in the 2.5 mL of188
sludge should be in the range of 15 to 150 μg/g. If the nitrate content in the fresh vegetable was expected189
to be 10–1000 μg/g, the sludge was spiked with 15 μg/g of 15
NO−
3 . If the nitrate content was expected to190
be higher, then the sludge should be treated with 150 μg/g 15
NO−
3 . In this way the response of nitrate in191
the sample (isotope ratio, rAB), was in the 0.1 – 6 interval and could be interpolated with the calibration192
8

10. M
ANUSCRIPT
ACCEPTED
ACCEPTED MANUSCRIPT
plot shown in Fig. 1. After addition of the internal standard, the sample blend was vortexed for 5 min and193
centrifuged at 5000 rpm for 5 min. A 2 mL aliquot of the supernatant was transfered to a headspace vial,194
treated with 100 μL 1% sulfamic acid and 50 μL of Et3O+
[BF4]−
acetonitrile solution (section 3.1). The195
sulfamic acid was used to eliminate nitrite which could give positive interference [32]. The reaction was held196
at room temperature for 30 min following GC–MS analysis (section 2.3). The resulting isotope ratio rAB197
(obtained dividing the peak area of the GC–MS signal at 46 Da by the one at 47 Da) was correlated to mass198
fraction of nitrate in the fresh vegetable matrix (wA) through the following formula:199
wA =
rAB − a0
a1
·
mB
mA
(1)
where a0 and a1 are the intercept and slope of the calibration plot showed in Fig. 1 and mB/mA is the ratio200
between the amount of internal standard solution and sample. In this case (Fig. 3.3, Prep. B) mB/mA201
could be calculated with a model reported in Supplemental Information.202
3.4. Figures of merit and benefits203
The nitrate ethyl derivative (EtONO2) was eluted with both a DB-FFAP high-polarity column and a DB-204
5.625. In the first case EtONO2 eluted at 5.85 min on the isotherm at 85 ◦
C, whereas in the second case205
after 1.28 min on the isotherm at 30 ◦
C. Since the derivatization entailed separation of the analyte from206
vegetable matrix to gas phase, the chromatography was clean and free from interferences. Furthermore, the207
method was very rapid (1.8 min of chromatography on DB-5.625) allowing GC–MS analysis of 18 samples208
in less then 45 min with a common instrumental setup; in this regard, the GC–MS method was much faster209
than a typical ion chromatography run (25 min in our case). In Fig. 3 the overlaid GC–MS chromatograms210
of 18 vegetable samples are reported: despite the short program, the baseline is flat and stable without other211
peaks beside the analytical one.212
The method has been optimized for measurement of nitrate in vegetable where detection power is not213
a concern given the high levels normally present in such matrices. As reported in Fig. 4, the method214
could measure a standard solution of 1 μg/g NO−
3 with a signal-to-noise ratio of 15, resulting in a method215
detection limit of 0.2 μg/g. The chromatographic baseline noise was calculated using both peak-to-peak216
and RMS×3 algorithm without significant difference in the signal-to-noise estimation. Since the method217
implies a 1:10 dilution of the sample with water, the detection limit of nitrate in fresh vegetable matrices218
is 2 μg/g, well below the regulatory limits. This figure of merit was obtained with an Hewlett-Packard219
5973 GC–MS in electron ionization mode. Using a more recent Agilent 5975c with an automated headspace220
incubator/injection system, the instrumental detection limit was better by an order of magnitude. For221
applications that require more sensitivity, negative chemical ionization can provide detection limits in the222
low part-per-billion range [32]. The novel GC–MS method could quantify NO−
3 in vegetable in the 10–10,000223
μg/g range with uncertainty better than 1% (see Supplemental Information and Table 3). Finally, also224
9

11. M
ANUSCRIPT
ACCEPTED
ACCEPTED MANUSCRIPT
Figure 3: Determination of nitrate in spinach leaf: 18 samples were analyzed sequentially within 45 min and
the SIM chromatograms collected on m/z = 46 Da are overlaid.
the method economics is favorable: for single analysis the costs for both 15
NO−
3 internal standard and225
Et3O+
[BF4]–
reagent does not exceed 0.5 USD considering current commercial prices for such chemicals.226
3.5. Validation227
The implementation of isotope dilution for routine measurements has significantly simplified analytical228
chemistry work contributing to traceable results of high-precision [44]. When the enriched internal standard229
is added to the sample at first and equilibrated with the incurred analyte [37], all analyte losses that can230
occur during analysis are compensated for. Furthermore, fluctuations of the experimental conditions and231
instrumental drifts, that could result in variation of the absolute analytical signal, are corrected by the232
10

12. M
ANUSCRIPT
ACCEPTED
ACCEPTED MANUSCRIPT
Figure 4: Three replicate measurements of a 1 μg/g standard nitrate solution and a reagent blank. Data
acquired on DB 5.625 column with the HP 5973 GC–MS system
Table 2: Time variation of 14
NO−
3 /15
NO−
3 ratio in
a blend of lettuce and internal standard
Time (min) 14
NO−
3 /15
NO−
3
0 1.236
5 1.242
10 1.246
20 1.249
60 1.243
120 1.246
1440 1.253
2880 1.248
internal standard normalization. To further confirm the consistency of isotope dilution for this application,233
a sample of lettuce was prepared accordingly to the procedure described in Fig. 2 (Prep A) and monitored234
over 48 hours. No variations of nitrate response were observed in this time frame (rAB = 1.245 ± 0.005,235
Table 2). The data shown in Table 2 also demonstrated that equilibration of sample and internal standard236
was instantaneous. Due to the lack of Certified Reference Material for nitrate in vegetable [45], validation237
was performed completing spike recovery tests and comparing the performance of novel method with an238
orthogonal approach based on ion chromatography.239
11

13. M
ANUSCRIPT
ACCEPTED
ACCEPTED MANUSCRIPT
Table 3: Determination of nitrate in vegetables (μg/g): comparison between GC–MS
and ion chromatography with UV–vis detection
Sample GC–MS IC–UV
Arugula 4823.8 ± 3.7 (uR = 0.1%) 4850 ± 174 (uR = 3.6%)
Lettuce 673.2 ± 1.6 (uR = 0.2%) 658 ± 53 (uR = 8.1%)
Spinach 633.4 ± 1.2 (uR = 0.2%) 620 ± 16 (uR = 2.6%)
Endive 166.16 ± 0.28 (uR = 0.2%) 163.8 ± 2.3 (uR = 1.4%)
Kale 73.64 ± 0.52 (uR = 0.7%) 73.0 ± 2.2 (uR = 3.0%)
Carrot 15.80 ± 0.18 (uR = 1.2%) 15.27 ± 0.13 (uR = 0.82%)
Table 4: Spike recovery test of nitrate in lettuce aqueous extracts.
Sample NO−
3 added NO−
3 by GC–MS NO−
3 by IC–UV
I.D. μg/g Total (μg/g) Rec. (%) Total (μg/g) Rec. (%)
Unspiked n/a 25.28 n/a 24.2 n/a
+ 2% 0.486 25.76 99% n/d n/d
+ 20% 5.044 30.43 102% 29.3 101%
+ 200% 51.52 78.05 102% 77.0 103%
3.5.1. Nitrate in vegetables: GC–MS vs IC240
The GC–MS method was applied for the determination of nitrate in several vegetables matrices. The241
experiments were performed in parallel with ion chromatography (IC–UV) and the results are reported in242
Table 3. Each vegetable matrix was extracted once and the aqueous extracts were divided to six aliquots:243
three were measured by GC–MS and three with IC–UV. The measurement precision was estimated on the244
basis of two standard deviations on the three replicate measurements. Agreement between IC–UV and245
GC–MS was excellent and the latter method provided higher precision.246
A spike recovery test on an aqueous lettuce extract was also performed. For this purpose, 4 g of lettuce247
sample were extracted with 40 mL of water as reported in Section 3.3. The centrifuged aqueous layer was248
then divided into eight portions. An unspiked aliquot was analyzed by GC–MS (Fig. 3.3, Prep. B) and249
IC–UV (Section 2.4). In this way we could quantify the amount of nitrate in the extract. The other six250
aliquots were spiked with a known amount of nitrate standard solution and analyzed in parallel. The test251
results are presented in Table 4. When the lettuce extracts were spiked with amounts of nitrate equal to252
20% and 200% of the endogenous NO−
3 level, both GC–MS and IC–UV provided quantitative recoveries.253
When the extract was spiked with 0.49 μg/g NO−
3 (equal to 2% of the endogenous NO−
3 ) only the GC–MS254
12

14. M
ANUSCRIPT
ACCEPTED
ACCEPTED MANUSCRIPT
Table 5: Spike recovery test of nitrate in a sample of freeze-dried
spinach. The results are reported on dried sample
Sample NO−
3 added NO−
3 by GC–MS
I.D. μg/g Total (μg/g) Rec. (%)
Unspiked-a n/a 14543 n/a
Unspiked-b n/a 14554 n/a
Unspiked-c n/a 14392 n/a
+ 10%-a 1388 15869 99%
+ 10%-b 1082 15569 99%
+ 10%-c 1285 15839 105%
+ 20%-a 2425 16778 94%
+ 20%-b 2912 17336 98%
+ 20%-c 1973 16431 98%
+ 50%-a 5738 20422 103%
+ 50%-b 6947 21744 104%
+ 50%-c 5447 20049 102%
could provide a quantitative recovery.255
3.5.2. Spike recovery test on a freeze-dried spinach sample256
A second spiked recovery test was performed on a spinach matrix. The sample was first freeze-dried as257
reported in Section 2.2 and 0.2 g of spinach powder were placed into each of twelve vials. The samples258
were then divided into four groups of three. A 1.5 mL volume of water was added to the first group named259
“Unspiked”. In the other three groups a 1.5 mL volume of aqueous nitrate was added to match the 10%,260
20%, and 50% level of endogenous NO−
3 . The aqueous sludge was frozen at −80 ◦
C and freeze-dried on a261
Thermo Scientific 5L ModulyoD Freeze Dryer for 48 hours. During the freeze-drying process, the spiked262
nitrate was more likely bound to the spinach matrix resulting in a better mimicking of endogenous nitrate.263
The samples were then analyzed by GC–MS (Fig. 2, Prep. B) and the results are reported in Table 5. In264
all cases the recovery of added nitrate was quantitative.265
4. Conclusion266
In this study we present the first isotope dilution GC–MS method for the determination of nitrate in267
vegetables. A 15
NO−
3 was employed for quantitation purposes allowing precision better than 1% on the268
13

15. M
ANUSCRIPT
ACCEPTED
ACCEPTED MANUSCRIPT
measurement results. The methods was optimized for quantitation of nitrate in fresh vegetables in the 10-269
10,000 μg/g range with a detection limit of 2 μg/g. This method was validated comparing its performance270
with ion chromatography and by several spike recovery tests. The method is overall very simple and271
entails the use of commercially available triethyloxonium tetrafluoroborate for conversion of NO−
3 to volatile272
EtONO2 in an aqueous environment at room temperature. The gaseous derivative is sampled from the273
headspace allowing for a clean and rapid chromatography (1.8 min on DB 5.625). The method is traceable274
to SI units and it could become a reference for the measurement of nitrate in vegetables.275
14

16. M
ANUSCRIPT
ACCEPTED
ACCEPTED MANUSCRIPT
References276
[1] P. Santamaria, Nitrate in vegetables: toxicity, content, intake and EC regulation, Journal of the Science277
of Food and Agriculture 86 (2006) 10–17. doi:10.1002/jsfa.2351.278
[2] W. Lijinsky, N -Nitroso compounds in the diet, Mutation Research/Genetic Toxicology and Environ-279
mental Mutagenesis 443 (1999) 129–138. doi:10.1016/s1383-5742(99)00015-0.280
[3] L. Knobeloch, B. Salna, A. Hogan, J. Postle, H. Anderson, Blue babies and nitrate-contaminated well281
water, Environmental Health Perspectives 108 (2000) 675–678. doi:10.1289/ehp.00108675.282
[4] A. Costagliola, F. Roperto, D. Benedetto, A. Anastasio, R. Marrone, A. Perillo, V. Russo, S. Papparella,283
O. Paciello, Outbreak of fatal nitrate toxicosis associated with consumption of fennels (Foeniculum284
vulgare) in cattle farmed in campania region (southern Italy), Environmental Science and Pollution285
Research 21 (2014) 6252–6257. doi:10.1007/s11356-014-2520-9.286
[5] J. Alexander, D. Benford, A. Cockburn, J.-P. Cravedi, E. Dogliotti, A. Di Domenico, M. L. Fern´andez-287
Cruz, J. Fink-Gremmels, P. F¨urst, C. Galli, P. Grandjean, J. Gzyl, G. Heinemeyer, N. Johansson,288
A. Mutti, J. Schlatter, R. van Leeuwen, C. V. Peteghem, P. Verger, Nitrate in vegetables - scientific289
opinion of the panel on contaminants in the food chain, The EFSA Journal 689 (2008) 1–79. doi:290
10.2903/j.efsa.2008.689.291
[6] N. G. Hord, Dietary nitrates, nitrites, and cardiovascular disease, Current Atherosclerosis Reports 13292
(2011) 484–492. doi:10.1007/s11883-011-0209-9.293
[7] S. Lidder, A. J. Webb, Vascular effects of dietary nitrate (as found in green leafy vegetables and294
beetroot) via the nitrate-nitrite-nitric oxide pathway, British Journal of Clinical Pharmacology 75295
(2012) 677–696. doi:10.1111/j.1365-2125.2012.04420.x.296
[8] E. Weitzberg, J. O. Lundberg, Novel aspects of dietary nitrate and human health, Annual Review of297
Nutrition 33 (2013) 129–159. doi:10.1146/annurev-nutr-071812-161159.298
[9] G. Pironti, N. Ivarsson, J. Yang, A. B. Farinotti, W. Jonsson, S.-J. Zhang, D. Bas, C. I. Svensson,299
H. Westerblad, E. Weitzberg, J. O. Lundberg, J. Pernow, J. Lanner, D. C. Andersson, Dietary nitrate300
improves cardiac contractility via enhanced cellular Ca2+
signaling 111 (2016) 34. doi:10.1007/301
s00395-016-0551-8.302
[10] C. D. Usher, G. M. Telling, Analysis of nitrate and nitrite in foodstuffs: A critical review, Journal of303
the Science of Food and Agriculture 2 (1975) 1793–1805. doi:10.1002/jsfa.2740261122.304
[11] R. H. Lowe, J. L. Hamilton, Rapid method for determination of nitrate in plant and soil extracts,305
Journal of Agricultural and Food Chemistry 15 (1967) 359–361. doi:10.1021/jf60150a002.306
15

17. M
ANUSCRIPT
ACCEPTED
ACCEPTED MANUSCRIPT
[12] L. Merino, Development and validation of a method for determination of residual nitrite/nitrate in307
foodstuffs and water after zinc reduction, Food Analytical Methods 2 (2009) 212–220. doi:10.1007/308
s12161-008-9052-1.309
[13] O. ¨Ozdestan, A. ¨Uren, Development of a cost-effective method for nitrate and nitrite determination in310
leafy plants and nitrate and nitrite contents of some green leafy vegetables grown in the Aegean region311
of Turkey, Journal of Agricultural and Food Chemistry 58 (2010) 5235–5240. doi:10.1021/jf904558c.312
[14] R. Andrade, C. O. Viana, S. G. Guadagnin, F. G. R. Reyes, S. Rath, A flow-injection spectrophotometric313
method for nitrate and nitrite determination through nitric oxide generation, Food Chemistry 80 (2003)314
597–602. doi:10.1016/s0308-8146(02)00508-3.315
[15] A. A. Chetty, S. Prasad, Flow injection analysis of nitrate-N determination in root vegetables: Study of316
the effects of cooking, Food Chemistry 116 (2009) 561–566. doi:10.1016/j.foodchem.2009.03.006.317
[16] D. Farrington, A. P. Damant, K. Powell, J. Ridsdale, M. Walker, R. Wood, A comparison of the318
extraction methods used in the UK nitrate residues monitoring program, Journal of the Association of319
Public Analysts 34 (2006) 1–11.320
[17] P. Kong Thoo Lin, A. N. Araujo, M. C. B. S. M. Montenegro, R. P´erez-Olmos, New PVC nitrate-321
selective electrode: application to vegetables and mineral waters, Journal of Agricultural and Food322
Chemistry 53 (2005) 211–215. doi:10.1021/jf049227u.323
[18] M. I. N. Ximenes, S. Rath, F. G. R. Reyes, Polarographic determination of nitrate in vegetables, Talanta324
51 (2000) 49–56. doi:10.1016/s0039-9140(99)00248-9.325
[19] J. R. Santos, J. L. M. Santos, J. L. F. C. Lima, Single interface flow system with potentiometric326
detection for the determination of nitrate in water and vegetables, Talanta 80 (2010) 1326–1332. doi:327
10.1016/j.talanta.2009.09.031.328
[20] Z. Szabo, K. B¨oddi, L. Mark, L. G. Szabo, R. Ohmacht, Analysis of nitrate ion in nettle (Urtica dioica329
L.) by ion-pair chromatographic method on a C30 stationary phase, Journal of Agricultural and Food330
Chemistry 54 (2006) 4082–4086. doi:10.1021/jf0524628.331
[21] M. D. Croitoru, Nitrite and nitrate can be accurately measured in samples of vegetal and animal332
origin using an HPLC-UV/VIS technique, Journal of Chromatography B 911 (2012) 154–161. doi:333
10.1016/j.jchromb.2012.11.006.334
[22] M. Zhou, D. Guo, Simultaneous determination of chloride, nitrate and sulphate in vegetable sam-335
ples by single-column ion chromatography, Microchemical Journal 65 (2000) 221–226. doi:10.1016/336
s0026-265x(00)00056-4.337
16

18. M
ANUSCRIPT
ACCEPTED
ACCEPTED MANUSCRIPT
[23] S. De Martin, P. Restani, Determination of nitrates by a novel ion chromatographic method: occurrence338
in leafy vegetables (organic and conventional) and exposure assessment for italian consumers, Food339
Additives and Contaminants 20 (2003) 787–792. doi:10.1080/0265203031000152415.340
[24] M. Iammarino, A. Di Taranto, M. Cristino, Monitoring of nitrites and nitrates levels in leafy vegetables341
(spinach and lettuce): a contribution to risk assessment, Journal of the Science of Food and Agriculture342
94 (2014) 773–778. doi:10.1002/jsfa.6439.343
[25] M. Jimidar, C. Hartmann, N. Cousement, D. Massart, Determination of nitrate and nitrite in vegetables344
by capillary electrophoresis with indirect detection, Journal of Chromatography A 706 (1995) 479–492.345
doi:10.1016/0021-9673(94)01290-u.346
[26] C. Merusi, C. Corradini, A. Cavazza, C. Borromei, P. Salvadeo, Determination of nitrates, nitrites and347
oxalates in food products by capillary electrophoresis with pH-dependent electroosmotic flow reversal,348
Food Chemistry 120 (2010) 615–620. doi:10.1016/j.foodchem.2009.10.035.349
[27] N. H. Snow, G. C. Slack, Head-space analysis in modern gas chromatography, TrAC Trends in Analytical350
Chemistry 21 (2002) 608–617. doi:10.1016/s0165-9936(02)00802-6.351
[28] G. J. King, C. Gazzola, R. L. Blakeley, B. Zerner, Triethyloxonium tetrafluoroborate as an ethylating352
agent in aqueous solution, Inorganic Chemistry 25 (1986) 1078–1078. doi:10.1021/ic00228a003.353
[29] G. A. Olah, K. K. Laali, Q. Wang, G. K. Surya Prakash, Onium Ions, Wiley, New York, 1998.354
[30] V. G. Granik, B. M. Pyatin, R. G. Glushkov, The chemistry of trialkyloxonium fluoroborates, Russian355
Chemical Reviews 40 (1971) 747–759. doi:10.1070/rc1971v040n09abeh001967.356
[31] E. Pagliano, J. Meija, R. E. Sturgeon, Z. Mester, A. D’Ulivo, Negative chemical ionization GC/MS357
determination of nitrite and nitrate in seawater using exact matching double spike isotope dilution358
and derivatization with triethyloxonium tetrafluoroborate, Analytical Chemistry 84 (2012) 2592–2596.359
doi:10.1021/ac2030128.360
[32] E. Pagliano, J. Meija, Z. Mester, High-precision quadruple isotope dilution method for simultaneous361
determination of nitrite and nitrate in seawater by GCMS after derivatization with triethyloxonium362
tetrafluoroborate, Analytica Chimica Acta 824 (2014) 36–41. doi:10.1016/j.aca.2014.03.018.363
[33] A. D’Ulivo, E. Pagliano, M. Onor, E. Pitzalis, R. Zamboni, Vapor generation of inorganic anionic364
species after aqueous phase alkylation with trialkyloxonium tetrafluoroborates, Analytical Chemistry365
81 (2009) 6399–6406. doi:10.1021/ac900865d.366
[34] E. Pagliano, J. Meija, J. Ding, R. E. Sturgeon, A. D’Ulivo, Z. Mester, Novel ethyl-derivatization367
approach for the determination of fluoride by headspace gas chromatography/mass spectrometry, An-368
alytical Chemistry 85 (2013) 877–881. doi:10.1021/ac302303r.369
17

19. M
ANUSCRIPT
ACCEPTED
ACCEPTED MANUSCRIPT
[35] R. J. Wells, Recent advances in non-silylation derivatization techniques for gas chromatography, Journal370
of Chromatography A 843 (1999) 1–18. doi:10.1016/s0021-9673(98)00986-8.371
[36] P. De Bi`evre, Isotope dilution mass spectrometry: what can it contribute to accuracy in trace analysis?,372
Fresenius Journal of Analytical Chemistry 337 (1990) 766–771. doi:10.1007/bf00322250.373
[37] J. Meija, Z. Mester, Paradigms in isotope dilution mass spectrometry for elemental speciation analysis,374
Analytica Chimica Acta 607 (2008) 115–125. doi:10.1016/j.aca.2007.11.050.375
[38] E. Pagliano, Z. Mester, J. Meija, Reduction of measurement uncertainty by experimental design in376
high-order (double, triple, and quadruple) isotope dilution mass spectrometry: application to GC-377
MS measurement of bromide, Analytical and Bioanalytical Chemistry 405 (2013) 2879–2887. doi:378
10.1007/s00216-013-6724-5.379
[39] E. Pagliano, Z. Mester, J. Meija, Calibration graphs in isotope dilution mass spectrometry, Analytica380
Chimica Acta 896 (2015) 63–67. doi:10.1016/j.aca.2015.09.020.381
[40] E. Pagliano, M. Onor, E. Pitzalis, Z. Mester, R. E. Sturgeon, A. D’Ulivo, Quantification of nitrite and382
nitrate in seawater by triethyloxonium tetrafluoroborate derivatization—headspace SPME GC–MS,383
Talanta 85 (2011) 2511–2516. doi:10.1016/j.talanta.2011.07.098.384
[41] J. F. Pickup, K. McPherson, Theoretical considerations in stable isotope dilution mass spectrometry385
for organic analysis, Analytical Chemistry 48 (1976) 1885–1890. doi:10.1021/ac50007a019.386
[42] P. H. Nam, B. Alejandra, H. Fr´ed´eric, B. Didier, S. Olivier, P. Andr´e, A new quantitative and low-cost387
determination method of nitrate in vegetables, based on deconvolution of UV spectra, Talanta 76 (4)388
(2008) 936–940. doi:10.1016/j.talanta.2008.04.048.389
[43] AOAC Official Method 976.14, Nitrate and nitrite in cheese, modified Jones reduction method.390
[44] K. G. Heumann, Isotope-dilution ICP–MS for trace element determination and speciation: from a391
reference method to a routine method?, Analytical and Bioanalytical Chemistry 378 (2004) 318–329.392
doi:10.1007/s00216-003-2325-z.393
[45] I. Castanheira, L. Oliveira, A. Valente, P. Alvito, H. S. Costa, A. Alink, The need for reference materials394
when monitoring nitrate intake, Analytical and Bioanalytical Chemistry 378 (2004) 1232–1238. doi:395
10.1007/s00216-003-2382-3.396
18

20. M
ANUSCRIPT
ACCEPTED
ACCEPTED MANUSCRIPT
Rapid determination of nitrate in vegetables by gas chromatography
mass spectrometry
Highlights:
• Novel isotope dilution GC-MS method for high-precision determination of
nitrate in vegetable.
• Stable calibration plot for more than one month
• Simple: based on aqueous chemistry at room temperature
• Rapid: only 1.8 minutes of chromatography
• Cost effective: less than 0.5 USD of chemicals per analysis

21. M
ANUSCRIPT
ACCEPTED
ACCEPTED MANUSCRIPT