Nitrate in different matrices, Poster Venice 1998

1
Determination of nitrate in different matrices
using capillary electrophoresis Theo P.E.M.
Verheggen, Jetse C. Reijenga, Fred Huf and Frans
M. Everaerts
Laboratory of Instrumental Analysis, University
of Technology, Eindhoven, the Netherlands
(corresponding author, tgtejr_at_chem.tue.nl)
Most teams were rather
optimistic about the confidence interval. By
observation it appeared that several mistakes
were made. The most important one was that the
standard deviation of the determination was
supposed to the one originating from the use of
the linear regression of the calibration graph,
without taking into account contributions from
sample preparation and dilution. As the
correlation coefficients were generally
excellent, the standard deviation from the fit
would be small. The standard deviation of the
extraction volume alone could easily have been 5
or more. Previous experiments in our laboratory
indicate that the amount of sample (50 g) and the
extraction volume (250 ml) should remain constant
from run to run. Deviating from these guidelines
even resulted in systematic differences in
nitrate content found. Some student teams used a
very small sample amount (lt5g) in order to ovoid
the dilution step. Most teams
realized that they had been a bit optimistic on
the confidence interval, but it must be pointed
out it was the statistics of the total results
(Fig.5.) that convinced them, probably not an
increased insight into the contribution from the
sample preparation (Fig.6.). Almost all of them
recognized the few outliers, either by common
sence, or by applying one of the established
outlier tests. All teams agreed that the
difference between the two potatoe brands was
significant (results from team j and l would not
support this). Systematic errors in the overall
averaged results cannot be excluded. In our view
the student's results, summarized in Figures 5, 6
and 7 are quite interesting from the didactical
point of view. We intend to use the data sets in
our present analytical chemistry course to
illustrate aspects of accuracy and
precision. Non-fertilized lettuce contained
significantly less nitrate (2000 vs 2500 mg/kg)
at only slightly smaller plant mass. The cadmium
electrolyte system allows the simultaneous
determination of nitrate and nitrite in the
presence of excess chloride, such as in
fysiological samples. Significant differences in
nitrate content of different potatoe brands were
found (80 and 270 mg/kg respectively). The sample
preparation of the vegetables was the main cause
of the total confidence interval and it would
seem that the sample preparation method for the
potatoes requires some adjustement so that an
optimised, constant mass of sample is mixed with
a constant extraction volume 1. Jandik P and
Jones W R, J Chromatogr 546 (1991)
431-443 2. Kaniansky D, Zelensky I, Hybenova A
and Onuska F I, Anal Chem 66 (1994)
4258-4264 3. Janini G M, Muschik G M and Issaq
H J, J Cap Elec 1 (1994) 116-120 4. Guan F Y, Wu
H F and Luo Y, J Chromatogr A 719 (1996)
427-433 5. Hargadon K A and McCord B R, J
Chromatogr 602 (1992) 241-247 6. Gebauer P, Deml
M, Bocek P and Janak J, J Chromatogr 267 (1983)
455-457 7. Kawamura Y, Takahashi M, Arimura G,
Isayama T, Irifune K, Goshima N and Morikawa H,
Plant and Cell Physiol 37 (1996)
878-880 8. Bondoux G, Jandik P and Jones W R, J
Chromatogr 602 (1992) 79-88 9. Jimidar M,
Hartmann C, Cousement N and Massart D L, J
Chromatogr A 706 (1995) 479-492 10. Verheggen,
Th.P.E.M. and Everaerts, F.M., J.Chromatogr., 638
(1993) 147-153 Potatoe analyses were carried
out by 2nd year students of the Department of
Chemical Engineering at the Eindhoven University
of Technology in May 1998. The students processed
their own data.
INTRODUCTION
under these conditions could be constructed, and
nitrate, nitrite and chloride could be quantified
simulaneously if necessary. Electropherogram of
the determination of nitrate in a typical
undiluted urine sample. 1nitrate, 2chloride,
3sulfate, 4nitrite (spiked), 5, 6 and 7 are
unknowns.
The analysis of nitrate and other inorganic
anions has been a significant field of
application of CE for years, and was reviewed in
1994 1. The optimization of the separation and
trace level detection has been the subject of a
number of papers 2-4. Applications to the
analysis to food related samples were also
reported 5-9. The present paper reports the
application of CE to analyse nitrate in different
matrices. Separation and detection Detection of
nitrate in capillary electrophoresis can be
carried out in the direct and the indirect mode.
We chose for the former. Larges excess of
chloride, such as in the case of biological
fluids, requires additional measures regarding
resolution. Because a chloride excess locally
induces a lower fieldstrength this leads to a
distorted peak of the co-migrating nitrate, which
normally had about the same effective mobility.
For this reason a background electrolyte was
chosen that selectively retards chloride and
sulfate by complexation with cadmium. The
buffering co-anion in that case was acetate at pH
4.5. equipment Lettuce and urine were
analysed in a home-made closed CE system,
described previously 10. The injection was 5
seconds at 2500 Pa pressure. The capillary was 75
?m inner diameter fused silica. Potatoe analyses
were performed using P/ACE 2000 and P/ACE 5500
(Beckman) electrophoresis systems, using 75 ?m
capillaries of different lengths, detection was
at 214 nm. Calibration graphs were made by
changing the injection time. chemicals Background
electrolytes consisted of a 0.01 M solution of
hydrochloric acid, adjusted to pH 3.0 with
?-alanine (for the students analyses of
potatoes), or to pH 8.1 with TRIS (for the
lettuce analyses). The background electrolyte for
the urine analysis was 0.005 Mol/l cadmium
acetate, buffered to pH 5.0 with acetic acid. EOF
was suppressed 10 by the addition to the BGE of
0.05 polyvinylalcohol (Mowiol 88-8, Hoechst,
Frankfurt, FRG) and 5.10-5 M cetyltrimethylammoniu
m bromide. The latter was anion-exchanged to
remove the bromide. procedures calibration IS,
sample pretreatment Lettuce samples were prepared
by taking either the whole plant (excluding
roots), or taking half or a quarter by vertically
cutting the plant in slices and thus mixing
several plants. An approximately constant weighed
amount was put in an ordinary kitchen blender
together with 250 ml of deionized water and
blended for 5 minutes. The resulting mixture was
filtered through a white band filter in a funnel
and and 10 ml of the supernatant was pipetted
into a 100 ml calibrated volumetric flask, 1 ml
of a 2.510-2 M internal standard solution of
potassium bromide was added and made up with
deionized water. This was then filtered through a
0.45 ?m syringe type filter and injected. The
potatoe analyses procedure for the students was
given as follows "Make a caplibration graph by
injecting a 2.10-4 Mol/l standard solution of
sodium nitrate with injection times between 1 and
20 seconds. Extract nitrate from a weighed amount
of a roughly sliced (peeled) potatoe with a
measured amount of deionized water in a kitchen
blender for 5 minutes. The blender works best
when using between 100 and 300 ml water. Make an
estimate of the amount of sample to extract on
the basis of an order of magnitude content of 500
mg/kg. Dilute the filtered extract if
necessary." Lettuce analyses The lettuce
analysis were carried out by constructing an
internal standard calibration graph and injecting
equal volumes of different nitrate standards. The
individual peak areas had a relative standard
deviation of 0.9 (Nitrate) and 1.5 (Bromide).
The peak areas ratios had a relative standard
deviation of 1.2, so that the use of the
internal standard was not necessary after all. A
calibration graph without internal standard over
one decade was excellent. Fig.1. nitrate
calibration graph for lettuce analysis.
Urine analyses Nitrate was determined in
urine also with direct UV detection. Although
chloride does not adsorb even at 200 nm, a large
excess comigrating with chloride will lead to
unacceptable peak distortion and poor
detectability, caused by the locally lower
fieldstrength. More selectivity between chloride
and nitrate was required and accomplished by
using complex formation 6, in this case using
Cadmium to retard chloride. Although mobility
matching is fair, a good calibration graph of
chloride






Potatoe analyses The student teams generally made
calibration graphs consisting of 6-12 points and
analysed the samples in duplicate or triplicate.
The one obvious outlier must be due to a
calculation error. Intervals indicate 95
confidence interval. Note the logarithmic
concentration axis.
EXPERIMENTAL
Individual results of the two different potatoe
samples were apparently correlated (Fig.6.) which
can only be explained by individually different
systematic bias in the sample preparation
procedure.
Finally all student teams processed all collected
data in order to give a final verdict as to the
nitrate content. The results are depicted in
Fig.7.
CONCLUSIONS
The non-fertilized plants ended up with a
slightly reduced mass at the end of an
exponential growth curve. Fig.2. Growth
curve of the lettuce plants.
REFERENCES
It would seem that in spite of the initial
nitrate amount, the concentration dropped
gradually after 8 weeks. When harvested after 12
weeks, concentrations were slightly less than
2000 mg/kg for the non-fertilized lettuce and
less than 2500 for the fertilized crop. The
standard deviations indicated in Fig.3. include
the total determination, where aspects of sample
preparation predominated.
ACKNOWLEDGEMENT