See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/275038705

Uncertainty of measurement in spectrometric analysis: a case study

Research · April 2015

DOI: 10.13140/RG.2.1.1527.7609

CITATIONS READS

0 756

1 author:

Justina Dobiliene
Kaunas University of Technology
12 PUBLICATIONS   28 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Rapid evaluation of fresh chicken meat quality by electronic nose View project

All content following this page was uploaded by Justina Dobiliene on 17 April 2015.

The user has requested enhancement of the downloaded file.

 Measurement 43 (2010) 113–121

Contents lists available at ScienceDirect

Measurement
journal homepage: www.elsevier.com/locate/measurement

Uncertainty of measurement in spectrometric analysis: A case study

J. Dobilienė *, E. Raudienė, R.P. Žilinskas
Kaunas University of Technology, Metrology Institute, Studentu str. 50-253, Kaunas, Lithuania

a r t i c l e i n f o a b s t r a c t

Article history: The methods of spectrometric analysis belong to a group of measurement methods that are
Received 17 June 2009 used for their simplicity, speed of analysis, and large selectivity to solve various problems
Accepted 13 July 2009 of analytes. Accuracy and reliability of any chemical analysis largely depend on the way the
Available online 17 July 2009
analyzed sample is prepared.
The objective of this work is to evaluate the uncertainty and metrological reliability of
Keywords: material concentration measurement considering sample preparation and chemical–phys-
Uncertainty
ical transformation of spectrometric analysis.
Spectrometry
Sample preparation
Metrological properties of spectrophotometric method have been evaluated experimen-
tally and influence of analysis conditions on the measurement result was described.
Ó 2009 Elsevier Ltd. All rights reserved.

1. Introduction and only then this value is measured. Different sample pro-
cedures are carried out and they can influence measure-
Spectrometric analysis is a widely used measurement ment result by chemical–physical transformation. The
method for qualitative and quantitative analysis. sample is prepared mechanically and chemically during
Qualitative evaluation of spectrometric analysis result sample preparation process. Mostly these procedures are
comprises influence of all possible sources and factors, not assessed in practice.
including sample preparation technique, measurement Numerous authors have published on the topic of
principle, analyte concentration, matrix properties, etc. uncertainty evaluation of chemical measurements results.
The measurement uncertainty depends upon all men- Also some guides are published on uncertainty estimation
tioned parameters [1]. [2,3]. The material reported in journal articles is rather
Technical specifications of measuring device play general because of every chemical measurement is exclu-
important role in any measurement. Quite often producers sively specific. So uncertainty cannot be excluded from
as well as users are proud of high accuracy of their measur- the working practices. It is very important to evaluate all
ing instruments. However, while evaluating the results of possible parameters that have influence on uncertainty
spectrometric analysis, sample preparation procedures and all of them have to be taken into account.
are considered insufficient. It takes place in spite of the fact This paper discusses qualitative evaluation of chemical–
that the newest, fast, automated, needed the minimum physical transformation and its impact on uncertainty of
amount of sample and chemical reagents technology, and measurement results in spectrometric analysis.
measuring equipment providing high repeatability are ap- The measurement of the concentration of iron in a
plied for analysis. water sample using spectrophotometric method was cho-
Uncertainty evaluation is not a simple task because of sen to show experimentally the influence of chemical–
the peculiarities of chemical measurements. The chemical physical transformation conditions on the measurement
parameter always is transformed into physical parameter uncertainty.
The physical basis of spectrophotometric measurements
* Corresponding author. Tel.: +370 68233673; fax: +370 67351252. is well known [4]. Uncertainty sources for measuring
E-mail address: justina.dobiliene@ktu.lt (J. Dobilienė). instrument (repeatability of spectrophotometer reading,

0263-2241/$ - see front matter Ó 2009 Elsevier Ltd. All rights reserved.
doi:10.1016/j.measurement.2009.07.003
114 J. Dobilienė et al. / Measurement 43 (2010) 113–121

spectrophotometer drift, stray light, etc.) are known as well Sample preparation
[5–7]. ‘‘Classical” uncertainty sources (weighing, volumetric
operations, etc.) are estimated enough [7–9]. Measurement Chemical
uncertainty associated with linear least squares calibration Mechanical
is estimated [2,7–9]. However, in evaluation of uncertainty Concentration
of measurement results in routine laboratory attention that
is paid to chemical–physical transformation conditions
(sample pH, content of chemical reagent, exposure time of
Physical parameter Formation of
prepared sample, etc.) are insufficient.
measurement calibration curve
Fig. 1. Diagram of procedures affecting the result.
2. Peculiarities of chemical measurements

Chemical analysis differs from other measurements. measurement procedures when the final expression of
There is no way to measure the ‘‘true value” of anything. the analysis is obtained by measuring a certain physical
The best we can do in a chemical analysis is to apply care- parameter. A series of preparation procedures must be per-
fully a technique that experience tells us is reliable. If the formed before measuring the concentration of analyte.
results of measuring the same quantity by different meth- Amount and complexity of operations depend upon the
ods agree with each other, then we become confident that chosen method and aim of analysis [11].
the results are near the ‘‘true” value. But sample prepara- Sample preparation is still one of the most time-con-
tion procedures always contribute their own input to mea- suming, labor-intensive, and error-prone steps in the anal-
surement result. ysis cycle [12]. The selection of a sample preparation
Three main stages that have to be evaluated are re- method is dependent upon several factors such as:
quired for chemical analysis:
1. The analyte(s).
1. Transformation of chemical parameter into physical 2. The analyte concentration level(s).
parameter (all sample preparation procedures, auxiliary 3. The sample matrix.
reactions, etc.). 4. The instrumental measurement technique.
2. Measurement of transformed parameter. 5. The required sample size.
3. Evaluation of the result including uncertainty.
Selected method will require specific sample prepara-
tion equipment and reagents.
2.1. Chemical–physical transformation in spectrometric So, to solve a particular task an appropriate method of
analysis spectrometric analysis, where the quantity of analyte is
determined by measurement of various material parame-
Spectrometric analysis involves series of methods. ters, such as absorption, emission, mass-to-charge ratio,
Parameters that have impact on the uncertainty of mea- etc., is used [4].
surement result for various spectrometric methods are dif-
ferent because measured parameter differs as well. 2.1.1. Mechanical sample preparation
External conditions and error of measuring device have im- Any treatments or procedures applied to the sample in
pact on every analysis. But it is not as strong as chemical– order to convert it in to a form suitable for the chosen ana-
physical transformation. lytical method are attributed to sample preparation [13].
Chemical–physical transformation stage comprises all Different methods require of conformable sample prepara-
mechanical and chemical sample preparation procedures. tion but there can be find common aims and preparation
During the transformation sample is prepared in such a procedures for some spectrometric analysis methods
way that it should be a possibility to measure physical (atomic absorption and atomic emission spectroscopy,
parameter that is directly related with analyte spectrophotometry, infrared spectroscopy). At first analyte
concentration. has to be converted into a physical/chemical form compat-
All methods of chemical instrumental analysis are ible with chosen method of analysis (most analytical meth-
based upon receiving of analytical signal and measuring ods do not work well for solid or inhomogeneous samples)
its intensity, i.e. upon chemical and physical characteristics also sample for analysis must be as pure as possible (re-
of any material [10]. Fig. 1 illustrates procedures which move the maximum amount of extraneous material from
have influence to uncertainty of measured concentration the sample so that it would not interfere with subsequent
of analyte. measurement of the analyte of interest). Table 1 shows
Sample preparation procedures comprise all treatments mechanical sample preparation procedures for different
or procedures from sample collection (place the sample is aggregative states for different spectrometric analysis
collected, expected concentration of analyte), storing (tem- methods.
perature, time-span till analysis), transportation conditions
(vessels, duration, temperature, etc.), sample preparation 2.1.2. Chemical sample preparation
for analysis (all mechanical treatments applied to the sam- The uncertainty of the chemical sample preparation is
ple in order to convert it to a laboratory sample form) to the component of the combined uncertainty of the result.
 J. Dobilienė et al. / Measurement 43 (2010) 113–121 115

Table 1
Mechanical sample preparation procedures for different aggregation states in spectrometric analysis.

Method of analysis Solid sample preparation for analysis Liquid sample preparation for Gaseous sample
analysis preparation for
analysis
Atomic emission Heating, dissolution, impurity separation, evaporation Preconcentration or dilution, Impurities separation,
spectroscopy impurities separation drying
Atomic absorption Drying, homogenization, impurities separation, milling, Impurities separation, filtration, –
spectroscopy sieving, extraction, filtration, atomization preconcentration or dilution,
atomization
Spectrophotometry Extraction, drying, preconcentration, sorption, precipitation, Preconcentration or dilution, –
impurities separation, filtration, dissolution impurities separation
IR spectrometry Milling, dissolution, suspendation, pressing Dehydration Filtration
Mass spectroscopy Milling, evaporation Direct testing Direct testing

Unfortunately this uncertainty is mostly forgotten or ig- It shows that the tested object and the measuring
nored and this then leads to an optimistic underestimation instrument cannot be separated during the evaluation
of the uncertainty of the resulting measurement. It is the process.
purpose of modern uncertainty evaluation that the uncer- According to the diagram, in order to evaluate a mea-
tainty associated with the chemical preparation of the surement result thoroughly, all the factors that can influ-
sample be a part of the uncertainty budget of the final ence the measurement results have to be analyzed.
result. These factors are: technical characteristics of measuring
The influence of chemical sample preparation and instrument, measurement conditions, measurement ma-
external factors that can affect the result of spectrometric trix, sample preparation procedures, the interaction of a
analysis on measurement of analyte concentration is pre- test item and measuring instrument, restrictions of the
sented in Table 2. Also there is given physical parameter, measurement method used, reading uncertainty, and data
which has to be measured in different methods. Data in processing. Only after the close analysis of the above-men-
Table 2 show that temperature of test sample is the most tioned factors and the evaluation of additional conditions
important factor for evaluation of results of above listed of a certain case, the uncertainty as accurate estimate of
analysis methods. Significance of other components de- the measurement result can be properly provided.
pends upon specific measurement method.
4. Methods and results

3. Uncertainty sources of spectrophotometric analysis 4.1. Elaboration of method and identification of uncertainty
sources
By force of spectrometric analysis simplicity, speed of
analysis, and large selectivity, these methods are used to The measurement of the concentration of iron in a
solve many problems of different analytes. Advanced water sample was chosen for the analysis of spectrophoto-
equipment appears to be simple, but the relation between metric method.
concentration of analyte in material and the signal that has The measurement of iron content by spectrophotomet-
been registered is rather complicated. For this reason the ric method was carried out using a calibration curve.
research of metrological characteristics becomes more The absorbance of standard solutions c1  cn was mea-
complicated. In addition, an analyte signal usually depends sured and the values A1  An were obtained. In this case
not only on analyte concentration, but on the concentra- the concentration is calculated from the data obtained by
tion of all other materials, material structure, sample form, the evaluation of regression equation parameters of a cal-
environmental conditions. Furthermore, using any mea- ibration curve and is given by following formula:
surement instrument, accuracy and reliability of chemical
analysis depend largely on the way the analyzed sample
Ai ¼ ci  B1 þ B0 ð1Þ
is prepared, because preparation mechanism is specific in where ci is content of analyte in the sample, calculated
chemical measurements. from a calibration curve, mg/l; Ai , measured absorption of
For the evaluation of measurement result it is necessary the test sample; B1 , slope of the calibration curve; and B0
to analyze all possible factors [14] that could influence the is intercept of the calibration curve.
result. The slope of the linear least squares calibration curve B1
Method of spectrophotometric analysis is commonly and intercept of the prepared calibration curve B0 are cal-
used group of measurement methods in laboratories of culated from the following equations:
chemical analysis. However, metrological assessment of Pn
photometric measurement results is insufficient. ci Ai  ncA
B1 ¼ Pi¼1
n 2
ð2Þ
A diagram for evaluation of uncertainty for results ob- i¼1 c i  nc
2
P 2 Pn
tained from the spectrophotometric analysis is provided A i ci  c i¼1 ci Ai
in Fig. 2. B0 ¼ P n 2
ð3Þ
2
i¼1 c i  nc
 116 J. Dobilienė et al. / Measurement 43 (2010) 113–121

Table 2
Impact of chemical sample preparation and external factors on results of spectrometric analysis.

Methods of Physical Uncertainty sources

spectrometric parameter
Sample Impact of Reaction Separation Amount Volume Dilution Temperature Stability Retention Pressure
analysis corresponding
pH impurities between of analyte of of of of time of
chemical value
sample from chemical prepared test test prepared
(concentration)
and sample reagent sample sample sample sample
reagents matrix
Atomic absorption Absorption + + + + + + + +
spectroscopy
Atomic emission Emission + + + + + + + +a
spectroscopy
Spectrophotometry Absorption + + + + + + + + +
IR spectroscopy Absorption + + + + + + + + +
Mass spectroscopy Mass-to-charge + +
ratio
a
Only for gases.

 formation of calibration curve,

 evaluation of the result.

These procedures are shown in Fig. 3.

It is very important to evaluate all possible factors that
can affect the result solving the question of measurements
reliability. For this reason detailed analysis of both chosen
methods of measurement of iron concentration in water is
accomplished and shown in Fig. 4. There are shown possi-
ble factors of chemical–physical transformation in differ-
ent measurement stages for methods A and B
respectively in the right and left sides of Fig. 4.
Analysis of measurement procedure showed that
expression (1) of concentration of analyte is not sufficient.
In order to evaluate the influence of all the stages of the
measurement procedure, the mathematical model for cal-
culation of iron quantity in the water sample has to be sup-
plemented with new correction coefficients:

Fig. 2. Diagram for evaluation of uncertainty of spectrophotometric c ¼ c0  d  fr  fT  fpH  ft  f/  fb  fV ð4Þ

analysis results.
where c is concentration of iron in a test sample, mg/l; c0 ,
where ci is the concentration of the ith iron standard solu- content of iron in a sample calculated from calibration
tion; Ai , the absorbance of the ith standard solution curve, mg/l; d, dilution factor; fr , correction factor related
ðA1 ; . . . ; Ai ; . . . ; An Þ; and n is the number of points in the line. to the content of chemical reagent; fT , temperature correc-
In order to ensure the comparability among the re- tion factor; fpH , correction factor related to the solution pH;
search results, two methods of measurement of iron con- ft , correction factor related to the exposure time of pre-
centration in water have been chosen: pared solution; f/ , correction factor related to the impuri-
ties in the solution; fb , correction factor related to the
1. ISO 6332:1988 Water quality – Determination of iron – thickness of absorption layer; and fV is correction factor
Spectrometric method using 1,10-phenanthroline related to the volume of vessels.
(hereinafter method A) [15]. In the case of measurement according to this mathe-
2. Measurement method of iron content using sulfosali- matical model, the sources of uncertainty and their influ-
cylic acid (hereinafter method B). ence on the measurement results are given in Fig. 5.
Comparative cycle of measurements was performed in
A range of determination of the concentration of iron order to evaluate variation of analysis conditions. Analysis
was 0.10–2.00 mg/l. This range covers the area of both cho- in both cases was carried out changing temperature of test
sen methods. sample, exposure time of prepared sample, content of
The procedure of measurement of analyte by spectro- chemical reagent, thickness of absorption layer, pH value
photometric method according to the chosen methods of prepared solution, or by adding additional substances
consists of some stages: as impurities into specimen. The matrix of measurement
procedure used for experiments is displayed in Table 3.
 sample preparation for measurement, Parameters given in Table 3: block letters show the nor-
 measurement of absorption, mal value of appropriate parameter and the lower letters
 J. Dobilienė et al. / Measurement 43 (2010) 113–121 117

associated with chemical reagent differs with the different

concentration of a test analyte.
Sample preparation
The permissible deviation of thickness of the absorbing
layer is (from 0.05 to +0.05) and this influence would be
not significant for both used methods. More considerable
Filling of test sample capacity deviations of the absorbing layer make an influence on
Procedure of sample measurement results over full range of concentrations.
preparation for Particularly large deviations of results are met in the range
measurement
Test sample preparation of small concentrations.
Deviations of the sample pH from the normal value
influences the measurement results of both methods in full
Filing of sample capacity range of concentrations. Increased or decreased pH value
determinate more intense deviations of the results.
Preparation of reference Deviations from the specified conditions do not influ-
Chemical preparation of sample ence measurement results if measurements are carried
Preparation of calibration
out using the method B and exposure time of prepared
curve standard solutions sample is kept (1; +2) min. The deviations of exposure
Exposure of prepared sample
time of prepared sample have a significant influence on re-
sults of method A in the range of concentration from 1 mg/l
Calibration curve solutions
Sample absorption measured by absorption measured by up to 2 mg/l, when the exposure time for prepared sample
spectrophotometer spectrophotometer is less than required according to the methodic.
If a solution contains impurities, they reduce the accu-
Result racy and reliability of measurement results. Experiments
proved that the influence of impurities decreases if the
Fig. 3. Procedure for measurement of iron concentration by spectropho- concentration of test material increases. When measuring
tometric method. iron concentration in polluted samples or the ones of un-
known composition, it is necessary to include the compo-
nent of sample pollution in the calculations of combined
relative standard uncertainty.
show the values that are lower or higher than the normal
value. 4.2. Evaluation of combined uncertainty and uncertainty
components
C/c is the temperature of test solution,
D/d is the exposure time of a prepared sample, The uncertainty of measurement of iron content in
E/e is the amount of a chemical reagent, water sample is calculated by combining all standard
F/f is the thickness of an absorption layer, uncertainties of input values c0 ; f r ; f T ; f pH ; f t ; f / ; f b ; f V .
G/g is the pH value in a test solution, Uncertainties have been combined using the errors propa-
H/h is the impurities in a test sample. gation rule [2,3].
Relation between the measurement result and input
Limits of experimental conditions of possible variants parameters is expressed by the following model:
for iron concentration measurement procedure are shown
in Table 4. c ¼ f ðx1 ; x2 ; . . . . . . ; xi . . . . . . :xN Þ ð5Þ
Performed experimental researches demonstrated that where x1 ; . . . ; xi ; . . . ; xn expresses input parameters
the deviations of temperature had the most significant ðc0 ; fr ; fT ; fpH ; ft ; f/ ; fb ; fV Þ.
influence on the results of both methods in the range of An uncertainty component due to the calibration curve
small concentrations (up to 0.5 mg/l). However, deviations uðC cal Þ comprises three constituents that depend on the
of temperature are significant in the range of concentra- parameters of regression equation, the characteristics of
tions over 0.5 mg/l as well, as deviations in particular measuring instrument and the concentration of calibration
points can reach up to 20% of the nominal value. Referring curve standard solutions. The uncertainty of the calibration
to the calculations of experimental results it can be con- curve uðC cal Þ is calculated as:
cluded that both methods require including uncertainty qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
component associated with temperature when calculating uðC cal Þ ¼ uðC cal ; regÞ2 þ uðC cal ; instÞ2 þ uðC cal ; stÞ2 ð6Þ
expanded uncertainty in the range of concentration up to
1 mg/l with test temperature of 20 ± 3 °C. where uðC cal ; regÞ is the standard uncertainty of regression
Deviations in the amount of chemical reagents from the equation for calibration curve; uðC cal ; instÞ, the standard
normal value are important for the results of both meth- uncertainty of measuring instrument characteristics; and
ods, particularly in the concentration over 1 mg/l, where uðC cal ; stÞ is the uncertainty of concentration of standard
the deviations of result gain up to 40% of the nominal va- solutions.
lue. In the range of smaller concentrations, deviations The standard uncertainty of regression equation for cal-
reach up to 10% from the nominal value. Analysis of mea- ibration curve uðC cal ; regÞ is calculated by the least squares
surement results showed that the influence of component method:
118 J. Dobilienė et al. / Measurement 43 (2010) 113–121

Procedure of sample
Method A measurement Method B

sample acidify Sample preparation sample acidify

50 ml flask Filling of test sample capacity 50 ml flask

boiling Test sample preparation 40 min boiling

50 ml flask Filling of sample capacity 50 ml flask

+1 ml hydroxylaminechloride +1 ml ammoniochloride
+2 ml acetate buffer Chemical preparation of sample +1 ml sulfosalicylic acid
+2 ml 1,10-phenanthroline +1 ml ammonia

15 min Exposure of prepared sample 5 min

absorption layer20mm Sample absorption measured by absorption layer 20mm

spectrophotometer wavelength 400nm
wavelength 510nm

Fig. 4. Procedure for measurement of iron concentration by methods A and B.

sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
s 1 1 ðc0  cÞ2 The uncertainty of the weight uðmÞ is stated as uncer-
uðC cal ; regÞ ¼ þ þ ð7Þ tainty of repeatability u(m, rep), calibration uncertainty of
B1 p n S
Pn  2 the weight u(m, cal) and uncertainty caused by rounding
j¼1 A j  ðB0 þ B 1  c j Þ of the digital reading u(m, roun) The equation for the uncer-
s¼ ð8Þ
n2 tainty of the weight is as follows:
X n
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
S¼ ðcj  cÞ2 ð9Þ
uðmÞ ¼ uðm; repÞ2 þ uðm; rounÞ2 þ uðm; calÞ2 ð12Þ
j¼1

where B1 is a slope of the calibration curve; p, the number The uncertainty associated with the purity of the re-
of measurements to determine c0 ; n, the number of mea- agent is calculated from the producer’s specifications. The
surements for calibration; c0 , determined concentration purity of iron
pffiffiffi is given in the certificate and the value is di-
of iron; c, the mean value of different calibration stan- vided by 3 to obtain the standard uncertainty (a rectan-
dards; i, the index of number of calibration standards; j, gular distribution is assumed).
the index of number of measurements to obtain the cali- The uncertainty of the volumetric flasks is stated as the
bration curve; and s is the residual standard deviation. uncertainty of filling the flask to the mark uðV fl ; fillÞ, cali-
The standard uncertainty for characteristics of measur- bration uncertainty of the flask volume uðV fl ; calÞ, and
ing instrument is: uncertainty of the volume caused by the temperature dif-
ference. The uncertainty of the volume of a flask is:
Dinstr
uðC cal ; instÞ ¼ ð10Þ qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
3 2 2
uðV fl Þ ¼ uðV fl ; calÞ þ uðV fl ; fillÞ þ uðV fl ; tempÞ2 ð13Þ
where Dinstr is the accuracy of spectrophotometer.
V fl  Dt  c
The concentration of reference solution and volume of uðV fl ; tempÞ ¼ pffiffiffi ð14Þ
standard solutions influences the uncertainty of standard 3
solutions concentration uðC cal ; stÞ. where V fl is volume of flask; Dt, the temperature difference
The uncertainty of reference solution concentration is between the value used for the calibration of the flask and
stated as uncertainty of the purity of reference material the temperature of the medium used in the laboratory; and
uðPÞ, weighing uncertainty of reference material uðmÞ and c is thermal expansion coefficient of water.
uncertainty of the volumetric flask of reference solution The uncertainty of the volume of pipette is stated as the
uðV fl ; refÞ. The uncertainty of the reference solution con- repeatability of the volume delivered by the pipette, cali-
centration uðcref Þ is calculated as: bration uncertainty of the pipettes uðV p ; calÞ, and uncer-
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi tainty of the volume caused by the temperature
uðcref Þ ¼ uðmÞ2 þ uðPÞ2 þ uðV fl ; refÞ2 ð11Þ difference between the value used for the calibration of
 J. Dobilienė et al. / Measurement 43 (2010) 113–121 119

V(et) m Fe ci(st)
temperature repeatability
rounding
calibration repeatability
Purity drift
repeatability rounding
drift
c(et)
Ai(st)
Vi(st)
c0
repeatability calibration
calibration curve
temperature
calibration uncertainty of spectrophotometer
coefficients of regression equation
c (Fe)

temperature
calibration
repeatability

d fT fφ f pH fr ft fb fVi

Fig. 5. Diagram of factors affecting the result.

Table 3 Table 4
Matrix of experiments for iron concentration measurement procedure. Limits of experimental conditions.

Parameter Possible variants Measurement condition Method Experimental conditions

J K L M N O Low Normal High

limit conditions limit
C/c c C C C C C
D/d D d D D D D Temperature (°C) A 10 20 30
E/e E E e E E E B 10 20 30
F/f F F F f F F Exposure time of a prepared A 5 15 20
G/g G G G G g G sample (min) B 3 5 10
H/h H H H H H h Amount of chemical reagent (ml) A 1 2 4
B 0.5 1 2.5
Absorption layer (mm) A 5 20 30
B 5 20 30
pH A 1 4.5 9
the flask and the temperature of the medium used in the B 4 P9 10
laboratory uðV p ; tempÞ. The uncertainty of the pipette is: Impurities:
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi Fe:Zn
uðV p Þ ¼ uðV p ; calÞ2 þ uðV p ; repÞ2 þ uðV p ; tempÞ2 ð15Þ Fe:Cu A and B – – 1:10
Fe:Co
The components of evaluated uncertainty according to
the experiments are the following:

 the component evaluating the amount of chemical The combined standard uncertainty of the measured
reagent; analyte in full range of measured concentrations is calcu-
 the component of the temperature of the test solution; lated as:
 the pH component of the test solution;
N 
X 2
 the component associated with the exposure time of ocFe
uðcFe Þ2 ¼  uðxi Þ2 ð17Þ
prepared solution; i¼1
oxi
 the component of the impurities in the test solution;
 the component of the thickness of absorbing layer. where uðxi Þ are standard uncertainties of input parameters,
ðocFe =oxi Þ are coefficients of sensitivity.
The standard uncertainties uðfi Þ of the above-mentioned The expanded uncertainty of iron measurement UðcÞ is
components are calculated from experimental curves of obtained by multiplying the combined uncertainty by a
each uncertainty component and taking into account pos- coverage factor k ¼ 2. A level of confidence is 95%.
sible real conditions of the analysis.
5. Discussion of results
aiþ  ai
uðfi Þ ¼ pffiffiffiffiffiffi ð16Þ
12 In carrying out measurements by method A different
where aiþ and ai are the lowest and the highest values of influence of uncertainty components to expanded uncer-
each component in chosen range. tainty has been determined for the following concentration
120 J. Dobilienė et al. / Measurement 43 (2010) 113–121

Source of uncertainty
vessels

absorbing layer 0.2 mg/l

impurities
0.1 mg/l
exposure time
pH
temperature
2 mg/l
amount of reagent
regression equation 1.5 mg/l

measuring instrument 1 mg/l

concentration of standard 0.5 mg/l

solutions
0 20 40 60 0 20 40 60
Influence, %
Fig. 6. The influence of uncertainty components on expanded uncertainty in the measurement of iron concentration using method A.

ranges: up to 0.2 mg/l and from 0.2 mg/l to 2 mg/l. This ponents on expanded uncertainty has been determined in
influence is illustrated in Fig. 6. the range of concentrations up to 0.5 mg/l and from
Relative combined standard uncertainty is up to 0.12 in 0.5 mg/l to 2 mg/l. This influence is shown in Fig. 7.
the range of small concentrations up to 0.2 mg/l and up to Relative combined standard uncertainty is up to 0.17 in
0.07 in the range of concentrations from 0.2 mg/l up to the range of concentrations up to 0.5 mg/l and up to 0.07 in
2 mg/l. The analysis of uncertainty components has shown the range of concentrations from 0.5 mg/l up to 2 mg/l.
that in the range of small concentrations the components Research results on uncertainty components demon-
depending on the temperature, the pH of test solution, strate that the component of solution exposure time has
the sample matrix, and the calibration curve have the most not any significant influence on measurement results in full
significant influence on the measurement uncertainty. range of tested concentrations. Predominant components
In the range of concentrations over 0.2 mg/l the compo- are different in particular concentration. In the range of
nents of the amount of chemical reagent, the pH of test small concentrations, components of the calibration curve
solution, the sample matrix, and the concentration of stan- and the matrix of a tested sample have the most significant
dard solutions of calibration curve have a dominant influ- influence on the measurement uncertainty of analyte. In the
ence on the measurement uncertainty of analyte. range of concentrations over 0.5 mg/l, the biggest uncer-
In carrying out measurements at specified conditions of tainty components are due to the amount of chemical
analysis using method B, the influence of uncertainty com- reagent, the pH of test solution, and the sample matrix.

Source of uncertainty
vessels

absorbing layer
impurities
exposure time 0.5 mg/l
pH 0.2 mg/l
temperature 0.1 mg/l 2 mg/l
amount of reagent 1.5 mg/l
regression equation
1 mg/l
measuring instrument

concentration of standard
solutions
0 20 40 60 0 20 40 60
Influence, %
Fig. 7. The influence of uncertainty components on expanded uncertainty in measurement of iron concentration by method B.
 J. Dobilienė et al. / Measurement 43 (2010) 113–121 121

Histograms in Figs. 6 and 7 confirm the statement that material concentration is insignificantly sensitive to the
characteristics of measuring instrument are not sufficient parameters of the used measuring instrument.
to evaluate analyte measurement uncertainty. Therefore
it is very important to ensure the conditions of sample References
preparation and measurement process, and it is necessary
to take into account correction factors in the measurement [1] R.V. Meyer, Minimizing the Effect of Sample Preparation on
Measurement Uncertainty, Agilent Technologies, USA, 2002.
model of analyte. [2] EURACHEM/CITAC Guide Quantifying Uncertainty in Analytical
Measurement, second ed., 2000.
[3] BIPM, IEC, IFCC, ISO, IUPAC, OIML, Guide to the Expression of
6. Conclusion Uncertainty in Measurement, International Organization for
Standardization, Geneva, 1995.
It is necessary to evaluate parameters that appear in [4] A.D. Skoog, F.J. Holler, T.A. Nieman, Principles of Instrumental
Analysis, fifth ed., 1998. 10. D. Mazej, V. Stibilj, Accredit. Qual.
three main phases: physical–chemical transformation,
Assur. 8 (2003) 117–123.
measurement and estimation of the result to make the fi- [5] L. Sooväli, E.I. Roõõm, A. Ku } tt, I. Kaljurand, I. Leito, Accredit. Qual.
nal chemical measurement model. Assur. 11 (2006) 246–255.
The influence of separate sample preparation proce- [6] R. Ramachandran, Rashmi, Analyst 124 (1999) 1099–1103.
} rgens, L. Paama, I. Leito, Accredit. Qual. Assur. 12 (2007) 593–
[7] J. Ju
dures and quality of their performance has direct impact 601.
on uncertainty of spectrometric analysis results. Most of [8] J. Traks, L. Sooväli, I. Leito, Accredit. Qual. Assur. 10 (2005) 197–207.
the hints are not new. They should already be considered [9] A. Drolc, M. Roš, Acta Chim. Slov. 49 (2002) 409–423.
[10] EURACHEM/CITAC Guide: Traceability in Chemical Measurement. A
in standard procedures in well-organized professional lab- Guide to Achieving Comparable Results in Chemical Measurement,
oratories. It is a new challenge to put them in the context 2003.
of measurement uncertainty, which arises from chemical [11] D. Mickevicius, Chemines analizes metodai, I, II dalys. – Vilnius,
1998.
and mechanical sample preparation. The knowledge of [12] E. de Oliveira, Sample preparation for atomic spectroscopy:
measurement uncertainty is an additional value to any evolution and future trends, J. Braz. Chem. Soc. 14 (2) (2003).
analytical result. [13] I. Leito, ISO GUM Uncertainty in Chemistry: Successes and Difficulties,
University of Tartu, Testing Centre, 2004. Available from: <http://
The analysis of measurement of iron content in water www.ut.ee/katsekoda/presentations/Leito_TMQAC_2004_presented.
by spectrophotometric method was performed and pdf>.
showed that uncertainty was basically formed by the first [14] S.K. Timothi, The Uncertainty of Measurements Physical and
Chemical Metrology: Impact and Analysis, American Society for
chemical analysis phases (measurement conditions, the
Quality, USA, 2002.
standard solutions, and the parameters of calibration [15] ISO 6322:1998 Water quality – Determination of iron – Spectrometric
curve). Therefore, the total measurement uncertainty of method using 1,10-phenanthroline.

View publication stats