Linear Analytical Calibration Curve, Analytical Determinations by Interpolation and Standard Additions*

Determination of unknow x and error estimates and LOD from measured y vs. x data by unweighted linear regression.
Homoscedastic data with normal error distribution is assumed.
Ivano Gutz - IQ-USP * If there is a significant increase in volume
during the addition of standards to the sampl
refer to cell P2 to apply corrections.
Calibration data defining the Analytical Curve
Enter x (Conc.) and y (Signal) data (n <=30, expandable).
Volumes should be first converted to concentrations.
Use new lines for replicates (if any).

DATA INPUT Regression Fit LINEAR REGRESSION PARAMETERS for y = a

Conc., x Signal, y Fitted y y residuals Relat. error 0.0110309523809524
0 0 0.011031 -0.01103095 >> 0.268884571428571
0.5 0.15397 0.145473 0.008496762 5.52% 0.9991409
1 0.28417 0.279916 0.004254476 1.50% Relative Standard Error RSE (%)
1.5 0.41771 0.414358 0.00335219 0.80% Std dev of regression
2 0.5505 0.5488 0.001699905 0.31% 3*S(residuals^2)/(n-2)
2.5 0.67647 0.683242 -0.00677238 -1.00% Limit of detection LOD
Limit of quantification LOQ
number of data pairs
two-tailed t-statistic, 95%

0.8
Analytical Calibration Curve or Standard Addition C
0.7
f(x) = 0.268884571428571 x + 0.0110309523809524
R² = 0.999140913233016
0.6

0.5
Signal

0.4

0.3

0.2

0.1

0
0 0.5 1 1.5 2
Concentration

0.01
y residuals
0.005

0
0 0.5 1 1.5 2
-0.005

-0.01

-0.015
 0
0 0.5 1 1.5 2
-0.005

-0.01

-0.015
on and Standard Additions*
nweighted linear regression.

re is a significant increase in volume

the addition of standards to the sample,
o cell P2 to apply corrections.

RESSION PARAMETERS for y = a + bx DETERMINATION OF ANALYTE CONCENTRATION IN SAMPL

a intercept sa 0.0059678342775 Sample Signal Number of Result
b slope sb 0.0039422215775 name yx measurem. xx
r2 correlation r 0.9995703643231 Enter 1 in M14 to activate Standard Addition Method in the next l
elative Standard Error RSE (%) 2.93% 0 0
sy/x 0.0082457480241 Determinations based on fitted linear analytical calibration curve
3*sy/x 0.0247372440722 River water A 0.3983 1 1.440280
Limit of detection LOD 0.0919994923499
Limit of quantification LOQ 0.3066649744998
n 6
t 2.78

on Curve or Standard Addition Curve

28571 x + 0.0110309523809524
3016

1 1.5 2 2.5 3
Concentration

1 1.5 2 2.5 3
1 1.5 2 2.5 3
 Standard Addition with variable volume - data preparation
The addition of a volume Vad of standard of concentration Cstd to a certain initial volume Vin of solution containing
xx, to be determined) and, consequently, attenuates its original signal (that is added to the signal due to the stan
colum Signalcorr. of the Data Output, used to find the Result xx (cell W10) by linear regression. To plot the Data
copy Cad std. and Signalcorr. and Paste Special Values in Conc. x and Signal y (columns A and B).

SAMPLE VOLUME AND STD. CONCENTR. REGRESSION

Sample Volume Vin 100 unit, e.g., mL Slope b 47.43555
Standard conc. Cs 300 unit, e.g., mmol/L Intercept a 100.2115
TRATION IN SAMPLES DATA INPUT DATA OUTPUT Result xx 2.112583
Std dev Uncertainty Vad std. Signal, y Cad std. Signal corr. Vin/Vtot correction
sxx (CI 95%) +/- 0 100 0.000 100.000 1.000 0.000
Method in the next line 0.3 143.3 0.897 143.599 0.997 0.299
0.6 184.8 1.789 185.396 0.994 0.596
cal calibration curve 0.9 224 2.676 224.892 0.991 0.892
0.033241 0.092291 1.2 269.1 3.557 270.286 0.988 1.186

Standard Addition with variable volume

There are at least three other methods to calculate the analyte
concentration by standard addition with variable volume. A simple one is
to ignore the dilution effect on the concentration (as if Vin remained
constant) and multiply the measured signals by Vtot/Vin (the reciprocal
of the dilution factor, as usual in conductimetric titrations).
𝑆_"cor" =(𝑆_𝑒×(𝑉_𝑖𝑛+𝑉_(𝑎𝑑 𝑠𝑡𝑑.)))/((𝑉_𝑖𝑛))
Two others can be found in a separate spreadsheet. Results and
uncertainties are undistinguishable from those obtained here.
of solution containing the sample dilutes the analyte (of conc.
ignal due to the standard). This dilution is compensated in
on. To plot the Data Ouput and find the uncertainty and LOD,
and B).

rrection
Some notes, definitions, examples and bibliography, in English and Portuguese, randomly paste
Most of the statistical equations employed here are available, for example, in Harvey's book [1] or in Hughes's spreads
(but not the results uncertainty) can also be obtained with the Analysis ToolPak supplement of Excel, that includes ANO
The parameters values (including the 'results uncertainty') where checked against examples in references [1] and [2] am
References:
[1] David Harvey, 5.4: Linear Regression and Calibration Curves
https://chem.libretexts.org/Bookshelves/Analytical_Chemistry/Analytical_Chemistry_2.1_(Harvey)/05%3A_Standardiz
5.04%3A_Linear_Regression_and_Calibration_Curves
[2] Jeff Hughes's (RMIT University) - Regression Statistics for Analytical Determinations, Excel spreadsheet, 2002 (no lo

Orientação sobre a Validação de Métodos Analíticos DOQ-CGCRE-008 Revisão 05 Agosto 2016 (INMETRO)

Comparar equações de LD e LQ com: Planilha de validação: uma nova ferramenta para estimar figura
Test data set for standard addition (LD = 3,3s) Validação - Uma revisão baseada em documentos e normas (na
http://www.portalaction.com.br/validacao-de-metodologia-analitica
http://www.cnpsa.embrapa.br/met/images/arquivos/14MET/n
Traz a diferenças nas definições adotadas pelas várias normas:

Para definições mais detalhadas, consultar http://www.portalac

Validação de Metodologia Analítica (mais detalhado, baseado n
O LD pode ser expresso em unidades de resposta e é tomada n
http://www.portalaction.com.br/validacao-de-metodologia-analitica/115-limite-de-deteccao
O critério adotado pela RDC Nº 166 [1] é 3,3 (avaliado para 10 b
Concentra Area http://www.in.gov.b
0.24 8597.852 Usualmente os valores em escala de sinal são convertidos
0.24 8597.258 usando a função de calibração
0.24 8596.783
0.24 8596.908 então
0.24 8597.301
0.24 8597.496 III - Determinação baseada em parâmetros da curva analítica:
0.27 9607.39 Obs.: ver demonstração da esc
0.27 9607.714 http://www2.dbd.puc-rio.br/pe
0.27 9607.443 cópia em https://www.maxwell.v
0.27 9608.133 O desvio padrão pode ser obtido de 3 formas:
0.27 9607.176 a) o desvio padrão do intercepto com o eixo do Y de, no mínimo
0.27 9607.247 contendo concentrações do analito próximas ao suposto limit
0.3 10617.69 b) o desvio padrão residual da linha de regressão
0.3 10617.97 c) a partir da curva de calibração proveniente da análise de um
0.3 10617.99
0.3 10617.95 <---------- Um conjunto de dados de teste é apresentado à esquerda
0.3 10617.8 Com a planilha Calibration & LOD obtém-se LOQ = LQ e LOD x 1,1 = LD
0.3 10617.8 Basta trocar 3 por 3,3 em I23 e 10/3 por 10/3,3 em I24 para atender à RDC
0.33 11627.84
0.33 11628.43 Limit of Blank, Limit of Detection and Limit of Quantitation
0.33 11628.01 Clin Biochem Rev. 2008 Aug; 29(Suppl 1): S49–S52
 0.33 11628.33 60 replicates required to establish each parameter and 20 to verify them fo
0.33 11628.34 https://www.ncbi.nlm.nih.gov/pmc/articles/PMC2556583/
0.33 11628.1 Limit of Blank (zero concentration) as defined in EP17
0.36 12637.93 LoB = meanblank + 1.645(SDblank)
0.36 12638.12 Limit of detection
0.36 12638.54 LoD = LoB + 1.645(SDlow concentration sample)
0.36 12638.19 Limit of Quantification
0.36 12638.63 LoQ ≥ LoD
0.36 12638.32 LoQ is the lowest concentration at which the analyte can not only be reliabl
“Functional sensitivity” is defined as the concentration that results in a CV=
Data from Harvey's book, containing detailed explanaition of unweighted and weighted linear regression
0 0 Summary
0.1 12.36 resultados conicidem Limit of Blank (LoB), Limit of Detection (LoD), and Limit of Quantitation (L
0.2 24.83 LoB is the highest apparent analyte concentration expected to be found w
0.3 35.91 LoB = meanblank + 1.645(SDblank)
0.4 48.79 LoD is the lowest analyte concentration likely to be reliably distinguished
0.5 60.42 LoD = LoB + 1.645(SD low concentration sample)
LoQ is the lowest concentration at which the analyte can not only be relia
Data from Cristian's book
http://bcs.wiley.com/he-bcs/Books?action=resource&bcsId=8257&itemId=0470887575&resourceId=32610
ch16_standard_deviation_of_sample_concentrations.xls
Conc., ppmAbsorbance
0.100 0.081
0.200 0.171
0.500 0.432
1.000 0.857 resultado coincide
Sample (yc) 0.463 Sample = 0.540+/-0.005 ppm

Example of standard addition with variable volume for fluoride ISE by Zellmer
V0 100 mL slope 59.16
Cstd 300 mg F/L Response=(B$61+A64)*10^((B$64-B64)/E$61)
Vstd, mL E, mV Cstd, mg/LResponse Cstd+Cx log (Cstd+Cx)
0 80.3 0.00 #VALUE! 2.12 0.325618
0.3 71.1 0.90 #VALUE! 3.00 0.477248
0.6 64.6 1.79 #VALUE! 3.88 0.588838
0.9 59.7 2.68 #VALUE! 4.75 0.677067
1.2 55.1 3.56 #VALUE! 5.62 0.749969
resultados ao lado desvia da reta, slope 2x59

Another example (app for android)

Results are coincident for all parameters but LOD, sligthly smaller with Calibrateman
https://terpconnect.umd.edu/~toh/models/ErrorPropagation.pdf

Ver tb: Metodologia Estatística para Validação de Métodos Analíticos Aplicável à Metrologia em Química
Sônia Maria de Freitas, dissertação de mestrado, PUC-Rio, 2003
https://www.maxwell.vrac.puc-rio.br/4058/4058_1.PDF (e cap. subsequentes _2, _3, etc.)

Validation Workshop – Validation Overview - Aug. 24, 2005 at NFSTC

https://strbase.nist.gov/validation/ValidationOverview.pdf
 randomly pasted around
n Hughes's spreadsheet [2], and many parameters Limit of Detection (LOD): Formally LOD is defined as the concentration of analyte req
l, that includes ANOVA an t-test. signal equal to the background (blank) plus three times the standard deviation of the
ences [1] and [2] among others. When repeated measures of the blank (10 measurements, in accordance with IUPAC)
can be estimated from the regression line assuming sblank(standard deviation of the bl
sy/x (standard deviation of regression).
05%3A_Standardizing_Analytical_Methods/ This assumes errors are normally distributed for each point and the variances of each
and estimated by sy/x.
sheet, 2002 (no longer available). - If the intercept is positive the LOD = 3*sB/b.
- If the intercept is negative this quantity is added
to the x intercept i.e -a/b + 3*sB/b

NMETRO) http://www.inmetro.gov.br/Sidoq/Arquivos/Cgcre/DOQ/DOQ-Cgcre-8_05.pdf

para estimar figuras de mérito na validação de métodos analíticos univariados

entos e normas (nacionais e internacionais) LOD and LOQ im Voltammetry and Stripping Aanalysis IUPAC 1997
arquivos/14MET/normasinternacional.pdf http://www.cma4ch.org/chemo/ftp/6902x0297.pdf
las várias normas: Eurachem, IUPAC, MAPA, Anvisa, etc. Limite de detecção e de quantificação - diversas definições
http://www.inmetro.gov.br/credenciamento/docs%5CLimites_Sistemas_an
ttp://www.portalaction.com.br/validacao-de-metodologia-analitica
alhado, baseado na RDC 166/17 (ANVISA) de 2017: Resolução da Diretoria Colegiada)
posta e é tomada normalmente como três vezes o nível de ruído para as técnicas com contínua gravação (por exemplo a cromatografia).
(avaliado para 10 brancos ou brancos enriquecidos na concentração mais baixa aceitável)
ttp://www.in.gov.br/materia/-/asset_publisher/Kujrw0TZC2Mb/content/id/19194581/do1-2017-07-25-resolucao-rdc-n-166-de-24-de-julh
são convertidos para escala de concentração
1.1 - RDC 166/17 (ANVISA): Resolução da Diretoria Colegiada
1.1 - RDC 166/17 (ANVISA): Resolução da Diretoria Colegiada
pela RDC 166 1.1.1 - Seletividade
1.1.2 - Linearidade For an example of validation, with definition
a curva analítica: 1.1.3 - Precisão http://www.scielo.br/scielo.php?script=sci
monstração da escolha de 3,3s em tese da PUC-Rio 1.1.4 - Exatidão http://www.scielo.br/pdf/qn/v42n1/0100-
2.dbd.puc-rio.br/pergamum/tesesabertas/0116460_03_cap_03.pdf 1.1.5 - Limite de Detecção
ps://www.maxwell.vrac.puc-rio.br/4058/4058_4.PDF 1.1.6 - Limite de Quantificação
1.1.7 - Robustez
do Y de, no mínimo, 3 curvas de calibração construídas 1.2- Referências Bibliográficas
as ao suposto limite de quantificação;
Bibliography
da análise de um apropriado número de amostras do branco. https://en.wikipedia.org/wiki/Detection_limit
https://strbase.nist.gov/validation/ValidationOverview.pdf
http://www.rsc.org/images/robust-regression-brief-50_tcm18-216960.pdf
e LOD x 1,1 = LD (adota-se 3 s, ao invés de 3,3 s da RDC nº 166)
Robust regression: An introduction
ara atender à RDC 166/17 (ANVISA) Comparison of results from different methods by the t test see: https://www.
Confidence interval in Regression Analysis in Anal. Chem. http://www.sciel
Equations, see in a\did\Calibration curve, Linear Regression, errors (s) calcu
and a\bib\chemometrics\A statistical overview of univariate calibration
 Worksheets for Analytical Calibration Curves
0 to verify them forn an analytical procedure in the Clinical Laboratory
https://terpconnect.umd.edu/~toh/models/CalibrationCurve.html
(95% probability) Linear calibration with error calculation
Linear interpolation (bracket method)
Quadratic calibration with error estimation
Cubic calibration. Download Excel
Log-log calibration

not only be reliably detected but at which some predefined goals for bias and imprecision are met.
hat results in a CV=20% (or some other predetermined % CV)

t of Quantitation (LoQ) are terms used to describe the smallest concentration of a measurand that can be reliably measured by an analytica
ected to be found when replicates of a blank sample containing no analyte are tested.

ably distinguished from the LoB and at which detection is feasible. LoD is determined by utilising both the measured LoB and test replicates

an not only be reliably detected but at which some predefined goals for bias and imprecision are met. The LoQ may be equivalent to the LoD

Obtained here Outro exemplo de Adição de padrão

Write only in the whitStandard JChemEd.chem.wisc.edu Vol. 77 No. 9 September 2000, page 1241
A12 to B41; I11 to S1 Addition Error Analysis Using the Variance–Covariance Matrix Carl Salter
Determined resulMethod
Measureme yx 100
Number ofm 1
Predicted rxx 2.116534
Std dev of sxx 0.03824
Result's un(CI 95%) + 0.121603

REGRESSION PARAMETERS
100.2406 a intercept 1.109634
47.36075 b slope 0.508369
0.999654 r2 correlati 0.999827
Relative St 0.006715
Std dev of sy/x 1.429688
3*S(residua3*sy/x 4.289064
Limit of de 0.090562
Limit of qu 0.301872
number of n 5
two-tailed t 3.18
 Tipos de soluç
Branco de mé
analito, a ser
Branco para r
todos os proc
Branco de cam
amostragem.
Brancos não n
por traços de
análise, ou do
Química
http://www.ufj

For similar definitions in Brazil, see http://www.in.gov.br/materia/-/asset_publisher/Kujrw0TZC2Mb/content/id/19194581

ration of analyte required to give a Sxo (or Sxx) is the standard deviation of the value of x(xo) determined from
ard deviation of the blank (C analyte=0). the measured y (yo). For standard addition, yo is zero.
rdance with IUPAC) are unavailable it Sy/x is the standard deviation of regression, equal to the root of the sum of
deviation of the blank) is equal to squares of the errors (residuals) divided by n-2. sy / x
To get the 'error' (more correctly the 95% confidence interval, CI) multiply s xo  b {
e variances of each point are equal sxo by the t statistic (n-2 d of f, 95%).
This value is obtained from a 'lookup' table or as function (INVT in Excel,
e.g., cell I27).
The two-tailed t test is required for the interpolated results (n-2 d.o.f.),
compared to the one-tailed test for the mean.

Analytical function – the relation between a measured quantity x and the Curva Analítica é o
concentration c.
Analytical curve – a graphical plot of the analytical function, whatever the Curva analítica de c
UPAC 1997 coordinate axes used. NOMENCLATURE, S
Analytical calibration function – the measured quantity expressed as http://publications.iu
function of c[i.e., x= g(c)].
Analytical calibration curve – a graph of the analytical calibration function.
imites_Sistemas_analiticos.pdf
IUPAC’s Compendium of Analytical Nomenclature [15] (Section 10.3.3.2)
http://www.iupac.org/publications/analytical_compendium/
Pure Appl. Chem., 2008, Vol. 80, No. 1, pp. 85-104
a cromatografia).
Calibration plot – alternative designation of analytical calibration curve of
an ISE (Section 8.3.2.1).
-166-de-24-de-julho-de-2017-19194412
http://dx.doi.org/10.1351/pac200880010085
ANALYTICAL CALIBRATION CURVES (CALIBRATION PLOTS) FOR pH AND
OTHER ION-SELECTIVE ELECTRODES
Colegiada To have unified presentation of the calibration plots of all ISEs, it is
recommended that the cell emf is ascribed to the ordinate (vertical axis)
with more positive potentials at the top, and the pa A or pcA values be
ation, with definitions, see also plotted on the horizontal axis with increasing pa A or pcA values to the right.
cielo.php?script=sci_arttext&pid=S0100-40422019000100099
About LOD, etc. of ISEs see Recommendations for nomenclature of ions-
elective electrodes (IUPAC Recommendations 1994) R. P. Buck, E, Lindner
df/qn/v42n1/0100-4042-qn-42-01-0099-suppl01.pdf
https://publications.iupac.org/pac/66/12/2527/index.html

tcm18-216960.pdf

st see: https://www.slideshare.net/DrBrijKishorePrasaAnalytical Methods Committee, AMCTB No 50

m. http://www.scielo.org.za/pdf/sajc/v69/27.pdf
sion, errors (s) calculation .pdf Guia de Validação e Controle de Qualidade Analítica
te calibration
Curve.html

ured by an analytical procedure.

B and test replicates of a sample known to contain a low concentration of analyte.

quivalent to the LoD or it could be at a much higher concentration.

DATA INPUT
, page 1241 Conc. x Signal y Write only in the whitStandard
0 0.24 A12 to B41; I11 to S1 Addition
5.55 0.437 Determined resulMethod
11.1 0.621 Measureme yx 0.24
16.65 0.809 Number ofm 1
22.2 1.009 Predicted rxx 7.008691
Std dev of sxx 0.178697
Result's un(CI 95%) + 0.568693

REGRESSION PARAMETERS
0.2412 a intercept 0.003763
0.034414 b slope 0.000277
0.999806 r2 correlati 0.999903
Relative St 0.00844
Std dev of sy/x 0.004858
3*S(residua3*sy/x 0.014574
Limit of de 0.423484
Limit of qu 1.411613
number of n 5
two-tailed t 3.182446

Statistics in Analytical Chemistry: Part 7 of 52

Figure 1 - Plot of reported concentration for
measurements of samples spiked with known
concentrations (obtained from the
appropriate calibration curve) versus true
concentration. Legend: 1) solid line—45° line,
representing the situation where reported
concentration always equals true
concentration; 2) dashed line—reported-
 Statistics in Analytical Chemistry: Part 7 of 52
Figure 1 - Plot of reported concentration for
measurements of samples spiked with known
concentrations (obtained from the
appropriate calibration curve) versus true
concentration. Legend: 1) solid line—45° line,
representing the situation where reported
concentration always equals true
concentration; 2) dashed line—reported- Níveis de Precisão
versus-true line; 3) dotted line—uncertainty De acordo com a definição do IC
interval around the reported-versus-true line;
“A precisão de um método ana
4) dot—individual reported-versus-true
measurement for a given spiked sample. entre uma série de medidas ob
https://www.americanlaboratory.com/913-Technical- sob condições predeterminadas
Articles/1704-Statistics-in-Analytical-Chemistry-Part-7-A- precisão intermediária e reprod
Review/

A incerteza nas medições e diluições envolvidas no preparo de soluções para calibração

pode ser estimada pela equação

Desvendando a medição nos ensaios químicos: 1. A curva analítica ou de calibração

http://doi.editoracubo.com.br/10.4322/sc.2011.015 DOI: 10.4322/sc.2011.015
Mais detalhes em http://www2.dbd.puc-rio.br/pergamum/tesesabertas/0116460_03_cap_06.pdf

Tipos de solução em branco usadas em métodos de calibração e determinação:

Branco de método: é uma amostra que contém todos os constituintes exceto o
analito, a ser usada durante todas as etapas do procedimento analítico.
Branco para reagente: é semelhante ao branco de método, mas sem ter passado por
todos os procedimentos de preparo de amostra. Análise de variância
Branco de campo: branco de método que, adicionalmente, foi exposto ao local de
amostragem. Fonte: BARROS NETO; SCARMINIO; B
Brancos não nulos podem indicar interferência de outras espécies ou contaminação
por traços de analito presentes nos reagentes usados na preservação, preparação e
análise, ou dos recipientes e instrumentos anteriormente expostos ao analito.
http://www.ufjf.br/nupis/files/2010/09/M%c3%a9todos-de-Calibra%c3%a7%c3%a3o.pdf

ntent/id/19194581/do1-2017-07-25-resolucao-rdc-n-166-de-24-de-julho-de-2017-19194412
No teste de significância da regre
a média quadrática residual (MQr
MQR e MQr a um certo nível de co
maior que o valor de F, conclui-se
O teste de falta de ajuste compara
(MQfaj) com a média quadrática d
valor do ponto de distribuição F, r
confiança, temos um modelo sem
 sy / x ( yo  y ) 2
s xo  { m1  1
 }0.5
b n b 2
 ( xi  x ) 2

urva Analítica é o nome recomendado pela IUPAC (e.g., em 1975)

urva analítica de calibração é o nome do gráfico correspondente
OMENCLATURE, SYMBOLS, UNITS AND THEIR USAGE IN SPECTROCHEMICAL ANALYSIS—Il.DATA INTERPRETATION
ttp://publications.iupac.org/pac/pdf/1976/pdf/4502x0099.pdf
1975
om a definição do ICH:
de um método analítico expressa a proximidade do acordo (grau de dispersão)
érie de medidas obtidas de múltiplas amostragens da mesma amostra homogênea
es predeterminadas. A precisão deve ser considerada em três níveis: repetibilidade,
rmediária e reprodutibilidade.”

(ANOVA)
NETO; SCARMINIO; BRUNS, Como Fazer experimentos – Pesquisa e Desenvolvimento na Ciência e na Indústria, 2010
gnificância da regressão, compara-se a razão entre a média quadrática devido à regressão (MQ R) e
ática residual (MQr) com o valor do ponto da distribuição F referente aos graus de liberdade de
um certo nível de confiança estatística, que é em geral, 95% de confiança. Se a referida razão for
alor de F, conclui-seque a regressão é estatisticamente significativa.
a de ajuste compara, para os vários níveis da variável X, a média quadrática da falta de ajuste
média quadrática devida ao erro puro (MQ ep). Assim, se a razão (MQfaj)/(MQep) for menor que o
de distribuição F, referentes aos graus de liberdade de MQ faj e MQep e para um nível de 95% de
mos um modelo sem falta de ajuste.