SIST-TP CEN/CR 10320:2004
(Corrigendum)Optical emission analysis of low alloy steels (routine method) - Method for determination of C, Si, S, P, Mn, Cr, Ni and Cu
Optical emission analysis of low alloy steels (routine method) - Method for determination of C, Si, S, P, Mn, Cr, Ni and Cu
This document specifies an optical emission spectrometry spark source routine standard method for multi-element analysis of unalloyed steel and iron.
Optische Emissionsanalyse von niedriglegierten Stählen (Reihenanalyse) - Verfahren zur Bestimmung von C, Si, S, P, Mn, Cr, Ni und Cu
Analyse des aciers faiblement alliés par spectrométrié d´émission optique (méthode de routine) - Méthode de détermination de C, Si, S, P, Mn, Cr, Ni et Cu
Optična emisijska analiza malolegiranih jekel (rutinska metoda) – Metoda za določevanje C, Si, S, P, Mn, Cr, Ni in Cu
General Information
Relations
Standards Content (Sample)
SLOVENSKI STANDARD
01-december-2004
2SWLþQDHPLVLMVNDDQDOL]DPDOROHJLUDQLKMHNHO UXWLQVNDPHWRGD ±0HWRGD]D
GRORþHYDQMH&6L630Q&U1LLQ&X
Optical emission analysis of low alloy steels (routine method) - Method for determination
of C, Si, S, P, Mn, Cr, Ni and Cu
Optische Emissionsanalyse von niedriglegierten Stählen (Reihenanalyse) - Verfahren zur
Bestimmung von C, Si, S, P, Mn, Cr, Ni und Cu
Analyse des aciers faiblement alliés par spectrométrié d´émission optique (méthode de
routine) - Méthode de détermination de C, Si, S, P, Mn, Cr, Ni et Cu
Ta slovenski standard je istoveten z: CR 10320:2004
ICS:
77.040.30 Kemijska analiza kovin Chemical analysis of metals
77.080.20 Jekla Steels
2003-01.Slovenski inštitut za standardizacijo. Razmnoževanje celote ali delov tega standarda ni dovoljeno.
CEN REPORT
CR 10320
RAPPORT CEN
CEN BERICHT
September 2004
ICS 77.040.20; 77.140.20
English version
Optical emission analysis of low alloy steels (routine method) -
Method for determination of C, Si, S, P, Mn, Cr, Ni and Cu
Analyse des aciers faiblement alliés par spectrométrié Optische Emissionsanalyse von niedriglegierten Stählen
d´émission optique (méthode de routine) - Méthode de (Reihenanalyse) - Verfahren zur Bestimmung von C, Si, S,
détermination de C, Si, S, P, Mn, Cr, Ni et Cu P, Mn, Cr, Ni und Cu
This CEN Report was approved by CEN on 3 June 2001. It has been drawn up by the Technical Committee ECISS/TC 20.
CEN members are the national standards bodies of Austria, Belgium, Cyprus, Czech Republic, Denmark, Estonia, Finland, France,
Germany, Greece, Hungary, Iceland, Ireland, Italy, Latvia, Lithuania, Luxembourg, Malta, Netherlands, Norway, Poland, Portugal, Slovakia,
Slovenia, Spain, Sweden, Switzerland and United Kingdom.
EUROPEAN COMMITTEE FOR STANDARDIZATION
COMITÉ EUROPÉEN DE NORMALISATION
EUROPÄISCHES KOMITEE FÜR NORMUNG
Management Centre: rue de Stassart, 36 B-1050 Brussels
© 2004 CEN All rights of exploitation in any form and by any means reserved Ref. No. CR 10320:2004: E
worldwide for CEN national Members.
Contents page
Foreword.3
1 Scope .4
2 Test sample preparation .4
3 Calibration of the instrument.4
3.1 Determination of calibration curve .4
3.2 Linear correlation.5
3.3 Quadratic correlation .6
3.4 Confidence limit (Syc) and standard error (Syx) in a linear correlation .7
3.5 Confidence limit (Syc) and standard error (Syx) in a quadratic correlation.7
3.6 Correlation accuracy .8
4 Calculation of interferences .8
4.1 Spectral interferences .8
4.2 Matrix interferences.10
4.3 Calculation of an interfered element due to either matrix effect or spectral interference.10
5 Determination of performance criteria .11
5.1 Determination of background equivalent concentration BEC .11
5.2 Determination of detection limit DL.11
5.3 Determination of repeatability as relative standard deviation RSD .13
5.4 Determination of accuracy SEA .13
6 Performing analysis .13
7 Ordinary maintenance.14
8 Quality control.14
8.1 Control chart .14
8.2 Calibration curve control .15
9 Examples .15
9.1 Determination of calibration curve .15
9.2 Calculation of interferences .20
9.3 Determination of background equivalent concentration.25
9.4 Determination of detection limit.26
9.5 Determination of repeatability.27
9.6 Determination of accuracy.29
9.7 Ordinary maintenance.31
9.8 Control CHART.31
9.9 Calibration curve control .34
10 Statistical results .35
Annex A Optical emission spectrometry.37
Bibliography .60
Foreword
This document (CR 10320:2004) has been prepared by ECISS /TC 20, "Methods of chemical analysis of ferrous
products".
1 Scope
This document specifies an optical emission spectrometry spark source routine standard method for multi-
element analysis of unalloyed steel and iron.
2 Test sample preparation
Prepare reference materials, test samples and setting up samples, by grinding to provide a uniform flat surface.
It is recommended a surface grinder with aluminium oxide, or zirconium oxide, abrasive to be employed with a
coarse grit of 60 - 100.
3 Calibration of the instrument
3.1 Determination of calibration curve
Analyse a series of reference materials (min. 5) for the element intensity/matrix intensity ratio according to the
pattern shown in Figure 1.
Calculate the average intensity for each reference material.
NOTE In (Figure 1) the first group values into parentheses refers to odd ns.
Correlate the average intensity ratio with the concentration of test element.
n
i
(1)
cA%= I
∑
i
i=0
where:
A means correlation constants;
i
i
I means unknown values (intensity ratio).
Figure 1
3.2 Linear correlation
For linear correlations, constants are calculated in the following way:
Ay=−Ax
(2)
01mm
()xy
∑
A = (3)
()x
∑
where:
x and y are average values (Σi /n);
m m m
n number of points;
x and y are intensity ratio readouts and concentrations, respectively, of test samples.
Besides correlation parameters, calculate the following:
[]n xy− x y
∑∑∑
r =
(4)
2 2
2 2
[]n ()x −()x ×[]n ()y −()y
∑∑ ∑∑
where r is the correlation coefficient.
A and A may also be calculated by single values according to the following equations:
0 1
yx − xy x
() ( )
∑∑ ∑ ∑
A = (5)
nx − x
()
()
∑ ∑
yA− ()x
∑∑1
A = (6)
n
nxy − x y
()
∑ ∑∑
A =
(7)
nx − x
()
()
∑ ∑
3.3 Quadratic correlation
Calculate constants in the following way, by resolving the system:
2 3 42
Ax++A x A x= xy (8)
() () () ()( )
∑ ∑ ∑ ∑
0 1 2
2 3
Ax++A ()x A (x)= (xy) (9)
∑ ∑ ∑ ∑
01 2
An++A x A ()x= y (10)
∑ ∑ ∑
01 2
where:
n is the number of tests.
cc−
()
∑ it ic
r =−1 (11)
()cc−
∑
it a
where:
c is the true value of the reference material i;
it
c is the value of the reference material i read on the calibration curve;
ic
c is the average concentration of all the c
a ic
3.4 Confidence limit (Syc) and standard error (Syx) in a linear correlation
Confidence limit indicates the area where the true value of y for a given x lies at 95 % of probability. In other
words, it indicates the area including the regression range. The slope error is an evidence of the method
sensitivity, in fact the wider x range the lower the slope error. The standard error in a correlation is the extent of
the deviation around the regression line. Calculate the standard error in a linear correlation by the following
formula:
yA−−yA xy
() ( )
∑ ∑∑
Syx = (12)
n− 2
and the confidence limit:
()x −x
i m
Syc=Syx× +
(13)
n
()x
∑
Calculate the slope error (Sb) as follows:
Syx
Sb= (14)
()x
∑
3.5 Confidence limit (Syc) and standard error (Syx) in a quadratic correlation
Calculate the standard error in a quadratic correlation by the following formula:
yA−−yA xy−A ()xy
() ( )
∑ 01∑∑∑ 2
Syx= (15)
n−2
and the confidence limit:
1()x −x
i m
Syc Syx
= × +
(16)
n
()x
∑
Calculate the slope error (Sb) as follows:
Syx
Sb= (17)
x
()
∑
3.6 Correlation accuracy
Among reference standards, either secondary or tertiary, select a low, a medium and a high standard and
determine their repeatability both in concentration and intensity ratio. The resulting values indicate the method
accuracy index at the specific concentration (intensity ratio).
If RSD is correlated with concentration, a change in repeatability is obtained with respect to concentration.
4 Calculation of interferences
4.1 Spectral interferences
In the line intensity/matrix ratio, analyse at least 5 reference materials on the interference line with a variable
concentration of the interfering element but possibly free from the interfered element. Analyses should be
repeated at least four times after performing the drift correction, according to the Figure 2 scheme. Calculate the
average intensity ratio for each reference material and compute the corresponding concentration of interfered
element.
Figure 2
Correlate the resulting concentrations to the concentration of the interfering element by the method of least
squares. The following equation is obtained:
n
i
cA% = c (18)
∑ ii
i=0
where:
c% is the concentration of the interfered element;
c is the concentration of the interfering element;
i
A is the correlation constants (it is advisable to calculate a straight line).
i
The angle coefficient of the straight line (A ) is the factor of spectral interference of the element interfering over
the analyte.
This factor is nonlinear only rarely and for wide concentration ranges of the interfering and/or interfered
elements. In this case it is advisable to work with analyte families and limit concentration ranges.
The following equation calculates the analyte concentration:
n k
i
cA%=+I fc (19)
∑∑i jj
i==00j
where:
f means the factor of spectral interference of element j over the analyte;
j
c means its concentration.
j
It is possible to calculate interference factors in intensity. In this case correlations must be made with intensities.
The following formula is used to calculate f easier and when only one interfering element has been detected:
j
n
i
cA%=+Ifc (20)
∑ i 11
i=0
n
i
cA%− I
∑ i
i=0
f = (21)
c
Suitable reference materials must be selected that have a variable concentration of the interfering element and
that have already been analysed during the calibration curve construction.
4.2 Matrix interferences
Having established that matrix interferences are minimized by subdividing test samples into spectral families, it
is however possible to calculate them by selecting at least 5 reference materials with a composition as close as
possible to the matrix, the interfering and the interfered element in variable concentration. Analyse materials
according to the previous scheme and calculate interference factors by the following equation:
n k
i
cA%=+I 1 fc (22)
∑∑i tt
i==00t
where:
f is the interference factor of element t over the analyte;
t
c is its concentration.
t
4.3 Calculation of an interfered element due to either matrix effect or spectral interference
The concentration of an interfered element, due to either matrix effect or line overlap, is calculated by the
following formula:
n k s
i
cA%=+I 1 fc +fc for t ≠ j (23)
∑∑ ∑
i tt jj
i==00t j=0
5 Determination of performance criteria
5.1 Determination of background equivalent concentration BEC
The background equivalent concentration corresponds to a net intensity whose value is equal to background
intensity. It is calculated as follows:
a) Calculate calibration curve;
b) Determine background intensity, i.e. the intensity corresponding to the intercept between curve and axis of
abscissas (I );
c) Multiply I by 2 (I x 2 = I );
0 0 B
d) With correlation parameters, calculate the concentration equivalent to I . The resulting value corresponds
B
to the BEC of the test element.
In a linear concentration, BEC is equal to the known term multiplied by -1.
5.2 Determination of detection limit DL
The detection limit may be assessed in two different ways according to whether the instrument DL for a given
analyte line or the method DL. is to be calculated
5.2.1 Detection limit of the method
5.2.1.1 Base straight line curve
a) Record the element intensity/matrix ratio for a series of reference materials (min. 4) in a matrix pattern as in
the following example:
RefMat 1 2 3 4
Conc. C C C C
1 2 3 4
Int. I I I I
11 21 31 41
I I I I
12 22 32 42
..........................................................................
I I I I
16 26 36 46
b) Calculate:
RefMat 1 2 3 4
Conc. C C C C
1 2 3 4
Int. I I I I
11 21 31 41
I I I I
12 22 32 42
..........................................................................
I I I I
16 26 36 46
Σ ΣI ΣI ΣI ΣI
1i 2i 3i 4i
mean I I I I
1m 2m 3m 4m
2 2 2 2 2
Σ (I ) Σ (I ) Σ (I ) Σ (I ) Σ (I )
ni 1i 2i 3i 4i
c) Calculate Sp (pooled estimate of standard deviation) according to the following formula:
2 2 2
I I I
() () ()
∑∑ ∑
2 1 2 k
I−+ +⋅⋅+
()
∑ i
n n n
1 2 k
2
Sp = (24)
N−k
where:
I represents all readings;
i
n is the number of readings for each reference material;
i
N is the total number of readings;
k is the number of test samples.
d) From the equation of the calibration curve
cA%=+IA (25)
calculate the DL for a discharge:
DL= A ×t×Sp
(26)
where:
t is the t of Student at 95 % for N - k degrees of freedom.
If r discharges are striken during an analysis:
DL
DLr = (27)
r
where DLr is the limit of determination.
5.2.1.2 Quadratic base curve
Calculate a straight line with the lowest 5 points in the curve and compute as per point 5.2.1.1.
5.2.2 Instrument detection limit DLi
Perform 10 analyses on a reference material with an analyte concentration equal to the background. Calculate
the standard deviation of the resulting concentrations. DL is obtained from:
DLi = 3× 2×s
(28)
where:
s is standard deviation of concentrations
5.3 Determination of repeatability as relative standard deviation RSD
Over a short time period, perform 10 analyses on a reference material without making any drift correction
between an analysis and the other. The instrument repeatability RSD, with a 95 % confidence limit, is expressed
by:
d
()
∑ i
RSD= 2 (29)
21n−
()
where:
d is the difference between a given value (intensity or concentration) and the average value of it;
i
n is the number of analyses.
Calculate repeatability for at least 3 reference materials with low, medium and high concentration within the
calibration curve validity range.
5.4 Determination of accuracy SEA
Perform at least 10 analyses, each time making a drift correction, with a reference material belonging to a
specific analyte family that has not been used in the calibration curve construction. The spectrometer accuracy,
with a 95 % confidence limit, is expressed by:
()d
∑
i
SEA= 2 (30)
21nI−−
()
where:
d is the difference between calculated and true concentration;
i
n is the number of analyses;
l is the degree of the calibration curve polynomial.
Calculate the instrument accuracy for at least 3 reference materials with low, medium and high concentration
within the calibration curve validity range.
6 Performing analysis
Place the sample on the sample table and position the ground surface over the counter electrode.
Excite the sample. At the end of the instrument analysis cycle, record the reading (if the intensity ratio is read),
or the concentration % m/m.
Reanalyze the sample, average the duplicate readings for each element.
Excite the samples in two opposite sides of the surface, half way between the center and the end of the grinded
surface.
7 Ordinary maintenance
It is suggested to carry out ordinary maintenance as per following schedule:
Table 1 — Maintenance schedule
Action Frequency
1.1
Check gas At least once a day
1.2
Check fluid At least once a day
1.3
Check parameters At least once a day
1.4
Clean stand Every 200 analysis, however once a day
1.5
Clean filter Every 200 analysis, however once a day
1.6
Sharpen point Every 15000 analysis
1.7
Clean lenses Every 15000 analysis
1.8
Check profile
1.8.1
Startup After 2 hours and every 8 hours during first week
1.8.2
Steady state Once a week during first month, then at least once a month.
Switching off and opening the spectrometer must be avoided, particularly in dusty environments and by
personnel other than maintenance staff.
During startup, profile must be checked at least against three wavelengths, low, medium and high respectively,
one of which should come from the matrix element, so that the whole spectrometer´s range is covered.
Subsequently, it will be sufficient to check profile against the analyte line of the matrix element.
Startup occurs when the spectrometer has been switched off for any reason and it is restarted after more than 8
hours. The experience suggests that for each hour off, one hour on is required to reach steady-state operation.
After 8 hours off, at least 24 hours on are required for steady state.
Frequencies as provided in the above table can be changed according to the manufacturer's instructions.
8 Quality control
8.1 Control chart
Soon after an instrumental drift correction is performed and then no later than every 4 hours, analyse reference
materials that have not been used in calibration curve construction. Record the test value of each certified
element on control charts as in the examples 9.8.
Acceptance limits are calculated from the correlation RDS vz. % of each element and concentration level.
Higher acceptance limit derives from nominal value + RSD, lower acceptance limit derives from nominal value -
RSD. The range +/- RSD is subdivided into 4 parts: the first quarter above nominal value is the higher control
limit, the first quarter below nominal value is the lower control limit.
The selected reference element is used for 2 analysis. The following cases may occur:
a) the average value lies within the HCL-LCL range;
b) the average value lies outside the above range, but within the HAL-LAL range;
c) the average value lies outside acceptance limits.
In case a) the instrument is calibrated and can be used for the following analysis.
In case b) it is advisable to carry out 2 analysis and record where the new average value lies. If the average
value lies as in point a), the instrument is calibrated. If it lies as in point b), two situations may occur:
b.1) the average value lies in the same area as formerly measured: the result is acceptable, but the risk exists
of one, though tolerable, systematic difference;
b.2) the average value lies in the opposite area: it is advisable to check the parameters affecting the
instrument repeatability, namely:
- point position and status;
- gas pressure;
- test surface;
- spectrometer temperature;
- profile;
- phototube stability;
- source stability.
Case c) means that the instrumental drift has probably not been corrected adequately. However, it is advisable
to repeat the analysis before correcting the instrumental drift.
8.2 Calibration curve control
Calibration curves in optical emission spectrometers tend to change over time. As an average, their validity as
established by experience is approx. 2 years. It is therefore advisable to check them every year. To this purpose,
a minimum 3 reference materials should be used, with low, medium and high analyte concentrations
respectively, that have undergone repeatability tests during calibration curve construction. After correcting the
instrumental drift, their intensity ratio should be measured. If the average value of 2 analysis per reference
material lies in the range Ia +/- 2s, where Ia is the average intensity ratio as obtained during the calibration
curve construction and s is the standard deviation from repeatability, the curve is still valid. If the average value
of 2 analysis with one refererence material lies outside the range Ia +/- 2s, the calibration curve is no longer
valid. It is therefore necessary to establish a new curve. The calculated values of I should be recorded on
calibration card, as in the example 9.9.
9 Examples
9.1 Determination of calibration curve
Table 2
Drift control Drift control Drift control
RefMat I RefMat I RefMat I
ii ii ii
1 13,23 5 30,031 13,56
2 18,83 4 26,865 30,48
3 16,07 3 16,012 19,60
4 27,09 2 18,824 26,22
5 30,13 1 13,383 16,06
Table 3
Drift control Drift control Drift control
RefMat I RefMat I RefMat I
ii ii ii
1 13,48 5 30,603 16,22
2 18,87 4 27,614 26,68
3 16,13 3 16,122 18,58
4 26,12 2 18,815 29,73
5 30,00 1 13,321 13,40
Table 4
(1) (2) (3) (4) (5)
RefMat NBS 1265 BCS 405/1 BCS 431 NBS 1270 BCS 456/1
C % 0,0067 0,0340 0,0190 0,0770 0,1010
I 13,23 18,83 16,07 27,09 30,13
ii
13,38 18,82 16,01 26,86 30,03
13,56 19,60 16,06 26,22 30,48
13,48 18,87 16,13 26,12 30,00
13,32 18,81 16,12 27,61 30,60
13,40 18,58 16,22 26,68 29,73
average 13,40 18,92 16,10 26,76 30,16
80,37 113,51 96,61 160,58 180,97
Σm I
ii
2 1076,62 2148,03 1555,61 4299,21 5458,88
Σm I
ii
average C % = 0,0475
average I = 21,07
ii
Table 5
x 2 3 4
x x x
13,40 179,562406,10 32241,79
18,92 357,976772,72128139,94
16,10 259,214173,28 67189,82
26,76 716,1019162,77512795,77
30,16 909,6327434,31827418,73
105,34 2422,46 59949,19 1567786,07
Σ
Table 6
y xy 2 2
xy y
0,0067 0,08978 1,2030520,000045
0,0340 0,64328 12,1708580,001156
0,0190 0,30590 4,9249900,000361
0,0770 2,06052 55,1395150,005929
0,1010 3,04616 91,8721860,010201
Σ 0,2377 6,14564 165,3106 0,017692
9.1.1 Calculation of a straight line correlation and its standard error
yA=+Ax (31)
yx − xy x
() ( )
∑∑ ∑ ∑
A = (32)
0 2
nx − x
()
()
∑ ∑
nxy − x y
()
∑∑∑
A = (33)
nx − x
()
()
∑ ∑
A = 0,0056005
A = 0,0704511
[]n xy− x y
∑∑∑
r =
(34)
2 2
2 2
[]() ()[]() ()
n x − x × n y − y
∑∑ ∑∑
r = 0,9969434
yA−−yA xy
() ( )
∑ 01∑∑
Syx = (35)
n− 2
(Syx) = {[0,017692 + 0,0704511 × 0,2377 - 0,0056005 × 6,1456]/(5-2)}
= 0,00000659789
Sxy = 0,002569
9.1.2 Calculation of a quadratic correlation and its standard error
yA=+Ax+Ax (36)
01 2
Resolve the following system:
2 3 42
Ax++A x Ax= x y (37)
() ( ) () ( )( )
0∑ 1∑ 2∑ ∑
2 3
Ax++A x A x= xy (38)
() ( ) ( )
∑ ∑ ∑ ∑
01 2
An++A x A x= y (39)
()
∑ ∑ ∑
01 2
where:
n is the number of data pairs
cc−
()
∑ it ic
r =−1 (40)
cc−
()
∑ it a
A = -0,0378397
A = 0,00234735
A = 0,0000741513
C 0,00693
ic:
0,03312
0,01917
0,07808
0,10041
C 0,047542
a:
r= [1 - (0,0000023707/0,00639163)] (41)
r = 0,99963
()yA−−yA (xy)−A xy
∑ ∑ ∑ ∑
01 2
Syx= (42)
n−2
(Syx) = {[0,017692 + 0,0378397 × 0,2377 – 0,00234735 × 6,14564 - (43)
- 0,0000741513 × 165,3106]/(5-2)}
Syx = 0,00094
9.1.3 Confidence limit and slope error in a straight line correlation
1()x −x
i m
Syc = Syx× +
(44)
n
()x
∑
Syx
Sb= (45)
x
()
∑
Table 7
x = x - x 2
i m (x -x )
i m
-7,67 58,8289
-2,15 4,6225
-4,97 24,7009
5,69 32,3761
9,09 82,6281
Σx = 203,1600
2 2
(Syc) = 0,00000659789 × [(1/5) + (-7,67) /203,16] = 0,00000323
Syc = 0,0018
Calculate Syc for all x .
i
(Sb) = 0,00000659789/203,16 = 0,0000000325
Sb = 0,00018
9.1.4 Confidence limit and slope error in a quadratic correlation
1()x −x
i m
Syc =Syx× +
(46)
n ()
x
∑
Syx
Sb= (47)
x
()
∑
(Syc) = 0,00000088855 ×{1/5,+,[-7,67 /203,16]}
(Syc) = 0,000000435
Syc = 0,0006596
Calculate Syc for all x .
i
(Sb) = 0,0000000043736
Sb = 0,00006613
9.1.5 Correlation accuracy
The data stemming from the following example 9.5 can be used. The standard deviation from repeatability of
concentration and intensity ratio is calculated in three reference materials together with their coefficient of
variation percent:
Table 8
Intensity concentration
Ref. mat. s cv% s cv%
I c
NBS 1265 0,155 1,16 0,00087 20,92
BCS 431 0,382 2,43 0,00215 12,21
BCS 456/1 0,609 2,07 0,00342 3,61
If a linear correlation is calculated with s values as a function of concentration, the resulting parameters show
c
how the instrument accuracy changes according to the analyte concentration. In this case:
s = 0,022643 c% + 0,00118303
c
9.2 Calculation of interferences
9.2.1 Line interference
Interference of molybenum on arsenic (analytical line 197,26 nm).
Four samples are considered having a variable molybdenum content and no arsenic. Analyse the apparent
arsenic concentration (in other words, analyse them against an analytical program containing arsenic and
record its concentration):
Table 9
Sample Mo % As % on curve readout
1 1,00 0,0068
2 2,00 0,0122
3 5,00 0,0331
4 10,00 0,0652
Correlating the % of arsenic readout with the molybdenum concentration; the angle coefficient of the straight
line is 0.0065. Therefore the arsenic correction factor with molybdenum is:
Mo
f =−0.0065 (48)
As
Let' s now consider the following equation:
n k
i
cA%=+I fc (49)
∑∑i jj
i==00j
In this case, there is one interfering element and the arsenic concentration in test materials is zero (c% = 0,
while c ranges from 1 to 10 %). Therefore:
n
i
cA%=+Ifc (50)
∑ i 11
i=0
n
i
cA%− I
∑
i
i=0
f = (51)
c
Table 10
Mo
Sample Calculation
f
As
0− 0,0068
1 -0,0068
0− 0,0122
2 -0,0061
0− 0,0331
3 -0,0066
0− 0,0652
4 -0,0065
Calculating the average of four values yields:
Mo
f = -0,0065 ± 0,0003 (52)
As
Obvioulsy with true samples c% is different from zero. Still the calculation technique is the same.
9.2.2 Matrix interference
This kind of interference does not concern with unalloyed steel. Just as an example Nickel interference over
manganese in stainless steels is showed. From the equation
n k
i
cA%=+I 1 fc (53)
∑∑i tt
i==00t
Detect reference materials having a single interfering element where:
n
i
cA%=+I 1fc (54)
()
∑
i 11
i=0
from which f is calculated:
n
i
c%− AI
∑ i
i=0
f =
n
(55)
i
AI ×c
∑ i 1
i=0
Table 11
Test sample Mn % Ni % Intensity
1 0,57 0,36 939
2 0,15 0,18 303
3 0,41 0,37 703
4 1,02 1,95 1664
5 0,50 0,06 810
6 0,38 0,21 639
7 0,42 2,12 755
8 0,598 0,25 959
9 0,38 0,40 686
10 0,91 2,16 1542
11 0,42 0,18 739
12 0,87 9,52 1651
13 0,77 20,70 1870
14 1,66 21,16 3565
15 1,10 20,10 2471
16 1,28 20,50 2815
17 1,17 20,60 2546
Using the above values to calculate a linear regression with the first 11 test samples yields:
Mn % = 0,000627732 I – 0,0323165
r = 0,9946
Calculate the factors for each of the remaining 6 samples:
Table 12
Ni
Sample Calculation
f
Mn
0,87−1,004
12 -0,0140
1.004⋅ 9,52
0,77−1,142
13 -0,0157
1,142⋅ 20,70
1,66− 2,206
14 -0,0117
2,206⋅ 21,16
1,10−1,519
15 -0,0137
1,519⋅ 22,10
1,28−1,735
16 -0,0128
1,735⋅ 20,50
1,17−1,566
17 -0,0123
1,566⋅ 20,60
By the average of the six values. the following is obtained:
Ni
f = -0,0134 ± 0,0013 (56)
Mn
If the standard deviation is ≤ 10 % of the factor, this can be considered as acceptable.
9.2.3 Calculation of the corrected curve
Once the interference factors are known, calculate the apparent concentrations (those that should be read in an
ideal curve) and correlate the different intensities to produce the corrected calculation of the correlation. Starting
from the equation:
n
i
cA%=+I 1fc (57)
()
∑
ii 11
i=0
The apparent concentration results from:
n
c%
i
AI = = apparent Mn (%) (58)
∑ ii
1+fc
i=0 11
As an example the computation of the sample 16 is:
1,28
= 1,765
apparent Mn (%) =
1− 0,0134× 20,50
Construct the following table:
Table 13
Test sample Mn (%) apparent Mn (%) Intensity
1 0,57 0,572 939
2 0,15 0,150 303
3 0,41 0,412 703
4 1,02 1,047 1664
5 0,50 0,500 810
6 0,38 0,381 639
7 0,42 0,433 755
8 0,598 0,600 959
9 0,38 0,382 686
10 0,91 0,937 1542
11 0,42 0,421 739
12 0,87 0,997 1651
13 0,77 1,066 1870
14 1,66 2,317 3565
15 1,10 1,506 2471
16 1,28 1,765 2815
17 1,17 1,616 2546
Curves as calculated with these new values are:
Base curve:
Mn (%) = 0,000650869 I – 0,0458012
r = 0,9967
Curve with all samples:
Mn (%) = 0,000654296 I – 0,0581408
r = 0,9987
9.3 Determination of background equivalent concentration
9.3.1 Straight line curve
From the equation
n
i
(59)
cA%= I
∑ i
i=0
If A = - 0,0704511 and A = 0,0056005 (see example)
0 1
BEC = (-0,0704511) ×(-1) = 0,0704511
9.3.2 Quadratic base curve
From the equation
n
i
(60)
cA%= I
∑ i
i=0
If A = -0,0378397, A = 0,00234735 and A = 0,0000741513
0 1 2
(see example)
For y = 0, x = 11,75
BEC = 0,0584 = A × (11,75 × 2) +A × (11,75 × 2) + A
2 1 0
9.4 Determination of detection limit
9.4.1 Determination of the detection limit of the method
Table 14
RefMat NBS 1265 BCS 405/1 BCS 431 NBS 1270 BCS 456/1
C % 0,0067 0,0340 0,0190 0,0770 0,1010
I 13,23 18,83 16,07 27,09 30,13
ii
13,38 18,8216,0126,86 30,03
13,56 19,6016,0626,22 30,48
13,48 18,8716,1326,12 30,00
13,32 18,8116,1227,61 30,60
13,40 18,5816,2226,68 29,73
Mean 13,40 18,9216,10 26,76 30,16
80,37 113,51 96,61 160,58 180,97
Σ I
ii
2 1076,62 2148,03 1555,61 4299,21 5458,88
Σ I
ii
2 6459,34 12884,52 9333,49 25785,94 32750,14
(Σ I )
ii
Tot Σ I 632,04
ii
Tot Σ I 14538,35
ii
Tot(Σ I ) 87213,43
ii
2 2 2
I I I
() () ()
2∑∑1 2 ∑ k
()I−+ +⋅⋅+
∑ i
n n n
1 2 k
2
(61)
Sp =
Nk−
2 2 2
II I
() () ()
2∑∑11 22 ∑ 55
I−+ +⋅⋅+
()
∑
ii
66 6
2
(62)
Sp =
30− 5
6459,34 12884,52 9333,49 25785,93 32750,14
14538.35− + + + +
6 6 6 6 6
2
Sp =
(63)
14538,35−14535,57
Sp =
(64)
Sp = 0,11
Sp = 0,33
t = 2,06
Straight line equation C% = 0,00560 × I – 0,0704511
DL = 0,33 × 0,005601 × 2,06 = 0,00381
9.4.2 Determination of the detection limit of the instrument
DL = 3× 2×s
= 0,00368 (65)
where:
s is standard deviation of background repeatability = 0,000868
9.5 Determination of repeatability
Table 15 — Standard NBS 1265. Carbon
I C dI dC 2 2
i i dI dC
13,40 0,0046 0,07 0,0004 0,0049 0,00000016
13,30 0,0040 -0,03 -0,0002 0,0009 0,00000004
13,20 0,0035 -0,13 -0,0007 0,0169 0,00000049
13,30 0,0040 -0,03 -0,0002 0,0009 0,00000004
13,10 0,0029 -0,23 -0,0013 0,0529 0,00000169
13,20 0,0035 -0,13 -0,0007 0,0169 0,00000049
13,40 0,0046 0,07 0,0004 0,0049 0,00000016
13,60 0,0057 0,27 0,0015 0,0729 0,00000225
13,50 0,0052 0,17 0,0010 0,0289 0,00000100
13,30 0,0040 -0,03 -0,0002 0,0009 0,00000004
13,33 0,0042 0,2010 0,00000636
( dl )
∑
RSD = 4×
(66)
2×()n−1
= 4 × 0,201 / 18 = 0,0447
RSD
RSD =0,211
()dC
∑
RSD = 4×
(67)
2×()n−1
RSD = 4 × 0,00000636 / 18 = 0,00000141
RSD =0,00119
Table 16 — Standard BSC 431. Carbon
2 2
I C dI dC
dI dC
i i
15,30 0,0152 -0,42 -0,0024 0,1764 0,000006
15,40 0,0158 -0,32 -0,0018 0,1024 0,000003
15,60 0,0169 -0,12 -0,0007 0,0144 0,000000
15,10 0,0141 -0,62 -0,0035 0,3844 0,000012
15,50 0,0164 -0,22 -0,0012 0,0484 0,000002
16,00 0,0192 0,28 0,0016 0,0784 0,000002
16,00 0,0192 0,28 0,0016 0,0784 0,000002
16,10 0,0197 0,38 0,0021 0,1444 0,000004
16,10 0,0197 0,38 0,0021 0,1444 0,000004
16,10 0,0197 0,38 0,0021 0,1444 0,000004
15,72 0,0176 1,3160 0,000041278
( dl )
∑
RSD = 4×
(68)
2×()n−1
RSD = 4 × 1,316 / 18 =0,292
RSD =0,541
( dC )
∑
RSD = 4×
(69)
2×()n−1
= 4 × 0,000041278 / 18 = 0,00000917
RSD
RSD =0,00303
Table 17 — Standard BSC 456/1. Carbon
2 2
I C dI dC
dI dC
i i
28,50 0,0892 -0,97 -0,0054 0,9409 0,000030
28,90 0,0914 -0,57 -0,0032 0,3249 0,000010
29,10 0,0925 -0,37 -0,0021 0,1369 0,000004
28,70 0,0903 -0,77 -0,0043 0,5929 0,000019
29,70 0,0959 0,23 0,0013 0,0529 0,000002
30,00 0,0976 0,53 0,0030 0,2809 0,000009
30,00 0,0976 0,53 0,0030 0,2809 0,000009
29,70 0,0959 0,23 0,0013 0,0529 0,000002
30,10 0,0981 0,63 0,0035 0,3969 0,000012
30,00 0,0976 0,53 0,0030 0,2809 0,000009
29,47 0,0946 3,3410 0,00010479
( dl )
∑
RSD = 4×
(70)
2×()n−1
RSD = 4 × 3,341 / 18 = 0,742
RSD = 0,862
( dC )
∑
RSD = 4×
(71)
2×()n−1
RSD = 4 × 0,000104793 / 18 = 0,0000233
RSD = 0,00483
9.6 Determination of accuracy
9.6.1 Chrome
l = 2
Table 18
detect.Cr theor.Cr d = (det-th.)
d
0,1012 0,1040 -0,0028 0,00000784
0,1019 -0,00210,00000441
0,1065 0,00250,00000625
0,1070 0,00300,000009
0,1028 -0,00120,00000144
0,1028 -0,00120,00000144
0,1017 -0,00230,00000529
0,1047 0,00070,00000049
0,1056 0,00160,00000256
0,1070 0,00300,000009
0,00004772
Σ
()d
()d
∑ i ∑
2 i
SEA = 4× SEA= 2×
(72)
2()n−I −1 2()n−I −1
SEA = 4 × 0,00004772 / 14 = 0,00001363
SEA =0,00369
9.6.2 Nickel
l = 2
Table 19
d = (det-th.)
detect.Ni theor.Ni
d
0,1331 0,1400 -0,0069 0,00004761
0,1365 -0,0035 0,00001225
0,1360 -0,0040 0,000016
0,1356 -0,00440,00001936
0,1354 -0,00460,00002116
0,1414 0,00140,00000196
0,1413 0,00130,00000169
0,1467 0,00670,00004489
0,1473 0,00730,00005329
0,1450 0,00500,000025
Σ 0,0002432
SEA = 4 × 0,0002432 / 14 = 0,00006949
SEA = 0,008336
9.7 Ordinary maintenance
Record actions, date and time on a card as follows:
Table 20
Date and Time
dd
mm
yy
oo
Check Gas x x x x x x x x x x x
Check Fluid x x x x x x x x x x x
Check Parameters x x x x x x x x x x x
Clean Stand x x x x x x x x x x x
Clean Filter x x x x x x x x x x x
Clean Lenses x
Sharpen Point x
Profile x x x x
9.8 Control CHART
Complete the following control chart according to the schedule indicated under Clause 8.
CONTROL CHART El % NV
DD/MM HH:MM
HAL HCL
RR MM, El % Team
LCL LAL
0,108
0,106
0,104
0,102
0,1
0,098
0,096
0,094
0246 8 10 12 14
N° tests
Figure 3
El %
CONTROL CHART
Laboratory LLL Element C
Team TTT Reference material BCS 456/1
Type of material SSS NNN Nominal value 0,101
Instrument OES QQQ Tolerance 0,0048
Month/year MM/YY
0,096 0,098 0,100 0,102 0,104 0,106
Date/time Test value
12/6:00 0,100 (1) x
12/6:03 0,102 x
12/10:00 0,097 (2) x
12/10:03 0,101 x
12/14:00 0,097 (3) x
12/14:03 0,105 x
13/6:00 0,107 (4) x
13/6:03 0,107 x
13/10:00 0,107 (5) x
13/10:03 0,101 x
13/10:06 0,102 x
14/6:00 0,096 (6) x
14/6:03 0,097 x
Figure 4
The above example reports some possible cases.
In cases (1) and (2) the instrument is calibrated; in case (3) it is advisable to check it for stability; in case (4) it is
necessary to check the instrumental drift; in case (5) the instrument is calibrated; in case (6) the instrument is
calibrated, but the risk of a systematic error exists.
9.9 Calibration curve control
The average intensity ratio values deriving from the analysis of reference materials must be recorded on
calibration cards according to the following example:
CALIBRATION CARD N° 88
st
ELEMENT STEEL CURVE NUMBER DATE 1
CALIB.
Carbon Low-alloyed steels CCNN MM/DD/AA
INSTRUMENT MANUFAC SERIAL IDENTIFIC. CALIBRAT. REVISION
TURER NUMBER PROCEDURE
O.E.S. QQQQ MMMM SSNN IDF NN CPR XX NN N
USER DEPARTMENT CALIBRATION BODY
Spectrometric Lab Chemical Lab
M E A S U R E S
REFERENCE NOMINAL STAN. DEV. CHECK DIFF. FROM CALIB.
MATERIAL READING x 2 READING NOMINAL VALID?
NBS 1265 13,4 0,31 13,2 - 0,20 YES
BCS 431 16,1 0,76 15,9 - 0,20 YES
BCS 456/1 30,16 1.22 31,0 + 0,84 YES
CURRENT CHECK DATE DATE OF NEXT CHECK OUT-OF-CALIBRATION
28/11/1993 28/11/1994 START END
NOTES: SIGNED
John Smith
Figure 5
Once completed, the above card will accompany the instrument during all its service life.
10 Statistical results
The method was tested submitting a set of reference materials to different laboratories of different countries.
The interlaboratory testing was conducted as outlined in ISO 5725-1:1994 using reference materials supplied by
BSC and JSS. Results of correlation of SEA and RSD versus reference values are shown in the following tables.
The equations are:
SEA (or RSD) = A El % + A
1 0
SEA and RSD are calculated according to 5.4 and 5.3.
(*) Home curves
SEA (as Reproducibility within the labs)
Element Range Equation r
% A El % A
1 0
Cr 0,01 - 0,35 0,0049 0,0006 0,910
Cu 0,01 - 0,5 0,0062 0,0002 0,991
Ni
...
Questions, Comments and Discussion
Ask us and Technical Secretary will try to provide an answer. You can facilitate discussion about the standard in here.
Loading comments...