CHEMICAL ANALYSIS
OPTICAL EMISSION
SPECTROMETER
Ir G HENDERIECKX
GIETECH BV
CONTENT
1. INTRODUCTION
2. TYPES OF SPECTROMETERS
2.1 CLASSIFICATION
2.2 COMMON TYPES
3. MECHANISM EMISSION OPTICAL SPECTROMETER
4. EQUIPMENT
4.1 GENERAL
4.2 MAINFRAME
4.3 SPARK TABLE
4.4 COMPUTOR
4.5 ROLE OF ARGON
5. SAMPLE REQUIREMENTS
5.1 TYPE OF SAMPLES
5.2 CHEMISTRY
5.3 USE
5.4 REMARK
6. LIMITS & TOLERANCES
6.1 LIMITS
6.2 TOLERANCES
7. INSTRUCTION
7.1 STANDARDISING & CALIBRATING
7.2 TESTING
8. INCORRECT RESULTS
8.1 INCORRECT STANDARDISING SAMPLES
8.2 INCORRECT PRODUCTION SAMPLES
8.3 INCORRECT TESTING
8.4 TESTING CONDITIONS
8.5 ARGON IS NOT CORRECT
8.6 REPRESENTIVITY OF SAMPLE TO CASTING
9. SPECTROMTERE RESULT COMPARED WITH…
9.1 CE – METER
9.2 WET CHEMICAL ANALYSIS
9.3 CONCLUSION
10. CONCLUSION
No data from this book can be used without verification; the author cannot and
will not accept any responsibility for possible consequences due to data used
from this book.
Nothing from this book may be copied without written approval of the author.
1. INTRODUCTION
Optical Emission Spectrometry (OES) is one of the oldest physical methods for
chemical analysis. The first equipment dates from around 1860, produced by Kirchoff
and Brussen. It became common practice in all foundries in the 1960’s.
The international standards do require mostly a chemical composition for the cast
metals, except for iron. But even if not required, the foundry does have an instruction
(or specification) about the chemical composition, adapted to the wall thickness and
required mechanical properties.
In order to prove that the delivered metal does comply with the requirements,
internationally accepted test equipment and procedures are required.
The spark light emission spectrometer is one of them.
But as common for all test equipment, a proper use is necessary. This means
calibration, standardisation and sampling.
To do this properly, good standardising and calibration samples are required. These
can be purchased with certificate (BAS London, England and other supplier labs) and
or produced by the foundry itself (but analysed by a certified lab).
The test samples (made by yourself) must be white solidified, grinded correctly
concerning surface and the sample must be flat. Sparks must be done on “fresh”
metal and not on top of an existing spark mark. No inclusion, nor gas nor slag… may
be present.
The spectrometer analysis requires a lot of care and even skill of the operator
concerning the taking and preparing of the sample, standardising the equipment and
performing the test. It is a “relative” testing method, which makes it sensible for
incorrect testing (especially the standardising sample and procedure are important).
2. TYPES OF SPECTROMETERS
2.1 CLASSIFICATION
2.2 COMMON TYPES
2.1 CLASSIFICATION
2.1.1 NATURE OF EXCITATION
2.1.2 MEASUREMENT PROCESS
1. Absorption spectroscopy
Use the range of electromagnetic spectra in which substances absorbs.
This includes “atomic absorption” and various molecular techniques
such as infrared, ultraviolet-visible and microwaves.
2. Emission spectroscopy
Use the range of electromagnetic spectra in which a substance
radiates/emits. The substance must absorb energy and its source
determines the name of the emission, like “luminescence”, which
includes “spectrofluorimetry”. Emission spectroscopy is often referred to
as optical emission spectrometry (OES), due to the light nature of what
is being emitted. It is a fairly accurate testing equipment.
3. Scattering spectroscopy
Measure the amount of light that a substance scatters at certain
wavelengths, incident angles and polarization angles. An example is the
Raman spectroscopy.
2.2.1 ABSORPTION
Absorption spectroscopy is a technique in which the power of a beam of light,
measured before and after interaction with a sample, is compared. Specific
absorption techniques tend to be referred to by the wavelength of radiation measured
such as ultraviolet, infrared or microwave absorption spectroscopy.
2.2.2 FLUORESCENCE
Fluorescence spectroscopy uses higher energy photons to excite a sample, which
then emit lower energy photons.
The technique has become popular for its biochemical and medical applications.
2.2.3 X-RAY
When X-rays of sufficient frequency (energy) interact with a substance, energy is
emitted as radiation (fluorescence) and will remove less-bound electrons from the
atom (Auger effect). The absorption : emission frequencies are characteristic for a
specific atom. The equipment can measure the characteristic X-ray frequencies or
Auger electrons.
These methods are capable of analysing metallic elements in parts per million, billion
or low concentration ranges. Light detectors are needed for analysing the flame.
2.2.5 VISIBLE
Many atoms emit or absorb visible light and in order to obtain a fine line spectrum,
the atoms must be in gas phase (substance must be vaporised).
It is often combined with UV absorption spectroscopy.
2.2.6 ULTRAVIOLET
All atoms absorb in the UV region because these photons are sufficient energetic to
excite outer electrons. If the frequency is high enough, photoionization takes place.
UV spectroscopy is used in quantifying protein and DNA concentration.
2.2.7 INFRARED
Infrared spectroscopy offers the possibility to measure different types of inter-atomic
bond vibrations at different frequencies.
It is used in organic chemistry, for analysing polymers and constituents like filters,
pigments and plasticizers.
2.2.9 RAMAN
Raman spectroscopy uses the inelastic scattering of light to analyse vibrational and
rotational modes of molecules.
3.1STEPS
3.2 ACCURACY
3.3 IMPORTANT REMARKS
3.1 STEPS
The spectrometer works by breaking apart individual atoms from the metal and then
working some quantum mechanical effects on the electrons to produce light
characteristic of the element.
Figure
Pre-sparked area under microscope
Black dots are graphite inclusions
There is a belt around the center
without any remaining free graphite.
During the sparking, the sample material is vaporized on the spark stand by an
arc or spark discharge. The atoms and ions contained in the atomic vapor are
excited into emission of radiation.
The tubes/photomultipliers are detecting the light, particular for their wave
length and transform the detected amount into electrical charge. This amount
of electricity is related to the amount of the element in the metal. Internally this
amount is compared with the stored set of calibration curves and the real
percentage concentration is shown on the output.
Figure
Principle of optical
spectrometer
With standard and calibration samples, the relation from the amount of
electrical charge (of the receivers) to the element concentration in the sample,
is set.
3.2 ACCURACY
The accuracy depends on:
Besides these factors, also the correct standardizing plays an important role.
See correct standardising, see next chapters.
This accuracy is depending on the quality of the detectors / receivers, which is very
important for the residual elements (Pb, Sb, Te, Ti, Al…).
For elements that can in very different percentages, there will be 2 (one for low and
one for high) detectors / receivers present. This is very common for Cr and Ni, but is
also necessary for S (very different for gray and ductile iron), which is mostly not
present.
The next problem is the carbon, especially for iron when free graphite is present. As
mentioned before, the spectrometer will always indicate a lower carbon content
compared to reality. To decrease the difference, some suppliers (AMETEK –
SPECTRO) have setup a double chamber system (with calculation of free graphite),
which gives better results.
Figure
Result with normal and modified carbon
analysis
2. The sample must be completely (at least at the spark area) white solidified (no
free graphite at all), which mostly requires copper moulds (depending on the
thickness)! The copper mould should be filled homogeneously and without
inclusions, nor gas, nor slag…
If the sample is thin, iron (or steel) moulds are sufficient. But the thickness
must be such that no cracks are appearing!
3. The sample may not be grinded too heavily to avoid heating. If the colour is
blue, it has been heating too much (some elements could be affected).
4. The grinding must be done with cooling fluid and it is advised to hold the
sample with a magnet-holder.
5. At the end of the grinding, the sample should be cleaned by alcohol and dried
properly.
6. Always at least 2 sparks are required because the test is of the comparative
type. It compares a secondary value (cell loading) with the result of standard
samples. There must be taken an average value.
4. EQUIPMENT
4.1 GENERAL
4.2 MAINFRAME
4.3 SPARK TABLE
4.4 COMPUTOR
4.5 ROLE OF ARGON
4.1 GENERAL
The spark light emission spectrometer has changed from large fragile equipment to a
much smaller and less demanding equipment.
Before, there was a necessity for constant temperature and humidity in very narrow
range, which required special precautions and extra investment.
Nowadays the required range for temperature and humidity are larger and a regular
executed standardisation can compensate the differences.
The stability of the spectrometer depends on the material of the base frame on which
the receiver tubes are mounted (thermal expansion coefficient of this metal) and on
the length of the internal chamber (the larger the dimensions, the more stable).
Figure:
Picture of current spectrometers
The main part of the spectrometer is a vacuum chamber, flushed with Ar (argon)
before use. The different parts are:
1. Optics (grating) that separate / isolate the specific wave lengths for the
elements to be measured
2. Detector system to measure the intensity of the light emissions
Figure:
Schematic view of
the main chamber
Figure:
Schematic view of the main chamber for carbon
measuring modified type
The light, emitted by the sparking of the sample, is detected by its entrance
concerning wave length per element. The photomultiplier tubes (receivers) are
positioned on a curved bar.
Figure:
Real view of main chamber
The number of elements that can be tested is depending on the number of tubes.
Some elements do need more tubes, especially when large ranges are to be tested.
Example: Ni in the range of 0 to 5 % and another from 5 to 40 %.
The second part is the spark table, consisting of an electrode and a connection with
the vacuum main part. It contains the excitation device (electrical discharge
electrode, arc or spark) to provide the required energy for excite the atoms in the test
sample.
The table must be cleaned after each spark and the sealing must be correct.
The sample should be flat and no light should escape (visible from operator side)
from the sparking area.
If the table area is not clean, this can involve the result too.
4.4 COMPUTOR
This is the combination of:
1. Electronics to acquire the detector signals and to control the functions of the
spectrometer
2. Software for calculating and displaying the emission spectra and concentration
values.
It can be that the computer also contain calculation programs for comparing “target”
and “real” composition and calculating the required melt additions to let them match
(within the tolerance range).
The argon flushing is done to clean the inside of the main chamber and the argon
pressure should be kept constant. After replacing an argon cylinder, high care should
be taken and the supplier instructions should be followed before further use.
When the Argon is closed at the end of the use, by restarting some flushing by
dummy sample sparks should be done to assure correct operation. With a new
spectrometer, this flushing can be 1 spark. If the spectrometer is somewhat older and
maintenance is not regularly performed, 2 to 3 dummy sparks can be required.
The argon must be of high purity (welding Ar is not sufficient), type 4.8, which means
99,9981 % Ar. On top of this, the pressure must be always > 0,3 bar to avoid oxygen
entering the cylinder and or the connection part. Oxygen must be < 3 ppm!
Some suppliers do recommend the use of a type of Argon with 2 % hydrogen to bind
the oxygen that is possibly present in the small (microscopic) cracks of the sample.
This is a good solution but the calibration and standardisation must be done with the
equal type of Argon to assure the highest test accuracy.
If there is air leakage, the test result will not be correct. So check the operation of the
vacuum pump and main frame sealing.
In recent times, some suppliers introduce a spectrometer with a closed circuit Ar-
system. In this system the Ar is pumped around and continuously cleaned and
purified by filters. If this purifying and cleaning can be done properly, it will be a large
safe of Ar consumption!
5. SAMPLE REQUIREMENTS
2. Standardising sample
This sample is certified with analysis and tolerance range. It is used for
the daily standardising and is less costly.
3. Reference sample
This is a sample, mostly produced by the foundry, which is tested
according to the calibration samples. It is less accurate but sufficiently
for the day to day use.
In case of doubt, the standardising sample can be used
4. Production samples
This is a sample that is produced by the foundry and tested by certified
labs. The accuracy is less than the certified lab samples but mostly
sufficient for the use in the production, especially the first test to find out
the metal bath chemical composition.
5.2 CHEMISTRY
The chemistry for the standardising and reference samples is important. They must
be selected in a way that they comply with the current production of the foundry.
Complying means that they have similar contents, preferred one below the lower limit
and one above the upper limit (see later).
This is important for every element, which should have at least a sample with the
minimum and maximum possible content and preferable with one in between these
extremes.
5.3 USE
The first very important remark is that iron can only be tested if it is white
solidified! It cannot be tested with a sample cut from the casting!
The samples must include the expected and tested concentration of every element in
the standardising and calibration range.
Pay attention for important residual elements, which are mostly present in a very low
concentration, which exceeds the range of the calibration and even sometimes the
standardisation samples.
Incorrect burns (overlapping, including moisture or oil, not white solidified sample)
must be removed immediately from the file.
SHAPE 2: best result
Figure:
Two most common
production samples
The spectrometer sample diameter (D) is at least 30 mm diameter and the sample
has a maximum thickness (t) of 6 mm.
Sample shape No 1 is the easiest to be grinded and prepared because the ingate-
sprue is easy to hold. The mould is made of 3 parts: down part for the disc and a split
mould for the ingate-sprue.
Due to the fact that this ingate-sprue is small (10 to 6 mm at the connection), the
”non-influenced” zone or “white solidified” zone is sufficiently large. No sparking in the
center area!
Figure:
picture showing part of the metal mould
and the resulting test samples
Sample shape No 2 is more difficult to hold properly during the grinding and
preparing. The mould is only in 2 parts. The sample itself can be sparked all over.
It is poured in a copper book mould for fast chilling of the sample. The sample must
be fully chilled. It must be clean and smooth, equal surface to avoid leakage during
sparking.
It is clear that the sample should be hold correctly; the best is a magnetic holder.
No heavy load can be applied and cooling is necessary to avoid blue colour, which
indicate that some elements are affected.
Figure:
picture showing variant
of sample shape 2
This sample is a variant of sample No 2, but having some extra metal pins in the
bottom. These metal pins are used for the wet chemical analysis (and or special C-
equipment as is LECO) and have the advantage, compared to chips and drilled
material, that they are solidifying white and by that are not loosing carbon.
Figure
Magnetic sample holder
For non-magnetic materials, it should been put into a holder (see next picture) and
then the holder with the magnet.
Figure
Holding tool for non-magnetic materials
5.4 CONCLUSION
The better the quality of the samples, the better the result quality.
6.1 LIMITS
6.2 TOLERANCES
6.1 LIMITS
6.1.1 Elements, oxides, compounds
6.1.2 Quality of metal
6.1.3 Correct carbon content
6.1.4 Structure and carbides
6.1.5 Needs time
The spectrometer has blind spots because they do not see if the elements measured,
are located in the matrix, in oxides, in compounds (compositions due to chemical
reaction of 2 or more elements).
Example: the spectrometer cannot tell you if the magnesium in the sample is free
element, oxidized (MgO) or present as a MgS.
The spectrometer will report the total magnesium in your production sample. Your
effective magnesium is the total minus the magnesium-sulfides minus the
magnesium-oxides. It is possible to calculate the magnesium-sulfides from the sulfur
(MgS contains 0,75 x Mg for 1 time S), but not the magnesium-oxides.
Mg on spectrometer Mgtotal
This is the reason that sometimes good nodularity is found with an 0,035% Mg, and
at other times, the nodularity is too low.
The lack of oxygen measuring capability is leading to the fact that the magnesium
concentration is more or less of an indicator of good ductile iron castings in steel and
the titanium / aluminium content for steel castings.
To make high quality constant castings, you do need to control your pearlite forming
constituents’ as well as the carbon and silicon. You will just need to add some
additional testing and calculation based on experience.
The Si-analysis from the spectrometer is correct as well as the carbon equivalent
(CE) measurement with the thermal analysis method. Combining this Si-content with
the CE gives the possibility for calculating the correct C-content.
6.2 TOLERANCES
Because the spectrometer testing is a “relative” testing, which means that the results
of a test bloc are compared to the results of a certified bloc (entered in the
spectrometer and checked every day, also with a tolerance). There will always be a
tolerance on the result compared to reality.
As a matter of fact, every test has a tolerance. In most of the international standards
for materials, there is a tolerance indicated.
EN 1559-2:2014 Founding – Technical conditions of delivery – Part 2:
Additional requirements for steel castings
The subscript “L” (L) indicates the limits of the element specified by the applicable
specification to calculate the tolerance for the upper and lower limit respectively.
Example:
Steel with 0,40 – 0,70 % Mn
Tolerance: 0,08 * 0,40 + 0,01 = 0,042 Lower limit adapted: 0,358
Tolerance: 0,08 * 0,70 + 0,01 = 0,066 Higher limit adapted: 0,766
1. Spectrometer capability
2. Sample preparing
3. Spectrometer standardizing
4. Testing.
7. INSTRUCTIONS
7.2 TESTING
7.2.1 SAMPLE TAKING
7.2.2 SAMPLE PREPARATION
7.2.3 OPERATOR INSTRUCTION
7.2.4 REPORTING
7.3 REMARKS
The operation is only possible if the equipment is properly standardised & calibrated.
7.1.1 CALIBRATION
7.1.2 STANDARDISING
7.1.3 CERTIFIED STANDARDS
7.1.4 WORKING STANDARD SAMPLES
7.1.1 CALIBRATION
Calibration is setting of the original shape of the curve as determined by the optical
alignment, and the electrical amplification.
When the equipment is installed, the calibrating is required and mostly done by
the supplier.
This involves that for each type of metal, the response curve from each tube
(receiver), one per element, is related to the real chemical composition.
Two items are important:
2. the performed calibrating is only valid for testing in the range between
the lowest and highest value, per element, from the used standards. A
little extrapolation can be done but with high care!
mV CALIBRATION
RECEIVER
ELEMENT "A"
Figure:
Calibration done by the
supplier with 4 samples
% ELEMENT A
Remarks
Newer electronics are much better than before, but recalibration should be
considered if the gain of the system has changed and caused a rotation of the
curves. If you standardize the spectrometer, and the analysis in the parts of your
curve away from the standards is off, a recalibration is indicated – the shape of your
calibration curve has changed. Many new spectrometers are not made to be easy to
recalibrate.
It is impossible to perform proper tests for chemical compositions way outside of the
used standards chemical composition range. This is mostly forgotten elements for
calibrating are the residual elements like Ce, Pb, Ti, As, Sb, B…
It is impossible to test elements in iron with a program that is only calibrated with
steel samples.
7.1.2 STANDARDISING
Standardization is the daily or more frequent correction of the “calibration curve” that
accounts for barometric pressure changes, and possibly other environmental
changes (temperature, humidity).
Each morning (each shift), or before each different material test, the equipment
must be standardised.
mV
STANDARDISING
Figure: Standardising done RECEIVER
% ELEMENT A
STANDARDISED EXTRAPOLATION
ZONE ZONE
mV STANDARDISING
RECEIVER
Figure:
ELEMENT "A"
Standardizing done with 1
sample; the test result in
the right part of the range
can be totally incorrect
% ELEMENT A
STANDARDISED EXTRAPOLATION
POINT ZONE
mV
RECEIVER STANDARDISING
EXTRAPOLATION EXTRAPOLATION
ZONE ZONE
This standardising can be done with other samples as for the calibrating, but keep in
mind that the accuracy of the test will depend on the accuracy of the analysis of the
standardising sample.
For standardizing the temperature of the equipment must be equal to the temperature
at which the next testing will be done! This requires:
1. heating and cooling device in the equipment to keep the temperature equal
2. room temperature should be kept equal during the complete time of testing.
After a weekend or in the morning at the start, the room temperature must
be set and constant for at least 2 hours.
The mistake that will be made if the temperature is changing, is depending on the
material of which the base plate and the vacuum box of the spectrometer is
produced. The normal iron and steel have a temperature gradient that is 4 times
higher than the special materials.
The accuracy of the test is depending on the accuracy of the standardising and
the calibrating. Both have an equal importance and should be done with the
highest care!
They are made in batches, and the certification is for the batch and not for the
standard unless you are dealing with a certified ALCO aluminum standard. In that
case each standard is certified.
The certification for the other standards of the world should be viewed with a little
doubt, and the best fit line of many standards should be used over just accepting a
few points.
Curves with multiple corrections for other elements are likewise suspect when only a
few standards are used. For example, if you are using 5 standards and have 3
corrections for other elements, then you have only 2 degrees of freedom left (5-3 ).
Many standards have had a mild heat treat to relieve stress and therefore have
some graphite in them making them un-usable for carbon analysis.
Others have a few elements that varied within the batch, and the block you received
may be different than the certified value by 2 or 3 sigma. The standard 1 inch block
like the MRDF standards was made in a mold of 495 pieces so 25 pieces would be
outside of the +/- 2 sigma value. For this reason alone, multiple blocks should be
used to build a calibration curve.
Working Standard samples are sometimes purchased certified standards, but many
times purchased uncertified standards or in-house produced secondary standards.
Using certified standards is expensive.
Every foundry should assure to stop using the standard when it reach the line on the
side. The line is where the standard becomes unreliable due to shrinkage, and/or
micro-segregation or graphite.
1. Find a thick copper plate (can also be iron) to chill the samples.
2. Make some rings with about 25 mm inside diameter out of ceramic or core
material with a 20 mm height.
3. Obtain some tellurium to suppress graphite formation. You will need to make
an initial trial run with a few rings. Place the tellurium in the bottom of a
pouring spoon and pour from a second spoon into the pouring spoon. Do not
overfill the spoon or too much tellurium can be lost.
4. Allow a second or two for the tellurium to finish reacting, and then pour the
rings.
5. Squish each ring quickly on the top to remove excess metal, and help promote
chill. On the first trial, it must be determined what amount of tellurium to add.
6. Now cut one of the standards from top to bottom and look for how far up from
the copper plate a good standard is made. By polishing the cut face, it can be
seen where the well chilled good iron ends. Small tight crystals are to be
preferred over large ones. The start of shrinkage or graphite will be the point
where the standard has to be thrown away. If half the height of the standard
can be used, it is done well.
7. Drill the top of one or two from each batch and analyze (chemical method) for
carbon and sulfur.
9. When you put the working standard into use, edit your spectrometer settings
and put in the printed results from that initial analysis plus the combustion
carbon/sulfur, and you have your very own traceable working standard!
Too much tellurium or not enough waiting time will cause your standards to be full of
gas holes.
Note that a dirty surface on the copper plate may also generate gas holes.
Each spoon of iron must be treated as a separate batch, so use a larger test bar
spoon rather than the normal spectrometer sample spoon.
7.2 TESTING
Is the sample taken at the same time, which means that the metal:
must be homogenised
is taken equal depth under the metal surface
is taken at about the same time after switching power off, or after
tapping or after a metallurgical treatment.
Is the sample taken with power off in case of electrical furnaces? If yes, how
long after switching off the power?
Is the sampling:
It can be wet or dry grinded, but the sample must be dried before using.
The grinding must remove surface scale, oxides and all oil and other residues.
Figure:
Rough grinding sample
Figure:
Grinding and polishing
equipment
Pay attention with using SiC-paper for grinding and polishing because small
remaining particles will render the testing incorrect.
The operator should check for “abnormal” results (very different from the other) and
try to find the cause. Make extra, at least 2, sparks and eliminate the “abnormal” one.
But never remove non-expected results because they are different.
Never keep sparking until 2 or good “good” results are appearing (removing more
than 1 or 2 non-expected ones).
Check the average result with the calculated (or expected) result.
Keep the sample (together with other test material) for verification by the customer.
7.2.4 REPORTING
The melting and or pouring crew must be reported very quickly in order to
enable them to correct the chemical analysis before pouring.
File the results per melting charge with the history (number and amount of
corrections performed during the operation).
The results can also be filed and used to make reports and or certificates. In
the reports, the required (if there is one) chemistry must also be mentioned.
Take care to correct the result with the additions of elements due to
metallurgical treatments like inoculation, de-oxidising… after the spectrometer
sample was taken.
7.3 REMARKS
Controlling the spectrometer is not an easy job:
1. First assure that the burns reflect the sample. Generally, a repeatability
target is set for each element and a sample is burned twice to check for
that repeatability.
The operator skill is needed to cope with the problems of bad samples due to cracks,
pinholes, and a lack of chill caused by hot or dirty sample molds.
The point needs to be cleaned very often, and the spark chamber and the optic lens
must also be periodically cleaned.
If the spark hole is not properly covered or the grinding of the sample is not even, air
(oxygen and water vapor) will get into the chamber and interfere with the lighter
elements.
Finally, if care is not taken in standardizing, the chemistry of different elements may
move up or down by operator.
Once a week, blind testing must be done. Take a previous sample, remove the date
and time from the sample and have each shift run and report the chemistry to the
leader. The results will tell who is the best operator and who might need further
training.
Another test that could only be down with SPC-software was to note the variability of
all the burns of a shift. Record the difference between the two accepted burns and
take the standard deviation of those burns by element and plot a histogram by shift
and by all shifts. Good operators and good samples produced a tight standard
deviation. Over time, and increase in the deviation of an element on all shifts might
indicate a problem with the element’s electronics or the standard(s) used for that
element.
Use a record of the amount of correction of each standardization combined with the
reason for the standardization. The reasons were:
1) cleaned the machine (brush for cleaning the electrode has lost its hairs),
2) routine standardization based on time, and
3) out of range chemistry causes a check on the standard and a
standardization was preformed.
8. INCORRECT RESULTS
The spectrometer is a very accurate instrument if it is used properly. If not, the most
stupid results will appear.
1. the two samples may differ due to the process and how the samples are
taken.
2. that there will be (very) different calibration curves / standards between the
two spectrometers. Often spectrometers are designed and built for a specific
metal type and have limited accuracy outside of that metal type.
If incorrect standardising samples are used, the result of one or more elements will
be continuously different with a similar “deviation”. But point is that no one can
recognise this unless another lab is doing the analysis and the difference is
recognised.
Many foundries do use their standardising samples incorrect due to the high cost of
them. They do overlapping sparks, do not remove previous sparks properly (not deep
enough).
Other foundries do use production samples (especially for steel), These samples are
produced by them and not tested by independent labs concerning the analysis.
The calibration should be done every shift, for ALL elements, according to the
procedure described in this book.
If the sparks are overlapping or the previous sparks are not removed completely, the
calibration is NOT CORRECT!
The overlapping area has less elements compared to the fresh one (not all elements
are burnt equally in amount and depth). This leads to the condition that the computer
is told that a lower amount (real from the calibration sample) is higher (theoretical
value if fresh surface is used). If calibrated like this, the production results will be
somewhat (depending on the element) higher. This can be embarrassing for the
element with low and maximised contents as are p, S, Al, Mg…
Similar will happen if calibration samples have left sparks (not completely removed).
1.2 It is possible that the thickness of the sample is too high in a way that the
solidification is not white.
1.3 It is possible that the thickness is too thin with the consequence that most
samples are cracked. In these cracks, if cooled in water are grinded and polished
with liquid, water and or other liquid is entering and if the spark does touch this
area, the test result will be incorrect.
Solution: 1. Redesign sample shape and verify the white solidification under the
sprue as well as the cracking tendency.
2. Dry the sample (if cooled with water) by hot air.
Figure:
Redesigned horizontal sample to avoid
non-white solidification of the center. At
the same time, it is impossible to use the
center part for sparking.
cut AA
Even though the chamber is pressurized, the turbulence of escaping gas around a
sample disk will pull in unwanted air.
If the sample is wet, the humidity (water) will be pulled into the testing chamber. The
sample should be dried.
Solution: instruct the sampling operator properly and make an instruction, which
contains the time (minimum is very important, but should always be equal)
between the end of the nodulising and sample taking as well as the depth
(in the liquid metal) of sampling.
Figure
Sample broken
Sample with cold run
1. Only 1 spark
2. Different sparks one upon the other
3. Sparks partly outside the sample
4. Sample not correctly grinded and cleaned
5. Sample not flat.
Because the spectrometer testing is a “comparison” test (compare the test result with
the previous introduced standard curves per element), it is always required to have at
least 2 sparks!
If more sparks are put one upon the other, the second one is testing a material that
was already influenced by the previous spark and the chance is high that the result is
not correct anymore.
Figure:
Sparks touching one another
Figure:
Pre-sparked sample micro – grinding too few
Figure:
One spark
Figure:
Correct sparking
Sparks outside the sample are definitely not ok because there is an influence of the
environment.
Samples that are not clean, are wet or grinded flat (this can be seen if the sparking
light is visible during testing), some air from the environment will influence the test
result.
The humidity should also be equal, but has less important effect on the result. The
rule is also that the humidity at the time of testing is equal to the one on the time of
the daily standardizing.
The humidity of the sample, sometimes wet due to the cooling during grinding, is
much more important. Always dry the sample just before using it.
The pressure in the bottle must be > 0,3 Bar to avoid that oxygen will enter the bottle
or the connector to the spectrometer.
The spectrometer is flushed with H-containing Ar, but not calibrated with the same Ar.
If the argon flushing / cleaning / functioning is not correct, the test results will be
scattering and no correct average can be calculated.
Solution: Check the vacuum pump and request maintenance of the sealing of the
main chamber and spark table (spark table can be done by the user) by
the supplier.
1. Sample is only a tiny fraction of the melt volume. The melt should be
homogeneous.
This is not the case if the sample is taken from an electrical furnace with
power on.
This is not the case if the slag (inclusions) are still floating and or still on top of
the metal (slag not removed).
2. If the furnace is still melting after the sampling, it can be that the chemistry is
changing (melt temperature above Tc-temperature will change C-content).
3. The melt can during tapping, transport and pouring still form oxides, forming
slag (if contact with the air) and change its chemistry. The Mg-fading in ductile
iron is a very good example.
4. The chemistry can change if some extra metallurgical treatments are done
after sampling:
Inoculation for iron
De-oxidising for steel.
The final sample should be taken in the pouring box, or at least in the ladle just
before pouring.
It is important to know how the spectrometer result is related to the results of other
chemistry testing and to reality.
9.1 CE – METER
9.2 WET CHEMICAL ANALYSIS
9.3 CONCLUSION
9.1 CE-METER
The CE-meter does measure correctly the CE (Carbon Equivalent).
Figure:
Double test cup holder
Figure:
Result double test
This test is done by chipping the test metal and burning it and collecting gas in
chemical liquids to enable the calculation of the element concerned.
The equipment is complex and the influence of the operator is high (weighing
samples, chipping properly, preparing chemicals…).
The chipping should be replaced by small pins / fins, thin material in order to avoid
free graphite loss. In this case, a correct C-content can be measured. There is also
special equipment for testing C and S.
Figure
Spectrometer sample with “pins” for
wet chemical analysis
Also only 1 spark, which is not
sufficient.
The testing takes a lot of time and cannot give a result before pouring.
The test material (taken with sufficient care) can be taken from the casting itself.
Figure:
Combustion and chemicals
9.3 CONCLUSION
The comparison between the spectrometer (and modified for C-analysis) and the CE-
meter and Wet chemical analysis is shown in next table.
Test sample White solidified White solidified Any Liquid metal cup
10. CONCLUSION
The result of the chemical analysis test with a spark light emission spectrometer is
depending on the proper use of the equipment, the correct standardisation and
calibration for the metal involved and the correct sample preparation.
The carbon content is the most critical one to test because all non conformities of the
machine will influence the carbon the most.
In a lot of foundries, the testing is done without the necessary care of calibrating and
performing a proper test.
The combination of a spectrometer and thermal analysis can assure that the best test
is performed, which is especially the case for carbon (element with the highest
deviation if something went wrong).