ICP-MS, or ICP-AES and AAS?

—a comparison ICP-MS-1
April 1994

Geoffrey Tyler
Varian Australia Pty Ltd
Mulgrave, Victoria, 3170, Australia

Introduction the selected ions reach the detector.

The attractiveness of the inductively coupled plasma The ICP-MS provides information for each atomic
atomic emission spectrometer (ICP-AES) has led mass unit (amu or Daltons; more accurately, the
many analysts to ask whether it is wiser to buy an ratio of the mass of the ion to its charge is
ICP-AES or to stay with their trusted atomic displayed, and labelled m/z), in the mass region
absorption technique (AAS)1. More recently, a new 3-250 amu. The isotope information can be used in
technique, inductively coupled plasma mass several ways; these include isotope ratio
spectrometry (ICP-MS), has been introduced2. measurements, often used for Pb and U which do
The ICP-MS offers at first sight, albeit at higher not have a constant natural abundance, and
cost, the advantages of ICP-AES and the detection analysis of samples having unnatural isotope
limits of graphite furnace atomic absorption abundances.
spectrometry (GFAAS). Isotope dilution is a method of spiking the samples
This article will briefly describe these three with a known concentration of a pure isotope to
techniques, and point out the important criteria by obtain a very accurate determination of the
which to judge their applicability to your own concentration of the element. A pre-requisite of this
analytical problems. technique is that the element of interest must have
more than one isotope.
For many people with an ICP-AES background, ICP-
MS is a plasma, with a mass spectrometer as a Detection Limits
detector. Mass spectroscopists would prefer to
describe ICP-MS as mass spectrometry with a ICP-MS detection limits are very impressive. Most
plasma source. Either way, the technique is capable detection limits are in the 1-10 parts per trillion (ppt)
of giving isotope information. This information can range for solutions. These are as good as, or better
help to overcome many of the “spectral” than, GFAAS detection limits for most elements in
interference problems that can occur in the mass pure water. ICP-MS can also detect many elements
spectrometer. that have very poor GFAAS detection limits.
ICP-MS has typically two to three orders of
Basically, the sample compartments and plasma of magnitude better detection limits than ICP-AES,
ICP-AES and ICP-MS look similar. In ICP-AES, the which has detection limits for most elements in the
optical spectrum (approximate range 165-800 nm), 1-10 parts per billion (ppb) range. Recently, some
is viewed and measured, either sequentially or ICP-AES spectrometers have shown impressive
simultaneously. The simultaneous ICP-AES is detection limits in the sub ppb region for selected
faster, but more expensive, than sequential elements in an optimized part of the spectrum for
ICP-AES. clean samples.
The ICP-MS extracts the ions produced in the It should be noted, however, that the comment
plasma into an interface consisting of a sampler above about ICP-MS detection limits is for simple
cone followed by a skimmer cone. This configuration solutions having low levels of other dissolved
enables the pressure to be reduced differentially material. For detection limits related to
from atmospheric pressure down to a final pressure concentrations in the solid, the advantage for
of between 10-5-10-7 Torr. The ions pass through the ICP-MS can be degraded by up to 50 times,
interface into ion optics, which optimize the ion because of the poorer dissolved solids capability.
paths. Neutral species are removed from the beam Some common lighter elements, (e.g. S, Ca, Fe, K,
at this stage, either by the vacuum pumps or by Se) have serious interferences in ICP-MS which
collision with a photon stop. The ions then pass degrade the detection limits considerably.
through a mass filter, usually a quadrupole, before
1
Interferences 4. Matrix effects
The three techniques present different types and Transport effects include spray chamber effects
complexities of interference problems. For this and differences in viscosity between sample
reason, we will look at each technique separately. solutions and calibration standards. This will
change the efficiency of aerosol production
ICP-MS interferences from one solution to another. Matrix matching is
usually required, although internal
1. Spectral
standardization can be used as an alternative
The spectral interferences in ICP-MS are method. The rapid scanning speed of ICP-MS
predictable and number less than 300. does give superior results when using an
Polyatomic and isobaric interferences are found internal standard.
where a species has a similar mass to the
5. Ionization
analyte, whereby the resolution of the
spectrometer (generally around 0.8 amu) will Ionization effects can be caused by samples
not resolve it, e.g. 58Ni on 58Fe, 40Ar on containing high concentrations of group I and
40
Ca,40Ar16O on 56Fe, or 40Ar-Ar on 80Se. group II elements. Matrix matching, sample
dilution, standard addition, isotope dilution,
Element correction equations (similar in
extraction or separation by chromatography
principle to inter-element correction in
may be necessary.
ICP-AES) can be used. In many cases
alternative isotopes with lower natural 6. Space charge effects
abundances may be employed. The use of
Space charge effects occur mainly behind the
mixed gases (small percentages of other gases
skimmer cone, where the net charge density
such as nitrogen, added to the main argon gas)
becomes significantly different from zero. The
can sometimes be effective in reducing
high ion density leads to interaction between
interferences.
ions present in the ion beam causing
2. Matrix acids preferential loss of the light ions in the presence
of heavy ions, e.g. Pb+ on Li+ 3. Matrix
It should be especially noted that HCl, HClO4,
matching, or careful choice of internal
H3PO4 and H2SO4 can cause considerable
standards across the mass range of analytes,
spectral problems. Polyatomic interferences are
will help to compensate for these effects,
caused by Cl+, P+, S+ ions in conjunction with
although this may prove difficult in practice.
other matrix elements like Ar+, O+, H+.
Isotope dilution will be effective though
Examples are, 35Cl40Ar on 75As and 35Cl16O on
51 expensive, but the simplest and most effective
V.
method is to dilute the sample.
The avoidance of HCl, HClO4, H3PO4 and
H2SO4 in ICP-MS is paramount for most ICP-AES interferences
analyses. Where this is not possible, separation 1. Spectral
chromatography (microcolumns) may be used
before the sample is introduced into the ICP-AES spectral interferences are more
plasma. This is a method many favour to get rid numerous and are more difficult to solve. There
of the unwanted species, and it also creates an are more than 50,000 ICP-AES spectral lines
opportunity to preconcentrate at the same time. documented, and the matrix can cause
Other techniques used to overcome these considerable problems, which makes a high
problems are: electrothermal vaporization resolution spectrometer mandatory for the
(ETV), and mixed gases. Another very analysis of samples such as steels, chemicals
expensive alternative is a high resolution and rocks. Inter-element correction, used
magnetic sector ICP-MS which can resolve extensively in simultaneous ICP-AES, can have
masses less than 0.01 amu apart. This enables only limited success.
many of the spectral interferences to be The background in ICP-AES may be elevated
eliminated. or structured, requiring an off line background
Solutions for ICP-MS analysis are normally correction. Sophisticated dynamic background
prepared in nitric acid. correction, if available, is very useful to improve
accuracy. Different molecular species such as
3. Doubly charged ions OH give peaks or bands which can cause
A doubly charged ion will cause a spectral analytical problems at low analyte
interference at half the m/z of the singly concentrations, degrading the detection limits in
charged ion, e.g.138Ba++ on 69Ga+ or 208Pb++ on real samples.
104
Ru+. These interferences are few and can be
The background in ICP-MS is so low, typically
considerably minimized, or effectively <20 counts/second, that it doesn’t pose a
eliminated, by optimizing the system before problem. This is a major reason for the superior
proceeding with the analysis.
detection limits of ICP-MS.

2
2. Matrix effects The trend to simplicity has been evident since 1993
and will continue in the future. One of the reasons
Like ICP-MS, ICP-AES can use internal
for this is full computer control of parameters stored
standards to overcome matrix effects such as
within a method. Another reason is the use of a
spray chamber effects and viscosity differences
multitasking graphical user interface, to show the
between samples and calibration standards4.
operator several indicators of data integrity on the
3. Ionization same screen. The use of such software also has a
very positive effect on method development time.
Interference from easily ionizable elements can
Before this software became available, ICP-MS
be minimized by careful choice of individual
method development was a highly complex and
element conditions6 or by adding an ionization
time-consuming task. GFAAS, although relatively
buffer, i.e. by adding an excess of a group I
simple for routine analysis, requires considerable
element.
skill in setting up the methods.
GFAAS interferences
Total Dissolved Solids (TDS)
The interferences in GFAAS have many sources.
Recent ICP-AES spectrometers have been able to
They include:
analyze routinely up to 10% TDS and even up to
1. Spectral 30% for simple salt solutions. Although the analysis
of 0.5% TDS for ICP-MS may be possible for a
There are a few spectral interferences in
limited timescale, most chemists are happier with
GFAAS when deuterium background correction
0.2% maximum TDS. This should be borne in mind
is used, but these can be eliminated by use of
when the original sample is a solid. The ultimate
Zeeman GFAAS.
detection limit for some elements in ICP-MS may
2. Background not be so impressive when expressed in the solid,
compared with ICP-AES. GFAAS can cope with
For many matrices careful programming of the
extremely high levels of dissolved solids.
ash stage is required to minimize the
background signal during the atomization. The
Linear Dynamic Range (LDR)
use of chemical modifiers can be helpful in
increasing the allowable ash temperature. For ICP-MS can have a LDR in excess of 105. Various
example, a Ni chemical modifier for Se methods for extending the linear range up to 107
determinations allows ash temperatures of up include de-sensitizing one of the ion lenses, use of
to 1000 °C before Se loss. The use of Zeeman detector analog mode, or use of a separate Faraday
background correction can give an cup as a second detector. These should be used
improvement in accuracy compared with D2 arc with caution, however, as high matrix component
background correction in many GFAAS concentrations may cause problems best solved by
applications. dilution. For this reason, and because of the
problems with high levels of dissolved solids,
3. Vapor phase interferences
ICP-MS should be mainly the domain of trace/
These can be caused by the atomization of the ultratrace analysis. The exception to this is when
analyte into a cooler gas environment. These using isotope dilution. With this technique very good
interferences have been minimized in recent results have been obtained with high concentrations.
years by isothermal tube design and use of
platforms so that the sample is atomized into a GFAAS has a limited LDR of 102-103. It can be used
hot inert gas environment. for higher concentrations if a less sensitive line is
selected.
4. Matrix effects
Trace to major element analysis may be performed
Matrix effects are caused by variable retention by ICP-AES because of its 105 LDR. ICP-AES is
of the analyte on the graphite tube, depending ideal for analysis up to and including percentage
on the sample type. The dry and ash stages levels. For this reason ICP-AES, in addition to
can have a dramatic effect on the shape of the ICP-MS or GFAAS, is often needed to fulfil
transient peak. The use of matrix modifers laboratory requirements.
(e.g. PdCl2) and hot injection can be quite
effective in minimizing these effects; also the Precision
use of peak area measurement can be
advantageous in some cases5. The short-term precision of ICP-MS is generally
1-3%. This is improved routinely by use of multiple
Ease of use internal standards. The longer term precision (over a
period of hours) is still <5% RSD. The use of isotope
For routine analyses, ICP-AES has matured in dilution can give results of very high precision and
automation to the point where relatively unskilled accuracy, although the cost can be prohibitive for
personnel can use methods created by the ICP-AES routine analysis.
specialist. Until recently, ICP-MS was still the
domain of the specialist chemist making fine ICP-AES has generally precisions of 0.3-2% RSD in
adjustments before performing routine analysis. the short term and again less than 5 % RSD over
several hours.
3
GFAAS, however, will generally have short term Hence, the cost of graphite tubes for the latter has
precisions of 0.5-5% RSD. Longer term precision is to be taken into account. In all three techniques the
a function of the number of graphite tube firings, cost of argon is a significant budget item, with the
rather than time. ICP techniques requiring more than GFAAS.

Sample throughput Capital cost

‘The ICP-MS has an incredible capacity to analyze a This is always a difficult subject to quantify because
vast number of samples for trace elements. Typical it will depend on the amount of automation, the
analysis time is less than 5 minutes/sample, for the accessories and the supplier. In very approximate
whole suite of required trace elements. For some terms, you can estimate that an ICP-AES will cost
applications this may only take a couple of minutes. twice as much as a GFAAS and 2-3 times less than
Consulting laboratories find the sample throughput a ICP-MS. It should be noted, however, that the
major advantage. accessories can distort these figures considerably.
While the speed of ICP-AES analyses will depend on Another cost that needs to be taken into account is
whether simultaneous or sequential instruments are that ultra trace analysis requires a clean laboratory
used, generally this can vary from 2 to 6 minutes and ultra pure chemicals. These are not cheap.
per sample. Simultaneous ICP-AES can be faster,
typically 2 minutes/sample, but sometimes its Accessories
accuracy can be compromised by spectral
interferences present with some types of samples Being a very rapid sequential method, ICP-MS can
(e.g. rocks). utilize transient signals in multi-element mode. This
opens the way for a host of accessories.
The speed of GFAAS is typically 3-4 minutes per Electrothermal vaporization, laser ablation, glow
element per sample (assuming 2 replicates). discharge and spark ablation can obviate the need
Automated overnight runs can be performed, and to dissolve the sample. Some accessories provide
this will improve throughput of samples. the means of separating the matrix from the sample
and/or to pre-concentrate. These include hydride
Total sample throughput can be a major factor
generation, and various forms of chromatography
favouring ICP-MS in the busy laboratory. The
(e.g. HPLC, ion chromatography, microcolumns).
following examples (expressed as solution
concentrations), will give a guide: The advantage of separation by chromatography for
speciation work has only been fully realized in
1. One to three elements/sample, at sub/low ppb
ICP-MS. This is due to the low concentration levels
concentration will generally be better by
of interest in environmental, toxicological, medical
GFAAS, assuming the elements of interest can
and food samples.
be determined by this technique.
Although ICP-AES can use some of the above
2. 5-20 elements/sample at sub ppm-% levels will
accessories, their cost and their marginal
generally be better by ICP-AES.
advantages have meant that we have rarely seen
3. 4 or more elements/sample at sub ppb and ppb many of them routinely used.
concentrations will generally be better by
ICP-MS, if the number of samples to be Summary
analyzed is high.
To advise anyone what to buy is always difficult.
Unattended operation Look at your present and future needs, and answer
the check-list questions in table 1. This should help
ICP-MS, ICP-AES and GFAAS can all operate you to decide.
unattended overnight, because of the modern
automated designs and the safety inherent in the It should always be remembered that no technique
use of inert argon gas in these techniques. For will satisfy all your requirements. The techniques are
highest productivity, overnight operation is complementary. There will always be samples where
mandatory. one technique is better suited for the analysis than
another.
Cost of ownership Table 2 shows a simplified comparison of the three
The running cost of ICP-MS is more than ICP-AES techniques. Table 3 compares detection limits.
because several components have a limited lifetime
and have to be replaced. These include the
turbomolecular pumps, the sampler and skimmer
cones and the detector. The torch and nebulizer
have similar lifetimes for both ICP-AES and ICP-MS
techniques. If ICP-AES is chosen instead of ICP-MS
the laboratory will probably require GFAAS as well.

4
References
1 G.Tyler, AA or ICP - which do you choose?
Chemistry in Australia, Vol 59, No 4,
pp 150-152, April 1992.
2 A.R. Date and A.L. Gray, Applications of ICP-
MS, Blackie, Glasgow, UK, 1989.
3 K.E. Jarvis, A.L. Gray, and R.S. Houk,
Handbook of ICP-MS, Blackie, Glasgow, UK,
1992.
4 M.Thompson, J.N. Walsh, Handbook of
Inductively Coupled Plasma Spectroscopy,
Blackie, Glasgow, UK, 1983.
5 J.E. Cantle, Atomic Absorption Spectroscopy,
Elsevier, 1982.
6 Analytical Methods for Liberty ICP
Spectrometer, Varian publication 85 100938 00,
Chapter 5, pp 81-82.
7 J.Olesik, Elemental Analysis Using ICP-OES
and ICP-MS, Anal. Chem. Vol 63 No 1, Jan 1
1991 pp 12A-21A.

Table 1. Checklist of analytical requirements

1. How many samples/week?

2. What are the sample types? (steels, rocks,
effluents, soils, etc)
3. What method of dissolution may be employed?
4. How many and what elements need to be
determined?
5. What are the concentration ranges?
6. What sample volume is typically available?
7. What other options/accessories are being
considered? Why?
8. How important is isotope information to you?
9. How much money is available to purchase or
lease costs/month?
10. What is the cost of ownership and running
costs for the techniques to fulfil the
requirements?
11. What skilled operators are available to you?

5
Table 2. Simplified comparison of ICP-MS, ICP-AES, GFAAS

ICP-MS ICP-AES Flame AAS GFAAS

Detection Excellent for Very good for Very good Excellent for
limits most elements most elements for some some elements
elements
Sample all elements 5-30 elements 15 seconds/ 4 mins/element
throughput 2-6 min/sample /min/sample element/sample /sample
Linear dynamic 105 105 103 102
range 8
(10 with range ext'n)
Precision
Short term 1-3% 0.3-2% 0.1-1% 1-5%
Long term (4hrs) <5%* <5%*
*
precision improves with use of internal standards
Interferences
Spectral few common almost none few
Chemical (matrix) moderate almost none many many
Ionization minimal minimal some minimal
Mass effects high on low NA NA NA
Isotopes yes no no no
Dissolved solids
(maximum tolerable 0.1-0.4% 2-25% 0.5-3% >20%
concentration)
No. of elements >75 >73 >68 >50
Sample useage low high very high very low
Semi-quantitative yes yes no no
analysis
Isotope analysis yes no no no
Routine operation easy easy easy easy
Method skill skill easy skill
development required required required
Unattended yes yes no yes
operation
Combustible
gases no no yes no
Operating cost high high low medium
Capital cost very high high low medium/high

6
Table 3. Detection limit comparison (µg/L)

Element ICP-MS ICP-AES Flame AAS GFAAS

As <0.050 <20 <500 <1

Al <0.010 <3 <50 <0.5
Ba <0.005 <0.2 <50 <1.5
Be <0.050 <0.5 <5 <0.05
Bi <0.005 <20 <100 <1
Cd <0.010 <3 <5 <0.03
Ce <0.005 <15 <200000 ND
Co <0.005 <10 <10 <0.5
Cr <0.005 <10 <10 <0.15
Cu <0.010 <5 <5 <0.5
Gd <0.005 <5 <4000 ND
Ho <0.005 <1 <80 ND
In <0.010 <30 <80 <0.5
La <0.005 <0.05 <4000 ND
Li <0.020 <1 <5 <0.5
Mn <0.005 <0.5 <5 <0.06
Ni <0.005 <10 <20 <0.5
Pb <0.005 <20 <20 <0.5
Se <0.10 <50 <1000 <1.0
Tl <0.010 <30 <40 <1.5
U <0.010 <30 <100000 ND
Y <0.005 <0.5 <500 ND
Zn <0.02 <1.0 <2 <0.01

ICP-MS, ICP-AES, Flame AAS: Detection limits (defined on the basis of 3 standard deviations of the
blank)
GFAAS: Sensitivity (0.0044 absorbance) measured with 20 µL of sample
ND: Not determined