Talanta 54 (2001) 1021– 1038

www.elsevier.com/locate/talanta

Review

Preparation steps in environmental trace element analysis

— facts and traps
Michel Hoenig *
Centre for Veterinary and Agrochemical Research (CERVA), Leu6ensesteenweg 17, B-3080 Ter6uren, Belgium

Received 8 November 2000; received in revised form 17 January 2001; accepted 17 January 2001

Abstract

The laboratory of CERVA is for several decades involved in the Belgian environmental research. The activity was
associated to national monitoring programs dealing with trace metal pollution of all compartments of the environ-
ment (sea and river waters, sediments and organisms but also soils, plants, animal and food samples). Such a
monitoring dealing with the total analyte contents in samples needed a comprehensive development of the whole
methodology associated to the analyses using atomic absorption and emission spectroscopy techniques. This includes
measurement but also preparation steps. The latter is the subject of this work. Long-term experience has shown that
precisely sample preparation is the most critical part of the analysis because it is responsible for the largest and often
hidden sources of errors. Errors due to contaminations may be usually overcome if necessary precautions are taken
concerning reagents, tools and the manner of working. The problem is different for analyte losses: in this case, the
responsible factor is an inappropriate methodology. This is particularly true for preparation of solid samples that
have to be brought in a solution in order to satisfy needs of introduction systems of most spectroscopic techniques
utilized in routine laboratories. For some types of samples (e.g. animal tissues), the dissolution is not a problem: it
may be readily achieved by several procedures. This is not the case for samples that contain silicates in their matrix
(e.g. soils, sediments, plants) because their complete dissolution cannot be ensured by a simple procedure. This review
describes the present knowledge regarding possibilities and errors that concern preparation steps. In addition, possible
effects of the preparation procedure on the quality of measurement are also systematically discussed. © 2001 Elsevier
Science B.V. All rights reserved.

Keywords: Sample preparation; Trace element analysis; Atomic spectroscopic methods

1. Introduction

The present paper has to be considered as a

* Tel.: +32-2-7692234; Fax: + 32-2-762305. tutorial: it does not necessarily present all last
E-mail address: m.hoenig@terv.var.fgov.be (M. Hoenig). novelties concerning the discussed topics but it is

0039-9140/01/$ - see front matter © 2001 Elsevier Science B.V. All rights reserved.
PII: S 0 0 3 9 - 9 1 4 0 ( 0 1 ) 0 0 3 2 9 - 0
1022 M. Hoenig / Talanta 54 (2001) 1021–1038
rather useful to help all ‘new’ analysts confronted
with the trace element analysis using atomic spec-
troscopy methods. Environmental aspects associ-
ated to the pollution problematic may be
modelized thanks to reliable data obtained by a
monitoring of several chemical and physical
parameters. It is evident that phenomena in play
are principally of chemical nature and that ways
leading to their identification and their possible
control are precisely associated to this discipline.
Therefore, chemical analysis plays a dominant
role and only adequately interpreted data ob-
tained under appropriate conditions can guaran-
tee the success of further investigations. In the
past, we witnessed situations where links between
different aspects of the environmental protection
under-estimated the analytical control. At present,
we can still not state that the utilization of chem-
ical analysis attains always the needed quality
level in this field. In this respect, it is first neces-
sary to adopt an analytical strategy, represented
by following points:
“ Rigorous definition of the problem to
investigate.
“ Selection and collection of samples related to
this problem (sampling).
“ Appropriate sample preparation.
“ Accurate determination of elements associated
to the initial problem.
“ Validation and evaluation of analytical results.
“ Interpretation of results as a function of the
investigated problem.
“ Relevant conclusions.
Regarding the initial situation, conclusions are
based on a set of analyses, which are character-
ized by a large number of elements determined in
various samples with an unknown and/or variable
matrix composition. It must be remembered here
that precisely the matrix will be responsible for
the degree of difficulties encountered during
analyses because it will impose constraints not
only in the analyte determination but also during
preparation steps. Due to these facts, samples
collected for environmental monitoring fall into
different categories as a function of the analytical
difficulty associated to the complexity of the ma-
trix and/or concentration levels to determine:
“ Waters (drinking, surface, ground, rain,
waste...).
“ Soils, sediments, suspended particulate matter
collected in natural waters, dusts.
“ Plant samples (variable content of insoluble
silicate compounds).
“ Seawater (ultra-trace concentrations), biologi-
cal fluids (very complex matrix: blood, milk, or
highly variable matrix: urine).
“ Wastes (urban or industrial): great difficulty to
ensure a representative sample to analyze [1].
This review is dedicated to the determination of
the total analyte content in samples. Problems
associated to the speciation of elements are an-
other aspect of the analysis: it will not be dis-
cussed here.
In most inorganic analytical laboratories, trace
element analysis is usually performed using
atomic spectroscopy techniques. The market of
this type of apparatus offers principally instru-
mentation initially dedicated to the analysis of
liquid samples. Only few manufacturers propose
equipments for direct analysis of solid samples:
X-ray fluorescence spectrometry (XRFS) and arc/
spark atomic emission spectrometry (AES). These
two techniques were intensively utilized between
the fifties and seventies. At present they cover,
with difficulty, needs usually required in environ-
mental applications necessitating determinations
of elements at trace or ultra-trace concentrations.
These remain, however, utilized in some industrial
domains because of the practical aspects of their
application, i.e., analyses can often be performed
directly on solid samples overcoming then prob-
lems associated to dissolution procedures.
In environmental and health diagnostics, deter-
mination of trace elements has been since the
seventies tackled using atomic absorption and
atomic emission techniques. Flame atomic ab-
sorption spectrometry (FAAS) offers a detection
power ranging between mg/l and mg/l depending
on the element to be considered, electrothermal
atomic absorption spectrometry (ETAAS) is able
to determine fractions in the mg/l average. How-
ever, AAS techniques are limited to determination
of metallic elements only. During the same period
appeared inductively coupled plasma atomic emis-
sion spectrometry (ICP-AES) exhibiting a detec-
 M. Hoenig / Talanta 54 (2001) 1021–1038
tion power situated between FAAS and ETAAS.
More recently during the eighties, equipments re-
sulting from a coupling of an ICP excitation
source with a mass spectrometer (ICP-MS) were
first commercialized. The detection power of ICP-
MS is very high, of the order of ng/l or lower.
Multielement analytical techniques, ICP-AES and
ICP-MS allow the determination of metals and
non-metals. Moreover, in comparison with AAS,
particularly electrothermal, the analytical
throughput using plasma-based techniques is con-
siderably improved.
Nowadays we assist to a general substantial
improvement of the utilizable performance of the
instrumentation. However, in such situation
where a dramatic lowering of detection limits is
observed, the risk of suddenly appearing errors
due to sample handling is increasing. Before the
commercial introduction of modern instrumenta-
tion, these ‘new’ errors were practically impercep-
tible in the determination of high concentrations
analyzed with less sensitive techniques. Dangers
of contaminations are now increasingly present:
the good choice of the sample preparation proce-
dure, the quality of its application and the ade-
quate laboratory environment become therefore
the most critical points for a successful trace
element analysis.
We will enumerate here some basic notions that
will allow a better understanding of the general
philosophy regarding existing trace element analy-
sis principles. Firstly, it is necessary to ascertain
that chemical analysis is a set of closely bound
steps: for example, the choice of the sample
preparation procedure will depend on the mea-
surement technique used and/or conversely. It is
then not sufficient to intuitively apply a non-vali-
dated procedure (mineralization, dissolution, mea-
surement technique...) to a sample with an
unknown composition. The set of analytical crite-
ria has to be chosen after a global consideration
of the final objective — reliable results in terms of
accuracy and precision. The topic is very vast: for
this reason we will discuss broad outlines and
general trends concerning sample preparation
steps, their principles, advantages and drawbacks.
Consequently, the reader will not find in this work
many ‘recipes’ but we hope that necessary ingredi-
1023
ents to resolve usual cases are present here or in
the references given. It remains for the analyst to
choose the appropriate preparation methodology
in function of sample types, of the available
equipment and of imperatives of the analysis and
of the whole study.
2. Definitions
Samples to analyze in environmental and health
studies occur as liquids or solids. They may be
divided into those that are already in an aqueous
solution (various water samples, beverages, milk,
blood, serum, urine...) or in other liquid form
(oils...). For routine analysis by atomic spec-
troscopy techniques, which are all dedicated to
work with aqueous samples, the analysis of other
liquid samples must be adapted (subject not dis-
cussed here). Solid samples can mainly be of
organic (plants and animal tissues...) or mainly
inorganic (soils, sediments...) nature. Prior to the
analysis, solid samples are generally solubilized by
an appropriate dissolution method. However,
each case necessitates a different analytical ap-
proach that has to be selected depending on the
sample composition. For this choice, the possible
behaviour of both main matrix and trace elements
has to be considered.
In order to simplify the problem associated to
the analysis of solid samples, we can consider at
the outset that the organic part of the sample will
be decomposed and eliminated during preparation
steps. The analyzed residue will then be consti-
tuted by mineral compounds only, but their con-
tent may be very variable: expressed in dry
matter, main elements (usually forming the ma-
trix) are present at concentrations superior to
0.1% (in environmental samples usually Ca, K,
Mg, Na and P), minor elements (B 0.1%, e.g. Fe,
Al...), trace elements are in mg/g and ultra-trace
elements below ng/g or lower. These definitions
are arbitrary but they reflect the sample reality
and difficulties encountered during an analysis.
Indeed, if errors of one percent are unacceptable
for results concerning main matrix elements, val-
ues with errors of ten percent or more are gener-
ally admissible in trace or ultra-trace element
1024 M. Hoenig / Talanta 54 (2001) 1021–1038
analysis. This fact is furthermore clearly perceiv-
able in results of any intercomparison exercise.
In view of elevated concentrations of main ele-
ments, the sample matrix is the factor responsible
for interference observed during measurements of
other elements present at lower concentrations.
The matrix represents also the most important
factor to consider for a successful dissolution of
the sample. Depending on the sample type, the
dissolution procedure generally includes several
steps. Here, the terminology is precise: the term
‘mineralization’ concerns samples with a totally or
partly organic matrix only (animal and plant tis-
sues, food samples, soils...). Prior to the analysis,
the organic compounds present have to be decom-
posed and/or eliminated precisely by a mineraliza-
tion procedure. Using various reagents, the
organic matter is decomposed into carbon diox-
ide, nitrogen oxides and water, thus liberating
into solution elements initially associated with it.
After a mineralization procedure, the resulting
sample residue should be essentially inorganic: it
will be subject to a final dissolution step in the
same way as a sample with an initially total
inorganic composition (rocks, metals...). For more
complex samples (organic+inorganic composi-
tion: soils, sludge, plant samples...), chemical
reagents and physical means used ensure most
often these two roles (mineralization and dissolu-
tion) simultaneously.
In a usual chemical analysis, the objective of
sample preparation stage is to bring all available
means into play in order to determine as readily
as possible the elements to be investigated. These
means are:
“ Conversion of the sample to a form compatible
with the measurement technique utilized (gen-
erally a dissolution).
“ Destruction and simplification of the matrix
(mineralization: wet digestion, dry ashing).
“ Analyte separation or preconcentration (topics
not treated here).
3. Principal sources of errors
Accuracy of results is often difficult to ensure
because of great number of analytical steps that
start with the sampling, continue with preparation
procedure and end by the determination itself.
Following the sample nature and other factors,
errors can occur at various steps of the analysis.
They result either in analyte loss (incomplete re-
coveries) or in its contamination (excessive
recoveries).
Analyte losses may be provoked by volatiliza-
tion, absorption, adsorption, transformation, pre-
cipitation or coprecipitation resulting from
treatments utilized during sample preparation
steps. Nevertheless, there are cases where losses
are apparent only. In other words, even in cases
without any losses due to preparation steps, possi-
ble interferences that occur during measurements
are often responsible for analyte signal suppres-
sion. This is attributed to matrix effects resulting
from differences between compositions of calibra-
tion standards and samples. In this case, the
measurement technique is responsible for the er-
ror observed: in practice, it is of paramount im-
portance to distinguish this type of error from
errors assignable to sample preparation
procedure.
Contaminations, frequently observed, are due
to systematic or random introduction of non-neg-
ligible amounts of the analyte during different
steps of the analysis. They result from reagents
and materials utilized or from ambient air; they
become the more troublesome as concentrations
to determine are lowered. Arising from reagents,
the contamination values are generally repro-
ducible from one sample to another (systematic
error). Other contamination sources, more easily
avoidable, are random and variable. It is neces-
sary to evaluate the possible global contamination
by several representative blanks and to consider
them in the calculation of results. On the other
hand, it is also necessary to clearly distinguish
contaminations due to sampling and sample stor-
age from those associated to preparation and
measurement stages.
3.1. Sampling
Sampling, the first and one of the most impor-
tant steps, has the role to ensure the representativ-
ity of the sample in the context studied. For this
 M. Hoenig / Talanta 54 (2001) 1021–1038
reason also, it represents the most potential
source of errors. The sampling, however, cannot
be considered as a real part of the analytical
protocol and we will present briefly its essential
aspects. If the sampling is not based on the
present knowledge and on the use of appropriate
tools with particular cautions, both systematic
and random errors may occur, attaining in some
cases (e.g. trace elements) several orders of magni-
tude. The reader interested by all problems associ-
ated with the sampling may find necessary details
in works [2– 9] dedicated to this very important
procedure. More detailed information concerning
the whole sample preparation may be found in
references [10– 19].
3.2. Collection and storage
Depending on the samples to analyze and ele-
ments to be determined, the sampling tools, filtra-
tion devices and storage vessels have to be
carefully chosen and cleaned to minimize risks of
a possible contamination. All vessels to be used
should thus be washed, rinsed and then soaked
overnight in 1– 10% (v/v) analytical grade nitric
acid. Finally, before use all vessels should be
rinsed in high purity deionized water. During
sample collection and storage, risks of contamina-
tion are important and become often critical when
they concern sampling of media presenting very
low concentration levels of analytes investigated.
This is for example the case of natural waters.
Also analyte losses may occur with such samples:
some cautions must then be taken into account
during their collection and storage.
In order to minimize analyte losses by adsorp-
tion of metal ions on the vessel or on the sus-
pended particles, the samples collected can be
stored for a short time in a refrigerator, and for
longer periods in a freezer. For the same purpose,
aqueous solutions are generally acidified immedi-
ately after sampling (generally at pHB1.5 with
nitric acid). This treatment is generally sufficient
to prevent the adsorption of trace elements on to
walls of the vessel for relatively long term. It must
be specified here that in practice it is very difficult
to remobilize back into solution previously ad-
sorbed trace elements, even using relatively drastic
1025
means such as strong acidification followed by
vigorous shaking.
Another mean to avoid adsorption losses is
immediate freezing of the sample collected and its
thawing just before measurements. This is the best
method for storage of natural waters because an
acidification may mobilize into solution some
trace elements associated with particulate matter
present in the sample. If the total content of trace
elements in the water sample is needed, dissolu-
tion of the suspended matter is desirable, but it
may become problematic if the analyte concentra-
tion in the aqueous phase only has to be deter-
mined. In this case and if means for freezing are
not available after sample collection, an aliquot
should be filtered (generally using a membrane
filter of 0.45 mm porosity) and subsequently aci-
dified. A global analysis of such water sample
may then be performed: aqueous phase and solid
residue remaining on the filter (after an appropri-
ate dissolution) are analyzed separately and a
calculation of their contribution in the initial sam-
ple may be done.
3.3. Preparation
In most cases, sample preparation includes sev-
eral stages (solid samples): drying (air, laboratory
oven...), homogenization (mixing, crushing...),
grinding (mills, mortars...), followed by a sub-
sampling, mineralization and dissolution of a sub-
sample. The obtained solution is finally filled-up
to a fixed volume. Some of these steps performed
in the laboratory may be sources of contamina-
tions, essentially due to the type of vessels and
purity of reagents, to water used but also to
ambient air. Additionally, grinding generally ap-
plied to solid samples prior to the dissolution
procedure, may contaminate the sample by abra-
sion of the mortar or milling parts with certain
hardness. Agate, for example, often used as mate-
rial for machining of laboratory mortars, presents
hardness similar to quartz, always present in soil
and sediment samples. In this case, it is more
appropriate to grind such samples in mortars of
superior hardness, e.g. boron carbide or corun-
dum. It must be noted that the use of the corun-
dum may affect the analysis by contamination of
the sample by Al, Mg, Ba, Cu and Zn.
1026 M. Hoenig / Talanta 54 (2001) 1021–1038
Sample preparation steps necessitate the use of
reagents and vessels made in porcelain, glass,
quartz, PTFE, platinum or various plastics. Impu-
rities present in reagents are important (and often
systematic) sources of contaminations, in particu-
lar if they are added in great amounts. For this
reason, in addition to usual analytical grade
chemicals, the market offers relatively expensive
reagents of high purity (Suprapur, Ultrapur,
Specpure...). Their use allows avoiding to a large
extent this type of contamination problems. It is
also possible to purify reagents (acids) in the own
laboratory: however, it must be remarked that
these procedures necessitate an appropriate ap-
paratus realized in expensive materials (quartz,
PTFE), they are delicate to master and require a
good skill of the operator.
Errors resulting in analyte losses and incom-
plete recoveries associated with mineralization
and dissolution procedures will be treated sepa-
rately in Section 4.
It is evident that sample preparation steps are
of paramount importance to ensure a good qual-
ity of the whole analysis. Contamination risks
increase with temperature, pressure, long-term
contact of solutions with the vessels and with
decreasing analyte concentration. To minimize
them, a number of principles must be fulfilled:
“ Consult established procedures specified in the
literature and take into account the real objec-
tive of the analysis. The most complex proce-
dure is not always the best.
“ Ensure a clean environment in the laboratory
(hoods, ovens, muffle furnaces, microwave
devices...).
“ For grinding, milling and homogenization, use
devices made of appropriate material to avoid
sample pollution.
“ Limit the mass of the sample to be analyzed
and volume of vessels used (to minimize the
contact area with the solution).
“ Use only water and reagents of high purity and
reduce the amounts used.
“ Clean carefully all vessels (soaking in acids
followed by an abundant rinsing with deion-
ized water).
“ Do not use old vessels to avoid adsorption
phenomena of trace elements on any worn-out
surfaces.
“ Simplify handling, avoid filtrations and trans-
fers of solutions if they are not absolutely
necessary.
“ Perform several blank procedures with the
same reagents, vessels and operating conditions
to evaluate possible contaminations and cor-
rect the results.
“ Check recoveries for the whole procedure using
reference materials of similar composition to
those of samples analyzed. If recoveries are
incomplete, find the reason distinguishing
preparation steps (responsible factor: proce-
dure) and measurement step (responsible fac-
tor: interference).
3.4. Measurements
Aqueous samples and dissolved solid samples
are generally introduced for analysis directly and
without any prior treatment. The only main prob-
lem with solutions is the collection and storage of
natural waters, already discussed in Section 3.2.
Concerning atomic spectroscopic analysis, no par-
ticular precautions have to be taken. If measured
concentrations satisfy the principal criteria of the
spectroscopic method used (sensitivity, detection
limits, dynamic range) and if possible interfer-
ences are under control, the analysis of solutions
may be performed automatically with all modern
spectroscopic systems. Of course, it is of great
help in the analysis if the operator knows the
approximate concentrations of the analyte and
the main matrix components in the sample in
order to decide whether or not some dilution is
desirable. Some types of aqueous liquid samples
necessitate particular cautions during the intro-
duction, e.g. blood coagulates in contact with
some chemical compounds like palladium chloride
or nitrate, often used as chemical modifiers in
ETAAS analysis. With modern apparatus, the
separate introduction of the modifier and sample
into the furnace is possible and their direct con-
tact is thus avoided.
Measurement apparatus may also be a source
of contaminations. The most known example is
associated to the ETAAS determination of a trace
element after a long period of its use as chemical
modifier. Nickel, for example, often used as effi-
 M. Hoenig / Talanta 54 (2001) 1021–1038
cient modifier in the As and Se determinations,
will progressively contaminate the whole atom-
izer. If it should be determined at trace levels after
such utilization of the furnace, an intensive decon-
tamination of the ETAAS system is imperative.
Similarly, the material and condition of tubings
that ensure the transport of solution to nebuliza-
tion system and of aerosol to the torch in ICP-
AES and particularly ICP-MS analyses may be a
source of adsorption/desorption phenomena lead-
ing to erroneous results.
Errors due to interferences are associated to the
measurement technique itself. They are defined, in
principle, as errors resulting from sample compo-
sition only. This kind of effects falls in another
field of interest than preparation steps and it is
not treated here. A documentation concerning
origins of interferences in spectrochemical meth-
ods and means of their correction may be found
in appropriate works [18– 25]. Nevertheless, we
will present some examples of problems resulting
from the use of particular reagents during prepa-
ration steps, because by this way they become an
inherent part of the sample analyzed.
The type of acid used in preparation procedures
can have important consequences during measure-
ment steps. It is commonly known that in all
techniques of atomic spectroscopy nitric acid is
the most desirable reagent. In spite of sometimes
observed signal suppressions in its presence (e.g.
in ICP-AES), no severe analytical problems with
nitric acid at concentrations up to 10%, some-
times higher, are practically observed in all tech-
niques of atomic spectroscopy as far as their
concentrations are similar in standards and sam-
ples. Also hydrogen peroxide added into most
mineralization procedures is rarely responsible for
analytical problems.
The presence of hydrochloric acid is not trou-
blesome in the ICP-AES analysis; on the other
hand, its single use is prohibited in ETAAS analy-
sis because of the possible formation of analyte
chlorides and of spectral and/or vapour-phase
interference that it can generate [20,21]. During
the atomization step a part of the analyte is
volatilized as undissociated chloride molecules,
resulting in a molecular absorption compensated
by the background correction device (deuterium,
1027
Zeeman) as a non-specific signal. The resulting
atomic signal of the analyte is consequently re-
duced. If the use of hydrochloric acid is unavoid-
able in a particular preparation procedure,
addition of nitric acid allows avoiding above de-
scribed problem (e.g. digestions with aqua regia).
For some applications, hydrochloric acid is also
undesirable in the ICP-MS analysis. It is, for
example, responsible for an isobaric interference
on the unique arsenic mass (m/z 75). In this case,
chloride in the sample combines with the argon
present in the plasma, forming thus 40Ar35Cl+ of
identical mass to those of As. Both masses are
detected simultaneously, leading to apparently
higher concentrations for arsenic.
Because of its high viscosity, utilization of sul-
phuric acid is often avoided in spite of its great
efficiency in decomposition procedures of organic
samples. Its presence in solutions is particularly
undesirable in analytical techniques where the
sample introduction is ensured by a nebulization
system (FAAS, ICP-AES, ICP-MS). Physical in-
terferences observed are attributed to changes in
the formation and/or transport of the aerosol if
there are physical differences between standard
and sample solutions. Especially for the case of
sulphuric acid, its concentration has to be identi-
cal in all solutions analyzed. This is often unreal-
izable because the part of acid really used to
achieve the mineralization remains unknown,
same as its residual concentration in resulting
solutions. Despite of its good efficiency in diges-
tion mixtures, the use of sulphuric acid is for these
reasons often avoided.
In dry mineralization procedures, various ash-
ing aids are sometimes added to the sample in
order to reduce the volatility of analytes (As, Se,
Te) but also to improve decomposition of the
organic matrix. This is generally achieved by addi-
tion of large amounts of magnesium nitrate and/
or oxide before ashing at high temperatures [10].
Except a possible high background absorption
during ETAAS determination of elements with
wavelengths near the principal atomic absorption
line of Mg, the high Mg-content in the final
solutions is generally not troublesome in analysis
of other elements. Unfortunately, it prevents the
often-required determination of the initial Mg-
1028 M. Hoenig / Talanta 54 (2001) 1021–1038
concentration in the analyzed sample and the
increased content of total dissolved salts will im-
pede the analysis by ICP-MS.
4. Dissolution of solid samples: principles
Collected solid samples are generally too het-
erogeneous to satisfy needs of the analysis. This
implies preliminary treatments in order to obtain
a more representative sub-sample with a finer
particle size. Prior to the following steps, samples
are dried at ambient temperature or in a labora-
tory oven (usually 75°C for mercury, 105°C for
other elements). For some types of samples,
lyophilisation procedures (drying by freezing) are
increasingly often utilized: they allow preserving
the initial sample texture and/or facilitating a
subsequent grinding.
For dry solid samples of certain hardness (soils,
sediments), several types of grindings (dry or wet)
may be applied manually (mortars) or mechani-
cally (ball, hammer, roll and disk mills or
grinders) machined in various materials of appro-
priate hardness (steel, agate, boron or tungsten
carbides, corundum...). Homogenization of other
types of solid samples (plant and animal tissues,
food samples...) is generally ensured using various
mixers. To facilitate the homogenization of fresh
organic samples, they may be ground after freez-
ing in liquid nitrogen.
In the following sections, we will mainly discuss
cases associated to the total analyte content in the
sample. This means that generally a quantitative
dissolution of solid samples is required. To ensure
this point for some types of samples (soils, sedi-
ments, plant material), the known procedures are
systematically too labour- or time-consuming to
be applied to routine analyses. For environmental
monitoring purposes for example, they are often
replaced by simpler and easily applicable proce-
dures. However, these substitutive methodologies
rarely lead to accurate determination of total
analyte contents but they are generally sufficient
to satisfy objectives of the study. This point will
be discussed later with more details.
We mention here also fusion procedures using
alkali metal hydroxides, carbonates or borates as
melting reagents. They are generally set apart for
applications in which the objective is the determi-
nation of main matrix elements in samples which
are insoluble in acids. Nevertheless, alkaline fu-
sions are often used for geological and industrial
purposes because they allow a subsequent efficient
dissolution of both matrix and trace elements.
Common methods of fusion generally require four
to eight times as much reagent as sample. The
amounts of fluxing reagents added are thus very
important and are consequently a potential source
of contamination in trace element analysis. For
this reason they must be of very high purity.
Concerning the analysis itself, the possible drastic
change of the initial matrix implies particular
caution because the ‘new’ matrix elements intro-
duced via fluxing reagents can also become a new
source of interference.
5. Mineralization: decomposition of organic
matter
Samples of organic or mixed nature are subject
to two distinct steps, which take often, place
simultaneously: mineralization and dissolution.
Samples of purely inorganic composition are sim-
ply dissolved.
The composition of environmental samples
varies from purely inorganic (e.g. fly ash) to
purely organic (e.g. fats), but generally, they are
an intermediate combination of these extremes.
This means that total dissolution of samples usu-
ally analyzed cannot be achieved in one step using
a single reagent. In practice, the necessary number
of steps and reagents is dictated by the matrix
composition. Purely organic or mixed samples are
usually brought into solution by some types of
oxidation process followed by an acid digestion of
the resulting residue, as well as of the initial
inorganic part of the matrix. In dry ashing proce-
dures, the organic matter is decomposed by calci-
nation at high temperature and the obtained ash
is dissolved in a strong acid. In most modern wet
procedures, various combinations and propor-
tions of strong acids with hydrogen peroxide en-
sure the mineralization and dissolution of the
sample.
 M. Hoenig / Talanta 54 (2001) 1021–1038
5.1. Dry ashing
Mineralization of the organic matter is done by
ashing at atmospheric pressure in a pro-
grammable furnace. The commonly admitted tem-
perature for this step is 450°C. In addition to
traditionally heated muffle furnaces generally em-
ployed for dry ashing purposes, the market pro-
poses now also microwave furnaces especially
adapted to attain elevated temperatures. Also a
low temperature ashing procedure (LTA) in a
radiofrequency oxygen plasma exists: unfortu-
nately, the necessary instrumentation is very ex-
pensive and not easily available on the present
market. In addition, the low temperature ashing is
a particularly time-consuming procedure.
The fresh or dried (usually 103– 105°C) sample
is weighed into a suitable crucible (generally plat-
inum) and placed in the furnace. Temperature is
progressively elevated following a heating pro-
gram to attain 450°C and maintained several
hours. The resulting inorganic residue (ash) is
dissolved by an appropriate acid. The obtained
solution is transferred into a volumetric flask,
filled up and analyzed. Depending on the initial
sample condition, results will be expressed on
fresh or dry weight. The application of dry ashing
methods is simple and large sample series may be
treated at the same time. This is not their unique
advantage: compared with wet digestions, the dry
ashing presents several other useful character-
istics:
“ The biggest advantage of dry ashing proce-
dures is the possibility of treating large sample
amounts and dissolving the resulting ash in a
small volume of acid (generally nitric or hydro-
chloric). This procedure permits the preconcen-
tration of trace elements in the final solution,
which is useful when very low analyte concen-
trations are to be determined.
“ The resulting ash is completely free of organic
matter, which is a prerequisite for some analyt-
ical techniques (e.g. data in ICP-MS or ICP-
AES with ultrasonic nebulization which may
be affected by the presence of some undissoci-
ated organic molecules).
“ The sample matrix is largely simplified and
resulting solutions are of very acceptable aspect
1029
(clear, colourless and odourless). This is rarely
the case of solutions obtained using wet diges-
tion methods.
“ Reagent volumes and the handling are reduced.
5.1.1. Procedures and de6ices usually employed
Dry ashing methods can be applied to mineral-
ization of organic materials, biological tissues,
plant and food samples, sludge, etc. Well mas-
tered, they ensure a total destruction of the or-
ganic matter; the associated elements are generally
transformed to carbonate or oxide forms.
Concerning the choice of ashing temperature, it
is therefore mandatory to ensure the quantitative
decomposition of the organic matter without a
partial or total loss of analytes by volatilization or
their incorporation into a residue insoluble in
usual reagents. The latter may result from forma-
tion of refractory oxides, from combinations with
other sample constituents present but also from
reactions with walls of the crucible. The most
appropriate vessel material for dry ashing is plat-
inum, which resists virtually to all acids except
aqua regia, but a factor to consider remains the
elevated cost of crucibles made in this material.
Dry ashing temperatures commonly admitted
for trace element analysis range between 450 and
500°C. They are generally high enough to ensure
a complete oxidation of the organic materials
avoiding dangers of volatilization losses for most
analytes (except Hg, As, Se). The losses are still
minimized if the ashing temperature is attained in
a slow gradient (9 4 h from ambient temperature)
that prevents any local hot spots or self-ignition
of the sample with consequent loss of analyte.
Losses can also result from overshooting of the
maximum temperature set.
The ashing temperature is maintained several
hours. If the ashing is performed under optimal
conditions, it leads to ashes of white or light grey
colour. After cooling, they are generally dissolved
by nitric or hydrochloric acid. Sometimes the
oxidation of organic matter is not complete: the
ashes present darker spots (dark grey to black)
attributable to insufficiently oxidized carbon. Be-
cause this contributes to a difficult subsequent
dissolution, such a residue must be moistened by
1030 M. Hoenig / Talanta 54 (2001) 1021–1038
1 ml nitric acid and recalcinated for 1 h at the
usual aching temperature. After this treatment,
ashes are generally clear and easily soluble.
5.1.1.1. Operating modes. Animal tissues (and all
samples of organic nature without presence of
silicates) [19]
“ Weigh in platinum crucible 1– 5 g dried sample
(105°C) or up to 20 g fresh sample.
“ Place the crucible into a cold muffle furnace
and elevate progressively the temperature to
attain 450°C in 4 h (6 h for fresh samples).
“ Maintain this temperature for 16 h.
“ After cooling, add 2 ml concentrated nitric
acid, heat the sample to boiling on a hot plate.
“ After cooling, transfer quantitatively the solu-
tion into a 50 or 100 ml calibrated flask and
fill-up to volume.
Plant samples - Method of the Comité Inter-In-
stituts d’Etude des Techniques Analytiques (CII)
with removal of Si [26– 28]. This procedure can be
possibly used for dissolution of soils and
sediments
“ Weigh in a platinum crucible 1– 3 g dried sam-
ple (105°C).
“ Place the crucible into a cold muffle furnace
and elevate progressively the temperature to
attain 450°C in 4 hours.
“ Maintain this temperature for 16 hours.
“ After cooling, moisten the residue with 2 ml
deionized water, add 3 ml concentrated nitric
acid and 2 ml concentrated hydrofluoric acid.
“ Evaporate slowly to dryness on a sand bath or
on a hot plate.
“ Repeat two times this step with addition of 2
ml nitric acid and 1 ml hydrofluoric acid.
“ Dissolve the residue with 2 ml nitric acid, wait
15 minutes, add 20 ml demineralized water and
heat up to a gentle boiling.
“ After cooling, transfer quantitatively the solu-
tion into a 50 or 100 ml calibrated flask and
fill-up to volume.
5.1.2. Problems
Nevertheless, dry ashing methods present some
limitations: even if it has been largely proved that
they satisfy indisputably the analysis of most ana-
lytes currently studied [18,26– 31], it is evident
that high temperatures applied will unavoidably
provoke more or less pronounced volatilization
losses of some elements. In analyses associated to
environmental studies, this danger concerns mer-
cury, arsenic and selenium [31,32]. Other rela-
tively volatile elements like cadmium, lead or
thallium tolerate, without losses, temperatures
usually applied for dry ashing.
The analysis of mercury, a particularly volatile
element, necessitates mandatory a sample prepa-
ration procedure based on wet digestion methods.
On the other hand, As and Se analysis can, in
some cases and under particular conditions, also
benefit of advantages offered by a dry ashing
procedure. The addition of ashing aids-generally
MgO and/or Mg(NO3)2-can sometimes allow to
form during the ashing procedure less volatile As
or Se compounds. The success of ashing aids
efficiency is of course strongly dependent on the
initial analyte form. For example, after a dry
ashing procedure of terrestrial plants, As and Se
recoveries are very consistent but this is not the
case for plants of aquatic origin [32]. On the other
hand, the utilization of ashing aids is particularly
delicate because some successful examples cannot
be generalized: for a routine use, the procedure
necessitates a serious and time-consuming valida-
tion for each type of samples analyzed. In addi-
tion, the utilization of ashing aids increases
significantly the total content of dissolved solids
in solutions, limiting then the application of this
approach for ICP-MS analysis. For all above-
mentioned reasons, the mineralization of samples
for arsenic and selenium analyses must be done,
as for mercury, using a good wet digestion
method (ensuring the quantitative decomposition
of organic matter in the case of ICP-MS analysis).
These three elements, furthermore, can benefit of
a common mineralization procedure.
Even for samples of organic nature, the dissolu-
tion mode of ashes described (i, in Section 5.1.1.1)
can sometimes lead to severe inadequacies. This
procedure does not ensure the dissolution of sili-
cate compounds and consequently of all elements
associated with them [31]. This problem is en-
countered mainly in plant analysis, often consid-
ered as easy to analyze as e.g. animal tissues.
Unfortunately, plant media contain variable con-
 M. Hoenig / Talanta 54 (2001) 1021–1038
centrations of silica that can attain several per-
cent. After a procedure without elimination of
silicon (= incomplete dissolution), poor recover-
ies for some other elements can be observed,
particularly traces. The best indication of this
phenomenon is the determination of aluminium,
partly associated with silicates and relatively
abundant in plant media [31,33]. If a poor recov-
ery is observed for this element (e.g. using a
reference material of similar composition), the
mineralization procedure ii, in Section 5.1.1.1 has
to be applied. In this case, the insoluble residue is
dissolved by hydrofluoric acid: silicates are de-
composed, silicon volatilized under SiF6- form
and the associated elements brought into the solu-
tion. This problem is not specific for dry ashing
procedures only: it is present to the same extent in
wet digestions of plant samples. Here also the
hydrofluoric acid step must be performed, necessi-
tating the use of PTFE vessels.
5.2. Oxidati6e acid digestions (wet methods)
Compared to dry ashing methods, mineraliza-
tion procedures using acid digestions present a
wide range of varieties that concern the choice of
reagents and their mixtures as well as devices used
for the procedure application. The large number
of procedures that may be found in the literature
(e.g. see the recent comprehensive revue of Lam-
ble and Hill [33] illustrates the fact that a lack of
a consensus exists, even inside a same sample
family. Therefore, any precise operating modes
cannot be presented here. However, basic princi-
ples exist: they must be applied after a serious
consideration that concerns the sample nature and
its composition as well as composition of the
reactive mixture and objectives of the analysis.
This is not always an easy task: consequently, for
routine analyses we often prefer the use of well
defined and easily mastered dry ashing
procedures.
5.2.1. Objecti6es of the analysis
Actually, the objective of an environmental
monitoring is most often concerned with the de-
termination of total content of analytes in samples
studied. In this work, we deal with this aspect
1031
only. Nevertheless, in some cases, the interest may
be oriented to the analyte form in the sample or
to its distribution in sample fractions (mineralogi-
cal, chemical, granulometrical...). The wet proce-
dures are only able to distinguish these very
different approaches because they may lead to
total dissolution as well as to a selective mobiliza-
tion of the analyte into solution. The success of
the analysis will depend among other factors on
the choice of reagents and of their combination
and concentrations. In environmental sciences, it
is then possible, using specific schemes of partial
and/or selective dissolutions, to better define the
analyte distribution in various constituents of the
sample. These possibilities concern mainly soils
and sediments. In practice, the below mentioned
wet digestion procedures will be chosen, adapted
and utilized following well-defined criteria [34]:
“ Total decomposition procedures are usually per-
formed in the presence of hydrofluoric acid
combined with other acids. They generally per-
mit the dissolution of all elements present in
the sample (except silicon which is volatilized
during the evaporation to dryness; some other
elements may be also partially lost, e.g. boron).
For the validation of a total attack, the use of
reference materials with recommended values is
of paramount importance.
“ Strong attacks are generally performed using a
mixture of strong acids, except hydrofluoric
acid. Better adapted to routine analysis, they
are easy to use but less complete in some cases
(soils, sediments...) than the total attack be-
cause of the presence of insoluble silicates.
These differences being generally without par-
ticular significance for geochemical purposes,
strong attack are satisfactory in several cases,
e.g. to evaluate the pollution level in environ-
mental monitoring: the dissolution of soils, sed-
iments and sludge is usually performed with
aqua regia (1 ml concentrated nitric acid+3
ml hydrochloric acid for a 1 g dry sample). On
the other hand, compared to a total attack, its
validation is more arguable concerning recov-
eries of some elements, which also depends
among other things on the composition of the
sample matrix. As far as we know, for valida-
tions of such partial procedures only few refer-
1032 M. Hoenig / Talanta 54 (2001) 1021–1038
ence materials are certified for many elements
and various types of attacks. Among them,
IRMM proposes BCR-141R to BCR-146R ref-
erence soils and sludge (aqua regia soluble
fractions) and 320-river sediment (informative
values for aqua regia soluble fractions). Simi-
larly, NIST offers SRM-2709 to SRM-2711
and SRM-2781 (soils and sludge). From a less
known supplier there are also available four
reference soils [35] certified for total, aqua re-
gia, hot nitric acid, cold nitric acid and CaCl2
contents.
“ Moderate attacks should, in principle, be used
to simulate transfers of elements in the envi-
ronment (e.g. assimilation of trace elements
from the soil by plants). At present, no interna-
tional normalization consensus exists for such
approaches because each country prefers to
keep his own methodology.
“ Particular cases are represented by selective
leachings in order to determine possible associ-
ations of the analyte with the main constituents
of the sample (e.g. with organic matter, iron
oxides, carbonates... in soils and sediments).
Here again no consensus exists but sequential
extraction procedures developed in the eighties
appear increasingly in the field of speciation
analyses.
5.2.2. Reagents
In the sample dissolution for the determination
of total contents of elements, the majority of wet
digestion procedures necessitate the use of six
reagents. Among them, four contribute princi-
pally to the destruction of organic matter (nitric,
sulphuric and perchloric acids and hydrogen per-
oxide). Hydrochloric and hydrofluoric acids
rather ensure the dissolution of inorganic com-
pounds. Before discussion of their utilizable mix-
tures, it is useful to consider reagents separately
and to know their action in dissolution proced-
ures.
“ Nitric acid is the most widely used primary
oxidant for the decomposition of organic mat-
ter. It reacts readily with both aromatic and
aliphatic organic materials, giving rise to oxi-
dation, esterification and nitration reactions
and leading to simple carboxylic acids, a factor
of a considerable importance in the destruction
of environmental materials [10]. Concentrated
nitric acid boils at about 120°C, a factor which
facilitates its removal after oxidation, but
which correspondingly limits its efficiency. Ni-
tric acid is commonly used in the presence of
sulphuric acid (which partly degrades the more
resistant material by elevating of the mixture
boiling point), with perchloric acid (which con-
tinues the oxidation after the nitric acid has
been removed), or with hydrogen peroxide for
its efficient oxidation properties.
“ Sulphuric acid is the main component of a very
efficient and often used mixture in wet diges-
tions procedures: nitric acid+ sulphuric acid+
hydrogen peroxide. The interaction of
sulphuric acid with organic compounds is very
complex; under suitable conditions, it can
cause oxidation, sulphonation, esterification,
hydrolysis of esters, dehydratation or polymer-
ization [10]. From the viewpoint of wet diges-
tion, the most important reactions are
probably oxidation and dehydratation. In addi-
tion to its function in partly degrading organic
materials by its own action, the presence of
sulphuric acid in mixtures containing other ox-
idizing agents also serves to raise the boiling
point (b.p. H2SO4 = 330°C) and so it enhances
the action of other oxidants. The main disad-
vantages associated with the use of sulphuric
acid are its tendency to form insoluble com-
pounds and, paradoxally, its high boiling
point. The insolubility of the alkaline earth
sulphates can cause difficulties when dealing
with samples rich in these elements, such as
bone or milk. Some trace elements form also
insoluble sulphates: the best known is the case
of lead. The high boiling point of sulphuric
acid makes it difficult to remove the excess acid
after completion of the oxidation. Another
drawback of sulphuric acid, as already men-
tioned, is the viscosity that may provoke trans-
port interferences in determinations performed
with analytical techniques where sample intro-
duction is ensured by a nebulization system.
Finally, in the presence of sulphates in samples,
also chemical interferences are sometimes ob-
served in ETAAS. For all the reasons specified
 M. Hoenig / Talanta 54 (2001) 1021–1038
here, it is advisable to avoid the addition of
sulphuric acid to digestion mixtures.
“ Perchloric acid: handled with knowledge and
care, this reagent is extremely efficient in the
destruction of organic material and from a
technical viewpoint, the mixture of nitric and
perchloric acids probably has less disadvan-
tages than any other oxidation mixture. Be-
cause of the occurrence of occasional explosion
with perchloric acid, it is reasonable to avoid
its use. In exceptional circumstances necessitat-
ing the use of perchloric acid, organic samples
have to be previously oxidized by nitric acid.
This preliminary treatment allows generally
eliminating the source of danger: an excess of
nitric acid has thus a protective role in diges-
tions with perchloric acid. General recommen-
dations concerning the use of perchloric acid in
digestion mixtures may be found in the litera-
ture [36].
“ Hydrogen peroxide: mixtures of acids with hy-
drogen peroxide are particularly efficient in
oxidation of organic matter. The reinforced
oxidizing power of hydrogen peroxide in the
presence of sulphuric acid is based on ‘in situ’
production of permono sulfuric acid H2SO5
that introduces oxygenated groups into many
kinds of organic molecules [10]. Combined with
the dehydrating action of sulfuric acid, this
type of reaction will rapidly degrade organic
materials to small species that readily volatilize
from the system.
5.2.3. Problems
As in dry ashing procedures, drawbacks may
appear in wet digestion methods. They are
assignable to adsorption, volatilization or
(co)precipitation phenomena. However, these
points have here different relative significations.
Firstly, temperatures attained are much lower
than those applied in dry ashing. Secondly, reten-
tion losses of elements on the walls of the mineral-
ization vessel are also less frequent. Reactions of
trace elements with other sample constituents are
also limited. Nevertheless, three mechanisms
should be noticed:
“ Coprecipitation of the analyte with a precipi-
tate formed by a main matrix element within
1033
the reactive mixture. The best-known example
is the coprecipitation of lead with calcium sul-
phate in the case of high sample calcium con-
tent and if the mineralization mixture contains
sulphuric acid [10]. A precipitate represents a
danger for the reliability of the analysis; it
should always be avoided even at the price of
changing the whole operational procedure.
“ Presence of chlorine in the sample or in the
digestion mixture. A digestion of the sample in
the presence of nitric acid ensures generally the
volatilization of chlorine ions as nitrosyl chlo-
ride. This occurs at lower temperatures than
the volatilization temperature of other ele-
ments. If nitric acid is not present in the reac-
tive mixture (e.g. in a sulphuric acid+ hydro-
gen peroxide mixture), volatilization losses may
be observed for several elements as germanium
and arsenic [10].
“ Incomplete dissolution (see also Section 5.1.2).
Following the order of importance, it is due to
inadequate choice or amounts of reagents, to
inappropriate heating program (temperature,
time) or to limited possibilities of the device
utilized. In other words, if the reactive mixture
is from a chemical viewpoint unable to decom-
pose the sample, no device, however sophisti-
cated, will achieve its dissolution in satisfactory
manner. This might occur in often-overrated
some modern heating devices, such as for ex-
ample microwaves.
5.2.4. Procedures and de6ices
The choice of a procedure and a device has to
be oriented as a function of type and nature of the
matrix. A comprehensive revue concerning wet
decomposition methods and their applications us-
ing microwave devices has recently been published
by Lamble and Hill [33]. This revue may also be
useful for selecting the choice of a conventional
wet digestion procedure since chemical processes
are similar in both cases.
5.2.4.1. Purely organic samples. The mineraliza-
tion and dissolution of these samples is generally
not problematic. Most wet digestion procedures
utilize a single acid (usually nitric) or a mixture of
acids and other oxidants (usually nitric acid+ hy-
1034 M. Hoenig / Talanta 54 (2001) 1021–1038
drogen peroxide). To reinforce action of these
reagents for more resistant samples, numerous
procedures add sulphuric or perchloric acids. The
mineralization is performed in open or close
systems.
Classically, open systems (digestions at atmo-
spheric pressure) employ any vessel, usually in
glass or PTFE (beaker, conical flask...), with or
without a refluxing condenser, and heated using a
conventional source (Bunsen, heating plate, sand
bath...). A previously dried or fresh sample is
weighed in the mineralization vessel and reagents
are added. The mixture is heated to boiling, which
is maintained the time necessary for the sample
decomposition and dissolution to occur. After
cooling, the resulting solution is quantitatively
transferred into a calibrated flask and completed
to volume. If an insoluble residue remains after
this procedure, it may be, following needs of the
analysis, separated by filtration or centrifugation.
If there are doubts concerning its composition,
the residue can be dissolved using another proce-
dure and analyzed separately.
During last years, open systems have pro-
gressed: usual mineralization ramps are composed
of several vessels equipped by reflux condensers to
limit possible volatilization losses of some ana-
lytes but also to avoid the evaporation of reactive
mixture. Such assembling is entirely satisfactory
for ensuring of current mineralizations of large
series of samples. In a parallel direction, open
systems heated by the microwave energy appeared
also on the market and are available in manually
operated or totally automated versions.
Closed systems (pressurized acid digestion) are
mainly represented by various models of PTFE-
lined acid digestion bombs with or without a
robust stainless steel body (in order to resist to
elevated pressures) or by high-pressure quartz ves-
sels. For several cases, the closed pressurized sys-
tems only can practically ensure the total
decomposition of organic matter present (e.g.
fats). Digestions performed in such devices benefit
of synergic effects of temperature and pressure.
These techniques are generally much more effi-
cient than conventional wet digestions in open
systems: the loss of volatile elements is avoided
and the decomposition of more difficult samples is
possible. The limiting factor of pressure bombs is
the small amount of organic matter that can be
treated (0.1–0.2 g). In pressure bombs, nitric acid
only (sometimes with a minimal addition of sul-
phuric acid) is generally used as the digestion
medium. In these procedures, the addition of
perchloric acid (and to a less extent also hydrogen
peroxide) is not recommended for evident risks of
explosion in the presence of organic matter.
Following their conception, the bombs may be
heated in laboratory ovens (usually at tempera-
tures ranged between 120 and 200°C), or increas-
ingly often, in dedicated microwave ovens.
However, advantages seem to be for the first
alternative, not only because of the significantly
lower price, but principally because of the possi-
bility of using PTFE bombs lined with a stainless
steel body that better resist to elevated pressures
generally to9 150 bars; in some cases up to 350
bars (Parr high pressure digestion bombs, mod.
4746). For bombs used in microwave oven appli-
cations, the body can only be made in plastic
materials: consequently, they must be operated at
lower pressures.
5.2.4.2. Samples of mixed composition. The uti-
lized procedures and reagents are very similar to
those used for the digestion of purely organic
samples. Only the mineral part of the sample
imposes a more complete dissolution, usually per-
formed by a combination of several acids. Nitric
acid, for example, often used alone (but some-
times with hydrogen peroxide) for decomposition
of a sample of organic nature, can be replaced
here by aqua regia that will ensure more easily the
dissolution of the mineral part of the sample.
However, it is unavoidable to recall the already
discussed problem of silicates. Their dissolution is
not ensured by aqua regia and supplementary
steps, e.g. by using hydrofluoric acid, become
mandatory if the analyst is interested by the total
concentration of analytes. From a practical point
of view, this problem is more complicated in wet
procedures than in dry ashing methods described
previously. Several factors depending on the
device utilized make the hydrofluoric acid step
more difficult to realize in wet digestions:
 M. Hoenig / Talanta 54 (2001) 1021–1038
“ In classical wet digestion procedures, the reac-
tion vessel is often in glass; this material itself
is decomposed by hydrofluoric acid. Dissolu-
tion procedures using such a step must then
be performed in PTFE vessels, not always
available in sizes needed. Dry ashing proce-
dures are generally made in platinum
crucibles, which resist well to this treatment.
“ The attack by hydrofluoric acid is usually
completed by evaporation to dryness with the
objective to decompose silica compounds and
volatilize silicon in order to mobilize associ-
ated elements. While this is easily realizable in
dry ashing methods, problems may be en-
countered in wet digestions. Most certainly,
the PTFE vessels used are not destroyed by
hydrofluoric acid, but during evaporation to
dryness (or to ‘near dryness’) there is a dan-
ger of a sample overheating that can result in
a more or less pronounced deterioration of
the PTFE inner surface [31]. The opacity of
PTFE does not facilitate a possible visualiza-
tion or control of this problem. Such hy-
drofluoric acid procedures are practically
applicable using small and large PTFE vessels
only (e.g. beakers), prohibiting then the use
of any condenser equipment.
“ Complications may be more severe in wet di-
gestion procedures employing a microwave
heating system. In pressurized bombs, the
evaporation to dryness cannot practically be
achieved. Some devices were recently commer-
cialized to overcome this drawback: they al-
low evaporation to dryness even in a ‘closed’
system (possibility of escaping and aspirating
vapours from each bomb to a common col-
lector).The open systems should be, in princi-
ple, better suitable for such applications
because a dedicated device can efficiently en-
sure the removal of vapours. In this case, a
hydrofluoric acid step can be achieved. Un-
fortunately, this is true only if this operation
is performed manually with a systematic
checking of residual volumes to avoid possi-
ble overheating. The automated open mi-
crowave systems allow, in principle, to
perform an unattended evaporation to dry-
1035
ness. Nevertheless, due to imprecision of dis-
tributed reagent volumes, to the impossibility
of sample visualization during treatment and
to a non reproducible positioning of the va-
pour aspiration head on the mineralization
vessel, tentative of unattended evaporation
steps often results in serious damages of
PTFE vessels [31]. On the other hand, open
digestion devices allow handling larger sample
volumes (open systems: up to 2 g organic
matter, close systems: up to 0.2 g organic
matter).
5.2.5. Choice of a procedure: an open problem
In spite of the lack of a consensus concerning
wet digestion procedures, there is an ample
choice of references in the literature dealing with
sample preparation steps [10,11,13–17,33,37,38].
Consequently, the analyst must evaluate their
efficiency to possibly modify some parameters
and validate its own procedure. The validation
has to be performed as thorough as possible
using reference materials. Presently the instru-
mentation market offers many devices making
the wet digestion more efficient and easier to
manage by means of possible automation. This
concerns principally microwave systems. In these
devices, the energy transfer from reactive mix-
ture to the sample is claimed to be better than
in the case of conventional means of heating. As
already quoted, this advantage is not signifi-
cantly perceivable in digestions of usually ana-
lyzed environmental samples. Nevertheless, in
some other cases, the microwave heating exhibits
a particularly high efficiency: the best-known ex-
ample concerns the mineralization of some plas-
tic materials where long carbonaceous chains
must be destroyed. It must also be admitted that
the use of microwave assisted mineralization
procedures is actually predominant. However,
this might be due to the absence on the market
of other modern systems or because of a dimin-
ished interest for ‘old’ and apparently less ele-
gant methods such as e.g. dry ashing?
Nevertheless, the lack of a generalized methodol-
ogy for wet digestions with or without mi-
crowaves shows clearly that a consensus is not
yet always reached in this domain.
1036 M. Hoenig / Talanta 54 (2001) 1021–1038
6. Direct analysis of solid samples
In the atomic spectroscopic instrumentation,
introduction of liquid samples is generally ensured
by means of various nebulization systems that
exhibit a poor efficiency and this is consequently
one main limiting factor concerning the actual
detection power of the instrument. Several ap-
proaches were nevertheless proposed to avoid the
mineralization of solid samples by introduction of
their slurried form into e.g. ICP-AES. Some ex-
perimental factors, such as particle size discrimi-
nation during nebulization and aerosol transport
and also incomplete dissociation of these particles
due to their low residence time in the plasma,
pose nevertheless practically insurmountable
drawbacks in a routine analysis. In other words,
there are too many differences in transport and
dissociation phenomena between the solid slurried
samples and solutions used for the calibration
[39].
On the other hand, solid sampling procedures
are particularly convenient for ETAAS analysis.
However, problems may arise because of danger
of unrepresentative sub sampling due to low sam-
ple mass analyzed. Direct analysis of solids by
ETAAS is initially handicapped by several restric-
tive factors: for this reason these applications
were practically abandoned already in the seven-
ties and replaced by a much more useful alterna-
tive called slurry sampling ETAAS. This method
seems to be the best approach to overcome some
of the difficulties associated with the direct sam-
pling of solids [12,40– 52].
The recent evolution of ETAAS using plat-
forms and adequate signal processing contributes
largely to routine applications of this solid sam-
pling alternative. Studies have also been extended
to the use of chemical modifiers to minimize the
effects of the matrix components. Slurry ETAAS
analysis is as fast as analysis of solutions and may
be performed using autosampler facilities. The
powdered sample is generally suspended directly
in autosampler microvials, in demineralized water
or in a diluted nitric acid solution (which ensures
a partial solubilization of the sample, thereby
improving its homogenity). Calibration is per-
formed using simple aqueous standards. The slur-
ried samples must be stirred periodically (mag-
netic stirring or ultrasonic mixing) just prior to
sampling by an autosampler capillary to avoid
sedimentation of the particles. In the absence of
stirring, the settling of the solid in the water-sus-
pended samples may be also overcome by the
preparation of a more stable slurry by using a
viscous medium or thickening agents [41,52].
To ensure a good repeatability of the measure-
ments, a representative number of particles must
be analyzed. For a usual analysis, this depends on
the particle density in the slurry as well as on the
size of the particles. In the most common environ-
mental analysis 1–50 mg of the solid sample is
slurried in 1 ml. For analytes with high concentra-
tions in the sample, the use of less sensitive alter-
native lines or maintenance of the internal purge
gas flow during the atomization step is recom-
mended.
It has repeatedly been shown that only very fine
granulometry of the slurry may ensure correct
results [45–51]. Concerning the representativeness
of the sampling, it is evident that the presence of
large particles in the sample was found to be the
most critical factor in the analysis. The impor-
tance of an intensive grinding of the sample prior
the analysis is thus obvious. The only exceptions
are sediment samples where trace elements are
systematically associated with fine granulometric
fractions (organic matter, clay minerals). The pos-
sible analytical errors due to the presence of some
large particles (generally quartz and other sili-
cates) in the analyzed aliquot are thus not prob-
lematic because they contain negligible amounts
of trace elements generally determined in environ-
mental samples. For slurry analysis of sediments
the grinding efficiency is consequently less impor-
tant than for other types of samples where the
analyte is distributed more homogeneously
[43,44].
At present, slurry sampling ETAAS may be
usefully applied to various analyses of environ-
mental samples (soils, sediments, atmospheric par-
ticles, ground plant sample, lyophilized animal
tissues, suspended solids collected in the natural
waters). On the other hand, it is probably not the
most reasonable approach for all cases: routine
analysis of large numbers of samples is difficult
 M. Hoenig / Talanta 54 (2001) 1021–1038
and necessitates, in principle, a robotized ultra-
sonic homogenization of the slurry (which is at
present distributed by one of the ETAAS manu-
facturers only). The mineralization of the sample
is often an easier way and the work with solutions
may take advantage of a complete apparatus au-
tomation. Usual large samples such as soils, sedi-
ments or plant material would thus be mineralized
before analyses rather than resorting to slurry
analysis. This is particularly true for cases in
which multi-element analysis is needed.
As compensation, slurry analysis may be ad-
vantageously applied to analysis of micro-samples
(atmospheric dust, suspended particulate matter
in natural water...) or of very insoluble samples in
acids (some minerals, e.g., titanium oxides...).
Practically, slurry analysis allows one to avoid all
gross sources of errors such as contaminations
(thanks to minimum sample handling) and losses
(no heating, no volatilization; no dissolution, no
insoluble residues). Consequently, this method
may be very useful as an alternative method for
checking other sample preparation techniques.
A similar methodology may also be applied to
ICP-AES or ICP-MS analysis where introduction
of the sample into plasma is also ensured after its
vaporization using a graphite furnace (electrother-
mal vaporization, ETV). Unfortunately, the com-
mercial distribution of dedicated ETV devices is
still limited.
The determination of mercury is a particular
case. Due to its high volatility, sample prepara-
tion steps are much more delicate than for all
other analytes usually studied. Nevertheless, the
market offers two systems developed exclusively
for a direct determination of mercury, excluding
preparation steps. After weighing in a nickel boat,
the solid fresh sample (or the solution) is intro-
duced into a combustion furnace flushed by a
stream of oxygen. Vapours produced pass
through a catalytic agent that retains undesirable
residues resulting from combustion of the organic
matter. The whole mercury is trapped on a gold-
sand amalgamator and subsequently released by
its heating. The determination is finally realized
by atomic absorption. Using this system, all solid
samples but also solutions and other liquid sam-
ples may be easily analyzed. The system may be
1037
automated: only weightings (for solids) or pipet-
tings (for liquids) precede the analysis.
7. Conclusions
We have seen that a large number of factors
associated to sample preparation steps may
strongly influence the reliability of the whole anal-
ysis. This concerns not only the efficiency and
practical usefulness of available preparation pro-
cedures but also their compatibility with the needs
imposed by measurement techniques. We can now
consider that appropriate tools for sample prepa-
ration are in the hands of analytical chemists but
their diffusion to practical users needs to be im-
proved. This is particularly perceivable for trace
elements, for which the diffusion of information is
a very important step to generally ensure the main
objective of the analysis: the obtaining of reliable
results.
References
[1] M. Hoenig, TrAC 17 (1998) 272 – 276.
[2] M. Stoeppler, Sampling and Sample Preparation,
Springer-Verlag, Berlin, 1997.
[3] Ph. Quevauviller, Quality Assurance in Environmental
Monitoring: Sampling and Sample Pretreatment, VCH,
Weinheim, 1995.
[4] P. Gy-Hétérogénéité, Echantillonnage, Homogénéisation.
Ensemble cohérent de théorie., Sciences de l’ingénieur,
Collection Mesures Physiques, E. Masson, Paris, 1988 (in
French).
[5] B. Markert, Environmental Sampling for Trace Analysis,
VCH, Weinheim, 1994.
[6] M.R. Carter, Soil Sampling and Methods of Analysis,
Lewis Publishers, 1993.
[7] A.M. Jose, M. Azcue, Manual of Aquatic Sediment Sam-
pling, CRC Press, 1995.
[8] A. Gomez, R. Leschber, P.L. Hermite, Sampling Prob-
lems for the Chemical Analysis of Sludge, Soils and
Plants, Elsevier Applied Science Publishers, London,
1986.
[9] R.G. Melcher, T.L. Peters, H.W. Emmel, Sampling and
sample preparation of environmental material, in: Topics
in current chemistry, Springer, Berlin, 1986.
[10] T.T. Gorsuch, Destruction of Organic Matter, Pergamon
Press, 1970.
[11] R. Bock, A Handbook of Decomposition Methods in
Analytical Chemistry, International Textbook Company
Limited, Glasgow, 1979.
1038 M. Hoenig / Talanta 54 (2001) 1021–1038
[12] U. Kurfürst, Solid Sample Analysis, Springer-Verlag,
Berlin, 1998.
[13] J.M. Kingston, L.B. Jassie, Introduction to Microwave
Sample Preparation: Theory and Practice, ACS Profes-
sional Reference Book, Washington, 1988.
[14] J.C. Van Loon, Selected Methods of Trace Metal Analy-
sis, John Wiley & Sons, New York, 1985.
[15] P. Brätter, P. Schramel, Trace Element Analytical Chem-
istry in Medicine and Biology, Berlin: Walter de Gruyter,
1980.
[16] Z. Sulcek, P. Povondra, Methods of Decomposition in
Inorganic Analysis, CRC Press, Boca Raton, 1989.
[17] L. Dunneman, J. Begerow, A. Bucholski, Sample Prepa-
ration for Trace Analysis, in: Ullmann’s Encyclopedia of
Industrial Chemistry, vol. B 5, Verlag Chemie, Weinheim,
1994, p. 65.
[18] M. Hoenig, M. Guns, Analysis of environmental and
biological samples by atomic spectroscopic methods, in:
P. Quevauviller, E. Maier (Eds.), Quality Assurance for
Environmental Analysis, Elsevier, Amsterdam, 1995.
[19] M. Pinta, Spectrométrie d’Absorption Atomique: Appli-
cations à l’Analyse Chimique (part II: Analytical Applica-
tions), Masson, Paris, 1979 (in French).
[20] E.J. Czobik, J.P. Matousek, Anal. Chem. 50 (1978) 2 – 12.
[21] B. Welz-Atomic, Absorption Spectrometry, VCH Wein-
heim, 1985 and 1999.
[22] S.J. Haswell, Atomic Absorption Spectrometry: Theory,
Design and Applications, Elsevier, Amsterdam, 1991.
[23] M. Hoenig, A.-M. de Kersabiec, L’atomisation Elec-
trothermique en Spectrométrie d’Absorption Atomique,
Masson, (in French), Paris, 1990.
[24] M. Thompson, J.N. Walsh, A Handbook of Inductively
Coupled Plasma Spectrometry, Blackie, Glasgow, 1983.
[25] L.H.J. Lajunen, Spectrochemical Analysis by Atomic Ab-
sorption and Emission, Royal Society of Chemistry, Cam-
bridge, 1992.
[26] M. Pinta et al CII-Analusis 3 (1975) 345-353 (in French).
[27] M. Pinta-Oleagineux, 28 (1973) 87-92 (in French).
[28] C.I.I Comitéinter Instituts D’etude des Techniques Analy-
tiques, Fresenius J. Anal. Chem. 345 (1993) 230 – 233.
[29] The Members of CII, V.J.G. Novozamsky, R.C. Houba,
I. Daniel, Fresenius’ J. Anal. Chem. 345 (1993) 198 – 201.
[30] M. Hoenig, H. Docekalová, H. Baeten, J. Anal. Atom.
Spectrom. 13 (1998) 195 – 199.
[31] M. Hoenig, H. Baeten, S. Vanhentenrijk, E. Vassileva,
Ph. Quevauviller, Anal. Chim. Acta 358 (1998) 85 –94.
[32] E. Vassileva, H. Docekalová, H. Baeten, S. Vanhentenrijk
and M. Hoenig Talanta, accepted.
[33] K.J. Lamble, S.J. Hill, Analyst 123 (1998) 103R – 133R.
[34] M. Hoenig, A.-M. de Kersabiec, Spectrochim. Acta 51B
(1996) 1297 – 1307.
[35] Certified Reference Soil Materials, CRM 7001, 7002,
7003, 7004, Analytika Co. Ltd., Prague, Czech Republic.
[36] Analytical Methods Committee, Analyst 84 (1959) 214.
[37] H. Matusiewitz, R.E. Sturgeon, Prog. Analyt. Spectrosc.
12 (1989) 21 – 39.
[38] R. Chakraborty, A.K. Das, M.L. Cervera, M. de la
Guardia, Fresenius J. Anal. Chem. 355 (1996) 99 – 111.
[39] G. Ploegaerts, H. Baeten, M. Hoenig, Analusis 25 (1997)
336 – 340.
[40] D.V. Brady, J.G. Montalvo, G. Joseph, G. Glowacki, A.
Pisciotta, Anal. Chim. Acta 70 (1974) 448 – 452.
[41] M. Hoenig, P. Van Hoeyweghen, Anal. Chem. 58 (1986)
2614 – 2617.
[42] R. Karwowska, K.W. Jackson, Spectrochim. Acta 41B
(1986) 947 – 950.
[43] M. Hoenig, P. Regnier, R. Wollast, J. Anal. Atom.
Spectrom. 4 (1989) 631 – 634.
[44] M. Hoenig, P. Regnier, L. Chou, J. Anal. Atom. Spec-
trom. 6 (1991) 273 – 275.
[45] N.J. Miller-Ihli, Fresenius’ J. Anal. Chem. 337 (1990)
271 – 274.
[46] N.J. Miller-Ihli, Spectrochim. Acta 44B (1989) 1221 –
1227.
[47] N.J. Miller-Ihli, Anal. Chem. A 64 (1992) 964A – 968.
[48] N.J. Miller-Ihli, Fresenius J. Anal. Chem. 345 (1993)
482 – 489.
[49] D. Bradshaw, W. Slavin, Spectrochim. Acta 44B (1989)
1245 – 1256.
[50] R. Dobrowolski, J. Mierzwa, Fresenius J. Anal. Chem.
344 (1992) 340 – 344.
[51] C. Bendicho, M.T. de Loos-Vollebregt, J. Anal. Atom.
Spectrom. 6 (1991) 353 – 374.
[52] D. Littlejohn, S.C. Stephen, J.M. Ottaway, Proc. SAC
86/3rd BNASS, Bristol, UK, 1986.