Talanta 76 (2008) 254258

Contents lists available at ScienceDirect

Talanta
journal homepage: www.elsevier.com/locate/talanta

Study of contamination sources in the process of cryogenic grinding

s Cernohorsk
Miloslav Pouzar a, , Anna Krejcova a , Toma
y a , Kveta Peskova b
a
b

legi 565, Pardubice 532 10, Czech Republic

Department of Environmental Protection, University of Pardubice, Cs.
Institute of Analytical Chemistry, Czech Academy of Sciences, Vever 97, Brno 611 42, Czech Republic

a r t i c l e

i n f o

Article history:
Received 21 November 2007
Received in revised form 18 February 2008
Accepted 20 February 2008
Available online 4 March 2008
Keywords:
Cryogenic grinding
Contamination
Cross-contamination
Cellulose
Electro-waste

a b s t r a c t
The study of contamination effect during cryogenic grinding of pure cellulose was carried out. The optimisation of important parameters of the grinding process (pre-cooling time, grinding time, cooling time
and number of cycles) was performed and the different sources of a possible contamination of samples
(earlier processed sample, metal parts of grinding tool) were evaluated. The results of ICP-oa-TOF MS analysis after microwave digestion of cellulose samples were used in this study. Signicant contamination of
cellulose samples by Fe at the level 130 g kg1 caused by wearing of steel stoppers and an impact bar was
detected. Cross-contamination by Fe, Cr, Mn and Cu at the level 400, 200, 200 and 2100 g kg1 , respectively was caused by previous grinding of electro-waste sample. This cross-contamination was possible
to be avoided by changing of a polycarbonate part of a grinding vessel.
2008 Elsevier B.V. All rights reserved.

1. Introduction
During the process of the cryogenic grinding, a material is deeply
frozen in order to be more fragile and easier to grind. Chilling
leads to an increase in hardness of a ground matter, an insertion
of failures in its structure and consequently a reduction of time
and energy vital to favourable sample pulverization. This technique
is favourable and used when a sample with the highly homogeneous particle size distribution and a diameter below 100 m is
required [1]. Typical application area of that modern sample preparation method is elemental analysis of various biological materials
or foods using the direct introduction of slurries into GF AAS. The
cryogenic grinding solves problems with agglomeration of samples
containing a high portion of fat [2]. It is benecial for pulverization
of brous specimens [3,4] as well as for powdering of soft, elastic
and pasty matters [5,6], or tough and hard ones contrariwise [7].
This procedure is efcient for a fast and effective comminution and
homogenisation of heavily heterogeneous and complex mixtures
if a small portion sampled have to represent its whole bulk [8].
Prevention of a thermo-labile species degradation and xation of
high volatile compounds in the sample matrix during the process
of mechanical des-integration are another benets of the cryogenic
processing [9].

Corresponding author. Tel.: +420 46 6037196.

E-mail address: milan.pouzar@upce.cz (M. Pouzar).
0039-9140/$ see front matter 2008 Elsevier B.V. All rights reserved.
doi:10.1016/j.talanta.2008.02.035

A hazard of sample contamination in the process of the cryogenic grinding strongly depends on a design of a mill used and
conditions selected for its operation. A type of ground matter
and desired limits of detection of applied analytical method are
another very important parameters usually mentioned in this context. Potential sources of the sample contamination in the process
of cryogenic grinding of plant materials were examined by Engelsen
and Wibetoe [2]. Yoshinaga and Morita [4] assigned the material of a grinding vessels as the probable source of a hair sample
contamination by Al, Fe and Mg. No signicant contamination of
cryogenically homogenised samples of breakfast cereals by Zn and
Fe was detected by Gouveia at al. [8].
The aim of this work was to evaluate main sources of contamination in the process of the cryogenic grinding. The bre cellulose
was chosen as a convenient material for this study because of
its physical character (elastic bres) and chemical composition
(poor organic matrix with minimum trace elements). Contamination caused by metal parts of grinding vessel was studied as well as
cross-contamination by previously ground sample.
2. Experimental
2.1. Instrumentation
The cryogenic impact grinder with a self-contained liquid nitrogen bath (4-5l) Model 6750 Freezer Mill (Spex, Certiprep, USA)
operated at the liquid nitrogen temperature was used in order to
comminute samples. The pulverization of frozen samples was car-

M. Pouzar et al. / Talanta 76 (2008) 254258

Fig. 1. Clots of crude brous cellulose.

ried out by a magnetically shuttling steel impact bar back-and-forth

against two stationary end plugs closing a cylindrical polycarbonate vial. Model 6751C20 grinding vials (volume 25 mL) were used
for all performed experiments.
The evacuable KBr die set with an inner diameter 16 mm (ICL,
Gareld, USA) and the hydraulic press with manual drive (Maassen,
Germany) were used for the cellulose pellet preparation.
The measurement of the cellulose bres length for the optimisation of grinding parameters was carried out using electron
microscopy (SEM Jeol 5600 LV, Japan), the sputter tool (MED 010,
Balzers union, Germany) and the software for image analysis (ACC
Version 6.0, Sofo, Czech Republic).
The qualitative and semi-quantitative analysis of possible
sources of contamination (pre-grinded sample of electro-waste,
impact bar and stoppers of grinding set) was performed using
energy-dispersive X-ray uorescence spectrometer Elva-X (Elvatech, Kiev, Ukraine).
The elemental analysis of ground cellulose samples was carried
out with the ICP-oa-TOF mass spectrometer Optimass 8000 (GBC
Scientic Equipment Pty Ltd., Australia) equipped with the concentric nebulizer Micro Mist coupled to a 70-ml thermostated (15 C)
cyclonic spray chamber (all Glass expansion Pty Ltd., Australia).
Microwave system SpeedwaveTM MWS-3+ with high-pressure PFA
vessels DAC 100 and the maximum total output of the microwave
generator 1450 W (Berghof, Germany) was used for digestions of
samples. The demineralised water was taken from the Milli-Q
Plus water-purication system (Millipore, Bedford, USA). Other
equipment used included calibrated volumetric glassware. The
calculations and statistical evaluation were carried out using commercial software ADSTAT 1.25 (Trilobyte Statistical Software Ltd.,
Pardubice, Czech Republic).
2.2. Optimisation of parameters of cryogenic grinding
Trying to grind cellulose in the original form of brous clots (provided by Department of Wood, Pulp and Paper University of Pardubice, Fig. 1), an impact bar broke through the sample and kept moving only through the channel formed. The main part of the sample
remained distributed along the walls of the polycarbonate vial and
stayed untreated in this case. To solve the above-mentioned problem, we decided to insert the sample in the form of pellets. Cellulose
was pressed in the evacuable die with inner diameter 16 mm by
force 10 tons for 2 min. Two pellets were positioned on the each side
of an impact bar in the space between its end and stopper. The total
mass of four pellets ground in every grinding program was 2.5 g.
The operating program used for the grinding equipment
included following parameters: the pre-cooling time (a period
before grinding, when a vial was immersed in the liquid nitrogen
for purpose of sample freezing), the grinding time (a period of

255

an impact bar motion in each working cycle), the re-cooling

time (a period between two working cycles, when the oscillation
of an impact bar was interrupted to avoid heating due to an
attrition of a magnetic bar with a sample and a sampling vial),
the number of working cycles (one cycle consists of a period of
grinding time followed by a period of cooling time), the impact bar
frequency (the number of back-and-forth movements of an impact
bar per second). Samples of brous cellulose were submitted to
different grinding procedures to optimise working parameters.
The pre-cooling time was set in the range 315 min. The grinding
and re-cooling periods took always the same time and both were
changed in the range of 26 min. The number of working cycles
was set with respect to value of the total grinding time (the sum of
all grinding times at particular grinding program) that was always
6 min. The impact bar frequency 10 Hz was applied at all tested
grinding programs. The statistical characteristics of bre length
distribution in the cellulose sample after grinding process were
adopted as the optimisation criterion.
2.3. Study of the contamination caused by the grinding set
A friction of an impact bar beating against an end stoppers in the
process of cryogenic grinding was considered as a possible source of
contamination. Working parts of the grinding tool were analysed by
ED XRF spectrometry to nd out probable contaminants derivable
from that source. Levels of elements detected as components of the
impact bar and the end stoppers were analysed in ground cellulose
by ICP-oa-TOF MS after microwave digestion and they were compared with these obtained from samples of crude non-ground bre
cellulose. The grinding was in that case carried out under optimal
conditions previously optimised. To avoid contamination from the
process of pellets pressing, ungrounded pellets were also analysed
by ICP-oa-TOF MS.
2.4. Study of a cross-contamination caused by previously ground
sample
To simulate the worst possible case of the cross-contamination
danger, electronic waste (parts of printed circuit board) was ground
just before cellulose samples using the same grinding set. After
processing of an electronic sample, the vial, stoppers and impact
bar used were washed in distilled water, dried and then cellulose
pellets were inserted. Three cellulose samples were ground subsequently and careful washing of the grinding set was performed
at the end of each grinding program. Thereafter, the polycarbonate part of vial was replaced by a new un-used one and next three
grindings of the cellulose were carried out. Elements found out at
ED XRF spectrum of electro-waste were consequently analysed in
all cryogenically ground samples of cellulose with using ICP-oa-TOF
MS after microwave digestion.
2.5. Analysis of cellulose samples by ICP-oa-TOF MS
2.5.1. Standards preparation
The single component standards of Fe, Cr, Mn, Cu, Pb, Ag and Mo
(1000 2 g mL1 , Analytika Co. Ltd., Czech Republic) were used
for the preparation of multi-element ones. Multi-element standards were prepared containing four levels of concentration: (i)
Cr, Mn, Cu, Pb, Ag and Mo all 200 g L1 and Fe 400 g L1 , (ii)
the same elements 10-fold times, (iii) 20 times and (iv) 40 times
diluted as (i). Calibration standards were prepared with 1 mL 65%
(w/v) HNO3 /100 mL of solution which was used for the microwave
decomposition in that volume. The water standard blank contained
1 mL 65% (w/v) HNO3 /100 mL of solution. The internal standard Y
50 g L1 was used in both calibration standards and blank.

256

M. Pouzar et al. / Talanta 76 (2008) 254258

Table 2
The bre length distribution of cellulose samples ground under different conditions

Table 1
The working parameters of ICP-oa-TOF MS
Parameter

Value

Ion optics
Skimmer
Extraction
Z1
Y mean
Y deection
Z lens mean
Z lens deection
Lens body

1 050 V
1 300 V
700 V
455 V
4V
1400 V
40 V
180 V

Pulse shaping
Fill
Fill bias
Fill grid
Pushout grid
Pushout plate
Blanker
Spectral frequency
Reectron

40 V
0V
40 V
390 V
492 V
200 V
33 kHz
690 V

Detection
Multiplier gain
Ion threshold
Integration window
Measurement
Mode

2800 V
8.4 mV
Auto
Pulse
counting/analog

2.5.2. Microwave decomposition of cellulose samples

Approximately 0.2 g of cryogenically ground cellulose sample
was accurately weighed into an acid washed TFM-PTFE digestion tube and 5 mL of nitric acid of analytical grade (65% (w/v)
HNO3 , Lach-Ner, Czech Republic) was added. The tube was heated
in a microwave oven at the power setting of 80% and 180 C for
5 min, at 90% and 220 C for 5 min. The maximum total output
of the microwave generator was 1450 W and maximum pressure in a digestion tube 100 bar. The digest was transferred into
a 50-mL acid washed volumetric ask and the ask was lled
up with demineralised water and stored in polypropylene container. Each sample was decomposed in four replicates. Two
water blanks were run with each batch of samples. Before analysis, digests were diluted 10-fold with demineralised water and
the internal standard Y was added in the nal concentration
50 g L1 .

Sample

C1

C2

C3

C4

C5

C6

Number of cycles
Grinding time (min)
Re-cooling time (min)
Pre-cooling time (min)
Mean length of bres (m)
Median length of bres (m)
P99 (m)

3
2
2
3
42.1
37.2
117

3
2
2
5
49.9
44.7
132

3
2
2
10
44.1
35.9
132

3
2
2
15
45.5
40.0
149

2
3
3
15
59.7
48.5
185

1
6

15
73.5
58.4
322

2.5.3. The ICP-oa-TOF MS method

The operating conditions of ICP-oa-TOF MS analysis were
adjusted according to the manufacturers recommendations: the
sample aspiration rate 1 mL min1 , power 1200 W, ows of plasma,
auxiliary and nebulizer gas 11, 1.0 and 0.9 L min1 and the
other parameters are summarised in Table 1. The sensitivity of
440 000 counts s1 per g l1 (mass integrated peak) and the resolution of 1600 was reached for 139 La. The same parameters for 238 U
were 675 000 counts s1 per g l1 and 1700. Mass calibration was
achieved using responses from 7 Li, 114 In and 238 U. The aqueous calibration with internal standard Y was used for quantication. Peak
area mode and 5-s data acquisition time and ve replicates were
used for measurement. Main analytical characteristics of ICP-oaTOF MS method were described for example by Sturgeon at al. [10].
3. Results and discussion
3.1. The optimisation of parameters of the cryogenic grinding
With the respect to the brous character of the cellulose matrix
we decided to monitor the bres length distribution as the parameter, which could describe the quality of a sample des-integration
in the best way. Three-electron microscopy shots (100 magnication) of each cellulose sample were treated by the software
for the image analysis. The length of each visible bre was measured as the main axis of a circumscribed ellipse. The minimum
number of evaluated objects at each microscopy shot was 100 and
the total number of evaluated objects for each particular sample
always exceeded 400 units. The main statistical characteristics of
the bre length distribution (arithmetical mean, median and 99th
percentile) for cellulose samples ground under different conditions

Fig. 2. The ED XRF spectrum of material of impact bar and end stoppers.

M. Pouzar et al. / Talanta 76 (2008) 254258

257

Table 3
Contamination caused by grinding set (samples C1 C3 ) and cross-contamination caused by previously ground sample (samples C4 C9 )
Fe (g kg1 )

Cr (g kg1 )

Mn (g kg1 )

Cu (g kg1 )

Pb (g kg1 )

Ag (g kg1 )

74 2.4
71 3.1

<LOD
<LOD

<LOD
<LOD

<LOD
<LOD

<LOD
<LOD

<LOD
<LOD

C1
C2
C3

195 5.9
187 5.4
209 7.2

<LOD
<LOD
<LOD

<LOD
<LOD
<LOD

<LOD
<LOD
<LOD

<LOD
<LOD
<LOD

<LOD
<LOD
<LOD

C4
C5
C6
C7
C8
C9

397 8.6
483 12
451 13
326 11
312 11
294 9.9

222 6.3
197 5.7
206 6.7
<LOD
<LOD
<LOD

241 6.5
188 5.9
224 5.8
<LOD
<LOD
<LOD

2146 66
2205 63
2035 58
<LOD
<LOD
<LOD

<LOD
<LOD
<LOD
<LOD
<LOD
<LOD

<LOD
<LOD
<LOD
<LOD
<LOD
<LOD

7.6
25.4

7.5
25.0

10.4
34.8

7.9
26.3

6.6
22.1

Ungrounded crude cellulose

Ungrounded cellulose pellet

LOD (3 a )
LQ (10 a )
a

5.6
18.8

: standard deviation of ve replicate measurements of the lowest calibration standard solution by ICP-oa-TOF MS multiplied by dilution factor.

are summarised in Table 2. From the data presented it is obvious

that the protraction of the pre-cooling time have no signicant
effect on quality of a ground cellulose sample. On the other hand,
the number of grinding cycles is possible to consider as an important parameter. We can conclude that a division of the total grinding
time on several shorter periods with an alternating re-cooling sentence gives better results than an uninterrupted processing.
3.2. Study of the contamination caused by the grinding set
The ED XRF spectrum of steel components of the cryogenic mill
is depicted in Fig. 2. As can be seen, main constituents of the impact
bar and the end stopper are Cr, Fe and Mo. Their contents at the
analysed material assessed by the fundamental parameters method
were 16.5, 83 and 0.5%, respectively. The material of KBr die set used
for preparation of pellets before cryogenic grinding was analysed
by the same way and its composition found out was very similar. The concentrations of above-mentioned possible contaminants
measured in crude ungrounded cellulose, in ungrounded pellets
and in grounded cellulose are summarised in Table 3. All detection
and quantication limits (LOD, LQ) for ICP-oa-TOF MS method used
were calculated as a concentration equivalent to 3 times (LOD) or 10
times (LQ) of the standard deviation of ve replicate measurements
of the lowest calibration standard multiplied by the dilution factor.

The concentration of Fe at that standard was 1 g L1 and concentrations of the other elements were 0.1 g L1 . A signicant content
of Fe on the level of quantication limit of ICP-oa-TOF method used
was observed at all ground cellulose samples (C1 C3 ). Lower concentrations of Fe on the level of the detection limit were found in
both ungrounded crude and pelletized cellulose samples. As can be
seen from the same concentrations of Fe found out in the crude and
the pelletized cellulose, no contamination by Fe can be connected
with the process of the pellet preparation. In this case, the detected
Fe is probably natural component of the cellulose matrix. On the
other hand, signicantly higher concentrations of Fe in the ground
samples pointed out the serious contamination from the process
of the cryogenic grinding. Next constituents of the grinding set Cr
and Mo was not detected neither in crude and pelletized cellulose,
neither in ground samples. The contamination by these elements in
the process of cryogenic grinding was hence considered to be negligible. Our results are inconsistent with that obtained by Goueya et
al. [8]. The same grinding tool (Model 6750 Freezer Mill from Spex
Certiprep, USA) was used for homogenisation of breakfast cereals in that work. Real samples were ground with a consequently
increasing time and the relation of Zn and Fe concentrations on the
duration of grinding step was evaluated. No contamination caused
by wearing of metal parts of the mill was detected. Incomparable
results were obtained probably through the difference in the study

Fig. 3. The ED XRF spectrum of electro-waste sample.

258

M. Pouzar et al. / Talanta 76 (2008) 254258

designs. We analysed ultra-traces of contaminants in a poor matrix

and the Fe content increase was about 130 g kg1 . Goueya et al.
treated a real matrix with high natural contents of Fe and Zn (30
and 20 mg kg1 approximately). A such high background content
of possible contaminants and corresponding standard deviations of
measurements (units or tens of mg kg1 ) made impossible to detect
elements content increasing in magnitude of hundreds g kg1 .
3.3. Study of the cross-contamination caused by previously
ground sample
The possible contaminants Fe, Cr, Mn, Cu, Pb and Ag were chosen
on the basis of ED XRF qualitative analysis of electro-waste sample.
The corresponding ED XRF spectrum is depicted in Fig. 3. Three
groups of samples were subsequently ground with using the same
grinding setrst, three samples of pelletized pure bre cellulose
(C1 C3 ), next two samples of electro-waste and than again three
cellulose samples (C4 C6 ). After that, the polycarbonate part of the
grinding set was replaced by new one and next three samples were
ground (C7 C9 ). The results of the analysis of above-mentioned cellulose samples are summarised in Table 3. As can be seen, Pb and
Ag were not detected at any sample. The concentrations of Cr and
Mn on the level of their quantication limits were found out at
samples ground after electro-waste samples. When a polycarbonate part of a grinding vessel was replaced, concentration of these
elements lowered to a level corresponding with limits of detection. The same behaviour was possible to observe in the case of
Cu, when its concentration in the samples ground before polycarbonate part changing was 10 times higher than the corresponding
quantication limit. The content of Fe at samples ground after
electro-waste was two times higher than its quantication limit
and two times higher than the content of Fe at samples ground
before electro-waste. After replacing of polycarbonate part, the
level of Fe decreased but stayed still signicantly higher, than in
samples ground before electro-waste.

4. Conclusion
Two main sources of sample contamination in the process of
cryogenic grinding were studiedthe contamination caused by
friction of a grinding set and cross-contamination caused by previously ground sample. In the rst case, the signicant contamination
by Fe on the level of hundreds of microgram per gram was observed.
Wearing of steel stoppers and an impact bar was not only one source
of contamination by Fe. This element was an important part of
cross-contamination by previous ground electro-waste sample as
well. In contrast to other cross-contaminants (Cu, Cr and Mn), the
replacing of a polycarbonate part of grinding set did not solve the
cross-contamination by Fe.
Acknowledgements

Financial support from the projects GACR203/06/0134

and MSM
0021627502 used in this work is greatly appreciated.
References
[1] D. Santos Jr., F. Barbosa Jr., A.C. Tomazelli, F.J. Krug, J.A. Nobrega, M.A.Z. Arruda,
Anal. Bioanal. Chem. 373 (2002) 183189.
[2] Ch. Engelsen, G. Wibetoe, Fressen. J. Anal. Chem. 366 (2000) 494503.
[3] K.G. Fernandes, A.R.A. Nogueira, J.A.G. Neto, J.A. Nobrega, Talanta 71 (2007)
11181123.
[4] J. Yoshinaga, M. Morita, K. Okamoto, Fressen. J. Anal. Chem. 357 (1997) 279283.
[5] C.S. Nomura, C.S. Silva, A.R.A. Nogueira, P.V. Oliveira, Spectrochim. Acta B 60
(2005) 673680.
[6] F.A. Silva, C.C.F. Padilha, L.E. Pezzato, M.M. Barros, P.M. Padilha, Talanta 69 (2006)
10251030.
[7] D. Santos Jr., F. Barbosa Jr., S.S. de Souza, F.J. Krug, J. Anal. Atom. Spectrom. 18
(2003) 939945.
[8] S.T. Gouveia, G.S. Lopes, O. Fattibello-Filho, A.R.A. Nogueira, J.A. Nobrega, J. Food
Eng. 51 (2002) 5963.
[9] D. Point, W.C. Davis, S.J. Christopher, M.B. Ellisor, R.S. Pugh, P.R. Becker, O.F.X.
Donard, B.J. Porter, S.A. Wise, Anal. Bioanal. Chem. 387 (2007) 23432355.
[10] R.E. Sturgeon, W.H. Lam, A. Saint, J. Anal. Atom. Spectrom. 15 (2000) 607616.