Powder Technology 319 (2017) 261–270

Contents lists available at ScienceDirect

Powder Technology

journal homepage: www.elsevier.com/locate/powtec

Performance of analytical centrifugation for the particle size analysis of

real-world materials
Christian Ullmann a,⁎, Frank Babick a, Robert Koeber b, Michael Stintz a
a
Institute of Process Engineering and Environmental Technology, Dresden University of Technology, 01062 Dresden, Germany
b
Joint Research Centre, Directorate F – Health, Consumers and Reference Materials, European Commission, 2440 Geel, Belgium

a r t i c l e i n f o a b s t r a c t

Article history: Analytical centrifugation (AC) is a powerful technique for the measurement of particle size distributions. This
Received 20 February 2017 study evaluates its performance for the grouping of particulate materials according to the European
Received in revised form 12 June 2017 Commission's recommendation on the definition of a nanomaterial, which bases on the number-weighted medi-
Accepted 23 June 2017
an diameter. The focus lies on cuvette-type AC with turbidity detector and on the performance in the case of poly-
Available online 24 June 2017
disperse, non-spherical real-world materials. Two grades of barium sulfate (nano and non-nano) were analysed.
Keywords:
This study compares the repeatability of the measurement and the method precision including sample prepara-
Analytical centrifugation tion, measurement and conversion into number-weighted particle size distributions. Experimental results show
Nanomaterial detection that the selected radiation wavelengths of 470 nm and 865 nm have only little impact on the number-weighted
Evaluation median diameter, but affect the intermediate precision of these values. However, the intermediate precision of
Intermediate precision the method is mainly determined by sample preparation up to 12.6% with respect to the number-weighted me-
Particle size analysis dian diameter (x50,0). This paper provides information on particle sizing on real-world materials for the first time.
© 2017 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/by-nc-nd/4.0/).

1. Introduction that the reliability of the method reaches a certain level of interest and
avoids false compliant results [5].
In 2011, the European Commission (EC) recommended a definition Any method description has to include sample preparation, mea-
of the term ‘nanomaterial’. This definition states that a material is con- surement procedures, data analysis and reporting. Sample preparation
sidered as nanomaterial if N50% of the particles in the number- is a crucial step for non-predispersed substances because it determines
weighted particle size distribution (PSD) have their smallest dimension the state of agglomeration of the particles. Hartmann et al. [6] published
below 100 nm [1]. The EC's recommendation of a definition includes ag- a critical review and an approach of harmonisation and standardisation
gregated, agglomerated and individual particles and will probably be of dispersion protocols for suspending nanoparticle powders in liquids.
implemented in several regulatory contexts, e.g. REACH and cosmetics Mawson et al. [7] discussed the risk of sample contamination with gen-
regulations. Currently, the different communities use different defini- erated particles from erosion of the dispersing tooI, e.g. ultrasonic trans-
tions [2]. The implementation of the EC definition is facilitated by a ducers. Moreover the suspension needs to remain stable during the
few measurement techniques that directly measure the number- analysis (e.g. throughout sample segregation in centrifugal fields) with-
weighted median particle diameter (x50,0) of all primary particles – out re-agglomeration or flocculation. Jillavenkatesa et al. [8] discussed
such as electron microscopy. Yet these techniques are rather expensive sources of error and variations of sedimentation-based particle sizing
in acquisition and operation, for which reason the performance of other techniques. Babick and Ullmann [9] reconsidered the conversion to
measurement techniques is highly interesting. Linsinger et al. [3] number-weighted size distributions theoretically and with regard to
reviewed widely available measurement techniques in the nano cuvette-type AC measurements results. Centrifugal analysis techniques
range. Recently, Babick et al. [4] published a systematic evaluation of are distinguished into AC operating with cuvettes and AC operating
commonly used particle sizing techniques based on experimental with disc-shaped rotors (disc-AC). Numerous publications examined
data. Analytical centrifugation is recommended as a screening method the sedimentation processes in these AC types theoretically and on
to implement the EC definition of a nanomaterial. In this context, the experimental data, e.g. in [10]. ISO already standardised both AC tech-
identification of nanomaterials by analytical centrifugation implies niques [11,12]. Cuvette-type AC does not require calibration for sedi-
mentation rate determination if the user operates the instrument in
⁎ Corresponding author at: Institut für Verfahrenstechnik und Umwelttechnik,
homogeneous mode. NIST recommends centrifugation instruments
Technische Universität Dresden, 01062 Dresden, Germany. typically for particle size measurements in the size range of 0.1 μm to
E-mail address: [email protected] (C. Ullmann). 300 μm [8]. According to ISO13318-2 [12], the typical particle size

http://dx.doi.org/10.1016/j.powtec.2017.06.057
0032-5910/© 2017 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
262 C. Ullmann et al. / Powder Technology 319 (2017) 261–270

range is from about 0.1 μm to 5.0 μm for AC with turbidity detectors were produced by chemical precipitation. Both materials passed tests
(photo centrifuges). Therefore, the reliability of optical AC techniques on homogeneity and stability. Roebben et al. [31] call these materials
needs to be evaluated systematically for particle diameters below “real-life materials”. If the substances met criteria on homogeneity and
100 nm. Turbidity AC intrinsically yields extinction weighted size distri- stability, Roebben et al. classify them as “representative test materials”
butions (Qext(x)). The correlation between particle diameter and the to have a third group of materials next to the groups of certified
extinction is especially in the size range between 100 nm and 10 μm reference materials and reference materials (non-certified). The term
non-linear and non-monotonic. Assuming Rayleigh scattering only ex- “representative test materials” is not used for IRMM-381 and IRMM-
tinction correlates for transparent nanoparticles to x6 and for light- 387 here because this substance cannot be purchased from JRC yet
absorbing nanoparticles to x3 (typically at particle diameters below and only Babick et al. [4] published first results on measurements of
100 nm), The conversion of extinction-weighted particle size distribu- IRMM-381 and IRMM-387.
tions (Qext(x)) to number-weighted ones (Q0(x)) will be prone to errors The solubility of BaSO4 is 0.24 mg in 100 ml H2O at 20 °C [32]. The
after applying the Mie theory [13]. The Mie theory contains several sim- manufacturer stated the particle density 4.4 g/cm3 for both grades of
plifying assumptions about the state of the observed system, e.g. spher- BaSO4. Both, the substance IRMM-381 and the substance IRMM-387
ical particle shape and high dilution to ensure absence of particle consist mainly of single constituent particles, but also contain small
interaction and multiple scattering. Interested readers are referred to amounts of agglomerates and aggregates. For the substance IRMM-
the relevant textbooks (e.g. van de Hulst [14], Bohren & Huffman [15]). 381, SEM images show that the particle diameters are between 80 nm
Most cuvette-type AC technologies detect space and time resolved and 2 μm (Fig. 1). The size ranges of the finer substance IRMM-387
transmitted light by an optoelectronic sensor (STEP® technology). vary from a few nanometres up to rare particles with diameters of
This principle was used in numerous applications to characterise sepa- about 200 nm. The morphology of the constituent particles is not
ration and stability of dispersed systems, e.g. for bio-analytical and med- spherical but polygonal. TEM images taken for the EU-project
ical reasons, in the late 90s [16,17]. After the turn of the millennium, the NanoDefine (www.nanodefine.eu) show a diffraction contrast that
manufacturer applied the technology additionally for particle character- indicates that the material is crystalline [33].
isation [18–22]. The detection of nanoparticles is possible when high
volume concentrations, large path length or small detection wave-
lengths are used. Sobisch et al. [23] applied a dual-wavelength AC
configuration successfully to the characterisation of gold and silver
nanoparticles. A number of studies have been published using STEP-
Technology®. These studies implies the use of space and time resolved
extinction profiles for stability testing, e.g. for beta-carotene emulsions
[24], nanocarriers for the natural antioxidant resveratrol [25], nano-
sized silica dispersions in presence of polycations [26] and long term
stability of engineered nanomaterials (ENM) in artificial freshwater
[27]. Several dispersibility studies were published, e.g. for carbon nano-
tubes (CNTs) in aqueous matrices [28] and organic solvents [29].
This paper evaluates the performance of turbidity cuvette-type AC
and the corresponding analytical chain with respect to the determina-
tion of number-weighted median particle diameter. AC instrumenta-
tions are considered with different incident radiation (470 nm and
865 nm) and materials with different nominal size ranges (nano and
non-nano, i.e. x50,0 b 100 nm and x50,0 N 100 nm, respectively). In
terms of physics, the use of the blue light (470 nm wavelength) is
more suitable than the use of NIR (865 nm wavelength) for detection
of nanoparticles and therefore the vendor of the cuvette-type AC recom-
mends 470 nm wavelength. However, the use of 865 nm wavelength
might be interesting in the case of broadly distributed particle size dis-
tributions, especially if there is a part of the size distribution in μm size
range. The evaluation of trueness is limited to the comparison of particle
size modes as there is no certified referenced material available with a
certified median diameter for AC methods. Usually AC instruments
were qualified with latex or silica suspensions and user compares the
size distribution modes in the volume-weighted size distribution.
Linsinger et al. [30] gave an overview on the requirements for reference
materials for measuring the diameter of nanoparticles.

2. Experimental

2.1. Materials

The Joint Research Centre (JRC) of the European Commission

provided two grades of BaSO4 powders IRMM-381 (non-nano grade)
and IRMM-387 (nano grade). The IRMM-387 powder was sourced
from an OECD sponsorship program repository and comes from the
same batch as used in the EU-project NANOREG. Both substances
IRMM-381 and IRMM-387 were provided for the NanoDefine EU-
project and represent conventional nano and non-nano particulate ma- Fig. 1. SEM images of BaSO4 particles, precipitated by “drop on grid” method A) IRMM-381
terials with kiloton to-megaton production quantities. These substances (magnification 10,000) B) IRMM-387 (magnification 25,000).
 C. Ullmann et al. / Powder Technology 319 (2017) 261–270 263

For qualification of the instruments, JRC purchased three monodis- to acceleration forces. The principle used here (STEP-technology®) is
perse suspensions NANOSPHERE™ 3050A, 3100A, 3350A from Thermo well explained in [10]. The radiation source of the optical cuvette-type
Scientific (Fremont, USA) and provided a tri-modal suspension of poly- AC emits parallel light that illuminates the sample cuvettes. A CCD-
styrene particles in an aqueous matrix. The certified diameter – 46 ± detector measures the transmission simultaneously over the entire
2 nm, 100 ± 3 nm and 350 ± 6 nm (k = 2) – were determined by trans- sample length for all samples attached to the rotor. The detected inten-
mission electron microscopy. The ratio of the fractions was 0.5:1:1 by sity is automatically normalised to the intensity that measured for
mass and the suspension was further diluted with deionised water. empty centrifuge. The normalised intensity is further normalised with
The final tri-modal suspension had a solid content of 2.5 g/kg that is re- intensity profile of the particle-free dispersion medium, which results
lated to 0.24 vol%. Values for the particle density and refractive index in sample transmission, which can take values between 0 (complete ob-
were taken from the material information sheet. The density value scuration) and 1 (absolute transparency) in the suspension phase. The
used is 1.05 g/cm3 and the refractive index is 1.59. locally resolved normalised signal intensity T(r,t) is converted into the
extinction E(r,t):
2.2. Preparation of suspensions
Eðr; tÞ ¼ − ln ðTðr; tÞÞ ð1Þ
Suspensions of the two different grades of BaSO4 were prepared by
suspending powder in a 2 g/l sodium hexametaphosphate solution For concentrated suspensions multiple scattering and dependent
(CAS: 10124-56-8, SHMP) [33]. A SHMP solution containing 5 g/l (Alfa scattering occurs. However, the Lambert-Beer-law can be predicted for
Aesar) and deionised water was prepared and stored for the study. dilute suspensions extinction where L denotes the optical path length,
DLS measurements (HPPS, Malvern Instruments) at a scattering angle cN the number concentration of particles and Cext their extinction
of 173° showed that the solution was almost particle-free. The derived cross section
count rate, i.e. the intensity of scattered light, was just below 150 kcps,
whereas the suspension with the lowest sample content showed Eðr; tÞ ¼ L  cN ðr; tÞ  Cext ð2Þ
a derived count much higher than this value. Sufficient substance
(0.001 g - 0.1 g) was given into a flask and 4 ml of a 5 mg/ml SHMP so- The extinction profiles E(r) can be regarded as scaled sum functions
lution as well as 6 ml deionised water was added to give a concentration Qext(xStokes)/Qext(x) of the extinction-weighted size distribution at a
of 2 mg/ml SHMP and a volume of 10 ml for every suspension. Destruc- given time and the extinction-time-curves E(t) for given radial position.
tion of agglomerates was realised by probe sonication with the system The extinction-weighted particle size distribution (PSD) can be trans-
UDS751 (Topas, Germany). Ultrasonication was conducted with a formed into mass-/volume- or number-weighted particle size distribu-
7 mm sonotrode at nominal amplitude of 80%. Sonication time was tions by an appropriate optical model. Typically, the optical model used
330 s for IRMM-381 and 323 s for IRMM-387, which corresponds to a is Mie's solution [13] of electromagnetic scattering at spheres. The opti-
volume specific caloric energy input of approximately 1600 J/ml. The cal model computes Cext as function of particle diameter, wavelength λ
energy input was determined according to the approach of Taurozzi of incident radiation and the (complex) refractive indices of particles
et al. [34]. The distance between the tip of the probe and the bottom and dispersion medium.
was kept to a few millimetre. The flask was placed in an ice-water
bath to ensure continuous cooling bath during sonication. The 2.5. Centrifuge instruments
NANOGENOTOX dispersion protocol [35] provides a comprehensive de-
scription of the setup. The cuvette-type AC measurements regarding The instruments used for centrifugal analysis of particle size distri-
concentration studies were carried out with a maximum solid content bution are LUMiSizer® 610 and LUMiSizer® 651 from LUM GmbH
of 0.25 vol% according to ISO 13318-2 [12]. Therefore, the suspensions with radiation at wavelengths of 470 nm (blue light) and 865 nm
were prepared to have a mass content of at least 0.1 g/kg up to (NIR). A CCD-line sensor with 2048 elements detects the intensity and
10 g/kg. For the intermediate precision studies, the solid content of has a detector dot pitch of 14 μm. The detector's extinction working
the suspensions was 1.0 g/kg for IRMM-381 and 3.5 g/kg for IRMM- range is 0.1–4.0. The principal functionality had been successfully
387. The pre-dispersed polystyrene suspension (0.24 vol%) was not di- checked by regular qualification tests with certified reference materials.
luted. A volume of 2 ml was treated in an ultrasonic bath for 1 min. Suspensions samples were measured in 2 × 8 mm rectangular polycar-
bonate cuvettes. For suspensions with very low sample opacity, the ven-
2.3. Morphological characterisation dor recommends to use 10 mm optical path to increase sensitivity and
to reduce errors. This recommendation is additionally tested with sus-
SEM studies gave an overview on the states of agglomeration and pensions of IRMM-387. The slim cuvettes were filled with a volume of
aggregation of BaSO4 particles. The authors used a SEM, model Gemini 400 μl suspension and the 10 × 8 mm rectangular cuvettes with 1 ml
982, Carl Zeiss AG, Jena, Germany. For this purpose, the suspensions suspension. The measurements were carried out at 25 °C and a light
were first prepared as described above. Afterwards, the particles were actor of 1. Table 1 gives an overview on substance specific settings.
precipitated at the SEM substrate by “drop on grid” method. The magni- The instrument determined the time of temperature equilibration auto-
fications were 10,000 for IRMM-381 and 25,000 for IRMM-387. matically. After that, the instrument normalised the sensors of each slot
automatically. In the case of repeatability testing, ten subsamples were
2.4. Analytical centrifugation with turbidity detectors measured in the rotor slots 1–5 and 6–11. In the case of intermediate
precision testing, each day three samples were tested in the rotor slots
The particle's mass and the hydrodynamic mobility influence the 1,4 and 7. The time between sub-sampling the suspension and inserting
particle's sedimentation behaviour. This behaviour is also depending the cuvettes into the instrument was kept as low as possible and did not
on the fluid parameters, viscosity ηF and density ρF, and the density of exceed one minute. The software SEPView® analysed the data in the
the (nano-) particles ρNP. Stokes' law describes the motion of small par- mode “constant position”. Detloff et al. [20] showed that this
ticles in the fluid for small Reynolds numbers (Re b 0.5). Since it relies mode gives nearly the same PSD like the mode “constant time”, but its
on the assumption of spherical particles, any size calculate is a diameter overall performance is superior for polydisperse samples with broad
of a sphere with equivalent settling behaviour as the particle. This distributions.
equivalent diameter is usually called Stokes diameter xStokes with regard In every case, five positions were taken for analysis to ensure that
to ISO-standards on centrifuge methods [11,12]. The homogeneous the average is as stable as possible against outliers and that the results
technique monitors the segregation of a homogeneous suspension due of the analysis are representative for the whole sample. The chosen
264 C. Ullmann et al. / Powder Technology 319 (2017) 261–270

Table 1
Substance specific settings for analytical centrifugation measurements.

Instrument settings Trimodal polystyrene mix IRMM-381 (non-nano grade) IRMM-387 (nano grade)

Rotational frequency 4000 rpm 2000 rpm 3000 rpm

Detection intervals 250 × 10s, 250 × 15 s, 250 × 167 s, 250 × 412 s 500 × 5 s, 500 × 10s 250 × 5 s, 250 × 10s, 250 × 20s, 250 × 45 s
Measurement time 1 d 17 h 57 min 2 h 5 min 5 h 34 min

positions are distributed over the whole cuvette length. These are used median in the recommendation of a definition of the term
approximately 112.5 mm, 115.0 mm, 117.0 mm, 120.0 mm and ‘nanomaterial’. Moreover, the 10%-quantiles and 90%-quantiles of the
123.0 mm for IRMM-387 and 115.0 nm, 118.0 nm, 120.5 nm, particle size distribution, x10,0 and x90,0, were determined to analyse
122.5 nm and 125.5 nm for IRMM-381. The coordinates were shifted the precision at the borders of the size distribution. The precision
± 2 mm for every run to ensure that the position depended PSDs are means the closeness of agreement between independent test results ob-
fitting to each other as good as possible for all subsamples. Reasons for tained under stipulated conditions (ISO 5725-1 [38]). The intermediate
positions shifts are scratches and fingerprints at the cuvette's surface. precision describes the reliability under conditions intermediate be-
These things lead to light diffraction and scattering of the transmitted tween reproducibility conditions and repeatability conditions.
light through the cuvette and occur unsteady transmission profiles. In The cuvette-type AC method was tested on five different days
general, these artefacts are normalised against the last transmission (p = 5) with three replicates each day (n = 3) with one sample per
profile taken at the end of the measurement, but there are still residuals run. Each day three samples of the substance BaSO4 were suspended
left that lead to unsteady PSDs at these positions. Moreover, dust at the and dispersed as reported in Section 2.2 of this paper and measured to-
cuvette's outer surface and temporary adsorbed particles at the inner gether in the instrument. The normalisation of the optical components
surface occur unsteady transmission profiles and lead to unsteady in the instrument was done individually for every cuvette slot. To en-
PSDs at the related positions. Therefore, the positions were chose for sure that all replicates represent a complete and independent applica-
every LUMiSizer run individually. For the intermediate precision stud- tion of the cuvette-type AC method, the data analysis was done with
ies, the positions were chosen individually for each sample. The reason SEPView® for each replicate. The obtained data were analysed with
for this is the individual determination of the PSD required for each the ANOVA tool in EXCEL®. The mean of squares were calculated be-
sample. tween days and between replicates. The repeatability of the method is
All profiles were chosen for analysis of IRMM-381, whereas the first the ability to determine a value as close as possible to the same value
profile was ignored for IRMM-387 regarding that the final rotational with the three replicates over the 5 days, whereas the “within group
speed was not reached yet. The software detected the meniscus position mean squares” (WMS) is the mean of the five daily variances and
automatically, but allowed manual corrections for every subsample if equal to σ2within [39].
necessary. No manual data fitting or smoothing was done at extinction
p n
2
profiles at any time. The conversion to number-weighted PSD was ∑ j¼1 ∑i¼1 xij −xi
done by the use of the SEPView® software provided by LUM GmbH WMS ¼ ¼ σ 2 between ð3Þ
pðn−1Þ
with the instrument. The refractive indices used for both grades of
BaSO4 are 1.668 (470 nm) and 1.617 (865 nm) [36]. The software data-
WMS was calculated by the squares of the deviations of the 15 mea-
base provided the values for water, which are the dynamic viscosity
surements with p = 5 and n = 3. xi,j is the measurement result of the
0.89 mPas and the density 0.997 g/cm3 at 25 °C. The extinction-
measurement j on the day I and xi is the average diameter of the day i
weighted PSD and the corresponding number-weighted PSD were cal-
(Eq. (3)). This is the lowest possible standard deviation associated
culated from the same positions. For analysis of the polystyrene mix
with the cuvette-type AC method.
measurements, the particle size modal values were determined by
The “between group mean squares” (BMS) is estimated with p = 5
choosing time intervals for every mode and performing single peak
and n = 3 where x is the arithmetic average diameter of all 15
analysis.
measurements.

2.6. Determination of limit of reliable detection, precision and repeatability p

n∑i¼1 ðxi −xÞ2
BMS ¼ ð4Þ
p−1
The request for single particle sedimentation defines the maximum
sample content limit. In this case, viscous interactions among particles
are negligible and single particle scattering, i.e. non-linear optical effects The BMS estimates the variance that is the sum of two independent
such as multiple scattering are excluded [11]. According to ISO 13318-2 variances. Therefore, BMS is also estimated as the sum of the variance of
[12] the maximum volume concentration can be raised to 0.25 vol% for repeatability σ2within and the normed variance between days σ2between
optical cuvette-type AC to ensure single particle sedimentation. This with n = 3 [39].
limit might by negligible because the exact number of large strongly
scattering particles may be not important in case of performing BMS ¼ σ 2 within þ n σ 2 between ¼ WMS þ n σ 2 between ð5Þ
measurements to consider a material as a nanomaterial. The need for
a sufficiently high turbidity defines the lower concentration limit, as a The contribution to intermediate precision derived from the varia-
sufficient is necessary to monitor the sedimentation. This study investi- tion between days is calculated from the WMS and the BMS (Eq. (6)).
gated this limit. Therefore, the variance between day σ2between is calculated with n = 3.
The intermediate precision was determined in accordance to ISO
2725-3 [37] regarding the determination of the number-weighted me- BMS−WMS
σ 2 between ¼ ð6Þ
dian diameter x50,0. The median diameter separates the fine and coarse n
halves of the particle population with respect to the type of quantity
(e.g. number, volume, mass) when particles are sorted according to di- The intermediate precision variance σ2intermed.Prec expresses the total
ameter. The median diameter of a size distribution is more stable variances within and between days.
against the influence of outliers than the arithmetic mean or geometric
mean diameter of the particle size distribution and therefore the EC σ 2 intermed:Prec: ¼ σ 2 within þ σ 2 between ð7Þ
 C. Ullmann et al. / Powder Technology 319 (2017) 261–270 265

Finally, the relative standard deviation of the intermediate precision suspensions and therefore two stages of clearing up are visible in the in-
is tensity profiles between the initial and the last intensity profile. The first
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi stage can be determined at 37% and represents the intensity of the sus-
σ 2 intermed:Prec: pension after the sedimentation of the fraction with a particle diameter
RSDintermed:Prec: ¼ 100% ð8Þ
x of 350 nm. Analogous to this the second stage at approx. 84% represents
the suspension without the fraction of the 100 nm particles. The shape
of the profiles indicates that the three fractions are narrowly distributed
3. Results and discussion and not interacting with each other. There is no sediment visible and
due to the overall particle content (0.24 vol%) single particle sedimenta-
3.1. Qualification tion was assumed. Despite of the rather low initial transmission/inten-
sity values (b5%), the suspension was not diluted. Hence, non-linear
The qualification was performed with the tri-modal polystyrene sus- concentration effects cannot be excluded, which impedes the accurate
pension. During the measurements, the software recorded the trans- quantification of the three size fractions. However, for the purpose of
mission in adjusted intervals according to the expected sedimentation qualification only the modal size values are validated of each size frac-
velocity (Table 1). tions. These values were accurately determined since hindered settling
Fig. 2A shows a screenshot of the measured intensity profiles of a was avoided. Fig. 2B shows the number-weighted modal diameters of
measurement at the wavelength 470 nm. The x-axis represents the ra- the corresponding measured particle size distributions and the compar-
dial position of the measurement. The y-axis represents the normalised ison to the certified values and these from measurements at a wave-
signal intensity. Green lines represent the latest measured intensity length of 865 nm.
profiles while red lines represent initial profiles. The air-suspension in- The comparison of modes in Fig. 2B shows a satisfying agreement
terface (meniscus) is located at approx. 109 mm and the bottom of the between measured particle diameters and certified values. For both de-
cuvette (sediment position) is located at approx. 129.0 mm. During the tection wavelengths a slight discrepancy is observed for the determined
measurement, the suspension cleared up due to the forced sedimenta- particle diameters of 46 nm and 100 nm. The diameters of 52.1 nm and
tion of the particles. The suspension consists of three monodisperse 54.3 nm lead to deviations of 13% and 15% against the certified value de-
termined by TEM. While particle sizer determination by TEM is associat-
ed with dry particles only and determined without hydrodynamic
effects, the number-weighted mode is an accepted value for the evalu-
ation of the cuvette-type AC measurement technique and the conver-
sion model. Therefore, some degree of discrepancy is expected and the
deviation goes in the expected direction. The second mode is detected
with 105.9 nm and 101.7 nm. These diameters are also larger than the
diameter determined by TEM. Detecting the particle diameter with
larger particles lead to smaller percent deviations, which is also visible
in the third mode. The relative frequencies (ΔQ0) of the three PS
fractions (83:17:b0.5 weighted on certified diameter values and
mixing ratio) could roughly be confirmed by cuvette-type AC with
74.3:25.5:0.3 at 470 nm. The measurements at 865 nm lead to a high
underestimation of the 46 nm fraction due to the low turbidity of the
suspension in the NIR detection mode and could be determined in an
advanced way of data analysis only.

3.2. Optical behaviour of BaSO4 suspensions

Fig. 3 correlates the initial extinction of the suspensions before the

measurement with the BaSO4 sample content. The initial extinction of
the samples obeys the Lambert-Beer-law (Eq. (2)) for small concentra-
tions of IRMM-381 and for all investigated concentration levels of the
finer substance IRMM-387. For IRMM-387 the borderline for validity
of the Lambert-Beer-law is at least at a sample content of 1 g/kg at a
wavelength of 470 nm and close to a sample content of 2.5 g/kg at a
wavelength of 865 nm. Above an extinction of about 2.0, multiple
scattering occurs which correlates to a sample transmission of 13.5%.
This could lead to an underestimation of large particles. For this reason,
the assessment of intermediate precision of cuvette-type AC measure-
ments was performed for sample contents of 1 g/kg for IRMM-381
and 3.5 g/kg for IRMM-387.
The suspensions of IRMM-381 gave higher extinction values with
blue light than they did for IR radiation. This behaviour is even more ob-
vious for suspensions with particles in the “nano” size range. The effect
is based upon the optical behaviour towards the different wavelengths:
The calculation (Fig. 4) of the size-depended extinction cross-sections of
the particles show that nano sized particles (i.e. smaller than 100 nm)
scatter blue light (470 nm) up to 15 times more intense than they
Fig. 2. Cuvette-type analytical centrifugation results for qualification with tri-modal PS
(2.5 g/kg) A) normalised signal intensity for extinction measurement at wavelength of
scatter NIR light (865 nm).
470 nm, B) comparison of modal values (peak sizes) of the number-weighted size In the sub-micrometre range, the influence of wavelength is steadily
distribution for cuvette-type AC at different wavelengths with certified values. reduced and for particles at diameters of 1 μm the extinction cross-
266 C. Ullmann et al. / Powder Technology 319 (2017) 261–270

Fig. 3. Initial extinction of the suspension vs. BaSO4 content of IRMM-381 and IRMM-387
in 2 mm PC cuvettes with varied wavelength of 470 nm and 865 nm.

section is almost independent from light wavelength. The calculation

also shows that particles with diameters above 1100 nm scatter NIR
light slightly more than blue light (470 nm). Small amounts of particles
in the μm-size range might lead to a higher turbidity and therefore to
higher initial extinction values for NIR radiation. In conclusion, lower
wavelength are recommended for the detection of particles in the
sub-micrometre and nanometre size range for optical detection, if
these particles have similar optical properties like BaSO4, e.g. SiO2,
Al2O3 and polymers.

3.3. Working range regarding particle content

The study investigated the working range between 10 g/kg and Fig. 5. Median diameter x50,0 and quantiles x10,0 und x90,0 of number-weighted PSDs of
0.1 g/kg sample content for both grades of BaSO4 and both wavelengths. IRMM-381 using 2 mm polycarbonate cuvettes A) analytical centrifugation results at
Fig. 5 shows the results for the coarse material IRMM-381 and Fig. 6 470 nm, B) analytical centrifugation results at 865 nm.
shows them for the finer material IRMM-387. The median diameter
x50,0 and the quantiles x10,0 and x90,0 of the particle size distribution
are plotted versus the initial normalised intensity. Measurements of
IRMM-381 with 470 nm and 865 nm show a dependency between the
sample content of the suspension, corresponding to the initial normal-
ised intensity (see Fig. 3). The number-weighted median diameter
was determined between 200 nm and 300 nm, except for the highly di-
luted samples. Multiple scattering influenced some median diameters
(Fig. 3). The normalised intensity of these suspensions are 2.7% and
4.3% at 470 nm and 4.5% at 865 nm. The median diameter are not fol-
lowing the linearity of the Lambert-Beer-law (Eq. (2). The borderline
for multiple scattering is estimated at a sample transmission of approx.
13.5%. At very low values, especially below 5%, PSD measurements with
optical cuvette-type AC underestimate the quantity of coarse particles
and accordingly the size of the 90%-quantiles systematically.
In contrast, suspensions with a very low sample content show that
excessive dilution yields the risk of underestimating the average particle
diameter and may lead to false positive results of the NM classification
according to the EC definition. The results with sample contents of
0.35 g/kg and 0.1 g/kg lead to lower quantity of larger and highly
scattering particles and therefore to an underestimation of the median
diameter. This loss of particles may occur in the size range of μm
Fig. 4. Dependency of extinction cross-sections Cext from particle diameter for BaSO4 particles during sub-sampling. Dissolution of BaSO4 particles can be ex-
particles for wavelength 470 nm and 865 nm. cluded as the lowest concentration chosen was about 40 times higher
 C. Ullmann et al. / Powder Technology 319 (2017) 261–270 267

content the quantiles x10,0 and x90,0 of the PSD are determined with a
RSD of 19.1% and 20.0%, respectively. Further dilution leads to PSDs
dominated by the largest particles of the sample, whereas the standard
deviation of the quantiles decreases. The nearly transparent suspension
with a sample content of 0.1 g/kg leads to noisy results. In contrast, the
working range at light sources of 865 nm wavelength is limited to con-
centrated samples, whereas the median diameter is determined with a
RSD below 13% at sample contents of 10 g/kg and 3.5 g/kg only. Below
this sample content, the RSD of the median diameter is determined be-
tween 28% and 42%. Therefore, a measurement with an initial normal-
ised intensity above 80.5%, which is close to the intensity of the
solution without particles, is not recommended for this kind of sub-
stance. This limit is depending on the cuvette type and instrument
used and the substance analysed. Nevertheless, the method clearly clas-
sified the material as a nanomaterial according to the recommendation
for a nanomaterial definition for all concentration levels.
The working range of IRMM-387 is given up to 80% initial normal-
ised intensity due to the small particle diameters. The large number of
particles in the suspension at a solid content of 3.5 g/kg leads to a
weak turbidity of the suspension, but the instrument is able to monitor
the change of the normalised intensity. To overcome this limit to a cer-
tain determination of the median value, another type of cuvette with
the optical path of 10 mm was used. The study was carried out for
1 g/kg, 0.35 g/kg and 0.1 g/kg and the results of five subsamples were
averaged (Table 2). The results of the last dilution level were very
noisy. After conversion, these values led to number-weighted median
diameter x50,0. The use of cuvettes with larger optical path lengths
shifted the limit of detection regarding the sample content of the sus-
pension from 1 g/kg to 0.35 g/kg for both wavelength.

3.4. Precision of cuvette-type AC method

Figs. 7 and 8 show the results of the studies with the real-world ma-
terials on intermediate precision for IRMM-381 and IRMM-387, respec-
tively. The RSD of measurement repeatability for ten subsamples of one
prepared sample shows no significant difference between the two
wavelengths for the median diameter x50,0 and the 10% quantile
(x10,0) in the intrinsic PSD and therefore no characteristic variance is
Fig. 6. Median diameter x50,0 and quantiles x10,0 und x90,0 of number-weighted PSD of
IRMM-387 using 2 mm polycarbonate cuvettes A) analytical centrifugation at 470 nm, visible in the number-weighted PSD.
B) analytical centrifugation at 865 nm. In contrast, the precision of the whole method with individually pre-
pared samples shows a significant increase of the RSD for the wave-
length of 865 nm in the intrinsic PSD, especially between days the
than the solubility (0.24 g in 100 ml H2O at 20 °C, [32]). For IRMM-381, effect is visible. The illumination with 865 nm wavelength causes a
measurements with an initial normalised intensity above 55% cannot be more sensitive detection of less abundant large particles than illumina-
recommended for determining the number-weighted median diameter. tion with 470 nm wavelength. These few large particles exhibit a large
The measurements with sample content of 0.6 g/kg that are related to variability due to sampling problems. Probably large particles result
intensities of 33% at 470 nm and 57% at 865 nm lead to disparate particle from the suspension preparation procedure, i.e. from erosion of the ul-
diameters: At a wavelength of 470 nm a median diameter was mea- trasonic probe during the agglomeration destruction. Mawson et al.
sured with 264.8 nm, whereas NIR light leads to median diameter of [7] showed that the size range of such particles is 200 nm up to
210 nm. The disparity is even more obvious for the value at the quantile several μm. They consist of metals like titanium that light more strongly
x90,0, which varies from 581.3 nm to 430.7 nm. One can assume that the than equally sized particles do due to their higher optical contrast to the
PSD is measured the best way if the span between the quantiles x10,0 dispersion medium. The amount of these containment particles can be
and x90,0 is very large and might cover most of the range between the determined with an element specific technique like ICP-MS in further
smallest and the largest particles. This range between the quantiles studies.
x10,0–x90,0 of the PSD is very large for measurements in the normalised
intensity range between 10.7% - 33.3% at 470 nm and for 12.1% - 30.0%
at 865 nm. Regarding the number-weighted median diameter, an initial Table 2
normalised intensity up to 44.7% is suitable for measurements at Extinction-weighted and number-weighted median diameter of IRMM-387 using 10 mm
polycarbonate cuvettes measured at 470 nm and 865 nm wavelength.
865 nm.
Fig. 6 shows the median particle diameter x50,0 of IRMM-387 and the Sample x50,ext. at x50,0 at x50,ext. at x50,0 at
borders the of the PSD, which are represented by quantiles at 10% and content 470 nm [nm] 470 nm [nm] 865 nm [nm] 865 nm [nm]
[g/kg]
90% of the number-weighted PSD. Measurements at 470 nm show me-
dian diameters in the expected size range with low relative standard de- 1.00 83.8 ± 0.2 nm 43.9 ± 0.1 nm 110.8 ± 0.9 nm 49.5 ± 1.1 nm
viations (RSD) up to an initial normalised intensity of 77.2%. This 0.35 94.6 ± 1.2 nm 43.6 ± 0.2 nm 110.1 ± 2.5 nm 43,7 ± 4.7 nm
0.10 81.8 ± 1.1 nm 35.9 ± 5.4 nm 143.1 ± 13.5 nm 18.3 ± 1.2 nm
intensity corresponds to a BaSO4 content of 0.1 g/kg. At this sample
268 C. Ullmann et al. / Powder Technology 319 (2017) 261–270

Fig. 7. AC measurement precision (relative standard deviation) of the median diameter

x50,0 and the quantiles x10,0 and x90,0 compared with the intermediate precision (IP) for
IRMM-381 and the respective contributions from within-day and between-day effects.
Fig. 8. Analytical centrifugation measurement precision (relative standard deviation) of
A) extinction-weighted PSD, B) number-weighted PSD.
the median diameter x50,0 and the quantiles x10,0 and x90,0 compared with the
intermediate precision (IP) for IRMM-387 and the respective contributions from within-
day and between-day effects: A) extinction-weighted PSD B) number-weighted PSD.

The precision between days and within days increases to a higher

uncertainty of the median diameter x50,ext. and the quantiles x10,ext. challenge of representative sampling for analysis and therefore leading
and x90,ext. For the quantile x90,0 the intermediate precision increases to a variability of particle diameters determined.
even up to 10.9%). The impact of this effect decreases after conversion The impact of rare large particles from suspension preparation
to the number-weighted PSD, resulting in nearly equal RSDs for both procedure is more visible in the studies with IRMM-387 than in
wavelengths. In other words, uncertainty due to few number of those with IRMM-381. The RSD of the upper limit of the extinction-
larger particles experienced on volume base has only small effect on weighted PSD, represented by the quantile x90,ext., increased to 54.9%
number-weighted median diameters x50,0. It is remarkable that the un- for the precision within days and 65.8% between days, which leads to
certainty of the intermediate precision increases at the lower limit of the a total RSD of 85.7% for the intermediate precision. According to the
PSD (quantile x10,0) for 470 nm, which is caused by the individual data studies with IRMM-381, the impact of rare large particles decreases
treatment for each sample. The study showed that this individual data after conversion to the number-weighted PSD. Turbidity measurements
treatment leads to a higher RSD but also to an accurate average of the at 470 nm are much more sensitive to nanoparticles than ones at
particle median diameter or diameter quantile. Hence, expert knowl- 865 nm are (cf. discussion in Section 3.2), which is due to the strong
edge is still required for individual data treatment and signal analysis impact of wavelength on the extinction behaviour of nanoparticles
when aiming at Q0. It ensures ‘trueness’ of measured Q0(x) but also re- (cf. Fig. 4). In total, the intermediate precision of the number-
veals relatively low levels of precision. Moreover, the borders of the weighted median diameter x50,0 was determined with 9.4% for
PSDs, represented by the number-weighted quantiles x10,0 and x90,0, 470 nm and 12.6% for 865 nm.
are determined more reliable than the median diameter x50,0 for a Measurements at both wavelengths of 470 nm and 865 nm lead to
wavelength of 865 nm in the precision studies. This is related to the very similar particle diameters for both IRMM-381 and IRMM-387
 C. Ullmann et al. / Powder Technology 319 (2017) 261–270 269

Table 3 Abbreviations
Results of AC measurements with intermediate SD and RSD. Values obtained by analysis of AC analytical centrifugation
variance (ANOVA).
ANOVA analysis of variance
Substance Radiation No.-weighted SD RSD BMS between mean squares
wavelength λ median [nm] [%] CAS chemical abstracts service
[nm] x50,0 [nm]
EXT extinction
IRMM-381 470 277.2 26.3 9.60 IRMM Institute of Reference Materials and Measurements
(non-nano grade) 865 278.3 31.3 11.23
ISO International Organization for Standardization
IRMM-387 470 42.46 3.99 9.38
(nano grade) 865 42.42 5.35 12.6 JRC Joint Research Centre
kcps kilo counts per second
NIST National Institute of Standards and Technology
(Table 3). The number-weighted median diameter x50,0 of the non-nano PS polystyrene
grade substance was 277.2 nm at a wavelength of 470 nm and this di- PSD particle size distribution
ameter was not significantly larger at a wavelength of 865 nm with a RI refractive index
particle diameter of 278.3 nm. The RSD was 9.6% for 470 nm and with RSD relative standard deviation
11.2% somewhat higher for NIR radiation. Moreover, the particle diam- SD standard deviation
eter of the finer substance IRMM-387 was 42 nm with similar RSDs of SEM scanning electron microscopy/scanning electron microscope
9.4 and 12.6%. One can conclude that the precision for the two forms SHMP sodium hexametaphosphate
of BaSO4 is independent from exact size range, but it depends on the SOP standard operating procedure
wavelength used for determination for the sub-micrometre range. TEM transmission electron microscopy
WMS within mean squares

4. Conclusions
Symbols
This paper examined the performance of cuvette-type analytical Cext extinction cross section
centrifugation with turbidity detection for grouping particulate sub- cN number concentration
stances according to the EC definition of nanomaterials. Two grades of E extinction
BaSO4 particles in the sub-micrometre range, IRMM-381 (non-nano) L optical path length
and IRMM-387 (nano) were characterised with AC employing NIR and n number of replicates within a group of analyses
blue light for turbidity measurement. The software converted the intrin- p number of independent groups of analyses (no. of days)
sically measured extinction-weighted particle size distributions into Qq sum function, TOQ indicated by q
number-weighted size distributions by applying Mie theory. Both mate- r radial position
rials could be classified positive right and positive false according to the T transmission
definition of a nanomaterial. The number-weighted median diameters t time
of the two grades of BaSO4 particles were determined with 42 nm and x particle diameter
277–278 nm with similar values at both wavelengths. Our study xP,0 P-% quantile of particle size distribution
showed that the relative standard deviation is 9.4% to 9.6% for a wave- σ standard deviation
length of 470 nm and varies from 11.2% to 12.6% for NIR with 2 mm cu-
vettes. The measurements at 865 nm wavelength were less reliable
regarding the number-weighted median diameter and these studies
were outside the vendor's instrument recommendation to use smaller Acknowledgements
wavelength. As expected, the studies show that smaller wavelengths
are recommended for detecting and quantifying nanoparticles, but The article presents results of the European project NanoDefine
they also show that these operating setting allow the classification of (Development of an integrated approach based on validated and
material as a nanomaterial or not on a screening level. A precondition standardised methods to support the implementation of the EC recom-
for this generalisation is that the particles have similar optical properties mendation for a definition of nanomaterial), which has received
like BaSO4 particles, e.g. SiO2, Al2O3 and polymer particles. In general, funding from the European Community's Seventh Framework
major reasons for the method uncertainty are the impacts of sample Programme (FP7/2007–2013) under Grant Agreement no 604347. In
taking for suspending, sub-sampling the suspension and containment addition, the authors would like to thank LUM GmbH for providing
particles caused by abrasion of the dispersing tools. These reasons lead the instrument LUMiSizer® 651.
to changes in the composition of the samples analysed. A further expla-
nation of observed variability is related with data analysis. Finally, the References
studies identify the sample preparation step as the main source for the
deviation. The incidence of measuring rare large particles, generated [1] European Commission, Commission recommendation of 18 October 2011 on the
definition of nanomaterial (2011/696/EU) Off, J. Eur. Union 54 (2011) 38–40,
during the agglomerate destruction with ultrasonic probe sonication http://dx.doi.org/10.3000/18770677.L_2011.275.eng.
devices, lead to uncertainty in the intrinsically measured extinction- [2] H. Rauscher, K. Rasmussen, B. Sokull-Klüttgen, Regulatory aspects of nanomaterials
weighted size distribution. This impact decreases after conversion into in the EU, Chem. Ing. Tech. 89 (3) (2017) 224–231, http://dx.doi.org/10.1002/cite.
201600076.
number-weighted size distributions. Sample content studies showed
[3] T.P.J. Linsinger, G. Roebben, D. Gilliland, L. Calzolai, F. Rossi, N. Gibson, C. Klein, Re-
that an initial normalised intensity between 10 and 30% provides max- quirements on Measurements for the Implementation of the European Commission
imum reliability, but any transmission value in the range of 5% to 80% is Definition of the Term Nanomaterial, Luxembourg, 2011, http://dx.doi.org/10.2787/
63490.
typically acceptable. If this is not possible because the sample content
[4] F. Babick, J. Mielke, W. Wohlleben, S. Weigel, V.-D. Hodoroaba, How reliable can a
would exceed 0.25 vol%, especially in the case of weakly scattering material classified as a nanomaterial? Available particle sizing techniques at work,
nanoparticles, the initial normalised intensity should be below 80%. J. Nanopart. Res. (2016) 18, http://dx.doi.org/10.1007/s11051-016-3461-7.
Our results demonstrate that analytical centrifugation can be employed [5] European Commission, Commission Decision 2002/657/EC of 12, Implementing
council directive 96/23/EC, Off. J. Eur. Union 221 (August 2002) 8–36.
on a screening level for classifying a material as nanomaterial according [6] B.N. Hartmann, K.A. Jensen, A. Baun, K. Rasmussen, H. Rauscher, R. Tantra, J.M. Riego
to the EC-recommendation of a definition on the term nanomaterial. Sintes, Techniques and protocols for dispersing nanoparticle powders in aqueous
270 C. Ullmann et al. / Powder Technology 319 (2017) 261–270

media - is there a rationale for harmonization? J. Toxicol. Environ. Health B Crit. Rev. [24] F. Liu, D. Wang, C. Sun, Y. Gao, Influence of polysaccharides on the physicochemical
18 (6) (2015) 299–326, http://dx.doi.org/10.1080/10937404.2015.1074969. properties of lactoferrin –polyphenol conjugates coated beta-carotene emulsions,
[7] R. Mawson, M. Rout, G. Ripoll, P. Swiergon, T. Singh, K. Knoerzer, P. Juliano, Produc- Food Hydrocoll. 52 (2016) 661–669, http://dx.doi.org/10.1016/j.foodhyd.2015.08.007.
tion of particulates from transducer erosion: implications on food safety, Ultrason. [25] C. Caddeo, M. Manconi, A.M. Fadda, F. Lai, S. Lampis, O. Diez-Sales, C. Sinico,
Sonochem. 21 (2014) 2122–2130, http://dx.doi.org/10.1016/j.ultsonch.2014.04.005. Nanocarriers for antioxidant resveratrol: formulation approach, vesicle self-
[8] A. Jillavenkatesa, S.J. Dapkunas, L.-S.H. Lum, NIST 960-1, NIST Recommended assembly and stability evaluation, Colloids Surf. B: Biointerfaces 111 (2013)
Practice Guide-Particle Size Characterization, 2001. 327–332, http://dx.doi.org/10.1016/j.colsurfb.2013.06.016.
[9] F. Babick, C. Ullmann, Error propagation at the conversion of particle size distribu- [26] G. Petzold, C. Gotzsche, M. Mende, S. Schwarz, W. Jaeger, Monitoring the stability of
tions, Powder Technol. 301 (2016) 503–510, http://dx.doi.org/10.1016/j.powtec. nanosized silica dispersions in presence of polycations by a novel centrifugal sedi-
2016.06.039. mentation method, J. Appl. Polym. Sci. 114 (2) (2009) 696–704, http://dx.doi.org/
[10] T. Detloff, D. Lerche, Centrifugal separation in tube and disc geometries: experi- 10.1002/app.30608.
ments and theoretical models, D, Acta Mech. 201 (2008) 83, http://dx.doi.org/10. [27] A. Brunelli, A. Zabeo, E. Semenzin, D. Hristozov, A. Marcomini, Extrapolated long-
1007/s00707-008-0074-y. term stability of titanium dioxide nanoparticles and multi-walled carbon nanotubes
[11] ISO 13318-1:2001, Determination of Particle Size Distribution by Centrifugal Liquid in artificial freshwater, J. Nanopart. Res. 18 (2016) 113, http://dx.doi.org/10.1007/
Sedimentation Methods – Part 1: General Principles and Guidelines, International s11051-016-3412-3.
Organization for Standardization, Geneva, 2001. [28] B. Krause, M. Mende, P. Pötschke, G. Petzold, Dispersability and particle size distribu-
[12] ISO 13318-2:2007, Determination of Particle Size Distribution by Centrifugal Liquid tion of CNTs in an aqueous surfactant dispersion as a function of ultrasonic treatment
Sedimentation Methods – Part 2: Photocentrifuge Method, International Organiza- time, Carbon 48 (2010) 2746–2754, http://dx.doi.org/10.1016/j.carbon.2010.04.002.
tion for Standardization, Geneva, 2007. [29] U. Staudinger, B. Krause, C. Steinbach, P. Pötschke, B. Voit, Dispersability of
[13] G. Mie, Beiträge zur Optik trüber Medien, speziell kolloidaler Goldlösungen, Ann. multiwalled carbon nanotubes in polycarbonate chloroform solutions, Polymer 55
Phys. 25 (1908) 377–452. (2014) 6335–6344, http://dx.doi.org/10.1016/j.polymer.2014.10.012.
[14] H.C. van de Hulst, Light Scattering by Small Particles, Dover Publications Inc., [30] T.P.J. Linsinger, G. Roebben, C. Solans, R. Ramsch, Reference materials for measuring
New York, 1981. the size of nanoparticles, Trends Anal. Chem. 30 (1) (2011)http://dx.doi.org/10.
[15] C.G. Bohren, D.R. Huffman, Absorption and Scattering of Light by Small Particles, 1016/j.trac.2010.09.005.
WILEY, New York, 1981. [31] G. Roebben, K. Rasmussen, V. Kestens, T.P.J. Linsinger, H. Rauscher, H. Emons, H.
[16] A.V. Homeyer, D.-O. Krentz, W.-M. Kulicke, D. Lerche, Optimization of the polyelec- Stamm, Reference materials and representative materials: the nanotechnology case,
trolyte dosage for dewatering sewage sludge suspensions by means of a new centri- J. Nanopart. Res. 15 (2013) 1455, http://dx.doi.org/10.1007/s11051-013-1455-2.
fugation analyser with an optoelectronic sensor introduction, Colloid Polym. Sci. 277 [32] R. Kaltofen, I. Pagels, K. Schumann, J. Ziemann, Tabellenbuch Chemie, 1st ed VEB
(1999) 637–645, http://dx.doi.org/10.1007/s003960050435. Leipziger Druckhaus, 1957.
[17] T. Sobisch, D. Lerche, Application of a new separation analyzer for the characteriza- [33] D. Gilliland, F. Pianella, I. Rio, K. Loeschner, M. Correia, E. Verleysen, J. Mast, C.
tion of dispersions stabilized with clay derivatives, Colloid Polym. Sci. 278 (2000) Ullmann, 7, NanoDefine Technical Report D2.3, NanoDefine Consortium,
369–374, http://dx.doi.org/10.1007/s003960050527. Wageningen, 2016 (URL: http://www.nanodefine.eu).
[18] D. Lerche, Dispersion stability and particle characterization by sedimentation [34] J. Taurozzi, V. Hackley, M. Wiesner, Preparation of Nanoparticle Dispersions From
kinetics in a centrifugal field, J. Dispers. Sci. Technol. 23 (5) (2002) 699–709, Powdered Material Using Ultrasonic Disruption, National Institute of Standards
http://dx.doi.org/10.1016/j.powtec.2006.10.021. and Technology, 2012, http://dx.doi.org/10.6028/NIST.SP.1200-2.
[19] T. Detloff, T. Sobisch, D. Lerche, Particle size distribution by space or time dependent [35] K.A. Jensen, Y. Kembouche, E. Christiansen, N. Jacobsen, H. Wallin, C. Guiot, O. Spalla,
extinction profiles obtained by analytical centrifugation, Part. Part. Syst. Charact. 23 O. Witschger, The generic NANOGENOTOX dispersion protocol: final protocol for
(2006) 184–187, http://dx.doi.org/10.1002/ppsc.200601028. producing suitable manufactured nanomaterial exposure media, NANOGENOTOX
[20] T. Detloff, T. Sobisch, D. Lerche, Particle size distribution by space or time dependent Joint Action, 2011.
extinction profiles obtained by analytical centrifugation (concentrated systems), [36] M.-J. Weber, Handbook of Optical Materials, CRC Press, Boca Raton, 2003.
Powder Technol. 174 (2007) 50–55, http://dx.doi.org/10.1016/j.powtec.2006.10.021. [37] 37ISO 5725-3:1994, Accuracy (trueness and precision) of measurement methods
[21] Patent DE 102004011387 B4 2010, US Patent US 9 019 493 B2, 2015. and results – part 3: intermediate measures of the precision of a standard measure-
[22] S. Han Ng, P. Meng Woi, M. Basri, Z. Ismail, Characterization of structural ment method, Including Technical Corrigendum 1:2001, 1994 (Geneva).
stability of palm oil esters-based nanocosmeceuticals loaded with tocotrienol, [38] ISO 5725-1:1994, Accuracy (Trueness and Precision) of Measurement Methods and
J. Nanobiotechnol. 11 (2013), 27. http://dx.doi.org/10.1186/1477-3155-11-27. Results – Part 1: General Principles and Definition, 1994 (Geneva).
[23] T. Sobisch, F. Emmerling, M. Girod, M. Menzel, D. Lerche, Advantages of dual wave- [39] B.J. Winner, Statistical Principles in Experimental Design, 3rd ed McGraw-Hill, 1991.
length detection for size determination of nanoparticles with analytical centrifuga-
tion, Dispers. Lett. 4 (2013) 9–11 (ISBN: 978-3-944261-36-2).