A Systematic Study of the Cryogenic Milling of Chrysotile Asbestos

For more than 40 years, intensive research has been devoted to shedding light on the mechanisms of asbestos toxicity. Given the key role of fibre length in the mechanisms of asbestos toxicity, much work has been devoted to finding suitable comminution routes to produce fibres in desired size intervals. A promising method is cryogenic milling that, unlike other mechanical size reduction techniques, preserves the crystal–chemical properties of materials. In this study, the effect of cryogenic milling on the physical–chemical properties of commercial Russian chrysotile was studied in order to produce precise size fractions with invariant properties compared to the pristine fibres. In particular, two batches with fibres > 5 µm and < 5 µm were prepared, as this limit sets their potential toxicity. The results are fundamental for future in vitro toxicity testing of this commercial product, widely used in chrysotile-friendly countries but not yet adequately studied. Results show that fibre length can be controlled by milling time under cryogenic conditions without inducing structural defects or amorphization; short fibres (95% L < 5 µm) can be obtained by cryogenic milling for 40 min, while 10 min is enough to yield long chrysotile fibres (90% L > 5 µm).


Introduction
The term asbestos refers to a group of six mineral fibres of significant industrial and economic importance that have been widely used since antiquity [1]. Members of the asbestos group are chrysotile (serpentine group), and five species of amphiboles: crocidolite (fibrous riebeckite), amosite (fibrous cummingtonite-grunerite) and the fibrous variety of anthophyllite, actinolite and tremolite [2,3]. Today, there is enough scientific evidence to state that inhalation of asbestos may induce fatal lung diseases, such as lung cancer, malignant mesothelioma (MM), and asbestosis [3,4]. Accordingly, asbestos is included by the International Agency for Research on Cancer (IARC) in Group 1 as a carcinogen for human [3]. The most insidious asbestos-related disease is undoubtedly MM [5][6][7]. This aggressive type of cancer originates from the pleural cavity (i.e., the serous membrane that separate the lungs and the wall of the thoracic cavity) to which asbestos fibres can reach due to their favourable aerodynamic diameter [8,9]. The carcinogenesis of MM is initiated by the inflammatory response to those fibres that are retained in the pleural cavity, being too long to exit through the stomatal openings in the parietal pleural surface (diameter as a potential route for the preparation of fibres with controlled lengths. Size reduction by mechanical milling occurs due to a combination of impact, compression and shear forces [28]. Both the type of equipment and experimental parameters highly influence the imposed forces and the consequent material's response. Many different types of instrumental setups have been used to systematically investigate the effect of mechanical action on chrysotile. These include dry milling in planetary ball mills [29][30][31][32], milling in vibration ball mills in dry conditions as well as in the presence of hydrocarbons or water [33][34][35], dry milling in a vibration disc mill [36], and rotary mill (also called ring and puck mill or shutter box) [37]. Generally speaking, size reduction was accompanied by plastic deformation of the chrysotile structure and prolonged mechanical action eventually led to complete amorphization. Hence, alternative methods have been proposed that were shown to avoid structural-chemical modifications with respect to the pristine fibres: prolonged ultrasonification [34,38], gentle hand-milling [26] and chopping [24]. A promising comminution method that leaves intact the chemical-structural properties of chrysotile is cryogenic milling [23,39], which exploits the brittle fracture of materials cooled down to cryogenic temperatures [40][41][42]. In the present study, the effect of cryogenic milling on the chemical-structural properties of a commercial Russian chrysotile (Orenburg minerals, Yasniy) was investigated for various milling times in order to prepare different length fractions of fibres for toxicological studies. The crystallinity, chemistry, particle size and morphology of chrysotile fibres were measured by X-ray diffraction, Fourier Transform Infra-Red spectroscopy, Scanning Electron Microscopy and Energy Dispersive X-ray.

References
Milling Methods Environmental Conditions Advantages/Disadvantages [23] Cryogenic vibro-milling with steel balls −150 • C and wet condition Preservation of chrysotile crystallinity [32] Mechanical friction with cylindrical milling elements Room temperature and wet condition Reduction in chrysotile crystallinity [32] Agate mortar Room temperature and dry condition Reduction in chrysotile crystallinity [33,35] Ball milling Room temperature and dry condition Reduction in chrysotile crystallinity [34,35] Ball milling Room temperature and wet condition Water-wet conditions preserves the chrysotile crystallinity [36] Ring mill equipped with a tungsten carbide bowl Room temperature and dry condition Reduction in chrysotile crystallinity [37] Rotary mill Room temperature and dry condition Reduction in chrysotile crystallinity

Sample Preparation and Characterisation
A commercial chrysotile sample from the Russian Federation, extracted by Orenburg Minerals from an open pit mine near the town Yasniy was used in this work. This particular sample was extensively characterised in our previous work wherefore only a short description will be given here [21]. The sample is composed of bundles of fibres up to 1 mm in length that easily split lengthwise into fine fibres and thin bundles. The length of a single fibre and thin bundles is in the rage of 1.36 to 188.1 µm (mean 33.9 µm), whereas the fibre width is in the 0.05-2.79 µm range (average 0.7 µm). Clino-chrysotile is the major phase, accompanied by minor amounts of ortho-chrysotile, lizardite, magnetite, hydromagnesite and calcite. The composition, determined by electron probe microanalysis ( The milling experiments were conducted, using a Retsch mixer mill MM 400 (Düsseldorf, Germany), equipped with two processing stations. This is the very same equipment that was used to prepare powder batches for various analyses in our previous work dealing with this Russian sample [21]. The milling jars (35 mL) are made of steel and each one is accompanied with a steel milling ball (diameter of 20 mm, weight 14.24 g). The lining of the jar and balls is made of polytetrafluoroethylene (PTFE), which serves to prevent contamination of the sample with metals from the jar and balls during milling. The loaded jar is submerged in liquid nitrogen until acquiring the temperature of the surrounding medium and subsequently mounted in the mixer mill. During operation, the jar oscillates from side to side with a pre-set frequency . The amplitude of the oscillations is fixed. The repeated impaction of the ball against the embrittled material leads to size reduction. In the experiments performed in our laboratory, the jars were loaded with 1 g of raw chrysotile and an oscillation frequency of 30 Hz was applied for various durations, i.e., 5, 10, 20, 30, and 40 min. The investigated range of milling durations were chosen based on preliminary tests, evidencing too-low efficiency following durations of less than 5 min and a non-significant improvement of the size reduction for duration longer than 40 min. From now on, samples will be denoted by CHR (pristine chrysotile) followed by the milling duration if processed.

Characterisation Techniques
Morphometric investigation of chrysotile fibres following various milling durations were performed, using a JSM-6010PLUS/LA Scanning Electron Microscopy (SEM) equipped with and Energy Dispersive X-ray (EDX) spectrometer (Oxford INCA-350). EDX is an analytical method useful for identify and quantify (only semi-quantitative results) the elementary compositions of the materials observed with the SEM. EDX analyses were conducted to verify the changes in the elemental composition undergone by the chrysotile fibres after prolonged milling. Specimens for these analyses were prepared as follows: some milligrams of fibres were suspended in 5 mL of ethanol and a drop of the resulting suspension was placed on conductive, double-stick carbon tape attached to an aluminium stub [43]. After complete evaporation of the solvent, the specimen was coated with a thin film of carbon (10 nm thick), using a Carbon Coater-Balzers CED-010. The length of the chrysotile fibres was determined by processing SEM images obtained by secondary electron imaging and appropriate magnifications (500-20,000×). A series of representative SEM images were analysed using ImageJ software, which provides accurate measurements of the lengths of both straight and curved fibres [44]. The fibre length distributions were constructed based on a statistically representative number of fibres ( Table 2). As mentioned above, Russian chrysotile forms irregular, curvilinear aggregates of fibres and fibril bundles. Therefore, the length of the fibres was measured for each sample, considering both single fibres and fibril bundles. Table 2. Fraction of chrysotile fibres comprised in a given interval of fibre length after each milling treatment and main statistical parameters of fibre lengths. Raw chrysotile (CHR); chrysotile after 5 min of milling (CHR5); chrysotile after 10 min of milling (CHR10); chrysotile after 20 min milling (CHR20); chrysotile after 30 min of milling (CHR30); chrysotile after 40 min of milling (CHR40). Number of fibres (N). X-ray powder diffraction (XRPD) data were collected using an X'Pert PRO PANalytical θ/θ diffractometer (CuKα radiation, 40 kV and 40 mA), equipped with a stripe detector (X'Celerator). Powders were mounted in aluminium sample holders, using a side-loading technique. A step scan of 0.0167 • 2θ was used with a time of 1 s/step in the 5 • -40 • 2θ range. The incident beam pathway included a 1/4 • divergence slit and 1/4 • anti-scattering slit. The pathway of the diffracted beam included a soller slit (0.04 rad), an antiscatter blade (5 mm) and a Ni filter. The XRPD patterns were analysed using the X-Pert High Score Plus software DATABASE [45].

CHR
Fourier Transform Infra-Red (FTIR) spectroscopy analyses of the chrysotile samples were performed in the spectral range 600-4000 cm -1 , using a Bruker Vertex 70 spectrometer (Ettlingen, Germany) equipped with a DTGS detector and a KBr beam splitter. The spectrometer was coupled to a horizontal attenuated total reflectance (ATR) device consisting of a diamond crystal of 2 mm in diameter (Platinum ATR-QL, Bruker Optics, Ettlingen, Germany). ATR-FTIR spectroscopy is a fast, non-destructive, non-invasive, reagent-free, sensitive, and highly reproducible physicochemical tool for characterisation of mineral fibres [46][47][48]. In addition, ATR-FTIR requires only a small amount of sample for analysis with easy and quick preparation [46]. Figure 1 shows SEM micrographs of pristine chrysotile fibres as well as fibres exposed to cryomilling for different durations. As already reported in our previous work [21], the pristine sample (CHR) is composed of very long fibre bundles (up to 1 mm in length) with split ends (Figure 1a) accompanied by clusters of intertwined fibres (Figure 1b). The fibres are very long and curly (mean L = 33.9 µm) ( Figure 1a). The SEM images show some important morphological changes following milling (Figure 1c-f). Although most fibres are still aggregated in bundles and clusters after 5 min of milling (CHR5), the bundle length is reduced (Figure 1c). Following 10 min of milling (CHR10), short bundles are observed together with single fibres that have apparently been separated from the bundles/clusters ( Figure 1d). Increasing the milling time (10 min steps) up to a total of 40 min leads to progressively shorter fibre bundles and single fibres with decreased curliness (Figure 1e-h).

SEM Characterisation
From the analyses of SEM images, the fibre length distribution (i.e., the fraction of fibres comprised in a given length interval) was determined for pristine chrysotile and for chrysotile exposed to cryogenic milling for various durations. The results are shown in Figure 2 and Table 2. Length intervals were fixed to 0-5 µm, 5-10 µm, 10-15 µm, 15-20 µm, >20 µm. As shown in Figure 2, the length of the fibres decreases as a function of milling time. Pristine chrysotile (CHR) exhibits 61.5% of the fibres longer than 20 µm (Figure 2a). After 5 min of milling, this length class is reduced to a value of 49.1%, while the percentage of fibres falling within the range of 5-10 µm increases from 7.5% to 21.7% (Figure 2b). After 10 min (CHR10), approximately 50% of the fibres are shorter than 10 µm (Figure 2c). Increasing the milling time to 20 min (CHR20) further increases the percentage of this length class to 67% (Figure 2d). The most substantial size reduction is observed following 30 min of milling (Figure 2e). After 30 min, the fibres that belong to the range 0-5 µm are 65%, while the percentage of fibres longer than 20 µm is dropped to 1% (Figure 2e). Finally, after 40 min, almost all fibres (95%) are shorter than <5 µm (Figure 2f).
Since chrysotile is mainly composed of Si and Mg, the SEM-EDX semi-quantitative analyses focused on these two elements. Table 3 shows the Si and Mg content (reported in weight percentages of the oxides of the elements) of raw and ground Russian chrysotile (40 min) fibres measured by SEM-EDX. The semi-quantitative composition of the two samples is comparable. Very similar is the Si/Mg ratio, 1.05 for raw chrysotile and 1.02 for long-term milled chrysotile. Therefore, no Mg depletion was induced during cryomilling.
Since chrysotile is mainly composed of Si and Mg, the SEM-EDX semi-quantitative analyses focused on these two elements. Table 3 shows the Si and Mg content (reported in weight percentages of the oxides of the elements) of raw and ground Russian chrysotile (40 min) fibres measured by SEM-EDX. The semi-quantitative composition of the two samples is comparable. Very similar is the Si/Mg ratio, 1.05 for raw chrysotile and 1.02 for long-term milled chrysotile. Therefore, no Mg depletion was induced during cryomilling.
.      Figure 3 shows the XRPD patterns collected from pristine and cryo-milled chrysotile. Qualitative mineralogical analysis of pristine chrysotile evidences the presence of clino-chrysotile as major phase, accompanied by some minor phases (i.e., ortho-chrysotile, lizardite, magnetite, hydromagnesite and calcite). This result confirms the previous published data [21].  Figure 3 shows the XRPD patterns collected from pristine and cryo-milled chrysotile. Qualitative mineralogical analysis of pristine chrysotile evidences the presence of clinochrysotile as major phase, accompanied by some minor phases (i.e., ortho-chrysotile, lizardite, magnetite, hydromagnesite and calcite). This result confirms the previous published data [21].

XRPD Characterisation
The typical anisotropic diffraction peak broadening due to a distorted unit cell originating from the cylindrical lattice is evident [23].
From Figure 3 it is clear that cryo-milling does not provoke any changes in the diffraction pattern, i.e., crystallinity and microstructure remain invariant. This is further evidenced in Figure 4   The typical anisotropic diffraction peak broadening due to a distorted unit cell originating from the cylindrical lattice is evident [23].
From Figure 3 it is clear that cryo-milling does not provoke any changes in the diffraction pattern, i.e., crystallinity and microstructure remain invariant. This is further evidenced in Figure 4 that shows the (002) reflection of chrysotile for all investigated samples (a) together with the Full Width at Half-Maximum (FWHM) as a function of milling time for the same reflection (b). In fact, both reflection intensity as well as peak breadth remain invariant.  The typical anisotropic diffraction peak broadening due to a distorted unit cell originating from the cylindrical lattice is evident [23].
From Figure 3 it is clear that cryo-milling does not provoke any changes in the diffraction pattern, i.e., crystallinity and microstructure remain invariant. This is further evidenced in Figure 4 that shows the (002) reflection of chrysotile for all investigated samples (a) together with the Full Width at Half-Maximum (FWHM) as a function of milling time for the same reflection (b). In fact, both reflection intensity as well as peak breadth remain invariant.

FTIR Characterisation
The FTIR spectra of the samples are reported in Figure 5. In particular, the entire investigated spectral range is shown in Figure 5a, whereas a restricted range (3600-3720 cm −1 ) is shown in Figure 5b in order to evidence the absorption peak assigned to the O-H stretching vibrations [48]. The intense peaks at 950, 1020 and 1074 cm −1 are related to the Si-O bond [49]. The two peaks at 1420 and 1480 cm −1 can be assigned to the MO-H (M = Mg 2+ or Fe 2+ ) stretching vibrations modes [50]. The OH-stretching region 3600-3720 cm −1 (Figure 5b) shows two well-defined absorption bands. 3682 cm −1 intense ban is assigned to external MO-H groups, whereas 3643 cm −1 medium ban is assigned to internal MO-H groups [51]. The most important information retrieved from Figure 5 is that the spectra (relative absorption intensities and peak positions) collected from the samples are nearly identical ( Figure 5). Only some minor differences can be observed in the absolute intensity of the bands, which is not surprising considering the difficulties in reproducing the contact area between the sample and the ATR crystal and, thus, the peak intensity [52].
The image shows the negligible effect of milling on the chrysotile (002) peak. (b) Variation of Full Width at Half-Maximum (FWHM) of (002) chrysotile peak with time of milling.

FTIR Characterisation
The FTIR spectra of the samples are reported in Figure 5. In particular, the entire investigated spectral range is shown in Figure 5a, whereas a restricted range (3600-3720 cm −1 ) is shown in Figure 5b in order to evidence the absorption peak assigned to the O-H stretching vibrations [48]. The intense peaks at 950, 1020 and 1074 cm −1 are related to the Si-O bond [49]. The two peaks at 1420 and 1480 cm −1 can be assigned to the MO-H (M = Mg 2+ or Fe 2+ ) stretching vibrations modes [50]. The OH-stretching region 3600-3720 cm −1 (Figure 5b) shows two well-defined absorption bands. 3682 cm −1 intense ban is assigned to external MO-H groups, whereas 3643 cm −1 medium ban is assigned to internal MO-H groups [51]. The most important information retrieved from Figure 5 is that the spectra (relative absorption intensities and peak positions) collected from the samples are nearly identical ( Figure 5). Only some minor differences can be observed in the absolute intensity of the bands, which is not surprising considering the difficulties in reproducing the contact area between the sample and the ATR crystal and, thus, the peak intensity [52].

Discussion
Considering the major role played by fibre length in determining the toxicological action of asbestos, controlling this parameter is of paramount importance in targeted studies. Preparing batches of chrysotile fibres only differing in length generally involve size reduction of pristine fibres followed by separation. The first step is crucial, as alternation of the fibre properties must be avoided. The scientific literature has discarded traditional milling by way of reducing fibre length, as the mechanical action leads to plastic deformation of the chrysotile structure (Table 1).
In the present study, cryogenic milling was systematically explored as a potential size reduction technique. More precisely, the structural integrity was verified following various milling durations, using XRPD and FTIR, which are powerful techniques in evidencing structural changes on a molecular level.
The summary graphs in Figure 6 show the percentage of long (L > 5 µm) and short (L < 5 µm) Russian chrysotile fibres as a function of milling time (0 to 40 min). According to Figure 6, increased duration of cryogenic milling resulted in a successive reduction in fibre length. The longest milling duration investigated in this work (i.e., 40 min) resulted in 95% of the fibres having a length < 5 µm (Figure 6). rated from the sorted fibres prior to use.
The results shown in this work evidenced that the two desired size classes for in vitro toxicology studies (i.e., L < 5µ m and L > 5 µ m), could be directly obtained by cryo-milling without a subsequent separation step [25]. In fact, milling for 40 min directly delivered a batch consisting only of short fibres (95% < 5 µ m), whereas 10 min of cryo-milling resulted in a batch containing mainly long fibres (mean L = 14.32 µ m and 90% of fibres with L > 5 µ m). These two sets of fibres are suitable as they are, without the need for an additional size separation step. Another important advantage of the cryo-milled fibres (especially considering the upper length size class) was that stable fibre suspensions, ready to be added to the culture media for in vitro testing, could easily be prepared by ultrasonification of the cryo-milled fibres in a biological solution. In fact, even though most of the fibres were still arranged in bundles following mechanical pre-treatment under cryogenic conditions, they were easy to split by ultrasonic treatment. This was shown to be difficult to achieve with the very long and curly pristine fibres that, instead, tended to agglomerate in large clusters (not shown here).

Conclusions
In this work, cryogenic milling was systematically explored as a comminution method of a commercial Russian chrysotile. The high potential of this route is due to the embrittlement of the material, which facilitates brittle fracture and thus size reduction without plastic deformation.
The aim was to prepare different length classes of fibres suitable for in vitro studies, i.e., <5 µ m and >5 µ m, without modifying the crystal-chemical properties of the original ones. This particular mineral is currently used in pro-chrysotile countries and studies aimed at shedding light on their potential health hazards are highly motivated. XRPD data ( Figure 3) and FTIR spectra ( Figure 5) did not evidence any structural changes imposed by milling, as the data remained substantially invariant. Instead, the contrary was observed in earlier studies dealing with traditional milling methods. For example, Suquet [31] studied the effect of dry milling on the crystal structure of chrysotile. The author observed important changes in the diffraction pattern following milling in terms of intensity loss and peak broadening [31]. IR analyses showed that dry milling led to important changes in the lattice vibrations, indicating a disrupted chrysotile structure and the formation of amorphous silica [31]. The IR data indicated the presence of adsorbed CO 2 and H 2 O on the surface of milled chrysotile, which, according to the author, was explained by newly formed reactive surface sites [31]. Iwaszko et al. [30] performed an FTIR investigation of asbestos-cement containing chrysotile before and after prolonged highenergy ball milling. In particular, changes in the intensity and position of the chrysotile bands provoked by milling were evaluated. The authors observed a gradual disappearance of the characteristic bands as the milling time increased, being virtually extinct following 3 h of ball milling [30]. Bloise et al. [37] conducted dry milling of chrysotile using a ring and puck mill and the XRPD data showed a continuous amorphization of the fibres, which, following 10 min of mechanical action, was near complete.
The temperature during milling is a key factor determining the different outcome obtained by cryo-milling in comparison to traditional milling. Lowering the temperature to cryogenic conditions changes the mechanical behaviour of the material from ductile to brittle. Hence, the stress imposed by the milling medium results in brittle fracture rather than structural changes. In addition, temperature-induced dehydroxylation of chrysotile observed during traditional milling is avoided [37]. Even though the macroscopic temperature of the system generally is moderate, the local microscopic temperature can be even higher than 1000 • C [53]. The reaction mechanism by which chrysotile transforms into an amorphous phase following temperature-induced dehydroxylation has been elucidated in previous work [54]. The rate-limiting step of the reaction was found to be one-dimensional diffusion with an instantaneous or decelerator rate of nucleation [54]. This result was explained by the peculiar structure of chrysotile: this mineral is a hydrated magnesium silicate with the ideal chemical formula Mg 3 (OH) 4 Si 2 O 5 . The double-layered structure is composed of an octahedral Mg(OH) 2 sheet (O) bonded to a tetrahedral SiO 2 (T) sheet, with the Mg cations fully occupying the octahedral cation sites. The fibrous nature of chrysotile is due to a lattice misfit between the two layers, giving rise to a strain gradient that leads to self-rolling and the formation of a fibre structures with the Mg(OH) 2 surface exposed externally. Cattaneo et al. [54] concluded that the most probable one-dimensional escape paths for the water molecules were along the fibre axis. In this work, the FTIR analyses suggested that this process does not occur during cryogenic milling ( Figure 5). As shown in Figure 5b, the FTIR spectrum of chrysotile milled for 40 min preserved the characteristic bands assigned to the OH groups (i.e., 3643 and 3682 cm −1 ).
In the literature, various methodologies have been proposed to produce size-selected fibre samples as discussed in the introduction. Two-step procedures are generally applied, including comminution followed by size separation. Currently, gravitational wet separation of hand-milled fibres is the most common method of obtaining groups of fibres differing in length. This process is relatively time-consuming due to slow settling velocity (ca. 3 h for asbestos fibres according to [55]) and the fact that the solvent must be evaporated from the sorted fibres prior to use.
The results shown in this work evidenced that the two desired size classes for in vitro toxicology studies (i.e., L < 5µm and L > 5 µm), could be directly obtained by cryo-milling without a subsequent separation step [25]. In fact, milling for 40 min directly delivered a batch consisting only of short fibres (95% < 5 µm), whereas 10 min of cryo-milling resulted in a batch containing mainly long fibres (mean L = 14.32 µm and 90% of fibres with L > 5 µm). These two sets of fibres are suitable as they are, without the need for an additional size separation step. Another important advantage of the cryo-milled fibres (especially considering the upper length size class) was that stable fibre suspensions, ready to be added to the culture media for in vitro testing, could easily be prepared by ultrasonification of the cryo-milled fibres in a biological solution. In fact, even though most of the fibres were still arranged in bundles following mechanical pre-treatment under cryogenic conditions, they were easy to split by ultrasonic treatment. This was shown to be difficult to achieve with the very long and curly pristine fibres that, instead, tended to agglomerate in large clusters (not shown here).

Conclusions
In this work, cryogenic milling was systematically explored as a comminution method of a commercial Russian chrysotile. The high potential of this route is due to the embrittlement of the material, which facilitates brittle fracture and thus size reduction without plastic deformation.
The aim was to prepare different length classes of fibres suitable for in vitro studies, i.e., <5 µm and >5 µm, without modifying the crystal-chemical properties of the original ones. This particular mineral is currently used in pro-chrysotile countries and studies aimed at shedding light on their potential health hazards are highly motivated.
Analyses of SEM images showed a successive length reduction when the duration of cryo-milling increased from 5 to 40 min (the investigated time range). XRPD and FTIR data collected from these samples did not evidence any changes in crystallinity or microstructure.
It was shown that the two desired size classes of fibres could be obtained by modifying the milling duration: the class of long fibres was obtained following 10 min of milling (i.e., 90% > 5 µm), whereas the class of short ones could be prepared by prolonging the milling time to 40 min (i.e., 95% > 5 µm).
Stable fibre suspensions, ready to be added to the culture media for in vitro testing, could be easily obtained by ultrasonification of the cryo-milled fibres in a biological solution in which the bundles were easily split. Compared to the most common method for the preparation of different size classes of fibres, i.e., gentle hand milling followed by gravitational separation, cryo-milling allows a rapid preparation of the two desired size classes without the need of further separation.