Access to this full-text is provided by Springer Nature.
Content available from Particle and Fibre Toxicology
This content is subject to copyright. Terms and conditions apply.
R E S E A R C H Open Access
Composition, Respirable Fraction and
Dissolution Rate of 24 Stone Wool MMVF
with their Binder
Wendel Wohlleben
1*
, Hubert Waindok
1
, Björn Daumann
2
, Kai Werle
1
, Melanie Drum
1
and Heiko Egenolf
1
Abstract
Background: Man-made vitreous fibres (MMVF) are produced on a large scale for thermal insulation purposes. After
extensive studies of fibre effects in the 1980ies and 1990ies, the composition of MMVF was modified to reduce the
fibrotic and cancerogenic potential via reduced biopersistence. However, occupational risks by handling, applying,
disposing modern MMVF may be underestimated as the conventional regulatory classification -combining
composition, in-vivo clearance and effects- seems to be based entirely on MMVF after removal of the binder.
Results: Here we report the oxide composition of 23 modern MMVF from Germany, Finland, UK, Denmark, Russia,
China (five different producers) and one pre-1995 MMVF. We find that most of the investigated modern MMVF can
be classified as “High-alumina, low-silica wool”, but several were on or beyond the borderline to “pre-1995 Rock
(Stone) wool”. We then used well-established flow-through dissolution testing at pH 4.5 and pH 7.4, with and
without binder, at various flow rates, to screen the biosolubility of 14 MMVF over 32 days. At the flow rate and
acidic pH of reports that found 47 ng/cm
2
/h dissolution rate for reference biopersistent MMVF21 (without binder),
we find rates from 17 to 90 ng/cm
2
/h for modern MMVF as customary in trade (with binder). Removing the binder
accelerates the dissolution significantly, but not to the level of reference biosoluble MMVF34. We finally simulated
handling or disposing of MMVF and measured size fractions in the aerosol. The respirable fraction of modern MMVF
is low, but not less than pre-1995 MMVF.
Conclusions: The average composition of modern stone wool MMVF is different from historic biopersistent MMVF,
but to a lesser extent than expected. The dissolution rates measured by abiotic methods indicate that the binder
has a significant influence on dissolution via gel formation. Considering the content of respirable fibres, these
findings imply that the risk assessment of modern stone wool may need to be revisited based on in-vivo studies of
MMFV as marketed (with binder).
Keywords: Man-made vitreous fibres, Stone wool, Occupational safety, Biopersistence, Dissolution, Binder, Coating,
Gel, Respirable fraction
Background
Man-made vitreous fibres (MMVF) are non-crystalline,
fibrous inorganic substances (silicates) made primarily
from rock, slag, glass or other processed minerals. These
materials, also called man-made mineral fibres, [1] in-
clude glass fibres (used in glass wool and continuous
glass filament), rock or stone wool, slag wool and refrac-
tory ceramic fibres [2]. They are widely used for thermal
and acoustical insulation and to a lesser extent for other
purposes. These products are potentially hazardous to
human health because they release airborne respirable fi-
bres during their production, use and removal [3]. Fibre
pathogenicity probably originates from a common mode
of action from all respirable fibres, [4, 5] and is deter-
mined predominantly by aspect ratio, length and bioper-
sistence [6, 7]. The traditional rock (or stone) wool was
classified by the World Health Organization as a car-
cinogenic hazard to humans in 1988 [3]. In response,
glass and stone wool compositions with increased bioso-
lubility have been developed and commercialized [8].
* Correspondence: wendel.wohlleben@basf.com
1
Department Material Physics and Analytics, BASF SE, Ludwigshafen,
Germany
Full list of author information is available at the end of the article
© The Author(s). 2017 Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0
International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and
reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to
the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver
(http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.
Wohlleben et al. Particle and Fibre Toxicology (2017) 14:29
DOI 10.1186/s12989-017-0210-8
Content courtesy of Springer Nature, terms of use apply. Rights reserved.
Based on the in vivo tests required by the nota Q of the
European CLP regulation, certain classes of MMVF are
exonerated from classification as (carc. 2) carcinogen, [9]
in accord with the conclusions of the World Health
Organization report of 2002 [10]. Baan et al. very concisely
review the considerations of the respective IARC Mono-
graphs Working Groups (1987, 2001) in reaching their
conclusions [11]. In order to ensure that the increased
biosolubility is maintained, the European insulation wool
manufacturers association (EURIMA) implemented moni-
toring schemes to ensure that the chemical compositions
are kept within defined ranges [12].
However, products should be tested as commercial-
ized. The MMVF production inherently uses organic oil
and binder (phenolic resin etc.) that is sprayed onto the
stone melt directly in the fibre spinning chambers. The
primary mat is layered to give the product the required
weight per unit area, and passes through an oven, which
sets the thickness of the mat, dries it and cures the
binder [13]. The product is then air-cooled and cut to
size before packaging [2]. Thus, MMVF without binder
is not a necessary intermediate of occupational or com-
mercial relevance. MMVF without binder is not repre-
sentative of the commercial MMVF product for which
safe use on construction sites must be ensured. Studies
in 1995 deliberately removed binder from the commer-
cial product, e.g. by ozone cold-ashing, [14] and only
then investigated biopersistence. For the in vivo studies
reported in 2000 –2002, which were decisive for the
WHO and IARC committees to exonerate the class of
high-alumina low-silica stone wool (synonymously:
HT, biosoluble MMVF) from classification as cancero-
gens, “tested fibres were produced without binder or
oil”[15–17]. The conclusions and comments to the
extensive BIA report
1
already raise concerns that both
in vitro data on dissolution and in vivo data on clear-
ance and effects relate to MMVF without binder “that
is rare in occupational settings”[18].
Here we took a pragmatic perspective motivated by
occupational safety on BASF construction sites: We
sourced MMVF directly from construction sites, and in-
vestigated their properties without further modifications.
The strategy of the present contribution is to screen the
safety-relevant physical-chemical properties –compos-
ition, respirable fraction, in vitro dissolution rates–on a
set of modern stone wool MMVF sourced from various
countries and producers. To the best of our knowledge,
this is the first study to report composition, respirable
fraction and dissolution of modern MMVF with their
binder. We aim to benchmark results against literature on
reference materials, which represent the low-biosolubility
(MMVF21) and high-biosolubility (MMVF34) materials,
respectively. Methodology for MMVF dissolution screen-
ing does not need to be re-invented, because it already is
highly established [14, 19–21]. The strong correlation of
in vitro dissolution rates vs. in vivo clearance rates, fibro-
genic and carcinogenic potential was instrumental to iden-
tify safer MMVF compositions in the 1990ies [10, 22].
Materials and Methods
MMVF were sampled in kg quantities predominantly
from various construction sites within BASF Ludwigs-
hafen, where contractor and/or BASF-employed workers
handled MMVF products. Additionally, selected mate-
rials were sourced from sites in other countries, incud-
ing Finland, UK, Denmark, Russia, China. In all cases,
the producer and product grade are known, but are
coded here for anonymity. Producers are coded A to E,
and materials are coded MMVF #1 to #28 (due to mul-
tiple determination of oxide composition, some mate-
rials have more than one internal MMVF code, but are
listed only once here.) Only one material (MMVF #17)
cannot be traced to a specific grade and producer, be-
cause it was sampled from the dismantling of a BASF air
separation facility, where it is known to have been in-
stalled at least for 30 years, hence with certainty before
1995. Thickness of all MMVF ranged from 40 to 150
mm, and density ranged from 47 to 180 kg/m
3
. Samples
for dissolution testing were taken from the middle. Note
that two most relevant historical reference materials are
designated by the established codes MMVF21 and
MMVF34. There is no specific relation between the his-
torical reference MMVF21 and the modern MMVF#21.
Sample pretreatment for Al, Ba, Ca, Cr, Mg, Mn, P,
S, Sr, Ti composition analysis: Analysis was performed
for all materials in duplicate. In each case, a blank was
run in an analogous manner. A sample aliquot of
approx. 20 mg was weighed, to the nearest of 0.01 mg,
into a platinum crucible, and mixed with both 0.8 g of a
K
2
CO
3
-Na
2
CO
3
mixture and 0.2 g Na
2
B
4
O
7
. The cruci-
bles were inductively heated to a maximum temperature
of approx. 930 °C. During the melt fusion, the crucibles
were rotated and tilted to obtain a homogeneous melt.
After cooling the melt cake was dissolved in approx. 88
ml of water and 12 ml semiconc. hydrochloric acid. The
solution obtained was weighed again, and the final vol-
ume was calculated from the density of the solution
(1.015 g/ml).
Measurement of Al, Ba, Ca, Cr, Mg, Mn, P, S, Sr, Ti:
The analytes were determined by inductively coupled
plasma-optical emission spectrometry (ICP-OES, Varian
725-ES). Prior to taking the measurement, the instrument
was optimized in accordance with the manufacturer’sspe-
cification. Three replicate measurements are taken and av-
eraged. We measured with 10 s integration time the
following wavelengths [nm]: Al 396.152; Ba 493.408; Ca
317.933; Cr 206.158; Mg 279.553; P 213.618; S 181.972; Sr
Wohlleben et al. Particle and Fibre Toxicology (2017) 14:29 Page 2 of 16
Content courtesy of Springer Nature, terms of use apply. Rights reserved.
216.596; Ti 336.122. The dilution factors were 6.67 for Al,
Ba, Ca, Cr, Mg, Mn; and 1 for P, S, Sr, Ti. External calibra-
tion used concentrations of 0 / 1 / 5 mg/l with matrix-
matched standards. The nebulizer (Meinhard 1 ml) had a
flow of 0.7 l/min at pump rate 15 rpm. Complete
reproduction on MMVF#4 through #11 confirmed better
than 0.5% reproducibility on SiO
2
and Al
2
O
3
contents.
The slightly different sample preparation and measure-
ment methods for optimal analysis of B (by ICP-OES) and
Na, K (by Flame Atomic Absorption Spectrometry
(F-AAS)) are described in full detail in the Additional file 1.
Binder content and optional removal: As a standard,
all MMVF were measured as-received, without any sam-
ple preparation. For comparison in selected cases, the
binder was removed by low thermal annealing or by oxy-
gen plasma. Specifically, the oxygen plasma was gener-
ated in a Diener electronic PCCE, using 10 min at 60 W
O
2
plasma. Alternatively, an oven (Heraeus thermicon
T), pre-heated to 500°C, was used to remove binder on
samples of edge length 10 cm. The gravimetric loss can
be accurately detemined on such large samples, and is
attributed to the organic phase (binder). Full TGA on
selected materials confirmed that 500°C is appropriate to
remove binder. TGA utilized STA449 F3 (Netzsch), op-
erated under air with 40ml/min, heated by 5K/min from
35 °C to 560 °C. The analysis software (Netzsch Proteus
Thermal Analysis 6.1) adheres to DIN51005.
Scanning Electron Microscopy (SEM) was detemined
both before and after dissolution testing. SEM samples
were fixed on an adhesive film, coated with 9nm Pt and
investigated on a JSM 7500TFE (Jeol Company) oper-
ated at 5 keV. The topographic images were taken with
secondary electrons (SE).
The BET specific surface area was determined on
Quantachrome Autosorb according to ISO 9277:2010
by volumetric static measurement of the nitrogen iso-
therm at 77.3 K with data evaluation according to the
BET theory in the relative pressure range p/p
0
be-
tween 0.001 and 0.3. Samples were prepared for ad-
sorption analysis in a degasser, here the samples were
heated up to 200 °C under vacuum for 30 min or
more to remove moisture and other contaminations.
For the specific equipment, we verified that specific
surfaces down to 0.1 m
2
/g can be accurately deter-
mined. This was confirmed by repeatedly measuring
Certified Reference Materials (Community Bureau of
Reference - BCR 169 Alumina, certified at 0.100 m
2
/
g, measured 0.095 m
2
/g; BCR 175 Tungsten, certified
at 0.180 m
2
/g, measured 0.185 m
2
/g). However, the
Certified Reference Materials have high powder dens-
ity, whereas MMVF is less dense, resulting in a lim-
ited accuracy of the BET values of MMVF, which can
deviate ± 0.15 m
2
/g, corresponding to about 30%
uncertainty.
Respirable fractions To simulate handling of MMVF,
between 100 g and 400 g were cut in an Alpine LU 100
rotating mill at around 1 kg/h throughput with an Ultra-
plex rotor at 94 m/s relative speed against a Conidur 0.2
mm sieve. Particle size distribution of the resulting
MMVF dust was determined by cyclone cascade mea-
surements. Dust samples were dispersed in a dosing
feeder (K-Tron) and an injector with the aid of pressur-
ized air (20 m
3
/h). 20 m
3
/h particle loading gas flow was
fed into a 40mm tube with a length of 1m (Additional
file 1: Figure SI 1). A part of the gas flow (1.7 m
3
/h) was
sampled through a cyclone cascade. The separation cut-
off size of the cyclone cascade is sub-divided in four
steps from 10 μm to 0.3 μm. The aerodynamic diameter
is defined as the diameter of a sphere with the density of
1 g/cm
3
which has the same separation behavior as the
measured sample. The adaption is conducted in accord-
ance with the following equation:
dae ¼dgffiffiffiffiffi
ρg
ρ
r
With d
ae
= aerodynamic diameter, d
g
= measured par-
ticle size, ρ= density 1 g/cm
3
,ρ
g
= density of the sample
substance. Weighing of the respective particle masses
deposited on the individual cascade stages reflects the
particle size distribution of the samples.
Dissolution testing replicated closely (Additional file
1: Figure SI 2) the well-established MMVF dissolution
methods of Guldberg, Sebastian, de Meringo et al. [14,
19–21] as discussed extensively by BAuA [22] and by
the BIA report [18]. The method is a dynamic (flow-
through) system (Additional file 1:Figure SI_2). Specific-
ally, the amount of fibres (as a standard, m = 50 mg) is
weighed with an accuracy of ±0.2 mg and is dispersed
evenly in the cell. The measured BET specific surface
area gives the tested surface area SA = m * BET. The
flow rate was V = 48 ml/d, but was varied up to 240 ml/
d. This corresponds to a ratio of the initial surface area
SA to volume flow V of SA/V = 83 h/cm on average
(min 38 h/cm, max 160 h/cm; in inverse metrics our
average is V/SA = 0.033 μm/s). As we screened up to 7
cells in parallel, controlled by the same peristaltic pump
(Ismatec IPC 8, Additional file 1:Figure SI 2), the differ-
ent BET specific surface areas of the materials result in
slightly different SA/V. All testing performed at 37 ± 0.5
°C. The effluent was collected and pH checked. The
programmable sampler drew 10 mL for ICPMS analysis
(with the exact weight of each sample documented) after
1, 2, 4, 6, 8, 11, 14, 18, 21, 25, 28 and 32 days. This in-
cludes the sampling times of Guldberg et al. [14] and
adds more to increase resolution and duration. Add-
itionally to earlier methodology, we also collected eluted
medium between the sampling times (Additional file 1:
Wohlleben et al. Particle and Fibre Toxicology (2017) 14:29 Page 3 of 16
Content courtesy of Springer Nature, terms of use apply. Rights reserved.
Figure SI 2). This enables a cumulative analysis including
all dissolved ions, and becomes independent of interpolation.
After the experiment the remaining fibres are rinsed in de-
ionized water and dried to constant weight. The weight loss
is compared to the value calculated from interpolation of
the time resolved sampling and to the cumulative dissolved
ions. The morphology of the corroded fibres in relation to
the initial fibres is inspected by means of SEM.
ICPMS and/or ICPOES was used to determine dis-
solved ions in the eluates. All samples were analyzed for
Si, Al and Mg. The between-sampling collection was an-
alyzed for Si, Al, Mg, K, Ti, Fe, Ca. The samples were fil-
trated and diluted by a factor of 2 with de-ionized water.
A higher dilution (factor 25) was used for elements with
higher assays (Ca, K). A blank was run in an analogous
manner. In the dilutions obtained, the analytes were de-
termined by inductively coupled plasma-optical emission
spectrometry (ICP-OES Agilent 5100). Prior to taking
the measurement, the instrument was optimized in
accordance with the manufacturer’s specification. Three
replicate measurements are taken and averaged. We
measured with 10 s integration time the following wave-
lengths [nm]: Al 394.401; Fe 259.940; K 766.491; Si
288.158; Ti 334.941 (axial observation); and Ca 396.847;
Mg 279.553; (radial observation). The internal standard
was Sc at 361.383 nm. External calibration used concen-
trations of 0 / 1 / 5 mg/l. The nebulizer (Meinhard 1 ml)
had a flow of 0.7 l/min at pump rate 15 rpm. The
analysis was performed in duplicate with less than 10%
difference as reproducibility criterion. Otherwise, the
analysis was repeated. The statistical error in all ion con-
centrations used for calculation of dissolution rates is
thus below 10%.
The pH 4.5 medium composition, aiming to simulate
the phagolysosome, replicated the “PSF”medium previ-
ously validated for the purpose of inhaled particle dissol-
ution by NIST laboratories: [23] sodium phosphate
dibasic anhydrous (Na
2
HPO
4
) 142.0mg/l; sodium chloride
(NaCl) 6650 mg/l; sodium sulfate anhydrous (Na
2
SO
4
)71
mg/l; calcium chloride dihydrate (CaCl
2
.2H
2
O) 29 mg/l;
glycine (C
2
H
5
NO
2
) 450 mg/l (as representative of organic
acids); potassium hydrogen phthalate (1-(HO
2
C)–2-
(CO
2
K)–C
6
H
4
) 4085 mg/l; alkylbenzyldimethylammonium
chloride (ABDC) 50ppm (added as an antifungal agent).
This medium is near-identical to medium “C”in a previ-
ous interlab comparison of MMVF dissolution at pH 4.5
[19]. The pH 4.5 ± 0.4 was verified before and after the ex-
periment, and was re-measured also on the eluted sam-
ples. Analysis of blind cells showed that Si and Al
elements are sufficiently rare in the pH 4.5 medium,
whereas the background levels of Ca interfere with the
MMVF analysis.
The pH 7.4 medium composition, aiming to simulate
the extracellular lung compartment, followed one of the
previously described Gamble’s fluids. [24] magnesium
chloride (MgCl
2
) 95 mg/l; sodium chloride (NaCl) 6,019
mg/l; sodium phosphate dibasic anhydrous (Na
2
HPO
4
)
298 mg/l; sodium sulfate anhydrous (Na
2
SO
4
) 63 mg/l;
calcium chloride dihydrate (CaCl
2
.2H
2
O) 368 mg/l; so-
dium acetate (C
2
H
3
NaO
2
) 574 mg/l; sodium hydrogen
carbonate (NaHCO
3
) 2,604 mg/l; sodium citrate dihy-
drate (Na
3
C
6
H
5
O
7
) 97 mg/l. We added sodium azide
(NaN
3
) 20 mg/l as biocide. The MMVF literature docu-
ments a variety of Gamble’s pH 7.4 fluids, and the
present version is consistent with others used earlier on
MMVF dissolution [20]. The pH 7.4 ± 0.3 was verified
before and after the experiment, and was re-measured
also on the eluted samples.
Results
Composition
The oxide composition is summarized in Table 1, listing
the 15 MMVF materials that were also subjected to dis-
solution screening. All products tested were within a
narrow range of SiO
2
content, ranging from 40% to 44%,
with an average of 42% SiO
2
of the inorganic part. Add-
itionally to the inorganics, organic components were de-
tected in all MMVF with a content of the total weight
from 0.9% to 4.2%. In TGA, the mass loss occurs in
peaks between 300°C and 500°C, with an average mass
loss of 2.8 ± 1.0 % below 500°C. The organic component
is identified with the binder, oil, resins etc. that coat the
surface of the fibres. The binder is observed also on
SEM micrographs, where it visibly glues fibres together.
Detailed analysis of binder chemical composition was
not performed. By SEM, the distribution of fibre diam-
eter is polydisperse with diameters from below 2 μmto
above 20 μm, and often included large nonfibrous “shot”
particles on the order of 200 μm. Thus, the majority of
fibres in MMVF is too large to penerate deep into the
lung, but every modern MMVF examined did have a
small fraction of respirable fibres. Full statistical analysis
of the fibre diameter distribution was beyond the scope
of the present contribution, but the following section ad-
dresses the airborne fibre fraction. Finally, the specific
surface area SA of the MMVF ranged between 0.2 and
0.6 m
2
/g. For orientation, using the density of 2.8 g/cm
3
and the known fibre shape, the SA can be converted to
an average diameter of the fibres, giving values between
2.4 μmand10μm. Considering the polydispersity, this
is consistent with SEM and with literature [10].
Nine additional MMVF materials from further countries
and producers were analyzed for their composition, binder
content, specific surface area and SEM morphology (avail-
able, not shown here). Their properties (summarized in
Additional file 1: Table SI_1) are consistent and remain
within the ranges observed on the main test set.
Wohlleben et al. Particle and Fibre Toxicology (2017) 14:29 Page 4 of 16
Content courtesy of Springer Nature, terms of use apply. Rights reserved.
Table 1 Composition of MMVF sourced from various countries and producers. Weight content of oxides and of binder
Country of
origin
Producer
code
MMVF code SiO
2
Al
2
O
3
CaO MgO Fe
2
O
3
TiO
2
Na
2
OK
2
O MnO P
2
O
5
Cr
2
O
3
BaO S SrO B
2
O
3
SUM Al / (Al
+Si)
KI-
Index
BET (m
2
/
g)
% binder
content
Germany unkown MMVF #17 (pre-
1995)
54 7 30 3 3 0.5 1.5 1.3 0.1 0.2 0.1 0.0 0.1 0.1 0.0 100 0.13 22 0.3 0.1
Germany A MMVF #1 42 18 18 9 7 2.1 2.1 1.2 0.4 0.4 0.1 0.1 0.1 0.1 0.0 101 0.33 -6 0.2 4.2
Germany B MMVF #2 42 18 18 9 8 1.8 2.4 0.9 0.2 0.3 0.1 0.1 0.1 0.1 0.0 102 0.38 -5 0.5 4.1
Germany C MMVF #4 44 24 15 2 6 0.7 6.2 3.8 0.2 0.8 0.1 0.1 0.0 0.1 0.0 102 0.32 -21 0.6 1.1
Germany A MMVF #5 43 18 17 9 7 1.9 1.3 0.7 0.3 0.3 0.1 0.1 0.1 0.1 0.0 99 0.34 -8 0.2 3.9
Germany A MMVF #7 40 20 16 10 1 1.2 2.3 2.2 0.9 0.3 0.3 1.0 0.1 0.1 0.1 95 0.36 -7 0.4 2.4
Germany A MMVF #8 41 19 18 10 8 1.4 0.8 0.4 0.7 0.4 0.3 0.1 0.1 0.1 0.0 100 0.34 -8 0.2 0.9
China A MMVF #11 42 19 19 8 7 1.3 1.4 0.3 0.2 0.1 0.0 0.1 0.1 0.1 0.0 98 0.35 -8 0.3 4.1
Germany A MMVF #12 43 20 19 10 7 1.9 2.6 1.4 0.4 0.3 0.1 0.1 0.1 0.1 0.0 105 0.34 -7 0.2 3.7
Germany D MMVF #14 44 16 23 9 6 1.6 2.7 1.7 0.6 0.3 0.1 0.2 0.1 0.1 105 0.35 6 0.5 3.7
Germany E MMVF #20 42 18 18 13 8 0.8 1.4 0.4 0.1 0.1 0.1 0.1 0.1 0.1 0.0 101 0.36 -2 0.2 1.9
Finland E MMVF #21 42 17 16 12 10 0.8 1.5 0.4 0.1 0.1 0.1 0.1 0.1 0.1 0.0 100 0.29 -4 0.4 2.4
UK A MMVF #22 40 17 22 9 9 1.3 2 0.4 0.4 0.2 0.1 0.1 0.3 0.1 0.0 102 0.30 -1 0.4 3.2
Russia A MMVF #24 41 16 24 9 8 1.3 1.3 0.7 0.3 0.1 0.1 0.1 0.5 0.1 0.0 102 0.34 2 0.4 2.4
Germany A MMVF #26 41 19 19 8 7 2.1 2.7 0.7 0.2 0.4 0.1 0.1 0.1 0.1 0.0 100 0.34 -7 0.3 1.9
Wohlleben et al. Particle and Fibre Toxicology (2017) 14:29 Page 5 of 16
Content courtesy of Springer Nature, terms of use apply. Rights reserved.
Respirable fraction of milled MMVF
MMVF #2, #5, #12 were chosen for screening fraction-
ation because their composition is roughly average
across the entire test set, and thus considered to be rep-
resentative. After milling, their dusts were dispersed and
separated by the aerodynamic diameter by means of a
cyclone. The weight collected on the impactor stages
shows that for all MMVF investigated, about 59-75% of
total dust mass has aerodynamic diameters >10μm, and
are thus not or only partially inhalable to humans. How-
ever, the cyclone fractions with aerodynamic diameters
below 7.4 μm vary from 0.29% to 6.29% of milled
MMVF mass for the modern MMVF (Table 2). The
spread of this fraction is large, but not different from the
value of 3.65% found in the dust from historic
MMVF#17 (Table 2). The next smaller fraction with aero-
dynamic diameters up to 4.2 μm is significantly lower with
values from 0.02% to 0.22% for the modern MMVF, to be
compared to 0.04% for the historic MMVF#17. Two of
our modern MMVF were similar, but one (MMVF#5) has
significantly lower dustiness, as it also was visually soft
and clumpsy already during milling.
SEM analysis confirms that the cyclone fractions consist
of thinner fibres, and their diameters are consistent with
the nominal aerodynamic diameter cut-offs (Fig. 1 and
Additional file 1: Figure SI_3). The fractions also contain
milling debris of short fragments with low aspect ratio.
Dissolution
Screening was performed on the as received MMVF ma-
terials. The ions detected at each sampling time are nor-
malized to the initial content of the specific element in
the specific MMVF, and are then interpolated with due
consideration of the different sampling intervals to fi-
nally obtain the percentage that has dissolved from this
oxide. The resulting kinetics are plotted in Fig. 2 for
both pH conditions. In pH 7.4, dissolution is consistently
accelerating over the 32 days of sampling, and Si and Al
dissolve with near-identical kinetics, despite their very
different content in these materials. In contrast, in pH
4.5 we observe a significantly faster dissolution. Further,
in pH 4.5 the initial dissolution rate tends to slow down
over time. Finally, at pH 4.5 we observe a higher fraction
of Al than Si in ionic form (Fig. 2), and an even higher
fraction of Mg (Additional file 1: Figure SI_4).
Additionally to the kinetics sampling, the entire elu-
tion medium is collected between the sampling times
and analyzed so that all eluted ions are known. The sum
of all ions provides the cumulated dissolved fraction, in-
dependently for Si and Al, which is then weighted by
their relative content to give the “ion”columns in Table
3, divided by the initial surface area SA and by the total
time of 32 days to obtain the dissolution rate k in units
of ng/cm
2
/h. Table 3 summarizes the results for a total
of 15 materials. For each dissolution experiment, the
quantitative assessment by “ions”is supported by a com-
plementary assessment by gravimetry of remaining
solids, which is shown for the two main screenings in
pH 4.5 and pH 7.4 at standard conditions as column “S”
in Table 3.
The remaining solids were further imaged by SEM,
and compared against the MMVF before aging (which is
not the identical sample taken for dissolution, as SEM
typically requires coatings). We make no attempt to
evaluate statistically the fibre diameters. Instead, the
SEM analysis shows that in general the untreated
MMVF surface is smooth with occasional inclusion of
100 nm to 1 μm sized particles. Judging from the
morphology at fibre junctions, the binder covers the en-
tire MMVF surface (see untreated fibres in Additional
file 1: Figure SI_5, especially MMVF #8, #14, #20). The
re-analysis by SEM after dissolution confirms that the
fibre morphology is persistent after aging, in general
without splicing or obvious shortening. For exemplary
detailed results, MMVF #7 (Fig. 3) was chosen because
its composition is roughly average across the entire test
set, and thus considered to be representative. MMVF #4
(Fig. 4) was chosen because it has the highest Al content
of the entire test set and is an innovative product with
process and benefit characteristics inherited from both
stone wool and glass wool. MMVF #7 and MMVF #4
both change their surface significantly after 32 days at
pH 4.5, showing pronounced gel formation. For instance
on MMVF #4, deep craters of approx. 400 nm diameter
with sub-100-nm cracks at their bottom are frequently
observed after dissolution in pH 4.5 (Fig. 4). In contrast,
after dissolution in pH 7.4, deposits that often appear
crystalline are found on the surfaces. For further com-
parison Additional file 1: Figure SI_5 shows the near-
absence of morphological changes on MMVF #17 (pre-
Table 2 Fractionation of airborne MMVF. Weight gain of the total initial MMVF mass per fraction
Impactor stages, aerodynamic diameters MMVF #2 MMVF #5 MMVF #12 MMVF #17 (pre-1995)
< 7.6 μm 2.40% 0.29% 6.29% 3.65%
< 4.1 μm 0.22% 0.02% 0.13% 0.04%
< 1.2 μm 0.08% 0.01% 0.08% 0.01%
< 0.3 μm 0.03% <<0.01% 0.04% <<0.01%
< 0.1 μm <<0.01% <<0.01% <<0.01% <<0.01%
Wohlleben et al. Particle and Fibre Toxicology (2017) 14:29 Page 6 of 16
Content courtesy of Springer Nature, terms of use apply. Rights reserved.
1995), in excellent accord with the very low dissolution.
Throughout the test set of modern MMVF, at pH 4.5 a
gel with leaching structures in the form of pits is fre-
quently observed, occasionally also leaching craters, plat-
eaus, microcracks and also ring-shaped hems
(potentially collapsed bubbles). Fibre breakage is rare.
The SEM observations are summarized in Table 3, with
high magnification scans shown in Additional file 1:
Figure SI_5. Full SEM results are documented in the
Additional file 2: SEM Annex with low-mid-high magni-
fication for all MMVF before and after dissolution.
Several materials were subjected to modifications of the
standard conditions, in order to explore the origin of the
unexpectedly low dissolution rates. To compare against lit-
erature, binder was removed by two different methods from
MMVF #7 and from MMVF #4, which represent high and
low binder content respectively. Then dissolution at pH 4.5
was performed and found a dramatic acceleration of Mg
leaching and of Si, Al dissolution from the high-alumina
fibre MMVF #4 (Fig. 5). The k rate based only on Si and Al
doublesfrom20to39ng/cm
2
/h (Table 3), and the
remaining solids even drop to 64% after 32 days. The effects
are less pronounced but equally a significant acceleration
from 40 to 59 ng/cm
2
/h is observed for MMVF #7. In
accord, also the dissolution morphology changes without
binder. The surface is much smoother with leaching pits re-
duced in size or completely absent. Plasma treatment has
an intermediate effect both on morphology (Figs. 3 and 4)
and on kinetics (Fig. 5a, b). We also performed nitrogen ad-
sorption before and after binder removal on MMVF#4, #5
and #7 (low, high and mid binder content). The dimension-
less BET isotherm fitting constant c reduces by a factor 2.8
±1.3 with the binder. Due to this change of physisorption
mechanisms, the net change of specific surface area by the
presence of binder has positive or negative sign, depending
on the evaluation model: -22% by BET evaluation, +15% by
Langmuir evaluation.
Additional dissolution studies were also performed on
the respirable fractions of MMVF #5 and MMVF #12,
finding an increase of dissolution rates for MMVF #5 up
to 122 ng/cm
2
/h, and an slight decrease for MMVF #12.
For both cases, the milled, not fractionated materials
had intermediate dissolution rates (Table 3).
Discussion
Composition
Man-made vitreous fibres (MMVF) are classified within
the European Union (EU) as carcinogen category 2
Fig. 1 SEM micrographs of MMVF #12. As received –after milling –only respirable fraction. See the Supporting Information for analogous results
on MMVF #5
Wohlleben et al. Particle and Fibre Toxicology (2017) 14:29 Page 7 of 16
Content courtesy of Springer Nature, terms of use apply. Rights reserved.
(suspected human carcinogens), but Nota Q and Nota R
specify criteria to exonerate fibres from this classification
[9]. The HT stone wool fibres are a range of MMVF
compositions that fulfill European regulatory require-
ments for exoneration from classification as a carcino-
gen and are registered by a chemical compositional
range for the CAS number 287922-11-6. This range –
synonymously designated as “High-alumina, low-silica
wool”or “HT stone wool”or “biosoluble stone wool”– is
defined by the range of dominant metal oxides shown in
Table 4, with silica in the range 33% –43%, alumina in
the range 18% –24%. In contrast, the MMVF class of
pre-1995 “Rock (stone) wool”has a significantly higher
SiO
2
content of 43% –50%, and lower Al
2
O
3
content of
6% –15% [10]. As rationale for the delimitation of the
HT class with high biosolubility, it has been proposed
that “an increase in Al/(Al + Si) ratio will result in a
more hydrated and less continuous remaining silica net-
work as aluminum is removed [….] As a result, the abil-
ity to form dense surface layers is reduced and hereby
the dissolution rate increases”[16]. The modern MMVF
were analyzed to have a content of SiO
2
highly con-
trolled within a narrow range, and with still relatively
similar contents of Al
2
O
3
, CaO
2
, MgO, showing in this
order an increasing spread of composition across the
test set. As defined at the time of introducing the HT fi-
bres “a maximum limit of 43% SiO
2
and a minimum
limit of 18% Al
2
O
3
and 23% CaO + MgO should secure
that the fibres are biosoluble”[16]. The condition on
Al
2
O
3
is fulfilled by nearly the entire test set, but the
conditions on SiO
2
and CaO+MgO are frequently not
fulfilled. Thus, most but not all of the present set of
modern MMVF belonged to the class of “High-alumina,
low-silica wool”. Part of the test set was on or beyond
the borderline to the class of pre-1995 “Rock (Stone)
wool”. In terms of the the Al/(Al+Si) ratio, and also in
terms of SiO
2
content, the average of the test set is half-
way between the references MMVF21 and MMVF34,
with a relatively wide spread of individual oxides, but a
very low spread in the Al/(Al+Si) ratio = 0.34 ± 0.05
(min 0.29, max 0.38).
Dissolution
The class of HT stone wool is specifically designed to
have relatively lower solubility at neutral pH (for tech-
nical performance) and high solubility at acidic pH (for
product safety). Lower solubility at neutral pH is advan-
tageous for technical durability for the intended use.
The alumina content, to replace silica, is advantageous
Fig. 2 Dissolution kinetics in neutral and acidic pH conditions, all at
initial mass 50 mg MMVF, flow 48 ml/day. Si (black lines), Al (blue
lines). pH 4.5 (dots), pH 7.4 (crosses). aMMVF #4, bMMVF #5, c
MMVF #12, dMMVF #14, eMMVF #22
Wohlleben et al. Particle and Fibre Toxicology (2017) 14:29 Page 8 of 16
Content courtesy of Springer Nature, terms of use apply. Rights reserved.
Table 3 Dissolution screening at pH 7.4 and pH 4.5. “Ions”= cumulated dissolved Si and Al based on ICPMS quantification of all
eluted ions, in % of initial Si and Al; “S”= remaining solid mass, in % of initial MMVF mass; “k”= dissolution rate determined from
cumulated dissolved ions, in ng/cm
2
/h
screening at
pH 4.5
screening at
pH 7.4
pH4.5,
(binder
removed)
pH4.5
(milled)
pH4.5
respirable
(milled +
cyclone)
pH4.5
(1/5 lower SA/V)
Ions
(%)
S
(%)
k Ions
(%)
S
(%)
k k k k k SEM morphology before and after after dissolution
(full data in SEM Annex)
MMVF
#17
(pre-
1995)
2 95 9 Untreated –smooth surface
pH4.5 - very limited change of fibre surface after
treatment: minimal roughening, no significant gel
formation, no leaching pits.
MMVF
#1
11 85 90 4 94 31 171 Untreated - binder covers entire fibre, including 100
nm to 1 μm sized particles.
pH4.5 - Very limited change of fibre surface after
treatment: occasionally approx. 50 nm small leaching
pits (pores).
pH7.4 –significant change of fibre surface, approx.
200nm to 1 μm gel/deposit/crater structures
MMVF
#2
9 96 23 Untreated - smooth surface, occasional lumps of 100
nm to 1 μm sized particles.
MMVF
#4
10 100 20 7 92 17 39 Untreated –smooth surface
pH4.5 –very significant change of fibre surface with
pronounced gel; frequent occurence of approx.
400nm sized deep craters with sub-100-nm cracks at
bottom.
pH4.5 –without binder, change of fibre surface with
limited gel, no cracks, no pits, but approx. 200nm
sized shallow structures. Thermal removal more
effectiove than plasma removal of binder.
pH7.4 - very significant change of fibre surface with
pronounced, inhomogenously structured gel and/or
deposits, up to approx. 400nm large crater
MMVF
#5
10 85 87 4 97 35 104 122 Untreated –smooth surface, occasional lumps of 100
nm to 1 μm sized particles.
pH4.5 - significant change of fibre surface with
extensive gel formation, leaching with sub-100-nm
sized pits and cracks. pH4.5 on respirable-only fraction
also induces gel formation, leaching with sub-100-nm
sized pits
pH7.4 - significant change of fibre surface,
inhomogeneous leaching through gel with 1-μm-
diameter honeycomb structures.
MMVF
#7
11 94 40 2 99 6 58 Untreated –smooth surface, occasional 200 nm sized
particles
pH4.5 –significant change of fibre surface with
extensive gel formation, up to 4 μm sized leaching
pits, occasional micro-cracks and approx. 100 nm large
pores.
pH4.5 –binder removed thermally: no significant gel
formation, occasional micro-cracks (intermediate gel
formation if binder removed by plasma)
pH7.4 –very significant change of fibre surface with
gel formation and/or deposits of up to approx. 2 μm
large crystalline particles
MMVF
#8
10 97 65 Untreated –smooth surface. Binder covers entire fibre
(see ruptured crossings).
Wohlleben et al. Particle and Fibre Toxicology (2017) 14:29 Page 9 of 16
Content courtesy of Springer Nature, terms of use apply. Rights reserved.
to reduce costs and to increase productivity via having a
melt viscosity at 1,400° C of 10—70 poise. Hence, even
patents specify dissolution at pH 4.5, and define a particu-
larly preferred class by SiO
2
< 42.0% and Al
2
0
3
> 20% [25].
Literature “supports the use of in vitro fibre degradation
at pH 7.4 and/or pH 4.5 as an indicator of SVF [synthetic
vitreous fibre] potential pathogenicity”[26]
WeobserveatpH4.5andalsoatpH7.4dissol-
ution rates that are very similar to stone wool
MMVF21, which is plausible considering the related
Table 3 Dissolution screening at pH 7.4 and pH 4.5. “Ions”= cumulated dissolved Si and Al based on ICPMS quantification of all
eluted ions, in % of initial Si and Al; “S”= remaining solid mass, in % of initial MMVF mass; “k”= dissolution rate determined from
cumulated dissolved ions, in ng/cm
2
/h (Continued)
pH4.5 –significant change of fibre surface with gel
formation, numerous up to approx. 100 nm sized
leaching pits (pores).
MMVF
#11
13 85 58 Untreated –smooth surface, occasional lumps of 200
nm to 1 μm sized particles
pH4.5 –very significant change of fibre surface with
gel formation and deep micro-cracks (3μm x 300nm).
Pitting (approx. 1 μm large spots) and approximately
200nm visible pores
MMVF
#12
6 93 36 4 94 27 31 29 Untreated –rough surface, lateral cracks
pH4.5 –significant change of the fibre surface,
increased smoothness. No significant gel formation,
no leaching pits; pH 4.5 on respirable-only fraction:
limited gel formation, no significant leaching pits.
pH7.4 –significant change the fibre surface with
pronounced gel formation, inhomogenous leaching,
approx. 4μm x 0.5μm sized leaching pits
MMVF
#14
8 90 24 4 93 12 Untreated –binder covers entire surface, frequent
inclusion of approx. 100 nm sized particles
pH4.5- significant change of fibre surface with gel
formation, inhomogenous leaching, up to approx. 500
nm large leaching pits, occasionally up to approx 200
nm wide pits/pores.
pH7.4 –very significant change of fibre surface with
fine grained deposits of approx. 100 to 500 nm sizes.
MMVF
#20
7 89 49 Untreated –smooth surface, binder covers entire
surface
pH4.5 –significant change of fibre surface with gel
formation and erosion by frequent approx. 50 nm
sized leaching pits/pores
MMVF
#21
7 99 23 Untreated –smooth surface, very rare inclusion of
approx. 200 nm sized particles
MMVF
#22
5 91 17 5 95 15 Untreated –smooth surface with inclusion of approx
250 nm sized particles.
pH4.5 –significant change of fibre surface with gel
formation; up to 1.5 μm diameter plateaus that have
an up to 1 μm long micro crack in their center.
pH7.4 –significant change of fibre surface with gel
formation, approx. 500nm diameter leaching crater
MMVF
#24
9 91 28 Untreated –smooth surface with occasional inclusion
of approx 500 nm particles
pH4.5- very significant change of fibre surface with 2
μm diameter plateaus with several approx 50 nm
leaching pits on the plateaus
MMVF
#26
4 93 18 Untreated –structured surface with approx. 200 nm
to 1μm diameter elevations.
pH4.5 –significant change of fibre surface with gel
formation, no leaching pits.
Wohlleben et al. Particle and Fibre Toxicology (2017) 14:29 Page 10 of 16
Content courtesy of Springer Nature, terms of use apply. Rights reserved.
composition. And yet, the decreased SiO
2
content,
halfway to the MMVF34 reference, should result in
intermediate dissolution rates as well. MMVF21 at
pH4.5, tested at same SA/V as in our screening, had
k
Si
=47ng/cm
2
/h, k
leach
=72ng/cm
2
/h, and at
pH7.4 a k
total
=23ng/cm
2
/h [19]. The k
leach
rate re-
ports Ca and Mg ions whereas k
total
integrates all
measured ions. Focusing on the disintegration of vit-
reous fiber structure , this is an acceleration by 2.04
in acidic vs. neutral pH. For comparison, the average
acceleration in our test set is 2.5. Considering the
wide span from <<1 to >>1 for different MMVF
types, this is a close match. Further, also for
MMVF21 the moderate contribution of leaching with
a 1.5 fold higher leaching rate than Si-based dissol-
ution rate is fully consistent with our observations of
1.1 to 1.6 higher leaching rates (based on Mg) as
compared to the Si-based dissolution rates.
Fig. 3 MMVF #7 (average composition MMVF) morphology by
various dissolution conditions at pH 4.5 and pH 7.4, with and
without binder: SEM analysis
Fig. 4 MMVF #4 (high-alumina) morphology by various dissolution
conditions at pH 4.5 and pH 7.4, with and without binder:
SEM analysis
Wohlleben et al. Particle and Fibre Toxicology (2017) 14:29 Page 11 of 16
Content courtesy of Springer Nature, terms of use apply. Rights reserved.
In the following we systematically discuss potential
sources of error in the dissolution methodology:
1) Earlier studies used sieved material without shot. We
might have underestimated k by slow dissolution of
thick shot particles. This hypothesis was tested
experimentally here. Indeed, comparison of our
dissolution of entire MMVF against our dissolution
of respirable fraction shows in one case an
acceleration but in another case moderate
deceleration (Figure 1, Additional file 1: Figure SI_3,
Table 3: +40% for MMVF #5, -20% for MMVF #12).
Overall the shot effect and diameter effect are not
enough to explain the slow dissolution. This is
supported by Potter, finding only a 17%
acceleration between MMVF34 and respirable-
separate MMVF34 [27].
2) The media are not completely standardized
throughout literature. Our pH 4.5 medium, also
designated as “Phagolysosomal Simulant Fluid”(PSF)
[23] is actually based on earlier MMVF media, and
has the identical ingredients as the medium “C”, also
designated as “Modified Kanapilly (phthalate)”in the
interlab comparison of MMVF in different pH 4.5
media [19]. It was concluded that “The type C liquid
with the phthalate buffer gives results which in most
cases are comparable with those obtained with the
acidified Gamble's liquid (type B)”[19]. Additionally,
observations match: regarding Mg (and Al) leaching
at pH4.5 but not at pH7.4, or regarding the ratio of
rates obtained at different SA/V ratios.
3) de Meringo et al. measured k for SA/V ranging
from 10 to 400 h/cm, which extends farther than
our SA/V range [21,28].Ourdataconfirmsthat
higher k can be observed at lower SA/V, and our
actual acceleration is consistent with factors
observed in their study (Table 3), but our average
SA/V of 83 h/cm (inversely, V/SA = 0.033 μm/s)
is fully consistent with earlier data. E.g., Guldberg
Fig. 5 Binder effects on dissolution kinetics at pH 4.5 of MMVF #4,
with binder (dots), binder removed by plasma (crosses), binder
removed thermally (boxes) for the three oxides Si (black), Al (blue),
Mg (grey)
Table 4 SUMMARY of modern MMVF (#1 to #28) compared to IARC reference ranges: results on composition and dissolution. The
compositional range of Rock (stone) wool (cancerogen classification, low biosolubility) is represented by the historical reference
MMVF21. The compositional range of high-alumina, low-silica wool (synonymously HT stone wool or high biosolubility stone wool
or CAS 287922-11-6) is represented by the historical reference MMVF34
High-alumina, low-silica wool
(exonerated from classif.)
MMVF34 (represents
high biosolubility)
Rock (stone) wool,
(cancerogen classification)
MMVF21 (represents
low biosolubility)
This test set (excluding
MMVF #17)
COMPOSITION [10][19][10][19] Average Min Max
SiO
2
33 –43 39 43 –50 46 42 40 44
Al
2
O
3
18 –24 23 6 –15 13 19 15 24
CaO 23 –33 15 10 –25 17 28 16 33
MgO 10 6 –16 9
Fe-oxides 3 –973–867110
Al/(Al+Si) 0.41 0.25 0.34 0.29 0.38
DISSOLUTION Average Min Max
k
Si
pH 4.5 in
ng/cm
2
/h
> 400
‡
831 (620)* 47 (72)* 41 17 171
k
Si
pH 7.4 in
ng/cm
2
/h
58 (59)* 23 (20)* 20 6 35
*Note: Comparison of IARC Table 65 to Guldberg et al (1998) Table 4clarifies that IARC chose to document for MMVF21 and MMVF34 the k
leach
value, which
reports Ca and Mg ions. The actual fibre disintegration is assessed by earlier and by the present studies via k
Si
. Hence we compare materials based on k
Si
and
provide the values from the IARC table in brackets for transparency [10,19]
‡The value of k > 400 ng/cm
2
/h for the exonerated CAS range is from [16]
Wohlleben et al. Particle and Fibre Toxicology (2017) 14:29 Page 12 of 16
Content courtesy of Springer Nature, terms of use apply. Rights reserved.
et al. specify V/SA = 0.030 μm/s (corresponding
to SA/V = 92 h/cm, fully consistent with our
screening) [14,19]. The BIA report even
recommends “a low [V/SA] ratio of 0.003 μm/s
proved to be the most favorable condition to
relate to in-vivo data”[18].
4) Concerning ion analysis, we follow the advice from
Guldberg et al. to calculate our k values based on
dissolved ions during 25-30 days [19](32daysin
our case). They recommend Si, optionally Al. We
follow Potter to add the oxides of Si and Al for k
determination, [27] and additionally report kinetics
of Mg to assess leaching. We did not differentiate
initial k vs. average k, as proposed by initial studies
of de Meringo [21]. The limited accuracy of BET
determination imposes an uncertainty of up to 30%
for the conversion from measured ions to k values.
However, BET is not required to compare ion dis-
solution as-marketed vs. binder-removed, or pH4.5
vs. pH7.4, or entire vs. respirable-only.
5) Finally our sampling concept is fully consistent
with the Guldberg et al. method, [14] but adds
more. We believe this to be an improvement of
reliability, as the between-sampling volume repre-
sents more than 90% of all ions, whereas earlier
concepts only analyzed the samples, which hold
about 10% of all ions. It allows us to cross-check
the mass balance between remaining solids (gravi-
metric) vs. interpolated ion samplings vs. cumula-
tive ion sampling. We find excellent consistency
between the two ion-based methods (Additional
file 1: Figure SI_6b), and very good consistency
with few % mismatch between either of the ion
methods and the remaining solids (Additional file
1: Figure SI_6a). One outlier of significantly lower
remaining solids (64%) than expected from 19%
dissolved Si, Al ions (MMVF #4, thermal binder
removal), may be related to additional leaching ef-
fects, as observed on Mg (Figure 5a), but may
also indicate that because this is the only case of
dissolution >> 10% a more complicated calcula-
tion would be beneficial. Thelohan and de Mer-
ingo support that this is not required for low
overall dissolution [28]. On average across the entire
test set, the mass balance is 100.1 ± 4 % (min 95%,
max 110%), which we take as strong support for the
validity of the unexpectedly low dissolution rates.
In summary, we believe our dissolution method-
ology to be valid and appropriate for comparison
against literature data. Thus, the reasons identified for
the slow dissolution rates are the oxide composition
(discussed above) and the presence of the binder (dis-
cussed below).
Binder
We agree that studies without binder are highly relevant
for mechanistic understanding of shape-induced efffects
(the “fibre paradigm”), but not for assessment of occupa-
tional hazards. Studies without binder do not address an
occupational scenario (such as a traded intermediate),
but performed a post-processing of the as-marketed
MMVF to remove the binder or tested non-commercial
materials [29]. In reality, modern stone wool MMVF
were found to be coated by 2.8 ± 1.0 % binder (Table 1).
This value is consistent with literature [2]. Our nitrogen
adsorption isotherms show that the binder reduces the
adsorption energy of the first nitrogen layer significantly
across the ensemble fibre surface, not only locally. The
reduction of specific surface area by the binder covering
fibre-fibre touching points is moderate and does not suf-
fice to explain the reduced release of ions in the pres-
ence of the binder.
By optical microscopy of one example of glass wool
MMVF, Potter and Olang found no change in diameter
shrinkage rates in pH 7.4 with or without a novel
carbohydrate-polycarboxylic binder [30]. Due to the sig-
nificant differences in the composition of glass wool and
binder (in their example: 67.9 % SiO
2
, 1.3% Al
2
O
3
with
droplets of a hydrophilic binder that “swells in water”),
extrapolation to stone wool MMVF coated by hydropho-
bic binder and oil is impossible. To the best of our
knowledge, the effect of binder coatings on stone wool
MMVF has not been reported. Our direct comparison of
dissolution kinetics (Fig. 5, Table 3) evidences a signifi-
cant acceleration of stone wool dissolution by removal
of binder by either of the two removal processes tested.
Especially leaching of Mg is suppressed in the presence
of the binder (Fig. 5). This can be a direct effect of the
binder layer or an indirect effect via a silica-rich gel
layer. The occurence of gel on dissolving MMVF was
observed already in the 1984 WHO proceedings, and its
slowing effect on dissolution was discussed in detail [1].
Guldberg et al. highlighted that in their dissolution stud-
ies on high-alumina-low-silica MMVF –tested without
binder–gel formation was reduced, and they specifically
attributed this to the increase in Al/(Al + Si) ratio [16].
It is intriguing to compare our SEM results in Fig. 3 and
4: If we remove the binder, we reproduce the observa-
tion of Guldberg et al., as the surface of our binder-
removed MMVF after dissolution testing is smooth, does
not show leaching pits, and no obvious gel layer. This
holds for both processes that we tested to remove binder,
and for both MMVF#4 and MMVF#7 (Fig. 3 and 4). In
contrast, the same materials with their binder show a
pronounced gel formation (Fig. 3 and 4). The gel for-
mation with leaching pit features was actually observed
for the entire test set of modern MMVF, tested with
binder (Additional file 1: Figure SI_5 summarized in Table
Wohlleben et al. Particle and Fibre Toxicology (2017) 14:29 Page 13 of 16
Content courtesy of Springer Nature, terms of use apply. Rights reserved.
3. Even more SEM results are documented in Additional
file 2: SEM Annex). Potentially the binder-induced gel for-
mation is a mechanism contributing to the dissolution rates
(tested with binder) being lower than in previous literature
(tested without binder). Highly resolving SEM scans were
unfortunately not reported for the one example of glass
wool dissolution with hydrophilic binder [30].
Classification based on composition and biopersistence
The reduction of dissolution for the specific composition
(and binder) is relevant, as the range of k we measure is
the borderline region correlated by IARC to the change
of pathogenicity from MMVF21 and MMVF34 in terms
of their fibrogenic potential (Table 4): MMVF21 caused
pulmonary fibrosis, but MMVF34 did not. In 107 rats
exposed to MMVF34, no carcinoma and five adenomas
were observed. In the 107 rats in the control group, one
carcinoma and three adenomas were found [10]. All the
modern MMVF tested here where significantly different
in their composition from MMVF34, which actually is an
extreme point already in the “biosoluble MMVF”CAS
compositional range (with highest Al/(Al+Si)). MMVF34,
tested without binder, also had the highest biosolubility
within the CAS range [16]. Only for MMVF34, tested
without binder, the absence of chronic inhalation effects
[17] and absence of cancerogenicity [15] were reported.
Thus, exoneration of the CAS range was a) extrapolated
from materials without binder and b) extrapolated from
MMVF34 motivated by dissolution tests [16]. Table 4
summarizes our results on composition and dissolution,
and benchmarks modern MMVF against the IARC / CAS
ranges and representative MMVF.
For visualization, Fig. 6 plots the k results in the Al/
(Al+Si) metrics. No trend can be identified because of
the low spread of the modern MMVF in Al/(Al+Si). As
an alternative visualization, the k results are plotted in
KI metrics. KI is defined by obsolete German regulation,
now overruled by CLP, as the sum of content of the ox-
ides of Na, K, B, Ca, Mg, Ba minus twice the content of
Al oxide [31]. Interestingly, KI separates MMVF #4 and
MMVF #17 apart from the other MVMF, and a trend of
k in the KI metric is suggested by the available results,
but overall KI is not helpful to predict k. The differenti-
ation of MMVF #4 is expected because it is an innovative
product combining the high temperature performance of
stone wool with the thermal, acoustic and low weight ben-
efits of glass wool. This new type of mineral wool with a
composition similar to that of stone wool, processed
through a high temperature version of glass wool fiberis-
ing spinner, does show the same binder-induced gel for-
mation as our other test materials, but reduced sensitivity
on pH (Table 3). Regardless, we do not aim to establish
any new predictive parameterization of composition. In-
stead, the composition analysis simply shows that modern
MMVF do not all fall into the CAS range of exonerated
MMVF. Independently, the dissolution rates are found to
be lower (average 41 ng/cm
2
/h) than expected for exoner-
ated fibres that are biosoluble with no effects in intraperito-
neal injection (ip) cancerogenicity studies and in chronic
inhalation studies (>400 ng/cm
2
/h, Table 4). The most
Fig. 6 Summary of measured dissolution rates k [Si,Al, ng/cm
2
/h] as a function of the compositional parameters a,bof the molar ratio Al / (Al +
Si) or c,dof KI. The flow rates are identical throughout, and the pH is a,cpH 4.5 and b,dpH 7.4. The materials with highest k at pH 4.5 (MMVF #1,
MMVF #5) were also screened at pH 7.4 and are included in the plot
Wohlleben et al. Particle and Fibre Toxicology (2017) 14:29 Page 14 of 16
Content courtesy of Springer Nature, terms of use apply. Rights reserved.
important systematic uncertainty of 30% on absolute k
values is our use of BET for surface area determination, but
that does not compromise significance against the exoner-
ated fibre values. Only the in vivo studies are relevant for
current MMVF classification [10, 32], but biosolubility is
decisive to prevent fibre-induced pathogenicity [6, 33, 34].
Specifically for MMVF, in vitro dissolution testing is
known to correlate well with biosolubility and ultimately
with pathogenicity [10, 15–17, 26]. Classification relies on
in vivo clearance rates and/or fibrogenic and/or carcino-
genic potential [9, 32]. Our present contribution hence re-
mains a screening. We emphasize the importance of
validating the present findings by appropriately designed
in vivo studies that also use high resolution counting
methodology to determine the dimensions of retained
fibers with and without binder in the lung.
Potential of exposure
To complement the hazard screening, the potential expos-
ure needs to be known to assess the urgency of action. Ac-
cording to the European Standard EN 481 “Size Fraction
Definition for Measurement of Airborne Particles”(1993),
the thoracic fraction is that portion of the inhalable parti-
cles that pass the larynx and penetrate into the conducting
airways (trachea, bifurcations) and the bronchial region of
the lung (D50 = 10 μm), whereas the respirable fraction is
the portion of inhalable particles that enter the deepest
part of the lung, the nonciliated alveoli (D50 = 4 μm).
Thus, our studies (Table 2) show that the respirable frac-
tion of modern MMVF, assessed here by a 4 μmcutoff,is
not less than in pre-1995 MMVF. The function of the
binder is obviously to bind the fibres together, and thus it
is beneficial to reduce exposure, but the IARC collected
evidence that both manufacturing at MMVF production
plants and installation at construction sites generates
respirable airborne dust containing WHO fibres in a wide
range approximately from 0.01 to 1 fibres/cm
3
[10].
Overall, relevant exposure cannot be excluded.
Conclusion
The compositional range of modern MMVF products
(Table 4 and the additional products in Additional file 1:
Table SI_1), is not compatible with the reference mater-
ial MMVF34, that was used as benchmark to assess
fibres with high-alumina, low-silica compositions, which
consequently were exonerated from carcinogen classifi-
cation. Instead, the compositional range of modern
stone wool extends between MMVF34 (biosoluble stone
wool) and MMVF21 (low biosolubility stone wool) refer-
ence materials as limiting cases, with a compositional
average matching quite exactly the midpoint between
MMVF34 and MMVF21. These results are not limited
to a single producer or to a single country of origin, but
cover 5 producers from 6 countries. The dissolution
rates at pH 4.5, measured by a replicate of setups that
were developed and validated in the 1990ies, are an
order of magnitude slower than those reported for bio-
soluble MMVF34. This is significant considering all
known sources of error. Despite the SiO
2
content of an
average modern MMVF being reduced vs. the historical
benchmark MMVF21, the measured average dissolution
rates at both pH 7.4 and pH 4.5 are within 20% identical
to MMVF21 (Table 4). This is explained, at least in part,
by the presence of up to 4% binder that coats the
MMVF and has a significant influence on dissolution,
probably by favoring gel layer formation. Here we tested
MMVF as they are marketed and handled, i.e. with
binder, whereas it appears that previous hazard assess-
ment relied on abiotic, in-vitro and in-vivo studies with
MMVF dissolution accelerated by deliberate removal of
the binder. Considering that modern MMVF, with their
binder, have actual dissolution rates ranging from 6 ng/
cm
2
/h to 171 ng/cm
2
/h, which is the borderline range
correlated to the onset of lung fibrosis and thoracic tu-
mors, [10] and considering further the content of respir-
able fibres, the risk assessment of modern stone wool
may need to be revisited. However, in vitro dissolution
studies remain indicative and cannot replace nor predict
in-vivo studies of MMFV as marketed (with binder).
Endnotes
1
“Es ist weiterhin notwendig, nicht nur binderfreies
Material zu untersuchen, das in der Arbeitswelt kaum
auftaucht, sondern Fasern in handelsüblicher Form, d.h.
mit Bindermaterial.“(BIA report 02/98, Conclusions p.
139, endorsed also by “Stellungnahme R.C. Brown, U.
Nebe, ECFIA: …Im übrigen teilen wir die Ansicht, daß es
weiterhin notwendig ist, nicht nur binderfreies Material
zu untersuchen, das in der Arbeitswelt kaum auftaucht,
sondern Faserprodukte in handelsüblicher Form.“p. 279).
Author’s translations: “Furthermore it is necessary to in-
vestigate not only binder-free material, which is uncom-
mon in the working environment, but fibres in the form
that is customary in trade, i.e. with binder material.“and
“…by the way we share the view that it is necessary to
investigate not only binder-free material, which is uncom-
mon in the working environment, but fibre products in
the form that is customary in trade,“
Additional files
Additional file 1: Online Supporting Information: Schematics of
respirable fibre fractionation and of dissolution testing. Additional results
on MMVF composition, on respirable fibre content and on dissolution
kinetics at pH4.5. Characteristic features in SEM scans before and after
dissolution testing. Cross-correlation of three metrics of dissolution analysis.
(PDF 967 kb)
Additional file 2: Annex: extensive SEM before and after dissolution testing.
(PDF 8398 kb)
Wohlleben et al. Particle and Fibre Toxicology (2017) 14:29 Page 15 of 16
Content courtesy of Springer Nature, terms of use apply. Rights reserved.
Acknowledgements
We thank the medical department for general introduction to MMVF in
thermal insulation applications. We are indebted to Edgar Leibold, Product
Stewardship, for clarifying discussions on WHO fibre risk assessment. We
thank Werner Wacker for performing the plasma process for binder removal,
Rolf Henn for MMVF sourcing, logistics and binder quantification, Frank
Mueller for coordinating the milling process and Christian Schatz for
performing the aerosol fractionation. We thank the ICP/MS/OES team,
specifically Petra Veitinger, Andreas Herzog, Walter Grasser for excellent
support.
Funding
This work was funded by BASF SE.
Availability of data and materials
SEM, gravimetry and ICPMS results for all MMVF tested, all conditions tested,
are shown in the SI.
Authors’contributions
BD and WW designed the study. BD coordinated and interpreted the aerosol
part, WW coordinated and interpreted the dissolution part. KW replicated
dissolution methods from literature and performed all dissolution
experiments. MD performed SEM analysis, HW interpreted the SEM results.
HE coordinated elemental analysis. WW, BD, HW wrote the manuscript with
contributions by HE. All authors read and approved the final manuscript.
Ethics approval and consent to participate
Not applicable
Consent for publication
Not applicable
Competing interests
All authors are employees of BASF SE, a company that commercializes
products competing with MMVF for thermal insulation purposes. BASF SE is
not itself a producer of MMFV. Maintenance operations by BASF SE
employees extensively handle both historic MMVF for demolition or
refurbishment and modern MMVF in new installations.
Publisher’sNote
Springer Nature remains neutral with regard to jurisdictional claims in
published maps and institutional affiliations.
Author details
1
Department Material Physics and Analytics, BASF SE, Ludwigshafen,
Germany.
2
Department of Aerosol Technology, BASF SE, Ludwigshafen,
Germany.
Received: 4 May 2017 Accepted: 27 July 2017
References
1. WHO (World Health Organization). Biological Effects of Man-Made Mineral
Fibres.Proceedings of a WHO/IARC Conference. 1984.
2. TIMA (Thermal Insulation Manufacturers Association), Man-made Vitreous
Fibres: Nomenclature, Chemical and Physical Properties, 4th Ed., W. Eastes,
Editor. 1993, Nomenclature Committee of Thermal Insulation
Manufacturers’Association, Refractory Ceramic Fibres Coalition (RCFC), :
Washington DC.
3. IARC. (INTERNATIONAL AGENCY FOR RESEARCH ON CANCER), Man-made
Mineral Fibres and Radon 1988. WORLD HEALTH ORGANIZATION.
4. Roller M, et al. Dose-response relationship of fibrous dusts in intraperitoneal
studies. Environmental Health Perspectives. 1997;105(Suppl 5):1253–6.
5. Miller BG, et al. Influence of fibre length, dissolution and biopersistence on the
production of mesothelioma in the rat peritoneal cavity. Ann Occup Hyg. 1999;43
6. Donaldson K, Seaton A. A short history of the toxicology of inhaled particles.
Particle and Fibre Toxicology. 2012;9(1):13.
7. Oberdörster G. Determinants of the pathogenicity of man-made vitreous fibres
(MMVF). International archives of occupational and environmental health.
2000;73(1):S60–8.
8. Guldberg M, et al. The development of glass and stone wool compositions
with increased biosolubility. Regulatory Toxicology and Pharmacology.
2000;32(2):184–9.
9. CLP, Classification, labelling and packaging of substances and mixtures,
amending and repealing Directives 67/548/EEC and 1999/45/EC, and
amending Regulation (EC) No 1907/2006,inREGULATION (EC) No 1272/2008,
THE EUROPEAN PARLIAMENT AND THE COUNCIL, Editor. 2008. p. 1-1355.
10. IARC (INTERNATIONAL AGENCY FOR RESEARCH ON CANCER), MAN-MADE
VITREOUS FIBRES. 2002, WORLD HEALTH ORGANIZATION. p. 1-433.
11. Baan RA, Grosse Y. Man-made mineral (vitreous) fibres: evaluations of cancer
hazards by the IARC Monographs Programme. Mutation Research/Fundamental
and Molecular Mechanisms of Mutagenesis. 2004;553(1–2):43–58.
12. RAL, Erzeugnisse aus Mineralwolle. 1999. Gütesicherung RAL GZ 388.
13. EIPPCB_(European_Integrated_Pollution_Prevention_and_Control_Bureau),
Reference Document on Best Available Techniques in the Glass Manufacturing
Industry. 2000, EC Joint Research Centre, Institute for Prospective
Technological Studies: Sevilla.
14. Guldberg M, et al. Method for determining in-vitro dissolution rates of man-
made vitreous fibres. Glass science and technology. 1995;68(6):181–7.
15. Kamstrup O, et al. Carcinogenicity Studies after Intraperitoneal Injection of Two
Types of Stone Wool Fibres in Rats. The Annals of Occupational Hygiene.
2002;46(2):135–42.
16. Guldberg M, et al. High-alumina low-silica HT stone wool fibres: A chemical
compositional range with high biosolubility. Regulatory Toxicology and
Pharmacology. 2002;35(2):217–26.
17. Kamstrup O, et al. Chronic inhalation studies of two types of stone wool fibres
in rats. Inhalation Toxicology. 2001;13(7):603–21.
18. Muhle, H., et al., Fasern - Tests zur Abschätzung der Biobeständigkeit und zum
Verstaubungsverhalten (BIA-Report 2/98). . 1998: Sankt Augustin.
19. Guldberg M, et al. Measurement of in-vitro fibre dissolution rate at acidic pH.
Annals of Occupational Hygiene. 1998;42(4):233–43.
20. Zoitos BK, et al. In vitro measurement of fibre dissolution rate relevant to
biopersistence at neutral pH: An interlaboratory round robin. Inhalation
toxicology. 1997;9(6):525–40.
21. De Meringo A, et al. In vitro assessment of biodurability: acellular systems.
Environmental health perspectives. 1994;102(Suppl 5):47.
22. Roller M. Bedeutung von In-vitro-Methoden zur Beurteilung der chronischen
Toxizität und Karzinogenität von Nanomaterialien, Feinstäuben und Fasern.
Dortmund, Berlin: BAuA; 2011. p. 1–365.
23. Stefaniak AB, et al. Characterization of phagolysosomal simulant fluid for
study of beryllium aerosol particle dissolution. Toxicology in Vitro. 2005;
19(1):123–34.
24. Marques M, Loebenberg R, Almukainzi M. Simulated Biological Fluids
with Possible Application in Dissolution Testing. Dissolution Technologies.
2011;8:15–29.
25. Jensen, S.L., V.R. Christensen, and M. Guldberg, Man-made vitreous fibres.
2002, Patent US 6346494 B1.
26. Hesterberg, T.W. and G.A. Hart, Lung Biopersistence and in Vitro Dissolution
Rate Predict the Pathogenic Potential of Synthetic Vitreous Fibres. Inhalation
Toxicology, 2000. 12(sup3): p. 91-97.
27. Potter RM. A method for determination of in-vitro fibre dissolution rate by
direct optical measurement of diameter decrease. Glastechnische Berichte.
2000;43:46–55.
28. Thelohan S, De Meringo A. In vitro dynamic solubility test: influence of various
parameters. Environmental health perspectives. 1994;102(Suppl 5):91.
29. Kamstrup O, et al. Subchronic Inhalation Study of Stone Wool Fibres in Rats.
The Annals of Occupational Hygiene. 2004;48(2):91–104.
30. Potter RM, Olang N. The effect of a new formaldehyde-free binder on the
dissolution rate of glass wool fibre in physiological saline solution. Particle and
Fibre Toxicology. 2013;10(1):13.
31. AGS (Arbeitsgemeinschaft Gefahrstoffe), TRGS 905 Anorganische Faserstäube
(außer Asbest), 1995.
32. Harrison P, et al. Regulatory risk assessment approaches for synthetic mineral
fibres. Regulatory Toxicology and Pharmacology. 2015;73(1):425–41.
33. Bernstein DM, et al. BIOPERSISTENCE OF SYNTHETIC MINERAL FIBRES AS A
PREDICTOR OF CHRONIC INTRAPERITONEAL INJECTION TUMOR RESPONSE IN
RATS. Inhalation Toxicology. 2001;13(10):851–75.
34. Lippmann M. Toxicological and epidemiological studies on effects of airborne
fibres: Coherence and public health implications. Critical Reviews in
Toxicology. 2014;44(8):643–95.
Wohlleben et al. Particle and Fibre Toxicology (2017) 14:29 Page 16 of 16
Content courtesy of Springer Nature, terms of use apply. Rights reserved.
1.
2.
3.
4.
5.
6.
Terms and Conditions
Springer Nature journal content, brought to you courtesy of Springer Nature Customer Service Center GmbH (“Springer Nature”).
Springer Nature supports a reasonable amount of sharing of research papers by authors, subscribers and authorised users (“Users”), for small-
scale personal, non-commercial use provided that all copyright, trade and service marks and other proprietary notices are maintained. By
accessing, sharing, receiving or otherwise using the Springer Nature journal content you agree to these terms of use (“Terms”). For these
purposes, Springer Nature considers academic use (by researchers and students) to be non-commercial.
These Terms are supplementary and will apply in addition to any applicable website terms and conditions, a relevant site licence or a personal
subscription. These Terms will prevail over any conflict or ambiguity with regards to the relevant terms, a site licence or a personal subscription
(to the extent of the conflict or ambiguity only). For Creative Commons-licensed articles, the terms of the Creative Commons license used will
apply.
We collect and use personal data to provide access to the Springer Nature journal content. We may also use these personal data internally within
ResearchGate and Springer Nature and as agreed share it, in an anonymised way, for purposes of tracking, analysis and reporting. We will not
otherwise disclose your personal data outside the ResearchGate or the Springer Nature group of companies unless we have your permission as
detailed in the Privacy Policy.
While Users may use the Springer Nature journal content for small scale, personal non-commercial use, it is important to note that Users may
not:
use such content for the purpose of providing other users with access on a regular or large scale basis or as a means to circumvent access
control;
use such content where to do so would be considered a criminal or statutory offence in any jurisdiction, or gives rise to civil liability, or is
otherwise unlawful;
falsely or misleadingly imply or suggest endorsement, approval , sponsorship, or association unless explicitly agreed to by Springer Nature in
writing;
use bots or other automated methods to access the content or redirect messages
override any security feature or exclusionary protocol; or
share the content in order to create substitute for Springer Nature products or services or a systematic database of Springer Nature journal
content.
In line with the restriction against commercial use, Springer Nature does not permit the creation of a product or service that creates revenue,
royalties, rent or income from our content or its inclusion as part of a paid for service or for other commercial gain. Springer Nature journal
content cannot be used for inter-library loans and librarians may not upload Springer Nature journal content on a large scale into their, or any
other, institutional repository.
These terms of use are reviewed regularly and may be amended at any time. Springer Nature is not obligated to publish any information or
content on this website and may remove it or features or functionality at our sole discretion, at any time with or without notice. Springer Nature
may revoke this licence to you at any time and remove access to any copies of the Springer Nature journal content which have been saved.
To the fullest extent permitted by law, Springer Nature makes no warranties, representations or guarantees to Users, either express or implied
with respect to the Springer nature journal content and all parties disclaim and waive any implied warranties or warranties imposed by law,
including merchantability or fitness for any particular purpose.
Please note that these rights do not automatically extend to content, data or other material published by Springer Nature that may be licensed
from third parties.
If you would like to use or distribute our Springer Nature journal content to a wider audience or on a regular basis or in any other manner not
expressly permitted by these Terms, please contact Springer Nature at
onlineservice@springernature.com
Available via license: CC BY
Content may be subject to copyright.
