Eur. J. Mineral.

1997, 9, 585-596

Contrasting chemical compositions in associated lizardite

and chrysotile in veins from Elba, Italy

CECILIA VITI and MARCELLO MELLINI

Dipartimento di Scienze della Terra, Via delle Cerchia, 3,

1-53100 Siena, Italy

Abstract: Serpentinitic veins consisting of associated chrysotile and well crystallized lizardite were extensively
sampled in the Monte Fico quarries, Elba Island, Italy.
The veins are mostly formed ofeuhedral lizardite-IT. After optical, IR and X-ray characterization, the specimens
were studied by transmission electron microscopy. TEM provides evidence for the occurrence of chrysotile, poly­
gonal serpentine and lizardite, with microstructural relations pointing to a continuous recrystallization from chry­
sotile to polygonal serpentine and to lizardite. No major fault structure occurs in lizardite.
Chemical data indicate the existence of a compositional gap between lizardite [mean composition (Mg2.79Fe2+o.o9
3+
Fe o.o7Alo.o7)Σ=3.02(Sii.9iAlo.o9)Σ=2.oo O5 (OH)3.9o] and chrysotile [mean composition (Mg2.8oFeo.nAlo.o6)Σ=2.97Si2.oo O5
(0H)4] with lizardite systematically enriched in aluminum and iron with respect to chrysotile. Therefore, the two
minerals should no longer be considered as polymorphs.

Key-words: lizardite, chrysotile, Elba, polymorphism, infrared.

Introduction On the contrary, no sharp distinction has been

Serpentine minerals (lizardite, chrysotile and
antigorite) have been schematically considered as
p o l y m o r p h s with the ideal c o m p o s i t i o n
Mg3Si2O5(OH)4. Nevertheless, important chemi­
cal differences exist. For instance, antigorite de­
viates from the ideal composition of serpentine
by showing limited but significant brucite loss;
this feature, rather than being random, is related
to polysomatic modulations that require precise
structural and chemical modifications (Uehara &
Shirozu, 1985; Mellini et al, 1987; Uehara &
Kamata, 1994).
universally accepted between lizardite and chry­
sotile, and contrasting opinions have often been
proposed (for instance, Page, 1968; Whittaker &
Wicks, 1970). On the one hand, lizardite and
chrysotile have been considered as proper poly­
morphs, while, on the other hand, great emphasis
has been given to the role of aluminum in stabi­
lizing lizardite with respect to chrysotile (Caruso
& Chernosky, 1979). According to the common
interpretation, substitution of aluminum for sili­
con and magnesium can resolve the dimensional
misfit between the larger octahedral sheet and the
smaller tetrahedral sheet (Whittaker & Zussman,

E-mail: mellini@unisi.it

0935-1221/97/0009-0585 $ 3.00
© 1997 E. Schweizerbart'sche Verlagsbuchhandlung. D-70176 Stuttgart
586 C. Viti, M. Mellini

1956; Krstanovic & Pavlovic, 1964; Wicks & within the quarry outcrop, under optimal condi-
Plant, 1979; Wicks & O'Hanley, 1988). Alterna- tions for observation. We think, however, that
tively, the substitution of Mg 2+ and Si 4+ by triple similar euhedral lizardite veins may also be pre-
charge cations (i.e., Al and Fe 3+ ) is able to sent in other serpentinitic outcrops.
polarize the layer and produce stronger electro- The veins are formed mostly of lizardite-\T
static interactions (hydrogen bonds) between crystals, typically 0.3 x 0.3 x 0.7 mm in size
adjacent layers, thus stabilizing the flat structure (Fig. la). The texture is generally granoblastic,
of lizardite (Mellini, 1982; Mellini & Zanazzi,
1987; Mellini & Viti, 1994; Gregorkiewitz et al,
1996). Fe 3+ has been demonstrated by Mossbauer
spectroscopy to be more abundant in lizardite
than in chrysotile (O'Hanley & Dyar, 1993).
Uncertainties on lizardite and chrysotile com-
positions arise from two main factors. Firstly, it
is difficult to identify serpentine minerals (Wicks
& Zussman, 1975). Secondly, there is often
mutual contamination between these minerals
(Cressey & Zussman, 1976; Cressey, 1979);
lizardite and chrysotile in low-grade serpentinitic
rocks commonly form intimate intermixtures, so
that the analysed volume is rarely monomineralic
(except in thin-film TEM-EDS analyses, which
produce more spatially resolved data, together
with the concomitant texturaJ control of the
specimen).
More favourable results may be obtained from
vein serpentine, as reported here: unusual fea-
tures, such as the large grain size and the good
crystallinity, make it possible to obtain uncon-
taminated analyses, both for lizardite and as-
sociated chrysotile.

Occurrence and specimen description

Serpentine veins are common within the lizar-

dite-chrysotile serpentinite, with mesh and bastite
pseudomorphic textures, of Elba Island, Italy.
Several kinds of veins have been recognized
there: chrysotile veins, antigorite veins (Viti &
Mellini, 1996), lizardite-chrysotile veins and
more complex veins, where chrysotile, polygonal
serpentine and lizardite are present in variable
amounts.
The present results refer to lizardite-chrysotile
veins, representing, as far as we know, the largest
recorded occurrence of euhedral lizardite. The
veins were sampled within the two Monte Fico Fig. 1. a) Granoblastic lizardite texture, as seen along
quarries, where they appear as dark-green mas- the vein strike. The crystals are mostly oriented, except
sive bodies that are irregularly developed along for the two large crystals that show evident (001) cleav-
age planes, b) In some cases, the lizardite crystals are
deformation zones. The mesoscopic resemblance intermixed with an irregularly extinct matrix consisting
to the surrounding massive serpentinites explains of chrysotile and polygonal serpentine, c) Isolated euhe-
why this kind of vein has been recognized only dral lizardite-XT crystals.
 Vein lizardite and chrysotile 587

Table 1. X-ray diffraction patterns (Å) of lizardite from
Monte Fico North veins (sample MFN3). M + SL =
monochromator and side loading; M + NL = monochro-
mator and normal loading; NL = normal loading and no
monochromator; Gand. = Gandolfi camera.
(Mellini & Viti, 1994) and for the Rietveld re­
finement of powder neutron diffraction data
(Gregorkiewitz et al, 1996).

M + SL M+NL NL Gand. X-ray powder diffraction

lhkl d (A) Int. d(A) Int. <* (A) Int. d ( A ) Int.
001 7.240 1000 7.217 1000 7.234 1000 7.278 iooo\
100 4.606 96 4.600 31 4.605 10 4.666 500 X-ray diffraction patterns (CuKa radiation)
101 3.890 69 3.885 23 3.894 10 3.917 400
002 3.625 593 3.620 695 3.623 600 3.693 100 were collected for a monomineralic sample of
102 2.852 23 2.850 11 2.852 4 2.844 100
110 2.663 12 2.660 5 2.659 2.663 100
lizardite (MFN3 = Monte Fico North, specimen 3),
111 2.499 290 2.497 108 2.500 22 2.518 800 using different experimental conditions: Gandolfi
003 2.419 32 2.417 36 2.418 28
201 2.196 6 2.195 3 camera, automated Philips PW1710 Bragg-Bren-
112 2.146 122 2.145 52 2.147 15 2.146 700 tano diffractometer (both normal and side load­
202 1.947 6 1.945 2 1.953 100
004 1.814 13 1.814 15 1.814 13 ing, with and without post-diffraction monochro­
113 1.791 72 1.790 36 1.791 11 1.794 600
210 1.744 13 1.742 5 1.743 1 1.750 100
mator). The patterns are reported in Table 1,
211 1.695 12 1.694 4 1.693 1 1.698 100 together with the unit cell parameters determined
203 1.670 6 1.669 3 1.670 1 1.670 100
212 1.572 6 1.572 3 by least-squares refinements of a selected number
300 1.538 91 1.537 28 1.538 7 1.536 400 of reflections. The unit cell parameters (a =
301 1.505 49 1.504 21
114 1.500 62 1.500 36 1.500 15 1.504 300 5.332, c = 7.267 Å) are in good agreement with
005 1.452 8 1.452 11 1.452 8
204 1.426 4 1.427 2
those obtained by single-crystal X-ray diffraction
302 1.416 15 1.415 5 1.416 2 1.417 200 (a = 5.338 , c = 7.257 A) and powder neutron
220 1.332 7 1.331 3
221 1.310 41 1.310 14 1.309 3 1.311 200 diffraction (a = 5.333 , c = 7.272 Å).
303 1.298 8 1.298 3
115 1.276 21 1.275 14 1.275 7
The X-ray powder diffraction pattern obtained
311 1.260 5 1.260 2 with side loading and post-diffraction monochro-
222 1.251 9 1.250 3 1.252 100
205 1.229 2 1.229 2 mator (Fig. 2) shows more lines than usual. In
223 1.167 15 1.167 6 fact, as previously shown for antigorite (Viti &
116 1.102 4 1.102 3
206 1.074 10 1.074 4 Mellini, 1996), this is the most informative pat-
\mm

305 1.056 10 4
0.997 15 4
tern: side loading, by reducing the [001] pre-
0.982 4 2 ferred orientation effects, produces general hkl
0.971 8 3
0.967 3 2 reflections whereas monochromator removes
0.952 6 3 anelastic diffusion and leads to favourable
a = 5J32Ä a = 5329A a = 5.328A a = 5J33Ä 1
c = 7.267Å c = 7.266Å c = 7.266Å c = 7.266Å | peak/background ratios.

with the lizardite crystals nearly oriented parallel

to [001] which corresponds to the vein strike.
Euhedral lizardite is pleochroic, being light-green
when parallel to and light-yellow when perpen-
dicular to [001]; crystals are length-fast with low
birefringence and sharp parallel extinction. Local
textural variations may be observed with the
grain dimensions attaining 0.4 x 0.4 x 2 mm.
The lizardite crystals are occasionally
embedded in a fibrous matrix of chrysotile and
minor polygonal serpentine (as later shown by
TEM analysis). The amount of fibrous serpentine
varies from sample to sample; optically, thin
ondulating extinction bands characterize the
fibrous matrix (Fig. lb).
The density of lizardite, determined by heavy
liquid methods, is 2.611(3) g cπr 3 .
The Monte Fico lizardite (Fig. lc) has been
used for single-crystal X-ray structure refinement
TEM observations
The Monte Fico veins were studied by trans-
mission electron microscopy (Philips 400T,
working at 120 kV), mainly as ion-thinned sec-
tions. A 50 µm objective aperture was used for
image formation (nominal point to point resolu-
tion close to 4 A); electron diffraction patterns
were obtained from selected areas 1 µm in
diameter.
The lizardite crystals correspond to highly
ordered 1T polytype (based upon both electron
diffraction data and high-resolution images):
(001) lattice fringes do not show any deforma-
tion, irregular spacings, defects or other disorder
features.
The interstitial matrix consists of chrysotile
and polygonal serpentine. However, from point
588 C. Viti, M. Mellini

90000

[counts]

*\JV-J uy W LjWΛuk I I I I I I I I I I
100 [ ° 2 β ]

Fig. 2. X-ray diffraction pattern of MFN3 lizardite; side loading and post-diffraction monochromator.

to point, this matrix varies in amount, mineral Microstructural relations between lizardite and
composition and microstructural features. chrysotile are variable and change with the phase
In some places, chrysotile is the dominant min­ boundary. Parallel to the (001) lattice fringes of
eral (Fig. 3); the individual fibres are typically lizardite, the boundary with the fibrous matrix is
around 300 Å in diameter, with a central hole of
60-80 Å (possibly filled with amorphous mate-
rial); when seen along the fibre axis, chrysotile
shows (00 l) lattice fringes corresponding to reg-
ularly rolled layers. The fibre bundle consists of
several short chrysotile fibres, arranged accord-
ing to a preferred orientation pattern, but with
several fibres crossing the main texture.
Two kinds of polygonal fibres are associated
with chrysotile. The first one (Fig. 4a), typically
550-2500 A in diameter, shows a characteristic
indentation due to incomplete fibre polygonaliza-
tion: planar (001) lattice fringes form polygonal
sectors near the fibre rim but remain rolled, as in
chrysotile, near the fibre core. The number of
sectors is usually thirty, as reported by Baronnet
et ai (1994). The (001) sequence is frequently
disturbed by interlayer partings, particularly
near the fibre core. A small central hole, again
60-80 Å in diameter, is seen to occur here. The
second kind of fibre (Fig. 4b), with diameter
larger than 2500 A, is made up of polygonal
serpentine; fibres always consist of 30 sectors,
each characterized by regular sequences of planar
(001) lattice fringes. No indentation pattern is
produced here and polygonalization with sharp
intersector contacts can be observed. Fig. 3. Short fibre chrysotile of the interstitial matrix.
 Vein lizardite and chrysotile 589

mm 1

200 Å 500 Å
:, ېW
Ma\ j
Fig. 4. a) Polygonal serpentine fibre along the fibre axis; the number of indentations is 30. b) Polygonal serpentine
fibre with 30 sectors clearly evident.

irregular and indented, particularly close to the
larger serpentine fibres that are incompletely
resorbed within the lizardite crystals (Fig. 5).
Chrysotile sometimes remains within lizardite as
isolated relic fibres. Nearly perpendicular to the
(001) lattice fringes of lizardite, the boundary is
more regular, but different textural relations are
observed. In particular, minor lamellae of lizar-
dite may grow from the crystal boundary towards
the fibrous matrix (Fig. 6a). Alternatively, equi-

Fig. 5. Reaction front between lizardite and chrysotile; the orientation of lizardite (001) lattice fringes is horizontal
and parallel to the reaction front. A relic fibre of chrysotile is present in the right-hand part of the figure.
590 C. Viti, M. Mellini

Fig. 6. a) Reaction front between lizardite and the fibrous matrix; the orientation of lizardite (001) is horizontal,
nearly perpendicular to the reaction front. Minor lamellae of lizardite grow towards the fibrous matrix; (001) lattice
fringes are continuous across the boundary between the main crystal and the newly-formed lamellae of lizardite.
b) Reaction front between lizardite and fibrous matrix, with lizardite (001) having the same orientation as above.
Characteristic half-fibres of chrysotile are present at the reaction front.

dimensional half-fibres of chrysotile may form at
crystal boundaries, resulting from regular bend-
ing of lizardite (001) (Fig. 6b).
The overall texture observed at the TEM scale
is consistent with the following interpretation:
chrysotile is the early formed phase, but is later
transformed to polygonal serpentine and then to
lizardite. Continuous lizardite growth finally con-
sumes also the polygonal serpentine (Fig. 7).
Chemical composition
The compositions of lizardite and chrysotile,
associated in eight samples from at least four
different veins, were determined by electron mi-
croprobe analysis (EMPA), using a Jeol JXA
8600 equipped with wavelength-dispersive
spectrometers (WDS), and by energy dispersive
spectrometry (EDS) in both scanning and trans-
mission electron microscope modes.
The two minerals were firstly recognized using
their optical features under the polarizing micro-
scope. This procedure identifies the euhedral
lizardite crystals in relation to the chrysotile and
polygonal serpentine matrix (shown to be mostly
chrysotile by transmission electron microscopy).
Mean WDS compositions are reported in Table 2
(a.p.f.u., atoms per formula unit, calculated on
the basis of seven anhydrous oxygens; aluminum
distributed between octahedral and tetrahedral
sites to obtain complete occupancy of the tetrahe-
dral site). The compositional ranges for lizardite
are 2.77-2.87 (Mg), 0.12-0.17 (Fe), 0.04-0.07
(octahedral Al) and 1.88-1.97 (Si), 0.03-0.12 (tet-
rahedral Al), to be compared with compositional
ranges for chrysotile of 2.70-2.88 (Mg), 0.08-
0.16 (Fe), 0.03-0.08 (octahedral Al) and 1.97-
2.02 (Si), 0.00-0.03 (tetrahedral Al). These data
represent minimal estimates of the chemical
differences between lizardite and chrysotile. In
fact, in spite of the careful selection of individual
points for probing, several features may still bias
the data by producing contaminated analyses. We
may distinguish artifacts (e.g., beam broadening
in the bulk specimen, inaccurate positioning of
the electron spots, vertically inclined phase boun-
 Vein lizardite and chrysotile
Fig. 7. Lizardite lamella deforming a polygonal serpen­
tine fibre.
daries) from real phenomena (optically unre­
solved occurrence of intermixed phases).
Therefore, associated lizardite and chrysotile
from Monte Fico have distinctly different com­
positions. In particular, lizardite is always Al-
enriched, with aluminum substituting both for
tetrahedral silicon and for octahedral magnesium.
In chrysotile, the aluminum substitution is signif­
icantly lower: a variable amount of aluminum
may enter the octahedral site, but the tetrahedral
site is almost completely occupied by silicon
alone. To a lesser extent, lizardite and chrysotile
also differ in their iron content (here calculated
as Fe 2+ ), lizardite being always iron-enriched
with respect to the coexisting chrysotile.
The chemical distinction between the two min­
erals, as well as the combined effects of alum­
inum and iron, are depicted in Fig. 8, where the
FeO and AI2O3 contents appear positively corre­
lated: lizardite occupies a wide and continuous
field covering never less than 3 wt% in FeO and
1.3 wt% in AI2O3. On the contrary, chrysotile
591
plots towards Al-depleted and, for most of the
points, Fe-depleted compositions. Within the
chrysotile field, we may distinguish a few
anomalous points, close to 4 wt% FeO, possibly
due to local heterogeneities within the matrix.
The main chrysotile field is apparently split in
more (two or three) distinct compositional sub-
fields. We interpret the overall pattern as result­
ing from distinct compositional difference be­
tween lizardite and chrysotile matrix and we
attribute the different subfields of the chrysotile
matrix to local mineralogical differences, as
pointed out by TEM observations (for instance,
variable amounts of polygonal serpentine with re­
spect to chrysotile).
Loss of ignition (LOI) of 12.6 wt% and
Fe 2+ /Fe 3+ ratio (by titrimetry with KMnO4, FeO
2.34 wt% and Fe2O3 l .98 wt%) were determined
for a monomineralic sample of lizardite (crystals
extracted from the MFN3 sample).
We tried also to select a monomineralic sample
of MFN3 chrysotile, in order to compare the con­
tent of triple-charged cations in associated
serpentines; unfortunately the available amount
of MFN3 chrysotile was insufficient to allow
Fe 2+ /Fe 3+ titrimetric determination or LOI analy­
sis.
Combining microprobe data with titrimetry
and loss of ignition, we obtain the following
mean composition for MFN3 lizardite:
(Mg279Fe2+0.09Fe3+o.07Alo.07)Σ=3.02
(Sii.9lAlo.09)Σ=2.00 O5 (OH)3.90.
This result indicates concerted substitution of
trivalent cations (Al and Fe 3+ ) in lizardite. By
contrast, the average composition for the as­
sociated MFN3 chrysotile matrix is (Mg2.80Feo.11
Alo.Oô)Σ=2.97Si2.00 O5(OH)4.
Infrared data
Infrared spectra (from 4000 to 400 cm-1) were
collected with a Perkin Elmer 1710 FTIR
spectrometer, using KBr disks with approxi­
mately 2 mg of dispersed monomineralic
powder; KBr disks were kept in the oven over­
night to eliminate the effects of atmospheric
water in the O-H stretching region. IR spectra of
Monte Fico lizardite and chrysotile are reported
in Fig. 9 and may be compared with results for
vein antigorites (Fig. 3 in Viti & Mellini, 1996).
Lizardite, chrysotile and antigorite from the
Elba veins yield results in agreement with
592 C. Viti, M. Mellini

Table 2. Mean compositions (WDS) for associated lizardite and chrysotile from the Monte Fico veins.

MF19b MFN9a M F N 9 b
pz MF25 MF22b MF22a MF23 MFN3
|Si02 41.21 40.93 41.15 41.78 42.44 42.85 43.48 43.52
AI2o3 3.44 3.16 2.83 2.39 2.97 2.29 2.01 1.42
FeO 4.27 3.58 3.50 3.20 4.28 4.42 3.56 3.62
MgO 41.03 41.18 41.35 42.13 41.52 40.91 41.65 41.47
Itotwt% 89.94 88.86 88.83 89.49 91.21 90.47 90.69 90.03

Si 1.88 1.89 1.90 1.91 1.91 1.94 1.96 1.97

Al IV 0.12 0.11 0.10 0.09 0.09 0.06 0.04 0.03
Al VI 0.07 0.06 0.05 0.04 0.07 0.06 0.06 0.05
Fe 0.16 0.14 0.14 0.12 0.16 0.17 0.13 0.14
Mg 2.79 2.83 2.84 2.87 2.79 2.77 2.80 2.80
tot 5.02 5.03 5.03 5.03 5.02 5.00 4.99 4.99

ICHR MF25 MF22b MF22a MF23 MFN3 M F 1 9 b MFN9a M F N 9 b |

|Si02 43.13 42.62 43.22 43.54 44.83 44.71 44.85 44.70
AI2O3 1.89 1.19 0.89 1.23 1.63 1.14 1.16 1.05
FeO 3.07 2.46 2.05 2.30 3.30 4.28 2.72 2.67
MgO 40.92 41.79 41.61 41.67 41.43 40.09 41.53 41.36
totwt% 89.00 88.05 87.76 88.74 91.19 90.22 90.27 89.77

Si 1.97 1.97 1.99 1,99 2.00 2.02 2.01 2.02

Al IV 0.03 0.03 0.01 0.01 0.00 0.00 0.00 0.00
Al VI 0.07 0.03 0.04 0.05 0.08 0.06 0.06 0.06
Fe 0.12 0.09 0.08 0.09 0.12 0.16 0.10 0.10
Mg 2.79 2.88 2.86 2.84 2.75 2.70 2.78 2.78
[tot 4.98 5.00 4.98 4.98 4.95 4.94 4.95 4.96 1

previous IR data (Brindley & Zussman, 1959;
Yariv & Heller-Kallai, 1975; Uehara & Shirozu,
1985). T h e IR bands in the 3 6 9 0 - 3 4 0 0 c m *
spectral region are due to O-H (inner and outer)
stretching vibrations, with low-frequency shifts
depending upon the amount of aluminum substi­
tution for silicon and magnesium (Serna et al,
1977, 1979; Velde, 1980).
In our spectra, the inner O-H produces strong
o chrysotile
5 Γ
^ ^ T 1 — 1 —
4.5
^x
4
o lizardite
absorption bands at 3677-3688 c m 1 (frequencies
in antigorite are lower than in lizardite and chry­
sotile). As regards the outer O-H vibration, the
lizardite spectrum is characterized by a broad
band at 3440 cm-1. This band, systematically ab­
sent in chrysotile and antigorite spectra, is due to
interlayer hydrogen bonds.
IR bands at 1072-1088 c m 1 have been at­
tributed to the Si-O stretching vibrations, perpen­
dicular to the basal plane (frequencies progres­
sively decreasing from lizardite to antigorite and
chrysotile). The weaker strength connected with
the increasing frequency can be tentatively ex­
plained as resulting from the increasing tetrahe-
dral sheet curvature. More significant differences
arise from Si-O vibrations in the basal plane:
lizardite and antigorite show only one strong
band (at 952 and 986 cm'1 respectively) whereas
chrysotile shows two bands, at 956 and 1020 cm 1 .
The weak band at 805 c m 1 in lizardite has
been often interpreted as arising from Al lv -O
stretching vibrations; however, from the relation

1.5 2 2.5 3 3.5 4

1 X
0.5 1
Al O Wt% 2πc MxMy
v Mx + My
Fig. 8. Fe0-Al203 plot for associated lizardite and chry­ where v = IR frequency, M - atom mass, c =
sotile in the Monte Fico veins. velocity of light, and / = bond strength constant,
 Vein lizardite and chrysotile
chrysotile
lizardite
Wavenumbers (cm1)
Fig. 9. Infrared spectra for associated lizardite and chrysotile (MFN3 specimen).
aluminum substitutions for silicon and for magne-
sium would produce frequency variations of
VAiiv-O/vsi-o = 1.007 and VAivi-o/VMg-o = 0.980,
respectively; accordingly, the 805 cm-1 band
cannot be attributed to Al-O, either in tetrahedral
or in octahedral coordination. At present, the
attribution of this lizardite band remains uncer-
tain.
The interpretation of infrared spectra at lower
frequencies (less than 750 cm 1 ) is more difficult,
because several effects overlap (for instance, Mg-
O stretching vibrations or different vibration
modes). In any case, lizardite shows two sharp
bands at 617 and 441 cm 1 ; in chrysotile, these
bands (608 and 437 cm-1) are complicated by the
presence of several shoulders. Similar bands,
with the same frequencies found in lizardite (618
and 445 cm-1), are also present in antigorite spec-
tra, but these latter, however, are characterized
by further bands at 568 and 400 cm 1 .
Discussion and conclusion
In this section, it is proposed that chrysotile
and lizardite, when properly analysed, reveal
chemical discrepancies as large as those found
with respect to antigorite. Proper analysis means
593
something more than merely accurate beam posi-
tioning or the absence of contamination from
microstructural biases; it also requires paired data
from associated minerals. We know (e.g., O'Han-
ley & Dyar, 1993) that "the limited cation sub-
stitution in natural lizardite is caused by the
limited range in bulk-rock composition of serpen-
tines rather than limits imposed on substitution
by the serpentine crystal structures". Therefore,
by performing a cumulative comparison of chry-
sotile and lizardite compositions from any given
occurrence in the literature, there is the risk of
masking the original crystalchemical differences
actually existing at the outcrop by adding some
unknown geochemical variability.
Otherwise, several authors (e.g., Trommsdorff
& Evans, 1972; Dungan, 1977) have suggested
the possible metastable relationship of chrysotile
with respect to lizardite. As a result of combina-
tion of crystallographic, energetic and kinetic
factors, the crystallization of metastable chry-
sotile with a poorly constrained composition may
be well expected.
A correct estimate of compositions is impor-
tant for understanding the real phase relationships
between the two minerals. In fact, in the case that
associated chrysotile and lizardite actually share
the same compositions, the two minerals may be
considered as real polymorphs, most probably
594 C. Viti, M. Mellini

with one of the two phases metastably persisting compositional field, but the AI2O3 content never
within the stability field of the other one. In fact, exceeds 2 wt%, independently of SiO2 content;
Whittaker & Wicks (1970) cited the apparently finally antigorite has a very low AI2O3 content,
overlapping compositions and equivalent struc- never higher than 1 wt%.
tural formulae as criteria for regarding lizardite Concerning iron, lizardite is also more exten-
and chrysotile as polymorphs. More recently, sively substituted than chrysotile which contains
O'Hanley et ai (1989) have used the same argu- very low amounts of iron; on the contrary, anti-
ment. However, demonstration of preferential gorite shows a wide compositional field,
partitioning of aluminum and - in general - of frequently at higher iron contents with respect to
trivalent cations into lizardite with a resultant lizardite.
compositional difference between the phases Furthermore, Fe 2+ /Fe 3+ determinations reveal
would preclude the possibility of polymorphic higher amounts of triple-charged cations in lizar-
phase transition (Chernosky, 1971). Furthermore, dite. The compositional data for lizardite are thus
the conclusive demonstration of a chemical dis- in good agreement with the structural and in-
tinction between associated chrysotile and lizar- frared evidence: both structural geometry and in-
dite would also help in several theoretical treat- frared absorptions indicate the presence of inter-
ments of serpentine minerals. For instance, layer hydrogen bonds, possibly stabilizing the
O'Hanley et ai (1989), in performing a chemo- lizardite flat layers.
graphic analysis of phase relations in MASH sys- Although antigorite may be iron enriched,
tem, were obliged to remove chemographic most of this element is bivalent. This composi-
degeneracy by imposing the condition that alum- tional pattern may be related to some still un-
inum partitions into lizardite rather than into
chrysotile. More recently, this conclusion was
also proposed by O'Hanley & Dyar (1993): o lizardite o chrysotile • antigorite
based upon Mossbauer data, they found that
"lizardite does have a stability field" and that "the
serpentine minerals are not polymorphs in
general, although lizardite, chrysotile and anti-
gorite may have similar compositions in specific
situations". Therefore, the two minerals should
not be strictly considered as polymorphs. Instead,
chrysotile corresponds to a metastable phase,
formed owing to kinetic effects. After extensive
nucleation of chrysotile, this would evolve to
polygonal serpentine first and later to lizardite,
according to a non-isochemical transformation.
From a general point of view, the crystallization
sequence chrysotile-polygonal serpentine-lizar-
dite may be regarded as an example of Ostwald's
ripening rule. Kinetically, chrysotile would be
the most favoured phase, progressively trans-
forming to the stable phase (lizardite) through
several reaction intermediates.
The data set for this argument is provided by
the Elba serpentines, particularly the present
chrysotile-polygonal serpentine-lizardite veins
plus, for comparison, the antigorite veins pre-
viously reported (Viti & Mellini, 1996).
Chemical differences between lizardite and
chrysotile mainly arise from the extent of
cationic substitutions, following the composi- Fig. 10. a) AhO3-SiO2 plot for lizardite, chrysotile and
tional patterns for aluminum and iron pointed out antigorite from the Elba serpentine veins, b) FeO-SiO2
in Fig. 10a and in Fig. 10b respectively. Lizardite plot for lizardite, chrysotile and antigorite from the Elba
is always Al-enriched; chrysotile occupies a wide serpentine veins.
 Vein lizardite and chrysotile
known details of layer corrugation in antigorite;
at tetrahedral inversions, octahedral sites may be
larger, thus favouring the substitution of Fe 2+
(ionic radius = i.r. = 0.77 Å) for Mg ( i.r. =
0.72 Å) with respect to Fe 3+ and Al (i.r. = 0.645
and 0.535 A, respectively; Shannon & Prewitt,
1969).
The present conclusion concerning chemical
compositions broadly fits "the average trends"
already pointed out by Whittaker & Wicks
(1970): "...antigorites have higher SiO2...chry-
sotile have MgO and SiO2 contents close to those
of ideal serpentine...substitution of iron and
aluminum in the lizardite structure appears mod-
erate to substantial". However, rather than an
average trend, the pattern now appears as a real
chemical difference.
Demonstration of this difference was favoured
by the availability of suitable specimens, with
well grown and separated minerals. We expect
that similar compositional distinctions will also
be found by performing fine-scale probe analyses
on carefully controlled rock-forming serpentines,
distinguished on the basis of adequate techniques
with a sufficient spatial resolution.
Acknowledgements: G N V - C N R (Gruppo
Nazionale Vulcanologia - Consiglio Nazionale
delle Ricerche) is acknowledged for installing
and servicing the electron microprobe used in
this work. The authors are indebted with Prof.
J. Zemann and an anonymous referee for careful
reviews of the manuscript.
References
Baronnet, A., Mellini, M., Devouard, B. (1994): Sec-
tors in polygonal serpentine. A model based on
dislocations. Phys. Chem. Minerals, 21, 330-343.
Brindley, G.W. & Zussman, J. (1959): Infrared absorp-
tion data for serpentine minerals. Am. Mineral, 44,
185-188.
Caruso, L.J. & Chernosky, J.V.Jr. (1979): The stability
of lizardite. Can. Mineral, 17, 757-769.
Chernosky, JV. Jr. (1971): Minerals of the serpentine
group. Canergie Inst. Wash. Year Book, 70, 153-
157.
Cressey, B.A. (1979): Electron microscopy of serpen-
tinite textures. Can. Mineral, 17, 741-756.
Cressey, B.A. & Zussman, J. (1976): Electron micro-
scopic studies of serpentinites. Can. Mineral, 14,
307-313.
595
Dungan, MA. (1977): Metastability in serpentine-
olivine equilibria. Am. Mineral, 62, 1018-1029.
Gregorkiewitz, M., Lebech, B., Mellini, M., Viti, C.
(1996): Hydrogen positions and thermal expansion
in lizardite-17 from Elba by Rietveld refinement of
neutron powder diffraction: a low-temperature
study. Am. Mineral, 81, 1111-1116.
Krstanovic, I. & Pavlovic, S. (1964): X-ray study of
chrysotile. Am. Mineral, 49, 1769-1771.
Mellini, M. (1982): The crystal structure of lizardite-
17: hydrogen bonds and polytypism. Am. Mineral,
67, 587-598.
Mellini, M. & Viti, C. (1994): Crystal structure of li-
zardite-17 from Elba, Italy. Am. Mineral, 79,
1194-1198.
Mellini, M. & Zanazzi, PF. (1987):Crystal structures of
lizardite-17 and lizardite 2H\ from Coli, Italy. Am.
Mineral, 72, 943-948.
Mellini, M., Trommsdorff, V., Compagnoni, R. (1987):
Antigorite polysomatism: behaviour during pro-
gressive metamorphism. Contrib. Mineral. Petrol,
97, 147-155.
O'Hanley, D.S. & Dyar, M.D. (1993): The composition
of lizardite 17 and the formation of magnetite in
serpentinites. Am. Mineral, 78, 391-404.
O'Hanley, D.S., Chernosky, J.V.Jr., Wicks, F.J. (1989):
The stability of lizardite and chrysotile. Can. Min-
eral, 27, 483-493.
Page, N.J. (1968): Chemical differences among the
serpentine "polymorphs". Am. Mineral, 53, 201-
215.
Serna, C.J., Velde, B.D., White, J.L. (1977): Infrared
evidences of order-disorder in amesites. Am. Min-
eral, 62, 296-303.
Serna, C.J., White, J.L., Velde, B.D. (1979): The effect
of aluminum on the infrared spectra of 7Å trioc-
tahedral minerals. Mineral Mag., 43, 141-148.
Shannon, R.D. & Prewitt, C.T. (1969): Effective ionic
radii in oxides and fluorides. Ada Cry St., B25, 925-
946.
Trommsdorff, V. & Evans, B.W. (1972): Progressive
metamorphism of antigorite schist in the Bergell
tonalite aureole (Italy). Amer. Jour. Science, 272,
423-437.
Uehara, S. & Kamata, K (1994): Antigorite with a large
supercell from Saganoseki, Oita Prefecture, Japan.
Can. Mineral, 32, 99-103.
Uehara, S. & Shirozu, H. (1985): Variations in chemi-
cal composition and structural properties of anti-
gorites. Mineral. Mag., 12, 299-318.
Velde, B.D. (1980): Ordering in synthetic aluminous
serpentine: infrared spectra and cell dimensions.
Phys. Chem. Minerals, 6, 209-220.
Viti, C. & Mellini, M. (1996): Vein antigorites from
Elba Island, Italy. Eur. J. Mineral, 8, 423-434.
Whittaker, E.J.W. & Zussman, J. (1956): The charac-
terization of serpentine minerals by X-ray diffrac-
tion. Mineral Mag., 31, 106-126.
596 C. Viti, M. Mellini
Whittaker, E.J.W. & Wicks, F J . (1970): Chemical
differences among the serpentine "polymorphs": a
discussion. Am. Mineral, 55, 1025-1047.
Wicks, F J . & O'Hanley, D.S. (1988): Serpentine min-
erals: structures and petrology, in "Hydrous phyl-
losilicates", Reviews in Mineralogy vol. 19, S.W.
Bailey Editor.
Wicks, F.J. & Zussman, J. (1975): Microbeam X-ray
diffraction patterns of the serpentine minerals. Can.
Mineral., 13, 244-258.
Wicks , F J . & Plant, A.G. (1979): Electron microprobe
and X-ray microbeam studies of serpentine tex-
tures. Can. Mineral, 17, 785-830.
Yariv, S. & Heller-Kallai, L. (1975): The relationship
between the IR spectra of serpentine and their
structures. Clays Clay Minerals, 23, 145-152.
Received 20 June 1996
Accepted 27 November 1996