162 F. Stoppa et al.
/ Gondwana Research 37 (2016) 152–171
alkaline intrusions are important, because in many similar intrusions,
late carbonatite dyke swarms are typically the most REE-enriched compo-
nents, even if not widely mined, and are typically associated with margin-
al clinopyroxenites that contain high REE-apatite and pyrochlore (Nb)
(Al Ani and Sarapää, 2013). It is suggested that advanced differentiation
processes are able to concentrate these elements, producing late-stage
fluorite-carbonate-barite-apatite assemblages that may be economically
significant due to their high content of fluorite, phosphates and LREE,
V-bearing phases and other critical metals. If these mineral systems are
the result of late-stage, large-volume carbonatites, their concentration of
LREE and V makes them promising for further and more economically-
oriented exploitation.
5. Tectonics and source of the magma
In the last 2 Myr, the Italian Tyrrhenian lithosphere has experienced
rapid thinning, decreasing from a thickness of N100 km to b60 km
(Lavecchia and Stoppa, 2013 and references therein). Ultra-alkaline
rocks occur in Middle-Late Pleistocene continental grabens in an approx-
imately 75 km-wide belt, and have been discontinuously traced for
350 km along the Apennines mountain range (Table 1 and Fig. 1a).
Carbothermal residua are located along the Tyrrhenian coast in an area
where crust and lithosphere are even thinner and there is a large volume
of explosive ultra-alkaline magma generation, producing caldera volca-
noes (Fig. 1a). There are two opposing geodynamic models: (1) litho-
spheric delamination driven by continental rifting; (2) back-arc basin
extension driven by various kinds of subduction/slab break-offs. There
are numerous papers on the ultrapotassic magmatism of the Roman
Volcanic Province that suggest a subduction geochemical imprinting on
mantle source (e.g. Beccaluva et al., 2002; Schiano et al., 2004; De Astis
et al., 2006; Peccerillo and Frezzotti, 2015). There is also an increasing
number of studies that directly or indirectly relate carbonatites with a
subduction environment (e.g. Goodenough et al., 2004). In the Italian
case, there is no direct involvement of carbonatites in a compressional
environment; carbonatites have never been found in back-arc basins
and are generally considered characteristic of intraplate or continental
rift tectonic positions. Mantle nodules do not have the high-pressure/
low-temperature mineral assemblage expected for subducted litho-
sphere, even if the magmatic source were located beneath the slab (e.g.
De Astis et al., 2006; D'Orazio et al., 2007). In fact, D'Orazio et al. (2007)
suggest a subduction environment for Vulture carbonatites by means of
a “window” in the slab. Stoppa et al. (2008) argued against their hypoth-
esis. Details of this discussion are addressed in Lavecchia et al. (2006);
many other studies support a rift-related origin for Italian carbonatites
(e.g. Lavecchia and Boncio, 2000; Lavecchia et al., 2002; Bell et al.,
2013). Whatever the geodynamic motor, rapid lithospheric thinning
causes adiabatic decompression of the asthenosphere, which results in
retrieving fluids from depth, a small degree of partial melting and the
release of CO2 -rich, alkaline magma towards the surface (Macera et al.,
2003; Bell et al., 2013; Lavecchia and Stoppa, 2013; Pirajno, 2015;
Goodenough et al., 2016; Lee et al., 2016). This geological context fits
well with the presence of large CO2 emissions in this area (e.g. Mefite
d'Ansanto, near the Vulture volcanic complex, Fig. 1a) and the ultra-
alkaline magmatism. The conditions of parental magmas during ascent
towards the surface have been estimated by thermometric experiments
with melt and fluid inclusions and by mantle nodules as proxy
geobarothermometers (Jones et al., 2000). The occurrence of associated
lherzolite xenoliths indicates that the primary magma detached from
the source at about 22 kbar (Jones et al., 2000). Minerals from this mantle
nodule suite contain a large number of carbonate veinlets and globules
enriched in BaO, SrO, Na2O and K2O and liquid/gaseous CO2 bubbles
(Rosatelli et al., 2007; Panina et al., 2013 and references therein). These
features testify to the presence of alkaline carbonate melt(s) in the litho-
spheric mantle coming from underlying asthenosphere (Rosatelli et al.,
2007). A mantle origin of carbonates is corroborated by Sr–Nd isotope ra-
tios, which are similar in the extrusive rocks and the peridotite nodules
dd
(Castorina et al., 2000). The presence of alkaline carbonates and inclusions
in veined mantle nodules suggests a metasomatic enrichment of the lith-
ospheric mantle in LILE, LREE, S, F by alkaline carbonatite infiltration
(Rosatelli et al., 2007; Stoppa et al., 2009). As a consequence, the low
degrees of partial melting magma formed from this mantle source are
strongly enriched in the same incompatible elements that have a partition
coefficient very favourable for their concentration in the melt formed. The
geochemical hallmark of such a process is the abundance of incompatible
elements and critical metals that are, in turn, frequently associated with
exploitable ore deposits (Woolley and Church, 2005; Woolley and
Kjarsgaard, 2008; Verplanck et al., 2014). These features suggest a mineral
systems scenario similar to thoset observed in a continental rift, such as
the Rhine Graben and other REE belts in Europe (Goodenough et al.,
2016).
6. Geochemistry
New analyses of Italian primary carbonatites together with associated
carbothermal residua rocks, and a few other relevant silicate rocks togeth-
er with analytical details, are given in Table 2. Rocks have been analysed
with multi-method (XRF, INAA, ICM-MS, Infrared, Titration and Ion
Electrode) chemical analyses and were carried out at Ancaster, Ontario,
Canada L9G 4 V5 (samples PIA03 and PIA06) and Institute of Geochemis-
try, SB RAS, Irkutsk, Russia (sample PO1a, PO1, PO5a, PO3, PO8, CU4, V106,
V106a, V108, V109, V107, V105, V103, V104, SV11B, SV1, OR1, OR2, OR5,
OR3). Detection limits are 0.01%, 0.05 and 0.1 ppm, for major oxides and
trace elements, respectively. The error of these analyses is b1% for
major elements, except Na, H2O and CO2, where the total error amounts
to ~3%; and for trace elements total error amounts to ~2% for concentra-
tion N 300 ppm, 5% for concentrations of 50–300 ppm and 10% for concen-
trations b 20 ppm.
Composition range and the average of selected analyses of extrusive
calcite–carbonatite (i.e. rocks having CaCO3 N 50% vol.) are given for
comparison in Table 2. Extrusive calcite-carbonatites show a very
wide compositional range. Italian extrusives show a limited variation
in comparison, but the Vulture carbonatites are very similar to the aver-
age extrusive calcite-carbonatites. In Italian primary carbonatites, silica
abundances range from 11.1–24.7 wt.% and Al2O3 ranges between
2.4–6.17% oxide (ox.), which is largely due to the presence of olivine,
amphibole, phlogopite, clinopyroxene xenocrysts, whereas high MgO,
up to 13.6 wt.%, is generally due to forsterite and/or diopside content.
CaO ranges between 27.5–42.1% ox., which is very similar to the range
of worldwide extrusive calcite-carbonatites. Alkalis are low and around
1% ox. with a slight potassic character. P2O5 is below 1% ox., but at
Vulture N2% ox. was found.
Carbothermal residua carbonatites are quite pure Ca–F carbonates
with CaO 56.1%, F 8.65%, SiO2 1.10% and 2.27 wt.% BaO, 0.87 S%, 0.26%
SrO and P2O5 0.39%. All the other oxides are negligible. Fluor-ore has
similar composition but inverse ratio of F/CO2 but much higher BaO
and REE and lower SrO2.
The Italian calcite and whole-rock carbonatite compositions are
shown in the (BaO + SrO) ∗ 50-ΣREE ∗ 100-CaO triangular diagram of
Fig. 5.
Mantle-normalised whole rock geochemistry shows negative spikes
for some HFSEs (e.g. Nb–Ta and Zr–Hf) and high LREE/HREE fraction-
ation (Fig. 6a-b). More evolved carbonatites with lower Mg# = (Mg/
Mg + Fe2+), lower compatible element contents (e.g. Vallone Toppo
del Lupo and Oricola), have higher LREE/HREE and greater HFSE4 +/
HFSE3 + fractionation (i.e. Nb + Ta/Zr + Hf). LREEs are up to
N2200 ppm in the Vallone Toppo del Lupo carbonatite (Stoppa et al.,
2009); HFSEs, having larger ionic radii, are less fractionated, but U and
Th values are particularly high at Oricola and Vallone Toppo del
Lupo. In addition, V is up to 500 ppm in the Vallone Toppo del Lupo
carbonatite. Fluor-ore concentrates LREE up to 2200 ppm and V up to
438, Pb up to 1010 and As 565 ppm.
 F. Stoppa et al. / Gondwana Research 37 (2016) 152–171 163

Fig. 5. (BaO + SrO) ∗ 50-ΣREE ∗ 100-CaO triangular diagram for calcite composition and whole carbonatites of Italy. Circles are carbonatite whole rock composition, while striped
areas represent the compositional variation of calcite analyses (Vallone Toppo del Lupo carbonatite by Stoppa (2007) and D'Orazio et al. (2007); Pianciano: this work and Locardi
(1990); calcite from Vulture sövite by Rosatelli et al. (2000); Cupaello calcite and carbonatite by Stoppa and Cundari (1995); Oricola and Polino calcite by Rosatelli et al. (2010);
Oricola and Polino
carbonatites plus Monticchio and San Venanzo calcites by Stoppa and Woolley (1997). Pianciano carbothermal residua show a strong concentration of Ba–Sr which are favoured
to enter the carbonate structure at lower temperature.

Average Italian carbonatites (primitive mantle-normalised; Sun and McDonough, 1989) have a very similar pattern compared to average
worldwide extrusive carbonatites, with a slightly smaller abundance of HFSE, but higher uranium (Fig. 6a).
The average Italian carbonatite chondrite-normalised REE pattern is very similar to the average worldwide extrusive calcite-carbonatites,
but with a smaller LREE/HREE ratio. Lacn/Lucn is 104 and 140 and respec- tively, but with similar HREE abundance (Fig. 6b).
After plotting HFSEs against a differentiation index, e.g. Mg#, it be- comes clear that some carbonatites are more differentiated than others
and concentrate Mn, V, LREE, Nb, in the final stages of melt evolution (Fig. 7a-i). Critical metal concentration, V, Nb and LREE in mafic
silicate rocks of the Vulture are V ~ 200 ppm (average of 138 samples) and up to 376 ppm; Nb is ~ 100 ppm (average of 138 samples) and up to
443 ppm; LREE 725 ppm (average of 75 samples), and up to about 1500 ppm. Th and U are 39 and 11 ppm (138 samples) respectively and up
to 119 ppm and 44 ppm in samples VST25 and V111, respectively. Howev- er, the highest amount of V, Nb, LREEs are from
ijolitic/clinopyroxenitic rocks, which derive from the intrusive complex beneath the volcano (Principe, 2006; Stoppa et al., 2006). The
concentration of LREEs, Ba, Sr, V and Nb establishes a firm link with Italian primary carbonatites and kamafugites. However, some specialty
elements, in particular V content, are characteristic of the Vulture primary carbonatite, as well as Pianciano carbothermal residua and,
coupled with high LREEs, pro- mote the crystallisation of Ce–V-bearing minerals. The trace-element distribution of carbothermal residua is
very similar to that of primary Italian carbonatites, whereas the carbonatitic samples of Pianciano are lower, but show a significant pathway to
the Vulture intrusive calcite carbonatites (Fig. 6a-b). The carbonatites from Mount Weld, Australia (Hoatson et al., 2011) and from Barra do
Itapirapuã from Brazil

(Andrade et al., 1999), have a very similar distribution of LILE and HFSE and lower REE (Fig. 6a-b).
Italian carbonatites have C, O isotope ratios ranging from typical mantle values to enriched heavy isotope values (Stoppa, 2003). Both
primary and carbothermal carbonatites are in the field of primary carbonates in extrusive carbonatites worldwide (Fig. 4). Radiogenic isotopes
are in equilibrium between carbonatites, associated alkaline rocks and mantle nodules (Castorina et al., 2000).

7. Discussion

There is a divergence regarding the origin of CaCO3 in the rocks of the Roman Volcanic Province. According to some workers (e.g. Peccerillo
et al., 2010; see also Lentz, 1999) the CaCO3 modal content is derived from the reaction between alkaline fluids and limestones, whereas ac-
cording to others this is a primary carbonatitic component (Bailey, 2005; Bell and Kjarsgaard, 2006; Rosatelli et al., 2010). We consider the
second hypothesis to be more likely within the general framework of Italian ultra-alkaline magmatism (Bell et al., 2013) and because of the
LREE + Ba + Sr abundance of these fluids. In addition, there is strong ev- idence that the Italian lithospheric mantle was infiltrated by
alkaline carbonatitic melts, resulting in complex metasomatic mantle assemblages (Rosatelli et al., 2007). Fluid inclusion studies indicate that
primitive parental magma was enriched in CaO and CO2, LREE, Zr, Sr, Ba, S, and F and that carbonatite-silicate immiscibility occurred over a
wide range of T/P conditions (Panina et al., 2003; Solovova et al., 2005). Evidence of silicate–carbonate immiscibility in Italian ultra-alkaline
rock is represent- ed by carbonate amoeboid blebs and ocelli, carbonate bubbles in melilitite glass, and is also documented by geochemistry
(Rosatelli et al., 2010). Immiscibility is considered the most effective evolutionary process in

Fig. 6. (a) Mantle normalised (Sun and McDonough, 1989) LILE, HFSE and compatible
elements distribution in Italian extrusive carbonatites compared with Pianciano carbonatite
and fluor-ore plus the intrusive carbonatite of the Mount Weld, eastern Goldfields
Province, western Australia (Nelson et al., 1988), Barra do Itapirapuã carbonatite, southern
Brazil (Andrade et al., 1999; Comin-Chiaramonti et al., 2005) and, WEC = generalised field
for extrusive carbonatites compiled by 19 selected samples of extrusive calcite–carbonatite
s.s. from Toro–Ankole and Katwe–Kikorongo, Uganda; Shombole volcano, Kenya;
Kaisersthul, Germany; Cerro Sapo, San Cristobal and Salta, Bolivia; Pechenga, Kola-Russia
(http://georoc.mpch-mainz.gwdg.de/georoc/Csv_Downloads/Rocks_comp/CARBONATITE.
csv). (b) C1-chondrite normalised (Sun and McDonough, 1989) REE of the same samples of
Fig. 6a.
the alkaline mafic/carbonatitic melts, with open system conditions and
variable volatile ratios explaining the complex mineralogy of the ultra-
alkaline rocks association. The complex rock mineralogy results from pa-
rental mantle magma crystallisation with strong variations in volatiles
content. In addition, variation in the solubility of U–Th, Zr–Hf, Ta–Nb, Ti,
V, Sc, As, Co in the alkaline/peralkaline liquids explains the extreme
HFSE changes observed in some Italian evolved carbonatites and
melilitolites (Chakhmouradian, 2006). The formation of sövite/melilitolite
relates to slow differentiation and de-carbonation under plutonic condi-
tions. The carbonate melt coexisting with the alkaline silicate melt should
contain substantial amounts of alkalies and sodium-calcium minerals
(i.e. nyerereite and other alkali carbonates that have been found as inclu-
sions in melilite and apatite). However, considering the rarity of alkali
carbonates in the IUP carbonatites, we postulate that alkalies could be
consumed at crustal pressures (fenitisation?) or be dissolved in aqueous
fluid separating during melt evolution, in addition to the elements that
are usually easily mobilised in these fluids (viz. Ba, LREE, V, Sc, As, Co).
The role of volatiles, especially fluorine, in subvolcanic conditions and
late activity in ultra-alkaline volcanoes both in and around the intrusive
complexes is the key (Bambi et al., 2012). We argue that fluids from
deep magmatic carbonatite that became enriched in H2O, CO2, S, F and
SiF4 as dominant volatiles in carbothermal residua may form at shallow
levels at submagmatic temperature. However hydrothermal colloidal so-
lutions of CaF and SiO2 containing minor amounts of H2S, CaCO3, BaCO3
and Ca2PO5, could precipitate the fluorites of the Pianciano deposits at
lower temperature. The problems of this type of complex evolution are
similar to those of the genetic process that leads to the formation of
various types of pegmatite (London, 2008). Some kind of parallelism
exists between the two magmatic systems as they have the same ability
to concentrate/select critical metals. However, so far no one has investi-
gated this aspect. Concerning an evaluation of potential prospectivity, it
is possible to consider the relationship between REE and Nb worldwide
economic deposits, tonnages and ore grades (Fig. 8a-c). Due to increasing
demand, it is likely that sub-economic deposits will undergo further eval-
uation, possibly leading to near-future exploitation. As a consequence, the
economic size and position of a given deposit, ore grade and tonnage, will
change in the near future. The Pianciano REE deposit is well positioned in
terms of European resources, but at present it is still in the sub-economic
field. There is no estimation about its grade due to very fine-grained ore
and comminution of V and REE minerals with lean fluorite ore. On the
other hand, wakefieldite (Ce–V) separation, or Ce, V extraction would
produce a beneficiation of the fluorite ore, which may pay all the costs
of extraction and processing. However, very fine-grained wakefieldite
would not be expected to beneficiate well with current methods. At the
moment, Nb exploration is quite cautious and remains as a low priority
for potential producers. It is likely that this field will expand below the
ore grade of 0.1% oxides. Fig. 8c shows that a limited number of producers
and resources are up to the economical limit for both REE and Nb. Nb and
the REE content of Vulture carbonatites and ejecta are interesting and
near the economic-subeconomic limits for worldwide ore grade deposits
(Fig. 8c). Volcanic carbonatite deposits are small, but a detailed study of
ejecta could give an overview of the depth and ore grade of subvolcanic
bodies for potentially large volumes, as inferred by the abundance of
ejecta and the size of the volcano. However, geological and economic
constraints on the mineralisation cannot be stated until such time as
adequate information is obtained, in order to improve the advantages of
geophysical data, core drilling and 3D computer modelling considerably.
7.1. Potential mineral exploration targets in Italian carbonatites
Aspects of carbonatites worldwide that are typically considered as
key exploration targets are: (1) the regolith–laterite or soil caps,
where weathering of primary carbonatite minerals produces supergene
concentrations of REE and Nb; (2) late-stage carbonatite dykes and hy-
drothermal roof zones where carbonate fluor-ore precipitates below
magmatic temperature (pegmatitic to hydrothermal condition) togeth-
er with carbothermal residua enriched in REE; (3) primary intrusive
carbonatite that may contain exploitable apatite, magnetite, pyrochlore,
baddeleyte; (4) fenites (i.e. alkali feldspars plus aegirine metasomatic
felsic rocks) are not an exploited resource, but their presence indicates
alkaline fluid movement from the igneous complex towards country
rocks, which can carry and concentrate critical metals elsewhere; (5) ex-
trusive carbonatites are usually not an exploitable resource, but their
occurrence in nephelinitic volcanoes may indicate the presence of an
intrusive carbonatite, clinopyroxenite–ijolite–syenite subvolcanic
complexes (e.g. Napak volcano in Uganda), which are the main target
of critical metal exploration. The challenge is that the majority of new
deposits of raw materials, such as LREE, Nb, Zr, V–Sc, Co, As and other
metals could be found in future at greater depths below these volca-
noes. This suggests that exploration indicators at the surface are crucial.
This is certainly the case for alkaline rocks and carbonatites in Italy,