Mineral-Melt Equilibria and Geothermobarometry of Campi Flegrei Magmas: Inferences for Magma Storage Conditions

Abstract

The eruptions of Campi Flegrei (Southern Italy), one of the most studied and dangerous active volcanic areas of the world, are fed by mildly potassic alkaline magmas, from shoshonite to trachyte and phonotrachyte. Petrological investigations carried out in past decades on Campi Flegrei rocks provide crucial information for understanding differentiation processes in its magmatic system. However, the compositional features of rocks are a palimpsest of many processes acting over timescales of 10⁰–10⁴ years, including crystal entrapment from multiple reservoirs with different magmatic histories. In this work, olivine, clinopyroxene and feldspar crystals from volcanic rocks related to the entire period of Campi Flegrei’s volcanic activity are checked for equilibrium with combined and possibly more rigorous tests than those commonly used in previous works (e.g., Fe–Mg exchange between either olivine or clinopyroxene and melt), with the aim of obtaining more robust geothermobarometric estimations for the magmas these products represent. We applied several combinations of equilibrium tests and geothermometric and geobarometric methods to a suite of rocks and related minerals spanning the period from ~59 ka to 1538 A.D. and compared the obtained results with the inferred magma storage conditions estimated in previous works through different methods. This mineral-chemistry investigation suggests that two prevalent sets of T–P (temperature–pressure) conditions, here referred to as “magmatic environments”, characterized the magma storage over the entire period of Campi Flegrei activity investigated here. These magmatic environments are ascribable to either mafic or differentiated magmas, stationing in deep and shallow reservoirs, respectively, which interacted frequently, mostly during the last 12 ka of activity. In fact, open-system magmatic processes (mixing/mingling, crustal contamination, CO₂ flushing) hypothesized to have occurred before several Campi Flegrei eruptions could have removed earlier-grown crystals from their equilibrium melts. Moreover, our new results indicate that, in the case of complex systems such as Campi Flegrei’s, in which different pre-eruptive processes can modify the equilibrium composition of the crystals, one single geothermobarometric method offers little chance to constrain the magma storage conditions. Conversely, combined methods yield more robust results in agreement with estimates obtained in previous independent studies based on both petrological and geophysical methods.

1. Introduction

The chemical composition of magmatic minerals, combined with the determination of intensive variables (temperature, pressure) recorded by crystals at equilibrium, allows investigation of the chemicophysical conditions of magmatic systems during storage and prior to eruption (e.g., [1]). The compositional variations in magmatic minerals can be useful to track chemical and physical changes in the magma from which they grew [2,3,4,5,6,7,8,9,10,11]. Using experimentally determined phase relationships, e.g., compositions of minerals in equilibrium with a melt as a function of pressure, temperature, water content and oxygen fugacity, the chemical composition of mineral assemblages allows reconstruction of the crystallization conditions. Determination of pre-eruptive conditions in magma plumbing systems through equilibrium relationships between melts and coexisting minerals is one of the major targets of modern petrology. In this context, geothermobarometry allows estimation of crystallization pressure and temperature by applying calibrated equations to the compositions of minerals, matrix glass, melt inclusions and whole rocks, or a combination thereof. The fundamental premise of geothermobarometry is that the mineral assemblage and compositions of a rock are sensitive to pressure and temperature of formation and that the events subsequent to mineral equilibration have not significantly modified these properties [12]. The theoretical basis for most geothermobarometers consists of determining the equilibrium constant for a reaction and estimating the conditions of equilibration based on that value [12,13]. Many geothermometers and geobarometers have been calibrated through experimental data and thermodynamic models. These models are mostly based on exchange reactions between minerals and melt [13,14,15,16], cation content in minerals and coexisting melt [17,18], pressure-dependent variations in crystal lattice structure [19,20,21] or phase relations of a set of minerals [22,23]. Since these models are calibrated for certain mineral phases and restricted magma compositional ranges, the application of a suitable geothermobarometric method depends on the chemical composition of the analyzed rocks and their minerals. For mafic magmas, for example, ‘OPAM’ (olivine–plagioclase–augite melt) geobarometry uses the composition of a melt in equilibrium with plagioclase, clinopyroxene and olivine to estimate a pressure of ‘last equilibration’ [22,24]. Some models make use of clinopyroxene components, such as geobarometers based on equations describing the pressure-dependent jadeite-liquid (Jd-liq) reaction, as well as geothermometers based on equations considering the temperature-dependent Jd into diopside–hedenbergite (Ca(Mg,Fe)Si₂O₆; Di-Hd) and calcium Tschermak’s components (CaAlAlSiO₆; Ca-Ts) into Di-Hd exchange reactions [13,14,15]. Temperature can be determined from coexisting alkali feldspar and plagioclase [13,23,25] and from clino- and orthopyroxene solid solutions [26]. Most of the commonly used geothermometers are based on the modeling of entropy and volume changes occurring in equilibrium reactions between melts and crystals [27]. Clinopyroxene-melt (e.g., [15]) and plagioclase-melt (e.g., [28]) geothermobarometers are well-known and commonly used by petrologists. Based on the composition of the volcanic rocks of interest, when applicable, multiple geothermobarometric models can be employed to provide independent tests. Clinopyroxene-melt geothermobarometry has been used for specific active Italian volcanoes, for example, to define magma storage conditions under Mt. Etna [29]. Likewise, Masotta et al. [30], with the aim of improving the suitability of clinopyroxene-liquid geothermobarometers for evolved alkaline systems, developed and applied recalibrated geothermobarometric equations to Mt. Vesuvius and Campi Flegrei products.

In this work, the geothermobarometers of Putirka [13] and Masotta et al. [30] have been used to estimate pressures and temperatures of crystallization of olivine, clinopyroxene and feldspar crystals from volcanic products belonging to different periods of Campi Flegrei activity. In the last decades, various studies that used different geological, geochemical and/or geophysical information have tried to estimate the depths of magma storage below Campi Flegrei (e.g., [31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50]). Moreover, geothermometric estimates have been performed in several studies (e.g., [51,52,53,54,55,56,57,58]) in order to reconstruct the pre-eruptive temperature conditions of the magmas feeding different eruptions (e.g., Campanian Ignimbrite, Agnano–Monte Spina, Astroni). The temperature values obtained in these studies show largely variable ranges even for rocks of similar composition; moreover, apart from a few studies [30,58], geobarometric estimates are scarce.

The Campi Flegrei volcanic field represents an interesting case study where different geothermobarometers can be applied in order to: (i) estimate the storage depths at which the magmas stationed before single eruptive events or a cycle of events; (ii) reconstruct periods of shift in magma storage conditions; and (iii) unravel the evolution through time of its magmatic feeding system. All these outcomes would have a deep impact on the understanding of the behavior of the magmatic plumbing system that feeds Campi Flegrei, one of the highest-risk active volcanic areas on Earth for its high population density.

2. The Campi Flegrei Volcanic Field

The Campi Flegrei volcanic field, together with the Somma–Vesuvius stratovolcano and Ischia volcanic island, is one of the three active volcanoes of the Neapolitan area. Its morphology is dominated by a 12 km-wide caldera depression (Figure 1), resulting from multiple collapses related to two highly explosive volcanic eruptions: the Campanian Ignimbrite (CI) occurring c. 40 ka [59,60] and the Neapolitan Yellow Tuff (NYT) occurring c. 15 ka [61].

The volcanic activity has been dominantly explosive through time and has mostly led to the emplacement of pyroclastic rocks with subordinate lava flows and domes. The age of the beginning of volcanism in the area is not well-constrained. Old ignimbrites, even if highly altered, at Durazzano (116.1 ka), Moschiano (184.7 ka), Seiano Valley (245.9 and 289.6 ka) and Taurano–Acqua Feconia (157.4, 183.8, 205.6 and 210.4 ka) localities, in different sites of the Campania Plain, have been characterized by De Vivo et al. [63], Rolandi et al. [64] and Belkin et al. [65]. At least twelve pre-CI units are recognized at the Trefola Quarry [66] spanning the 59–39 ka period [67]. The CI, emplaced during a catastrophic explosive event having a magnitude of 7.2 [68], is considered the most powerful eruption ever to occur in the Neapolitan area. The CI Plinian eruption emplaced a large volume (~80 to 300 km³ dense rock equivalent; [68,69,70,71]) of pyroclastic fall and pyroclastic-density-current products, which resulted in a very complex sequence in proximal, medial and distal outcrops [69,72,73,74,75,76,77,78], as well as a co-ignimbrite ash fall dispersed in the eastern Mediterranean Sea and eastern Europe (e.g., [75,79,80]). Post-CI/pre-NYT volcanic activity was confined inside the CI caldera and the majority of the eruptions were produced by explosive, mostly hydromagmatic activity [66]. ⁴⁰Ar/³⁹Ar-dated eruptions range in age from 30.3 ka to 14.6 ka [67]. Moreover, Albert et al. [81] recently revealed the high magnitude (M 6.6 or VEI 6) of one of the pre-NYT deposits: the Masseria del Monte Tuff, dated at 29.3 ± 0.7 ka. The NYT (14.9 ka; [61]) was produced during the last largest eruption of the Campi Flegrei caldera, and it is by far the largest trachytic phreatoplinian deposit known to date. The NYT outcrops in scattered localities over an area of ~1000 km², with a conservatively estimated volume of ~40 km³ (DRE; [82,83,84]).

During the last 15 ka, intense volcanism and deformation affected the caldera. This period has been divided into three epochs of activity: epoch 1—c. 15–10.6 ka; epoch 2—c. 9.6–9.1 ka; and epoch 3—c. 5.5–3.5 ka [85]. During epoch 1, at least 32 magmatic-to-phreatomagmatic explosive eruptions took place, with a mean frequency of one eruption every 70 years [86]. Among these, the Minopoli 1, Soccavo 4, Minopoli 2, Fondo Riccio and Montagna Spaccata eruptions have been subject of recent studies (e.g., [87,88,89]). Close to the end of this phase, the Plinian deposit of the Pomici Principali eruption was emitted from the Agnano area (11.9–12.1 ka; [85]). During epoch 2, six low-magnitude explosive eruptions took place with a mean frequency of 65 years [86,90]. During epoch 3, an intense monogenetic explosive and subordinate effusive activity took place; the principal events that modified the morphological setting of the volcanic field were the emplacements of Solfatara (4.1–4.3 ka; [85]), Astroni (4.1–3.8 ka; [85,86,87,88,89,90,91]), Averno (5.4–4.1 ka; [85,92,93]), Nisida (4.1–3.2 ka; [85,94]) and Monte Nuovo (1538 AD; [48,95,96,97]) pyroclastic cones, the Agnano Monte Spina (4.4–4.6 ka; [85,98,99,100,101,102]) Plinian magmatic-to-phreatomagmatic eruption and Monte Olibano and Accademia lava domes (4.36 ± 1.13 ka; [103]). The last event took place in 1538 AD, with the formation of Monte Nuovo scoria cone ([48] and reference therein).

The intense fumarolic activity and the ground deformation episodes that occurred in recent decades (e.g., [104]) testify to the persistent activity of the Campi Flegrei magmatic system that remains in a state of unrest (e.g., [105,106,107]). The presence of 350,000 inhabitants in the central part of the caldera raises the risk level to very high [108,109].

The Campi Flegrei volcanic field erupted alkaline potassic rocks ranging in composition from trachybasalt to phonotrachyte, with a predominance of trachyte (Figure 2; [67,110,111]).

Most of the products older than 15 ka, except for a few rocks that were erupted during the pre-NYT period (Torregaveta and Masseria del Monte Tuff), exhibit the most evolved compositions (trachytes and phonotrachytes; Figure 2). At the end of the first epoch of the last 15 ka of activity, less-differentiated magmas (trachybasalt and latite) were erupted along NE–SW regional tectonic structures (e.g., [66,110]).

3. Materials and Methods

New compositional data on olivine, clinopyroxene, plagioclase and K-feldspar phenocrysts from products representative of some pre-CI eruptive units and of the last 5 ka’s Agnano–Monte Spina and Astroni eruptions have been collected (Supplementary Materials Table S1). Major- and minor-element (Si, Ti, Al, Fe, Mn, Mg, Ca, Na and K) contents of phenocrysts were acquired at the HP-HT Laboratory of Experimental Volcanology and Geophysics of the Istituto Nazionale di Geofisica e Vulcanologia in Rome (Italy), using a Jeol-JXA8200 electron microprobe equipped with five wavelength-dispersive spectrometers. Crystals in carbon-coated resin mounts were analyzed under high-vacuum conditions, using an accelerating voltage of 15 kV, with a beam diameter of 5 μm. The electron-beam current was set at 7.5 nA. Elemental counting times were 10 s on the peak and 5 s on each of two background positions. Corrections for interelemental effects were made using a ZAF (Z: atomic number; A: absorption; F: fluorescence) routine. The range of standards for calibration was taken from Micro-Analysis Consultants (MAC) and variable diffraction devices: albite (Si-PET, Al-TAP, Na-TAP), forsterite (Mg-TAP), augite (Fe-LIF), apatite (Ca-PET), orthoclase (K-PET), rutile (Ti-PET) and rhodonite (Mn-LIF). Accuracy was better than 1–5% except for elements with abundances below 1 wt%, for which it was better than 5–10%. Precision was typically better than 1–5% for all analyzed elements.

For the aim of this work, the newly acquired data were integrated with a suitably created database of the chemical composition of mineral phases from different periods of Campi Flegrei’s volcanic activity. We retrieved major and minor element contents of olivine, clinopyroxene, plagioclase and alkali-feldspar phenocrysts and microphenocrysts from previous works [51,52,53,55,56,57,65,69,72,83,87,88,89,94,95,98,101,113,114,115,116,117,118,119,120,121,122] in which these mineral phases were analyzed. When available, the whole-rock, matrix-glass and melt-inclusion compositions have been also included in the database. These were used as representative of melt compositions for the evaluation of equilibria and for the application of geothermobarometers.

The collected data belong to different periods of activity of the Campi Flegrei volcanism, that have been divided into: old ignimbrites—c. 290–115 ka [65]; pre-CI—c. 59–47 ka [67,103]; CI—c. 40 ka [59,60]; pre–NYT—c. 30–16 ka [67,103]; NYT—c. 15 ka [61]. As for the recent period of activity (last 15 ka), since there are no data on volcanic products erupted during the 15–13 ka time interval, we consider the last 12 ka, starting with the Pomici Principali eruption (12.1–11.9 ka; [85]).

4. Results

4.1. Mineral Chemistry

4.1.1. Olivine

Olivine crystals mostly occur in the products belonging to the last 12 ka of volcanic activity. In particular, this mineral phase is found in volcanic rocks emplaced during the Minopoli 1, Pomici Principali, Minopoli 2, Fondo Riccio and Astroni eruptions. Their MgO and FeO contents range from 50 to 43 wt% and from 19 to 10 wt%, respectively. Their CaO contents range between 0.5 and 0.2 wt%. The olivine phenocrysts range in composition from Fo₉₀ to Fo₈₀ (Forsterite mol %; Figure 3), showing a limited compositional variation.

The Forsterite (mol %) contents of olivine show a bimodal distribution (Figure 3), characterized by (1) the most frequent compositional population ranging between Fo₈₇ and Fo₈₄ and (2) a less frequent compositional population ranging between Fo₉₀ and Fo₈₈. On the other hand, olivine microlites occurring in the groundmass of several lava domes (Cuma; 45.9 ±3.6 ka; Wu et al. [103]; Punta Marmolite; 47 ka; Accademia; 3.9 ka; Melluso et al. [111] and references therein) show a wider compositional range (Fo_90–1) with respect to those of the phenocrysts belonging to explosive volcanic activity, studied in this work. The olivines belonging to such relatively scarce Campi Flegrei effusive products also show fayalite- and tephroite-rich compositions [111].

4.1.2. Clinopyroxene

Clinopyroxene occurs in all volcanic products belonging to the Campi Flegrei eruptions, and mostly classifies as diopside and Fe-rich diopside (Wo_52–41En_51–29Fs_25–4; Wollastonite–Enstatite–Ferrosilite; Figure 4a–c). Those belonging to rocks erupted after the NYT eruption show compositions also richer in Fe (Wo_53–44En_50–14Fs_32–4), e.g., hedenbergite.

The Mg# [molar Mg²⁺/(Mg²⁺ + Fe_tot) × 100] ranges from 92 to 40 (Figure 4d). This parameter reflects a compositional polymodality in the clinopyroxene phenocrysts of volcanic rocks belonging to all periods of activity. Clinopyroxenes of rocks belonging to the different periods of activity show variable range of Mg# (

Table 1. Ranges of Mg# values of clinopyroxene crystals belonging to different periods of Campi Flegrei activity.

Eruptive Period	Clinopyroxene Mg#
Old ignimbrites	89–67
Pre–CI	91–61
CI	92–62
Pre–NYT	91–55
NYT	92–61
Last 12 ka	92–41

), in particular those from rocks emplaced during the last 12 ka cover the whole range of values (Mg# = 92–41; Figure 5).

The correlations between Mg# and Al, Ti, Na, Mn and Cr contents show some changes/kinks along the whole continuous compositional variation. Moreover, the clinopyroxenes of rocks erupted during the last 12 ka show chemical variation trends different to those of clinopyroxenes of rocks erupted during previous periods (Figure 5). These features are described in detail below.

TiO₂ and Al₂O₃ contents of clinopyroxenes range from 2.90 to 0.10 wt% and from 9.62 to 0.73 wt%, respectively (Figure 5a,b). Na₂O and MnO contents range from 1.36 to 0.03 wt% and from 3.23 to 0.01 wt% (Figure 5c,d). Cr₂O₃ content ranges from 1.00 to 0.01 wt% (Figure 5e). Na₂O, TiO₂ and MnO contents increase with the decrease in Mg#. In particular, the Na₂O and MnO contents increase linearly with the decrease in Mg# from 92 to 80 and then increase exponentially with the decrease in Mg# from 80 to 55 (Figure 5c,d). An exception are some clinopyroxenes belonging to rocks of the last 12 ka, in which the Na₂O and MnO contents are on average lower and continue to increase linearly with respect to the whole Mg# range. Cr₂O₃ and SiO₂ contents decrease as Mg# decreases (Figure 5e,f). The Cr₂O₃ contents decrease exponentially with the decrease in Mg# from 92 to 80 and then decrease linearly as Mg# decreases from 80 to 41. The Al₂O₃ contents increase with the decrease in Mg# from 92 to 80, then decrease with the decrease in Mg#, except for the crystals of volcanic products belonging to the last 12 ka, in which the Al₂O₃ contents continue to increase as Mg# decreases (Figure 5a).

Similarly to olivine, the clinopyroxenes occurring in the groundmass of the Campi Flegrei lavas show wide compositional variations (Mg# = 89–1), covering the complete spectrum from Mg- to Fe-rich compositions and reaching Na- and Zr-rich compositions (e.g., aegirine; [111]).

4.1.3. Feldspars

Feldspar is a ubiquitous phase in the volcanic products of Campi Flegrei. K-feldspar is the most abundant phase in the pyroclastic rocks, belonging to all periods of activity, mostly in trachytes. Plagioclase mostly occurs in volcanic rocks representative of poorly evolved magmas. The plagioclase and alkali-feldspar crystals belonging to the different periods of activity cover wide compositional ranges, except for plagioclases from the pre-NYT and NYT periods (Figure 6a–c;

Table 2. Plagioclase and K-feldspar composition of products emplaced during different Campi Flegrei periods of activity. An = anorthite; Ab = albite; Or = orthoclase.

Eruptive Period	Plagioclase Composition	K-Feldspar Composition
Old ignimbrites	An88–49Ab44–10Or6–1	Or74–50
Pre-CI	An82–24Ab64–16Or11–2	Or67–54
CI	An90–25Ab62–8Or14–1	Or88–42
Pre-NYT	An89–76Ab20–9Or3–1;	Or87–51
NYT	An86–47Ab43–12Or8–2	Or86–72
Last 12 ka	An94–40Ab51–5Or18–1	Or87–39

).

The Campi Flegrei plagioclase shows a bimodal distribution, reflected in the An (mol %) content, characterized by a main compositional population exhibiting An_90–63 and a second population exhibiting An_62–40 (Figure 6d). Likewise, the frequency histogram of sanidine composition (Or mol. %; Figure 6e) shows two main compositional populations: the most frequent is in the range Or_88–68, whereas a second population is characterized by composition in the range Or_67–40.

4.2. Mineral-Melt Equilibrium

The equilibria between melt and selected minerals were investigated using (i) the Fe-Mg exchange coefficient for olivine and clinopyroxene, hereafter referred to as equilibrium test 1; (ii) the comparison between observed and predicted normative diopside–hedenbergite (DiHd) components for clinopyroxene, hereafter referred to as equilibrium test 2; (iii) the Or-Ab partitioning coefficient for alkali feldspar, hereafter referred to as equilibrium test 3; and (iv) the An-Ab partitioning coefficient for plagioclase, hereafter referred to as equilibrium test 4. For the evaluation of the equilibrium, Campi Flegrei melt-inclusion, whole-rock and matrix-glass analyses were used as representative of the composition of various melts.

4.2.1. The “Classic” Method for Assessing Equilibrium between Olivine or Clinopyroxene and Their Melt: The ^Fe-MgKd_min-liq Exchange Coefficient (Test 1)

Equilibrium test 1 is the most commonly used test for assessing equilibrium between clinopyroxene or olivine and melt pairs. It consists of the evaluation of the Fe-Mg partitioning between mineral and liquid, known as the Fe-Mg exchange coefficient, defined as ^Fe-MgKd_min-liq = (MgO_liqFeO_min)/(MgO_minFeO_liq), where “liq” is the liquid; “min” is the mineral; and MgO and FeO are molar fractions (e.g., Roeder and Emslie, 1970; Putirka, 2005, 2008). In this case, the equilibrium conditions are verified when the Fe-Mg partitioning between olivine and host rock (^Fe/MgKd_Ol-liq) is 0.30 ± 0.03 [123,124], and when the Fe-Mg partitioning between clinopyroxene and host rock (^Fe/MgKd_Cpx-liq) is 0.27 ± 0.03 (e.g., [15,125]). In a Rhodes diagram (Mg#_mineral vs. Mg#_melt), the lines joining points satisfying these conditions define the theoretical equilibrium field (Figure 7).

Concerning the Campi Flegrei olivines, only a few olivines from Fondo Riccio and Minopoli 2 are in the equilibrium field (Figure 7a). As for the clinopyroxenes, a few crystals from rocks belonging to different periods of activity are in equilibrium with their melts (Figure 7b–i).

4.2.2. An Alternative Equilibrium Test for Clinopyroxene: The Measured versus Predicted Components (Test 2)

An alternative and more robust test for assessing the equilibrium between a mineral and coexisting melt is what we call equilibrium test 2, which consists of the comparison between normative mineral components predicted for a mineral phase from melt composition, and those measured in the analyzed crystals (e.g., [13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50,51,52,53,54,55,56,57,58,59,60,61,62,63,64,65,66,67,68,69,70,71,72,73,74,75,76,77,78,79,80,81,82,83,84,85,86,87,88,89,90,91,92,93,94,95,96,97,98,99,100,101,102,103,104,105,106,107,108,109,110,111,112,113,114,115,116,117,118,119,120,121,122,123,124,125,126]). Since all the studied clinopyroxenes are diopsidic in composition, we take into account the Di and Hd components. These were calculated following the scheme proposed by Putirka et al. [14] and Putirka [13]. The predicted clinopyroxene components based on melt composition were calculated using equation 3.1a in Putirka [127]: ln [DiHd^cpx] = −9.8 + 0.24ln [Ca^liq (MgO + FeO^liq)(Si^liq)²] + 17,558/T + 8.7ln(1670/T) − 4.61 × 10³ [(EnFs^cpx)²/T].

Only one crystal from the old ignimbrites (Taurano ignimbrite) results in equilibrium (Figure 8a); nevertheless, we discarded these products since they were altered and, therefore, the rock composition was almost surely affected by intense weathering processes [65]. A few clinopyroxenes from all deposits belonging to the pre-CI period are in equilibrium with their melts (Figure 8b). Some crystals from the CI also are in the equilibrium field (Figure 8c). For the pre-NYT period, some clinopyroxenes result in equilibrium conditions (Figure 8d). Several crystals from the NYT are also in the equilibrium field (Figure 8e). For the last 12 ka, some crystals from Minopoli 1, Minopoli 2, Soccavo 4, Averno and Monte Nuovo eruptions and a few crystals from Pomici Principali, Fondo Riccio, Capo Miseno, Agnano–Monte Spina, Astroni and Nisida eruptions are in equilibrium with their melts (Figure 8f–h).

4.2.3. Equilibrium Tests for Alkali-Feldspar and Plagioclase (Tests 3 and 4)

For K-feldspar, equilibrium test 3 used here is based on Or–Ab partitioning between K-feldspar and melt (e.g., [13,128]), known as the Or-Ab exchange coefficient, defined as ^K-feld-meltKd_Or-Ab = Na_Feld × XAl_liq × XCa_liq/XCa_Feld × XNa_liq × XSi_liq vs. predicted ^K-feld-meltKd_Or-Ab = −0.67 + (K_liq/K_Feld² + ln (exp (K_liq²/Na_liq + K_liq)/10, according to the approach by Mollo et al. [128]. In a measured vs. predicted Kd_Or-Ab diagram, the equilibrium field is shown inside the 1:1 line ± 0.25 (Figure 9).

We do not consider the equilibrium relationships between K-feldspars and host rocks from the old ignimbrites that are altered (Figure 9a). For the pre-CI period, crystals from the S. Severino 1 and S. Severino 2 eruptions are in the equilibrium field, whereas those from the Torre di Franco eruption are not (Figure 9b). Several crystals from the CI eruption are in equilibrium with their melts (Figure 9c). All K-feldspars erupted in the pre-NYT period, except those from Belvedere Miliscola 1 and some from Trentaremi eruptions, are in the equilibrium field (Figure 9d). All K-feldspars from the NYT eruption are in equilibrium with their melts (Figure 9e). For the products erupted during the last 12 ka, apart from some crystals from the Montagna Spaccata, Astroni and Monte Nuovo eruptions, all the K-feldspars are in equilibrium conditions (Figure 9f–h).

The equilibrium conditions for plagioclase were tested via equilibrium test 4, which is based on the partitioning of An-Ab between mineral and melt. This is known as An-Ab exchange coefficient defined as ^pl-meltKd_Ab-An = (XNa_Plag × XAl_liq × XCa_liq)/(XCa _Plag × XNa_liq × XSi_liq) where “liq” is the liquid composition, “Plag” is the plagioclase composition, and all components are expressed as molar fractions (e.g., [13]). The variation diagram of An (mol %) vs. calculated ^pl-meltKd_Ab-An shows the plagioclase-melt stability field drawn using a value for ^pl-meltKd_Ab-An of 0.1 ± 0.05 (Figure 10; [13]).

Even in this case, we do not take into account the plagioclase-melt equilibria for the old ignimbrites (Figure 10a). For the pre-CI period, all crystals from the S. Severino 1 and S. Severino 2 eruptions and a few crystals from other pre-CI units are in equilibrium with their melts (Figure 10b). Several plagioclases from the CI are also in the equilibrium field (Figure 10c). For pre-NYT units, all crystals belonging to the Torregaveta eruption are in equilibrium, whereas those from Trentaremi eruption are not (Figure 10d). Most of plagioclases in the range An_85–50 from the NYT eruption are in equilibrium (Figure 10e). For the products of the last 12 ka, some plagioclase crystals from Minopoli 1, Minopoli 2, Pomici Principali, Montagna Spaccata, S. Martino, Agnano–Monte Spina, Averno, Astroni and Nisida eruptions result in equilibrium with their host melts (Figure 10f–h).

4.3. Geothermobarometric Estimates

We applied various geothermometers and geobarometers to all the mineral-melt couples which passed the equilibrium tests based on the ^Fe-MgKd_min-liq (olivine and clinopyroxene) and on the comparison between measured and predicted components (clinopyroxene and feldspar). Some geothermobarometers require the H₂O (wt%) content as an input parameter, which can affect the output T–P estimates. Hence, in cases where melt-inclusion compositions have been used as representative of melts in equilibrium with crystals, the volatile contents used are exactly those detected in melt inclusions of the erupted products; in cases where the whole-rock or matrix-glass compositions have been used as representative of melts, the H₂O analyzed in the whole rock or the water content obtained by difference has been used.

As shown before (Section 4.2), a few olivine crystals passed the equilibrium test. On these, we applied Equation (4) [T = (15,294.6 + 1318.8P + 2.4834 (P)²)/(8.048 + 2.8352 ln D^ol-liq_Mg + 2.047 ln 1.5 (X^Mg_liq + X^Fe2+_liq + X^Ca_liq + X^Mn_liq) + 2.575 ln (3X^Si_liq) − 1.41 7/2ln(1 − X^Al_liq) + 7ln(1 − X^Ti_liq) + 0.222 H₂O_liq + 0.5 P] of the olivine-melt geothermometer of Putirka et al. [129], commonly used for hydrous melts, and whose standard error of estimate (SEE) is 29 °C (e.g., [13]). Two olivine-melt pairs from Minopoli 1 yield temperatures of 1065 and 1060 °C (Figure 11a). The output temperature estimates for the Minopoli 2 olivines range between 1122 and 1045 °C, with an average of 1102 ± 20 °C. The Fondo Riccio olivines in equilibrium with their melts yield temperature estimates in the range of 1175–922 °C, with an average of 1005 ± 84 °C.

For clinopyroxenes resulting in equilibrium with their melts via equilibrium test 1, we used the geothermobarometers of Putirka [13] and Masotta et al. [30]. Specifically, we applied the Equation (33) [10⁴/T = 7.53 − 014 ln (X^cpx_JdX^liq_CaOX^liq_FeO+MgO/X^cpx_DiHdX^liq_NaX^liq_Al) + 0.07 (H₂O^liq) − 14.9 (X^liq_CaOX^liq_SiO2) − 0.08ln (X^liq _TiO2) − 3.62 ln (X^liq_NaO0.5/X^liq_KO0.5) − 1.1 (Mg#^iq) − 0.18ln (X^cpx_EnFs) − 0.027P] of Putirka et al. [13], which is a clinopyroxene-liquid geothermometer based on Jd-DiHd exchange and has been implemented with respect to the previous models of Putirka [14,15] through a larger experimental dataset and an increased number of regression parameters; the SEE for this geothermometer is 31.4 °C. For pressure estimates, we used Equation (32c) [P = −57.9 + 0.0745 T − 40.6(X^liq_FeO) − 47.7(X^cpx_CaTs) + 0.0676 (H₂O^liq) − 153(X^liq_CaO0.5X^liq_SiO2) + 6.89(X^cpx_Al/X^liq_AlO1.5)] of Putirka et al. [13] which represents a geobarometer based on the partitioning of Al between clinopyroxene and liquid and whose SEE is 2.9 kbar. The estimated temperatures and pressures (Figure 11b) are reported in

Table 3. Temperature and pressure output obtained from the Putirka [13] geothermobarometers applied to clinopyroxene-melt couples whose equilibrium has been verified with equilibrium test 1.

Eruptive Period	Melt Composition	T (°C)	Average, s.d.	P (kbar)	Average, s.d.
Pre-CI	trachyte-phono-trachyte	994–874	946 ± 34	10.6–3.4	6.2 ± 2
CI	trachyte-phono-trachyte	1015–894	953 ± 28	10–1.8	6.9 ± 1.6
Pre-NYT	latite-trachyte	1043–891	964 ± 42	12.5–3.8	8 ± 2.2
NYT	trachyte-phono-trachyte	1018–914	983 ± 23	9.2–3.3	6.9 ± 1.5
Last 12 ka	shoshonite-latite-	1110–884	976 ± 46	14.6–2.9	7.6 ± 2.1
	trachyte-phono-trachyte

.

The geothermometer of Masotta et al. [30] was applied by using the equations Talk2012 [10⁴/T = 2.1 − 0.4 ln (X^cpx_JdX^liq_CaX^liq_FeO+MgO/X^cpx_DiHdX^liq_NaX^liq_Al) + 0.038 (H₂O) − 1.64 (X^liq_Mg/X^liq_Mg + X^liq_Fe/X^cpx_DiHd) + 1.01 (X^liq_Na/X^liq_Na + X^liq_K) − 0.22 ln (X^liq_Ti) + 0.47 ln (X^cpx_Jd/X^liq_NaX^liq_Al(X^cpx_Si)²) + 1.62 (K_D^liq-cpx_Fe-Mg) + 23.39 (X^liq_CaX^liq_Si)] and Palk2012 [−3.89 + 0.38 (X^cpx_Jd/X^liq_NaX^liq_Al(X^cpx_Si)²) + 0.074 ((H₂O) + 5.01 (X^liq_Na/X^liq_Na + X^liq) + 6.39 (K_D^liq-cpx_Fe-Mg)], for estimating temperatures and pressures, respectively, which exhibit SEE lower than those of previous models (SEE_Talk2010 = 18.2 °C and SEE_Palk2012 = 1.15 kbar).

For each eruption, we plotted the T–P estimates obtained through this geothermobarometer specific for alkaline magmas (Figure 11c–e). The geothermobarometers of Masotta et al. [30] combine 10 clinopyroxene compositions with 10 melt compositions; in this case, we calculated P-T estimates only for a numerically appropriate set of data, i.e., only for those eruptions whose products have at least 10 mineral-melt couples in equilibrium. For example, for the pre-CI period, we applied geothermobarometry only to the Torre di Franco products (Figure 11c). Moreover, this geothermobarometer was used only for evolved compositions, since—as stated by Masotta et al. [30]—any attempt to use it on compositions different from those of their calibration dataset would produce high errors in estimation. The estimated temperatures and pressures are reported in

Table 4. Temperature and pressure output obtained from the Masotta et al. [30] geothermobarometers applied to clinopyroxene-melt couples whose equilibrium has been verified with equilibrium test 1.

Eruptive Period	Eruption	Melt Composition	T (°C)	Average, s.d.	P (kbar)	Average, s.d.
Pre-CI	Torre di Franco	trachyte	994–958	978 ± 8	2.6–0.4	1.7 ± 0.6
CI	CI	trachyte-phono-trachyte	1055–921	992 ± 22	4.0–0.1	1.3 ± 0.8
Pre-NYT	Trentaremi	trachyte	1081–968	1025 ± 24	5.3–0.3	1.8 ± 1.14
NYT	NYT	trachyte-phono-trachyte	1058–973	1023 ± 22	1.9–0.1	1.0 ± 0.4
Last 12 ka	Averno	trachyte-phono-trachyte	1014–933	960 ± 16	2.8–0.5	1.6 ± 0.5
	Nisida	trachyte-phono-trachyte	1027–961	994 ± 15	3.2–0.1	1.3 ± 0.8

.

We also applied the same two geothermobarometers (Putirka, [13] and Masotta et al., [30]) to the clinopyroxene-melt couples that resulted in equilibrium through equilibrium test 2. When using Equations (33) and (32c) of the Putirka [13] geothermobarometer, a few clinopyroxene-melt couples yield negative values of the output pressures, which have not been taken into account. The estimated temperatures and pressures (Figure 11f) are reported in

Table 5. Temperature and pressure output obtained from the Putirka [13] geothermobarometers applied to clinopyroxene-melt couples whose equilibrium has been verified with the equilibrium test 2.

Eruptive Period	Rock Composition	T (°C)	Average, s.d.	P (kbar)	Average, s.d.
Pre-CI	trachyte-phono-trachyte	918–823	946 ± 34	7.5–0.1	2.4 ± 2
CI	trachyte-phono-trachyte	886–811	848 ± 19	5.3–0.1	2.0 ± 1.4
Pre-NYT	latite-trachyte	996–830	892 ± 38	5.4–0.2	2.8 ± 2.3
NYT	trachyte-phono-trachyte	961–819	892 ± 42	4.8–0.1	2.2 ± 1.2
Last 12 ka	shoshonite-latite-	1179–817	946 ± 92	8.5–0.1	1.9 ± 1.5
	trachyte-phono-trachyte

.

When using the Talk2012 and Palk2012 equations of the Masotta et al. [30] geothermobarometers, most of the obtained pressures are negative, hence these values must be considered meaningless. In fact, the overall estimated pressures range from 4.9 to −1.6 kbar, with most of the values being <0 kbar (Figure 11g). The estimated temperatures and pressures are reported in

Table 6. Temperature and pressure output obtained from the Masotta et al. [30] geothermobarometers applied to clinopyroxene-melt couples whose equilibrium has been verified with the equilibrium test 2.

Eruptive Period	Rock Composition	T (°C)	Average-s.d.	P (kbar)
Pre-CI	trachyte-phono-trachyte	1055–889	945 ± 43	1.9–(−1.6)
CI	trachyte-phono-trachyte	963–900	934 ± 16	1.6–(−0.8)
Pre-NYT	trachyte	1022–875	937 ± 47	1.6–(−0.9)
NYT	trachyte-phono-trachyte	1105–961	1031 ± 31	1.5–(−1.3)
Last 12 ka	trachyte-phono-trachyte	1125–875	953 ± 70	4.9–(−1.3)

.

Lastly, Equation (27b) [T = (−442 − 3.72P)/(−0.11 + 0.1ln(X^K-feld_Ab/X^pl_Ab) − 3.27(X^K-feld_An) + 0.098(X^K-feld_An) + 0.52(X^pl_An X^pl_Ab))] of the two-feldspar geothermometer of Putirka [13] was used on alkali feldspars and plagioclases resulting in equilibrium through tests 3 and 4, respectively. The estimated temperatures (Figure 11h,i) are reported in

Table 7. Temperature output obtained from the two-feldspar geothermometer applied to plagioclase- and K-feldspar-melt couples resulted in equilibrium.

Eruptive Period	Eruption	Melt Composition	T (°C)	Average, s.d.
Old ignimbrites	Taurano	-	1193–1022	1071 ± 79
Pre-CI	S.Severino 1	trachyte	879–833	855 ± 8
	S.Severino 2	trachyte	880–828	853 ± 20
CI	CI	trachyte-phono-trachyte	1066–713	829 ± 70
Pre-NYT	Torregaveta	latite	1064–881	944 ± 45
NYT	NYT	trachyte-phono-trachyte	992–790	883 ± 30
Last 12 ka	Minopoli 1	shoshonite-latite	1093–959	1035 ± 43
	Pomici Principali	shoshonite-latite-trachyte	959–826	870 ± 33
	Soccavo 4	trachyte	928–842	871 ± 37
	Minopoli 2	shoshonite-trachyte	1047–952	1007 ± 27
	Montagna Spaccata	latite-trachyte	1068–807	871 ± 67
	S.Martino	trachyte	920–839	864 ± 14
	A-MS	trachyte-phono-trachyte	928–812	857 ± 23
	Paleoastroni 3	trachyte	964–905	930 ± 16
	Averno	trachyte-phono-trachyte	923–772	858 ± 30
	Astroni	latite-trachyte	1032–817	879 ± 29
	Nisida	trachyte-phono-trachyte	923–738	862 ± 21

.

5. Discussion

5.1. Magmatic Environments Reconstructed Based on Campi Flegrei Mineral Compositions

Mineral compositions strongly depend on pre-eruptive (i.e., P-T) conditions in magma (e.g., [10,11,130]). Hence, crystals can preserve information about the set of parameters (pressure, temperature, oxygen fugacity and volatile content) of the environment where they formed. The chemical composition of the Campi Flegrei olivines, clinopyroxenes and feldspars show a polymodal distribution. This is evident in the forsterite (mol %) contents of olivines (Figure 3), in the Mg# of clinopyroxenes (Figure 4d), in the orthoclase (mol %) contents of alkali feldspars (Figure 6e) and in the anorthite (mol %) contents of plagioclases (Figure 6d). The occurrence of two main compositional populations detected in the Campi Flegrei minerals suggests two prevalent “magmatic environments” in which crystals have grown. A “magmatic environment” does not necessarily represent a physical environment and can be defined as a specific set of intensive thermodynamic variables (e.g., [131]) which determines the composition of a mineral. Among these two prevalent compositional populations, one, pertaining to olivines with Fo_90–88, clinopyroxenes with Mg# in the range 90–78, plagioclases with high anorthite content and K–feldspars with low orthoclase content, can be ascribed to mafic or poorly differentiated magmas; whereas the other, pertaining to clinopyroxenes with Mg# in the range 77–40 and K-feldspars with Or_88–68, can be associated with evolved magmas. The latter is the most abundant component and represents the typical magmatic environment in which minerals found in the Campi Flegrei trachytes and phonotrachytes form. The compositional bimodality occurs in the olivine, clinopyroxene and feldspar crystals from Campi Flegrei rocks emplaced over all the periods of activity, without significant differences. This suggests that through time, magmas formed reservoirs located at two different, barely constant depths in the Campi Flegrei plumbing system, where they stagnated/equilibrated. All the studied olivine, clinopyroxene and feldspar crystals show similar ranges of chemical variation over the different periods of activity. For clinopyroxene, the element variations (Figure 4) define continuous compositional trends that are consistent with fractional crystallization processes, responsible for most of the detected compositional variations. The element variation diagrams show a deviation in the correlation trends between Mg# and other elements, specifically when Mg# decreases below the 80 threshold. These deviations are likely due to the beginning of crystallization of other mineral phases that determines the different partition of elements into different phases. As an example, when feldspar begins to crystallize together with the pre-existing clinopyroxene, the Al content of the melt is preferentially distributed in the feldspar. Furthermore, the Al₂O₃, TiO₂, Na₂O, Cr₂O₃ and MnO contents of clinopyroxenes from rocks belonging to the last 12 ka show different trends with respect to those of clinopyroxene from rocks of previous periods (Figure 4). In addition to the concomitance of crystallization of various phases, other processes able to determine the decoupling of elements should be considered. With respect to the previous periods, during the last 12 ka mafic magmas of deeper origin more frequently reached the upper crustal reservoirs. In fact, it is known that Mg-olivine-bearing (Section 4.1.1) mafic magmas (trachybasalt and latite) were erupted at Campi Flegrei only during the last period of activity, through vents located along a NE–SW regional fault system that probably tapped the deeper least-evolved reservoir (e.g., [66,110,120,132]). Diffusional modifications due to mixing between less and more evolved magmas can explain the observed different trends in the chemistry of minerals. The interaction between mafic and pre-existing evolved magmas or between melts and crystal mushes [94,133] would also explain part of the mineralogical disequilibria observed in the Campi Flegrei minerals (Figure 7, Figure 8, Figure 9 and Figure 10). Open-system magmatic processes such as mixing/mingling (e.g., [94,98,101,110,134]), crustal contamination (e.g., [135]) and CO₂ flushing [88,101] have been hypothesized for the Campi Flegrei magmas, especially during the last 12 ka of activity. Such pre-eruptive processes could have decoupled earlier-grown crystals from their equilibrium melts. These findings are supported by the equilibrium tests themselves. In particular, in some of these (e.g., Figure 10), the linear and parabolic trends are possibly due to the occurrence of closed- and open-system processes, respectively. In fact, similarly, in addition to providing information on the possible occurrence of these processes, the ^Fe-MgKd_min-liq equilibrium test is also useful for the identification of xenocrysts and/or antecrysts, late crystallization, crystal removal and closed-system crystallization (e.g., [13,136]).

5.2. Reliability of the Equilibrium Tests for the Campi Flegrei Minerals

For clinopyroxene, two different approaches (equilibrium test 1 and equilibrium test 2) were used here to test equilibrium between minerals and their melts. By applying equilibrium test 1 based on the Fe-Mg exchange coefficient to all the Campi Flegrei clinopyroxenes and screening all the mineral-melt pairs resulting in equilibrium (Figure 12a), we observe that almost all these couples are not in equilibrium, if they are tested by comparing the measured and predicted components (equilibrium test 2; Figure 12b).

Vice-versa, by applying this last method to the entire clinopyroxene-melt dataset and taking into account only clinopyroxene-melt couples in the 1:1 ± 0.05 line, it is possible to observe again a misfit with the method based on the ^Fe-MgKd_cpx-liq (Figure 12c,d).

The mismatch between the two tests can be explained in terms of the different parameters used in these methods. It is well-known that the equilibrium test based on the Fe–Mg exchange coefficient has certain limitations, and alone it is not sufficient to testify equilibrium crystallization: the ^Fe-MgKd_min-liq takes into account the behavior of the Fe-Mg exchange without considering that of minor elements, i.e., Al, Ca, Na, Ti, that can exhibit a wide range of variation in clinopyroxene (e.g., [13,126,137]). Putirka [13] showed that ^Fe-MgKd_cpx-liq is not a suitable tool for assessing equilibrium for a wide range of compositions. Similarly, the ^Fe-MgKd_ol-liq is sensitive to melt composition and hence fixed values for such coefficients cannot be applied (e.g., [138]). Therefore, an alternative and more suitable test for assessing equilibrium at the time of crystallization is based on the deviation of mineral components measured in a mineral such as clinopyroxene (Di, Hd, En, Fs, CaTs, CaTiTs, CaCrTs and Jd) from those predicted from the melt composition (e.g., [127]). Using this model, Putirka [13] and Mollo et al. [126,137] suggested that the difference between predicted and observed components provides a more robust test for equilibrium with respect to the method which takes into account the ^Fe-MgKd_min-liq.

In this regard, equilibrium test 2 represents a more adequate tool for the evaluation of equilibrium conditions. Nevertheless, the obtained pressure estimates on clinopyroxene-melt pairs resulting in equilibrium through this method cast some doubt on its reliability. In fact, for clinopyroxene-melt pairs whose equilibrium conditions have been verified by equilibrium test 2, negative values of the pressure estimates (Table 6) are obtained by applying the Equation (32c) of Putirka [13] and the Palk2012 of Masotta et al. [30]. This latter geobarometer has been chosen because it is specific for alkaline magmas and, in theory, more suitable for Campi Flegrei rocks; moreover, Equation (33) of Putirka [13] overestimates the temperature below 850 °C [30]. Still, a great number of negative pressure values have been obtained with the use of the Palk2012 equation. Such results raise doubts on the appropriateness of using this test for equilibrium coupled with the geothermobarometers of Masotta et al. [30]. On the other hand, both geobarometers, when applied to clinopyroxene-melt pairs in which the equilibrium has been verified through the ^Fe-MgKd_cpx-liq, do not yield negative pressure estimates.

5.3. Reliability of the Temperatures Estimated for the Campi Flegrei Minerals

A reliable magma temperature estimate goes beyond the information on how hot that magma was. The determination of temperature is critical, for example, in the application of diffusion modeling for timescale estimates. This allows for the retrieving of information on the residence times of crystals in subvolcanic plumbing systems with an important impact on the assessment of the current state of an active volcano during unrest (e.g., [57,139,140,141,142,143]). In particular, the obtained duration is exponentially dependent on the temperature value, since the latter is the crucial parameter used to calculate the diffusion coefficient (e.g., [144]).

For the temperature estimated through the geothermometers used in this work, regardless of the combination of equilibrium tests and geothermometer usage, the mineral-melt couples of the last 12 ka rocks always yield the widest temperature ranges. This is probably due to the occurrence of rocks with poorly differentiated compositions besides those with evolved compositions, unlike previous periods (Figure 2).

In order to evaluate the robustness of the geothermometer–equilibrium test combinations, we compare the temperature output values obtained here with those estimated in previous studies. Forni et al. [55] estimated a pre-eruptive temperature for the CI in the range 1070–879 °C through the clinopyroxene-liquid geothermometers of Masotta et al. [30]. In previous works, the pre-eruptive temperatures of the CI magma were estimated in the range 980–800 °C, based on various methods: (i) comparing whole-rock compositions with the Nepheline (Ne)–Kalsilite (Ks)–Quartz (Qz) system [51,145]; (ii) two-feldspar geothermometry [51,52,53]; and (iii) homogenization temperatures of melt and fluid inclusions in clinopyroxene and K-feldspar [53]. The variable results obtained by different studies highlight the necessity to find a method allowing precise estimates of the temperatures for the Campi Flegrei minerals. Those obtained here for the CI are various, depending on the geothermometer used, the equilibrium test and combination thereof: the clinopyroxene-melt geothermometer of Putirka [13] based on equilibrium test 1 yielded temperatures in the range of 1015–894 °C; the geothermometer of Putirka [13] based on equilibrium test 2 yielded temperatures in the range of 886–811 °C; the geothermometer of Masotta et al. [30] based on equilibrium test 2 yielded temperatures in the range of 963–900 °C; the geothermometer of Masotta et al. [30] based on equilibrium test 1 yielded temperatures in the range 1055–921 °C; the two-feldspar geothermometry yielded temperatures in the range of 1076–713 °C. Compared to the previous estimates, the application of the clinopyroxene-melt geothermometers to mineral-melt couples whose equilibrium has been verified through the equilibrium test 2 narrows the temperature ranges. On the other hand, the two-feldspar geothermometer yields a wide range of temperatures. Nevertheless, it should be kept in mind that the equilibrium temperature of feldspars can be lower with respect to that of clinopyroxene, in the crystallization sequence. Regardless, the temperature ranges obtained with clinopyroxene-melt geothermometers based on equilibrium test 1 in this work are similar to those obtained in previous studies.

In Orsi et al. [83], at P of 1 kbar, the ternary-feldspar geothermometer [146] applied to the NYT magmas gave temperature estimates of 838–746 °C. More recently, Forni et al. [89] estimated a pre-eruptive temperature for the NYT magmas in the range of 1095–910 °C through the clinopyroxene-liquid geothermometers of Masotta et al. [30]. Even in this case, our results are different, depending on the used methods and combination thereof: the clinopyroxene-melt geothermometer of Putirka [13] based on equilibrium test 1 yielded temperatures in the range 1018–914 °C; the geothermometer of Putirka [13] based on equilibrium test 2 yielded temperatures in the range of 961–819 °C; the geothermometer of Masotta et al. [30] based on equilibrium test 2 yielded temperatures in the range 1105–961 °C; the geothermometer of Masotta et al. [30] based on equilibrium test 1 yielded temperatures in the range of 1058–973 °C; the two-feldspar geothermometer yielded temperatures in the range of 992–790 °C. Hence, even in this case, the temperature ranges that most match the literature estimates are those obtained by geothermometers applied to mineral-melt couples whose equilibrium has been verified through equilibrium test 1.

For the Agnano–Monte Spina eruption trachytes, temperatures are constrained by both experimental petrology [54] and two-feldspar geothermometry [57] between 973 and 870 °C. In this work, no Agnano–Monte Spina clinopyroxene-melt couple passed equilibrium test 2. Hence, our estimates for pre-eruptive temperatures come from the clinopyroxene-melt geothermometer [13] applied to clinopyroxene-melt couples whose equilibrium has been verified with the ^Fe-MgKd_cpx-liq and from the two-feldspar geothermometer: the temperatures estimated with these geothermometers are in the ranges of 946–893 °C and 928–812 °C, respectively. The former estimates are in agreement with those of previous works, whereas the latter only partly fit. However, our results are more robust because we used a greater amount of data with respect to those of previous studies, in which the equilibrium conditions were not verified before applying geothermometry.

Astbury et al. [58] applied the Masotta et al. [30] geothermobarometers to clinopyroxene crystals of the Astroni eruption and obtained temperatures in the range of 980–960 °C and pressures in the range of 2.2–0.2 kbar. They also applied two-feldspar geothermometry (Equation (27); [13]) to feldspar rims that gave temperatures ranging from 970 to 800 °C. Here, we obtained T–P estimates for the Astroni eruption through the Putirka [13] clinopyroxene-melt geothermobarometers. The latter method, when applied to mineral-melt couples which passed equilibrium test 1, yielded temperatures in the range of 981–884 °C and pressures in the range of 13.4–5.8 kbar, while when applied to mineral-melt couples resulted in equilibrium via test 2 yielded temperatures in the range of 919–830 °C and pressures in the range of 1.4–0.7 kbar. The estimated T partially match the values obtained by Astbury et al. [58]. Nevertheless, the pressures obtained with the Putirka [13] geobarometer applied to mineral-melt couples that resulted in equilibrium via test 1 are unrealistically high (13.4–5.8 kbar).

D’Oriano et al. [116], Piochi et al. [95] and Arzilli et al. [47], based on phase relations and geothermometry, estimated that in the magma(s) feeding the Monte Nuovo eruption, phenocrysts formed at equilibrium temperatures of ~890–800 °C, whereas the microlite equilibrium temperatures were even higher (~1100–900 °C). Since no clinopyroxene-melt couple of Monte Nuovo rocks passed equilibrium test 1, our temperature estimates come from the two clinopyroxene-melt geothermometers applied to mineral-melt couples whose equilibrium has been verified through test 2. In particular, the Putirka [13] geothermometer yielded temperatures in the range of 899–834 °C and the Masotta et al. [30] geothermometer yielded temperatures in the range of 911–895 °C. Since our estimates are based on phenocrysts only, we can observe a good fit of the temperature values obtained through the Putirka [13] geothermometer with those of the literature.

5.4. Reliability of Pressures Estimated for Campi Flegrei Minerals

Calculated pressures together with bedrock density allow for the estimation of depths of crystallization, and hence, potential magma storage depth. Determining magma storage depths is essential for various reasons. For example, understanding the distribution of magma storage depths is important for linking magmatic processes to the expressions of ongoing unrest such as seismicity, ground deformation and gas emission, especially in volcanically active areas [147,148,149], as well as to provide information about mechanisms of crustal formation [150,151,152].

Regarding pressure estimates, in order to evaluate the reliability of the different equilibrium test–geobarometer couples used here, we compare our results with those obtained with different methods used in literature (e.g., geophysical investigations, melt inclusions, phase equilibria) to infer the storage depth of Campi Flegrei magmas. In recent decades, geochemical and geophysical investigations allowed for the assessment that the Campi Flegrei plumbing system is characterized by deep and shallow reservoirs, [31,32,33,34,36,37,38,39,40,41,42,43,44,45,47,49,87,88,95,101,121,132,135,153,154,155,156,157,158,159,160,161,162]. However, although the detachment of magma at a depth ≥ 8 km is widely accepted, there is no consensus about the structure of the plumbing system at shallower levels. In fact, some authors hypothesized the presence of permanent small magma chambers feeding the last 5000 years of activity, including the last event, at ~4–1 km of depth (Monte Nuovo eruption; [48,58,163]), while other authors suggested the development of an ephemeral localized storage zone during the magma ascent, where magma shortly resides until erupting or cooling (the so-called failed eruptions, e.g., [97]).

Regarding the large-volume CI eruption, Fabbrizio and Carroll [35] estimated that the magma reservoir was stored between 5 and 8 km below the surface, through experimental constraints on phase relations. Using thermodynamic modeling, Bohrson et al. [164] revealed that in the magmas feeding the CI eruption, major element variations were dominated by crystal–liquid separation at pressures corresponding to ∼7–5 km. Melt inclusions in crystals from CI analyzed by Marianelli et al. [117] yield a relatively wide range of pressure for the magmatic storage and degassing located at 2–6 km depth. Similarly, the phase-equilibria calculations performed by Fowler et al. [118] on CI glass and mineral compositions indicate isobaric fractionation at 0.15 GPa (∼6 km depth). Pappalardo et al. [72], using geochemical and textural analyses, suggested that a wide sill-like trachytic magma chamber was active under the Campanian Plain at ca 2.5 kbar before CI eruption. Fanara et al. [46], through a combination of natural and experimental data, proved that the CI magma could have been stored or ponded during its rising path at two different levels: a deeper one corresponding to a depth of about 8 to 15 km and a shallower one at about 1 to 8 km. Moretti et al. [165], through melt-inclusion-based studies of gas-melt saturation, pointed out that the huge volume of magma that extruded during the CI eruption differentiated and mixed at ∼6–3 km.

Hence, several studies provided different estimates of the storage conditions for the Campi Flegrei magmas that are also very variable for the same case study (e.g., the CI eruption). In most of these studies, the main inferred estimates range between ∼12 and ∼4 km, corresponding to a pressure range of ∼3–1 kbar, using a crustal density of 2.6 g/cm³ according to Berrino et al. [166]. This estimate is not precise and there is no possibility to really discriminate among two or more main magma storage depths, considering the SEE of the geobarometers. However, this pressure range seems reasonable since the anomalous layer with the shape of a wide sill extending below the Campi Flegrei area at ∼11–7 km depth identified in different geophysical studies (e.g., [31,32,33,34,36,37,38]) is generally accepted as the main storage region of Campi Flegrei magmas. By considering the wide range of pressures (~14.7–1 kbar) obtained with the Putirka [13] geobarometer based on equilibrium test 1, we cannot consider it as a good method for estimating the crystallization pressures of Campi Flegrei magmas. In fact, the most frequent pressure value obtained with this method is ∼7.5 kbar (Figure 13a), which corresponds to a depth of ∼29 km, by using a crustal density of 2.6 g/cm³, which is quite unreasonable for evolved magmas.

Equation (32c) of Putirka [13] applied to clinopyroxene-melt couples resulting in equilibrium with test 2 yields estimates that are more in agreement with those obtained in the literature. In this case, the pressures range from ∼5.6 to 0.1 kbar, with two outsider values at 8.4 and 7.5 kbar (Figure 13c). The pressure values show a bimodal distribution characterized by two main peaks at 1.5 kbar and 4.1 kbar. These values correspond to depths of ∼6 km and 16 km, by using a crustal density of 2.6 g/cm³. The first value is in agreement with depths of shallow reservoirs (≤8 km) frequently estimated in previous studies. The second value could be attributed to region of magma storage below 8 km depth. Nevertheless, it should keep in mind that precise depth estimates can be under- or over-estimated for the uncertainty due to the lack of reliable data on the average density of rocks at various depths (km).

5.5. Reliability of the Different Geothermobarometers Based on Different Equilibrium Tests for the Campi Flegrei Minerals

Based on the comparison with T–P estimates published in the literature, we can evaluate which combination of geothermobarometer–equilibrium test can be most suitable for the Campi Flegrei magmas. For temperatures, the two-feldspar geothermometer yields values commonly slightly lower compared to those obtained in the literature through independent studies. Overall, the geothermometers applied to clinopyroxene-melt couples whose equilibrium has been tested through test 2 do not yield temperature values fitting those estimated in previous works. On the other hand, the geothermometer of Putirka [13] applied to clinopyroxene-melt couples whose equilibrium has been tested through the Kd can be considered the most reliable combination of methods, being able to match the pre-eruptive temperature values obtained in previous studies. This finding is valid for all the eruptions fed by compositionally heterogeneous magmas, such as CI, NYT, Agnano-Monte Spina and Astroni 6, which show evidence of pre-eruptive interaction between less- and more-differentiated magmas (e.g., [55,56,59,83,88,91,94,98,101,120]). For magmas showing homogeneous and highly differentiated compositions (e.g., Monte Nuovo), the clinopyroxene-melt geothermometers applied to mineral-melt couples whose equilibrium has been verified through equilibrium test 2 can be considered the most adequate tool for the estimation of temperatures.

For the pressure estimates, as previously shown, when coupled with the equilibrium test 2, the geobarometer of Masotta et al. [30] yields a great number of negative output pressures. The most reliable pressure estimates, not affected by negative values and reasonable being comparable to those obtained through independent studies, are those obtained with the geobarometer specific for alkaline magmas [30] applied to clinopyroxenes-melt couples, whose equilibria have been verified with test 1. However, this geobarometer was calibrated on a set of experimental data at low pressures; this allows for the investigation of crystallization depths only in the shallower part of the Campi Flegrei magmatic system, casting doubts on depth estimates of its deeper portion. In this regard, the use of such a geobarometer can be an efficient tool for estimating crystallization pressures of minerals in shallow reservoirs and for obtaining useful information about syn-eruptive conditions (magma degassing, magma rising velocity, e.g., through study of microliths, crystal size distribution, etc.), but it does not give the possibility to investigate processes occurring at high depths.

6. Conclusions

Most of the findings which are discussed are based on the comparison between our geothermobarometric estimates and those obtained in other works through different methods, i.e., melt inclusion and phase relation studies. However, it is largely demonstrated that melt-inclusion data can be affected by post-entrapment modification, as well as that the inclusion-bearing crystals can have a wide range of origins and ages, further complicating the interpretation of magmatic processes [167]. Thus, melt inclusions’ volatile contents do not univocally record pre-eruptive storage depth but can follow syn-eruption degassing paths.

The mineral chemistry of the Campi Flegrei olivine, clinopyroxene and feldspar is characterized by two main compositional populations that are recurrent over all the periods of activity. Moreover, the chemical variation trend detected in minerals belonging to the last 12 ka and their numerous mineral-melt disequilibria testify, during this period, to a frequent occurrence of open-system processes (mingling/mixing, crustal assimilation, CO₂ flushing) able to produce diffusive effects and to decouple crystals from their equilibrium melts. Nevertheless, such prevalent compositions suggest that two main sets of thermodynamic variables (T, P) rule the growth of crystals and, hence, the storage conditions of the Campi Flegrei magmas. These two magmatic environments can be ascribed to mafic and evolved magmas, which presumably characterize deep and shallow reservoirs, respectively. Nevertheless, the current geothermobarometric methods do not allow precisely constraining and distinguishing the depth and temperature of the two main magmatic environments. Interestingly, the combination of equilibrium tests considered less robust with geothermobarometric methods considered less appropriate for the Campi Flegrei magma compositions yields estimates that are in agreement with results obtained in independent studies. In fact, our results highlight that (i) there is a low reliability for the combinations of the most recent equilibrium tests and geothermometric and geobarometric methods, which theoretically are the most suitable for the Campi Flegrei alkaline magmas; (ii) although equilibrium test 1 has certain theoretical limitations and is considered less suitable than equilibrium test 2, the mineral-melt couples resulting in equilibrium with the former test produced geothermobarometric estimates that are more in agreement with previous work estimates; (iii) the right choice for the best approximation of the geothermometric and geobarometric estimates depends, case by case, on the compositional features of the erupted products. In fact, the occurrence of frequent open-system magmatic processes, which influenced the chemical-physical conditions (temperature, volatile content, etc.) of the plumbing system, makes the application of classic geothermobarometric approaches critical.

From all these considerations, it is clearly possible to obtain information on the architecture of the Campi Flegrei magmatic system from the geothermobarometric estimates, but the critical issues highlighted in this work suggest that in complex systems such as the Campi Flegrei one, the application of combined methods must be preferred over the single approach.