The Kapsiki Plateau (altitude ≈ 1000 m) is a volcanic area located at the far north of the “Cameroon Volcanic Line” [13] . The Precambrian basement of this zone is recovered by numerous trachyte needles and dykes, phonolite and rhyolite, and partially overlain by basalt flows. The age of the basalts is between 33.2 ± 1.3 Ma [14] and 27.0 ± 0.5 Ma [12] . Trachytes have been dated at 29.6 ± 0.6 Ma, [15] and 35.3 ± 2.4 Ma [15] . The age of the rhyolites is between 32 ± 0.5 Ma and 29.0 Ma [14] . The needles are more than fifty and rise above the Kapsiki plateau. Many of them are emplaced for the benefit of feeder dykes (Teki dyke). Needles from Roumsiki, Gouldé, Mogodé, Roumzu and Re Dilili, Omtémalé and Mchirgui are the most typical [16] .

The presence of xenoliths in the Liri sector (south of the Kapsiki Plateau) was initially reported by [17] . According to the work of [16] , it is proved that the structural and Geophysical (gravimetric and seismological) features of the granite-gneissic base complex are still poorly defined. The Liri sector is located along the Cameroon Line with a basement constituted of granite and gneiss. This basement is crossed by faults and locally covered by basaltic lava flows. The lava flows are broken up into the small centimetric blocks. The spinel-lherzolite xenoliths (size: 7 - 30 cm) with sub-rounded shapes were discovered in these dark basaltic lavas.

Satellite documents [14] have made it possible to highlight alignments and lineamentary structures on the Adamaoua plateau. A main direction N70˚E emerges, called the “Adamaoua beam”. This direction is a megaline several hundred kilometers long, which reflects the surface conditions and more or less materializes the deformations of the lithosphere under the influence of the stresses. It’s formed during the Tertiary and then was uplifted up to 1 km relatively to the surrounding areas [18] . The plateau is limited by the N70˚E Adamaoua and Mbéré-Djerem faults [19] . The establishment of volcanism in this region is probably linked to these pan-African fractures [20] . Xenoliths crop out in basalt flows in the shape of balls or metric-sized blocks. A few rare clinopyroxenite xenolites are encountered at the summits of the massifs. It is also necessary to note the presence of dolerites which form a planeze in certain sectors (case of Likok).

Mineral phases of the xenoliths and host basalt were analyzed using the CAMECA SX50 electronic microanalyzer at the Pierre and Marie Curie University in Paris (France) and in Canada. The samples analyzed are thin sections of rocks of different natures (basalts and xenoliths) and the analyzes are made on polished sections. Before each analysis, the samples are metallized with carbon. Radiation is then analyzed by spectrometers, each of which explores a specific range of wavelengths. Finally, the intensity of the radiation measured at the detector is compared with that produced under the same conditions by standards containing the desired elements at a known concentration. The final quantification of the elements (atoms, % by weight) is carried out by the PAP program of [22] which allows the correction of the analyses. The sensitivity threshold depends on the atomic number of the elements measured, the counting time (s), the voltage (Kv) and the current (nA) used. The analytical conditions are as follows: olivine: 15 kV, 40 nA, 20 s per element, except Si: 10 s; clinopyroxene: 15 kV, 40 nA, 20 s per element, except Ti: 30 s (Fe³⁺ recalculated according to [23] ; orthopyroxene: 15 kV, 40 nA, 20 s per element, except Ti: 30 s (Fe³⁺ recalculated according to [23] ; Fe-Ti oxides: 20 kV, 40 nA, 40 s per cell except Al and Cr: 30 s (Fe³⁺ recalculated from [24] ; plagioclase: 15 kV, 10 nA, 10 sec; spinel: 10 kV, 80 nA, 40 s per element except Al and Cr: 30 s, and Zn and Ni: 15 s.

In the same way, the mineral phases of the xenoliths were also produced under the same conditions on the Cameca SX 100 microprobe of the Microsonde Ouest service in Brest (France).

1) The mineralogy of the xenolites of the Cameroon Volcanic Line and the others localities based on the following four mineral phases.

Olivine

Olivine crystals from lherzolites (Table 1) are not zoned. Olivine crystals display a wide range of Fo (88.58 - 92.15). They are as rich in MgO (Fo~89.5) as that of the peridotite nodules of Sao Tomé (Fo_90-92) [25] ; Bioko [26] and of the continental sector of “the Cameroon Ligne” [27] [28] [29] . The chemical compositions corroborate with recent data from Ngao Voglar in Adamaoua [30] which show that olivine crystals are relatively homogeneous and varies in composition from Fo_88.58to Fo_92.15 with Mg# of 89.1 - 89.8 [Mg# = 100 × Mg/(Mg + Fe²⁺)] and their NiO content varies from 0.03% - 0.5% and a low MnO value (0.11% - 0.2%). The CaO contents vary between 0.00% and 0.14%.

Clinopyroxene

The lherzolites containing chromium diopside (Table 2, Figure 3), nomenclature according to [31] , low in TiO₂ (0.50% - 0.75%) with medium content in Cr₂O₃ (0.63% - 0.77%) and the content en MgO ~ 49%. Diopside compositions vary between (Wo_40.9-42.2En_50.4-51.6Fs_7.1-7.6).

Orthopyroxene

The lherzolites contain enstatite Wo_2.6En_86.2Fs_11.2 (Table 3, Figure 3), nomenclature according to [31] low in TiO₂ (<0.27%), Al₂O₃ (2.27% - 5.98%) and Cr₂O₃ (0.28% - 0.59%). Similarly, the chemical compositions of the orthopyroxenes from Mazélé are similar to the Ngao Voglar data with an enstatite rich in MgO (Mg# = 90.1; En_89-91Wo₁Fs_8-10), low in TiO₂ (0.02% - 0.12%) and Cr₂O₃ (0.28% - 0.43%), and a high content of Al₂O₃ (5.71% - 6.01%) and CaO (0.48% - 1.14%) (Table 3)

Spinel

The spinels (Table 5, Figure 3) of the different samples have contents which vary slightly in magnesium and chromium from one sample to another. The Cr₂O₃ contents vary between 9.0% and 9.6% and those of MgO are close to 20%. Their TiO₂ content is ~0.34%. The edges of the spinels in the sample show in places an enrichment in FeOT and a decrease in Cr₂O₃. The TiO₂ contents are low but these spinels are rich in Al₂O₃ (55.28 - 56.07).

2) Mineralogy of Mazele spinel lherzolites

The mineralogical analyzes are almost identical to those observed in other localities which constitute the north and the Cameroon Volcanic Line.

Olivine

The Fo contents (91%) are relatively high, characteristic of mantle olivine (88 < Fo < 92) (Table 2). Nevertheless, their CaO contents are also high (0.0% and 0.13%) and are rather typical of phases having crystallized under low pressure and symptomatic of a volcanic origin [32] . The temperatures calculated using the thermometer of [33] are relatively high (1250˚C ± 50˚C) but consistent for the distribution of Mg, Fe and Mn.

Pyroxenes

The clinopyroxenes of xenoliths are diopsides (Wo_41.09-49.14En_47.25-51.82Fs_2.74.-7.82: [31] (Table 3), with Mg# ranging from 93.76 - 100. Their Cr₂O₃ contents (0.64% - 1.56%) qualify them as chromiferous diopsides (Cr a.f.u: >0.01: The fact that Mg#Cpx > Mg#Opx > FoOl, leaves to believe that these xenoliths are typical of residual peridotites [25] [26] ). The TiO₂ and Al₂O₃ concentrations in the clinopyroxenes of these peridotites are high by 0.5 and, respectively with Al^VI < 0.2 and Al^IV < 0.2.

Orthopyroxenes (Table 4) are enstatites [31] with a composition varying from _Wo2.5-2.6En_86.3-86.5Fs_10.9-11.2. The Mg# of orthopyroxenes correlates with the Fo of olivines and indicates that Mg#Opx > Mg#ol [27] [34] . The orthopyroxenes of the peridotites have a high Al₂O₃ content (≥5%) compared to the contents obtained in the peridotites of TOb, GA and Liri.

Spinel

spinel (Mg, Fe²⁺) (Cr, Al, Fe³⁺)O₄ (Table 5) is a mineral sensitive to chemical variations and plays, in particular, an important geochemical role in the partitioning of chromium. In the spinels analyzed, the Cr# content varies between 9.04% and 10.07%; the TiO₂ contents vary between 0.30% and 0.45%. Spinels occur in various habitus: in atoll, in automorphic grain, in complex association with other mineral phases.

Note that the alkaline nature of these basalts has been confirmed by the study of certain trace elements and rare earths. The diagram in Figure 3 shows the spectra of trace elements and rare earths normalized to chondrites from [35] of the host lavas of Ganguiré, [30] , Mount Cameroon, [36] and the Kapsiki Plateau.

It is observed that the trace element and rare earth spectra of these different lava units show similar trends. But we note the presence of a negative anomaly in Th and positive anomalies in U and Sr in the host lavas of Ganguiré. They all show a slight anomaly positive in Eu and negative in Dy. The spectra of these lavas are more enriched in light rare earths (La, Ce, Pr and Nd) compared to heavy rare earths (Dy, Yb and Lu) which is a characteristic of alkaline lava (Figure 4).

Fe-Mg exchange between olivine and orthopyroxene

In the case of these peridotites, visual observation under a microscope reveals that olivine and orthopyroxene constitute more than 70% - 90% of the modal composition, and are chemically very close to the FeO-MgO-SiO₂ (FMS) system. We can therefore characterize the chemical variation of the Fe and Mg systems with the olivine/orthopyroxene couple.

According to the work of [37] , the Fe-Mg exchange between olivine and orthopyroxene can be described by assuming a monosite ideal solution behavior for these two phases at high temperatures (>1000˚C), but such a model also remains valid for lower temperatures (800˚C - 900˚C) especially when studying the field of interest around Fo = 90, except for a few cases of the values of TOb and TO1 which slightly exceed (Fo = 92) and that of MAZ whose values are lower (Fo = 88). We can therefore use the following reaction to express this exchange

Mg₂SiO₄ + Fe₂Si₂O₆ <==> Mg₂Si₂O₆ + Fe₂SiO₄

Fo(ol) Fs(opx) En(opx) Fa(ol)

The partition coefficient is expressed as follows for the exchange of one atom for another:

K_D = ((X_Fe)_ol * (X_Mg)_opx)/((X_Mg)_ol * (X_Fe)_opx

For application to the natural system, considering that the exchange between Mg and Fe is limited to the octahedral sites not occupied by Ca, Na, Cr, Al (the other elements can be neglected), the concentrations corresponding to the FMS system can be estimated as being equal to:

(X_Mg)_ol = (Mg/Mg + Fe²⁺)_ol (X_Fe)_ol = 1 – (X_Mg)_ol

(X_Mg)_opx = (Mg/Mg + Fe²⁺)_opx (X_Fe)_opx = 1 – (X_Mg)_opx

Knowing the ratios (Mg/Mg + Fe) of olivine and orthopyroxene of peridotites we can construct the Roozeboom diagram (Mg/Mg + Fe) olversus (Mg/Mg + Fe) opx and determine the partition coefficient at compare with the theoretical model. Figure 5 shows a correlation between olivine and orthopyroxene, indicating partition coefficients greater than 1 with the exception of peridotites from Ganguirét having a mean value ≤0.94). The forsterite content of olivine ranges from 88 - 92. This range of values is typical of mantle rocks.

Al-Cr exchange between pyroxene and spinel

Since Al Indeed, [38] experimentally found by measuring the interatomic distance that, in pyroxene, Al occupies both tetrahedral and octahedral sites, while Cr only occupies octahedral sites. The presence of Na in pyroxene also requires an equivalent quantity of trivalent ions (Al, Cr and Fe³⁺) to maintain the electroneutrality of the phase. The proportion of these three species is relatively constant in the presence of a given aluminous phase for the peridotites of the mantle [39] . In our case, for the spinel facies, the concentration of trivalent Fe in the pyroxenes is very low and no significant value can be calculated from the results of the microprobe analyses.

Reaction in order of importance is:

(Mg, Fe)₃Al(AlSi₃O₁₂) + 1/2(Mg, Fe)Cr₂O₄ <==> (Mg, Fe)₃Cr(AlSi₃O₁₂) + 1/2(Mg, Fe)Al₂O₄

(opx) (sp) (opx) (sp)

The experiment shows that by measuring the interatomic distance in pyroxene, Al occupies both tetrahedral and octahedral sites, while Cr only occupies octahedral sites. The presence of Na in pyroxene also requires an equivalent quantity of trivalent ions (Al, Cr and Fe³⁺) to maintain the electroneutrality of the phase. The proportion of these three species is relatively constant in the presence of a aluminous phase for the peridotites of [40] . In our case, for the spinel facies, the concentration of trivalent Fe in the pyroxenes is very low and no significant value can be calculated from the results of the microprobe analyses.

In summary, to express the Al-Cr exchange between orthopyroxene and spinel, we should calculate the amount of Al in the octahedral site not associated with Na, i.e. VI^Al* = (Al − Cr − Na)/2. Tetrahedral aluminum is equal to IV^Al = (Al + Cr − Na)/2. A Roozeboom diagram (Cr/Cr + Al)Sp – (Cr/Al^IV) Opx is used to describe this exchange appropriately. The trend in Figure 4 implies spinel non-ideality at the Y site for a Cr-Al substitution. This observation is in agreement with the experimental and theoretical data which show that the simple model of regular solution is well suited to describe this Al-Cr partition between spinel and orthopyroxene.

X2 = X1/(X1 + K(1 – X1) exp(WG(2X1 – 1)/RT for fixed T with X1 = (Cr/Cr + Al)spX2 = (2Cr/Al + Cr − Na)opx; X1 and X2 being respective concentrations of Cr in spinel and orthopyroxene. WG(2X1 – 1)RT being the excess energy with a value of 11.3 kJ for the Margules parameter WG, from [41] . This expression allows us to calculate the theoretical Al-Cr partition between spinel and orthopyroxene which is represented in the inset in Figure 6 by asymmetrical curves contrasting with those of Figure 6. The majority of the points lie on a curve corresponding to an equilibrium constant practically equal to 2.5.

Al-Cr exchange between the two pyroxenes

The Al-Cr exchange between the two pyroxenes is well described by an ideal solution model [42] , using the couple (Cr/Al^IV)cpx and (Cr/Al^IV)opx, and an equilibrium constant written as follows:

K_D = (X_Al)_opx(X_Cr)_cpx/(X_Cr)_opx(X_Al)_cpx

avec (X_Cr)_opx = 2Cr/Al + Cr − Na

(X_Al)_opx = 1 – (2Cr/Cr + Al − Na)

Figure 7 indicates that the equilibrium constant is significantly different from unity and varies from 1.5 to 3.5 except for the samples from Ganguiré and Tello 2 whose distribution coefficients Kd are greater than 3.5. The Cr and Al contents of clinopyroxene are about twice as high as those of orthopyroxene. The correlation observed here confirms what we have seen in the previous diagrams: all the samples of peridotite show a good state of equilibrium, and are therefore completely characteristic of the normal mantle.

Diagram (Mg/Mg + Fe²⁺)sp versus (Cr/Cr+Al)sp

It is well known that spinel cannot be considered as a single-site ideal solution with respect to Fe―Mg substitution due to interactions between X and Y sites, called reciprocal effects. The interactions of the cations (Al, Cr) in the Y sites of spinel with the cations (Fe, Mg) in the X sites can be described by the following reaction:

FeAl₂O₄ + MgCr₂O₄ == FeCr₂O₄ + MgAl₂O₄

Since spinel is a Cr-rich mineral compared to other phases in which Cr is almost trace, the ratio (Cr/Cr + Al)sp reflects that of the system and controls (Mg/Mg + Fe²⁺)sp. An autocorrelation is observed in Figure 8. This figure which shows the reciprocal nature of the spinel solution, also illustrates a significant variation in (Cr/Cr + Al) of the system.

We can clearly see a well-defined domain for the xenoliths studied which is placed above that corresponding to the ophiolitic massifs (field of ophiolites according to [41] . This is the normal position of spinel peridotites typical of a lithospheric mantle sampled by alkaline magma quenched within hours. This characteristic leads us to believe that these peridotites are not fragments of an ultrabasic body (tectonic scale) in the continental crust, but that they have a deep origin in the mantle. This observation is completely analogous with that of other xenoliths in the alkaline basalts of Cameroon [27] . The difference between the position of xenoliths and ophiolites was explained by the kinetics of this exchange reaction [43] .

Balance state

Exchange reactions are well suited for verifying the global equilibrium state of paragenesis, whereas transfer reactions are more suitable for estimating equilibrium conditions. There are two types of transfer reactions classically applied to pyroxenes for spinel facies peridotites: one concerns the mutual solubility of orthopyroxene and clinopyroxene, and the other concerns the solubility of aluminum in buffered orthopyroxene by the spinel:

En(opx) <=> En(cpx) et Di(opx) <=> Di(cpx)

and Py(opx) + Fo(ol) <==> En(opx) + Sp(sp)

The theoretical equilibrium curves were calculated under different pressures according to the different thermometers (Bertrand et al., 1987; Bertrand and Mercier, 1985; Benoît, 1987).

The combination of these reactions is also a particularly suitable tool for testing the state of equilibrium of the pyroxenes of ultrabasic parageneses [44] .

Equilibrium conditions

Previous studies have shown us that the xenoliths studied are completely in equilibrium. This therefore allows us to estimate the physical conditions to which they were subjected just before the eruptions, using the transfer reactions described by the thermodynamic laws.

Temperature

Several geothermometers have been calibrated from experimental data and/or data from the natural system [40] , but the choice among these thermometers is not particularly easy. The fundamental problem for thermometry is the extrapolation of P-T from the experimental region when determining the pressures and temperatures of fairly simple synthetic systems to the complex natural compositions of xenoliths. Here we will arbitrarily consider three thermometers (one of which is for an exchange reaction) to determine the equilibrium conditions and check the consistency of the results.

[45] [46] calibrated a thermometer for the Fe―Mg exchange reaction between olivine and spinel:

T (˚K) = (4250(Y_Cr)_sp + 1343)/(LnK_D + 1.825(Y_Cr)_sp + 0.571)

où (Y_Cr)_sp= (Cr/Cr + Al)sp; K_D is an equilibrium constant

K_D = (X_Mg)_ol * (X_Fe)_sp/(X_Mg)_sp * (X_Fe)_ol

Given that the forsterite content of the olivine of the peridotites in the samples is between 88% and 92% and that the Fe³⁺ concentration of the spinel is on the whole very low, it seems reasonable to use this thermometer. The calculation gives a temperature of 845˚C - 1050˚C.

It is well known that as with all thermometers based on exchange reactions, this thermometer is easily disturbed. The temperatures calculated from this thermometer therefore reflect the equilibrium conditions of the last thermal state, that is to say when the peridotites were sampled by the magma which brought them up.

Pression

Pressure estimation remains problematic. Although some calibrations using the solubility of aluminum in orthopyroxene [39] or of calcium in orthopyroxene [40] have been proposed, there is no reliable and precise geobarometer for spinel facies mantle rocks.

As we did not observe garnet in the samples, it is plausible to consider a maximum pressure of 2.5 GPa and a minimum pressure of 0.9 GPa can also be estimated [46] due to the absence of plagioclase.

Other parameters are also sensitive to pressure, in particular the constant Na (Opx)/Na(Cpx) for the sodium exchange reaction between pyroxenes, and Ca in olivine buffered by the couple of reactions between the two pyroxenes. No calibration exists for Na but the Ca (Ol) geothermobarometer has been studied experimentally for CMS and CFMS systems for Fe = 0.16. The problem is the quality of analyzes of this trace element. As the silicon of spinel is a good thermometer, it is practical to study these two traces simultaneously with spinel, despite its empiricism. Calculation indicates that pressures from diopside are outside the allowable range (0.9 - 2.5 GPa) we mentioned, while pressures from enstatite are more reasonable (1.4 - 2 GPa). To calculate the pressure, we used the following [47] formula:

P(kbar) = 771.48 + 4.956.AlT – 28.756.Fe²⁺M1 – 5.345.Fe³⁺ + 56.904.AlM1 + 1.848.Ti + 14.825.Cr – 773.74.Ca – 736.57.Na – 754.81.MgM2 – 763.20.Fe²⁺M2 + – 759.66Mn – 1.185 (MgM2)2 – 1.876.(FeM2)2.

Here we observe a characteristic negative correlation of melt residue between the TiO₂ and SiO₂ content for the clinopyroxenes (Figure 8). All samples follow the evolution of the trend. The TiO₂ contents decrease from the high cpx values of MAZ2 and MAZ1 ≥ 0.7, to the average values of LO, TO₂ and Liri varying between 0.2 and 0.5. The lowest TiO₂ values occupying the bottom of the slope are those of TiO₂ present in the Cpx of Tob and To1 whose values are almost zero < 0.1. The same trend is observed with the evolution of the Al₂O₃ content as a function of SiO₂ which shows a negative correlation leaving higher values (7 - 8) of MAZ1 and MAZ2. On the other hand, the TiO₂ and Al₂O₃ contents are roughly anti-correlated with SiO₂ (behavior of a fusion residue). Their anticorrelations reflect their low levels. The proportion of Cpx can vary significantly within the same series or between the xenoliths chosen. The M1 and M2 octahedral sites of the Cpx have a very marked difference in terms of morphology. The variation of V(M1) as a function of Mg1 makes a positive correlation (Figure 9c). This testifies to a state of equilibrium between the minerals and their appropriate sites [47] [48] . Similarly for the other chemical elements Mg1, Al^iv which fill the same sites, we observe tendencies of a positive correlation contrary to the elements which do not share the same site (Al^iv, Mg2). A high Mg1 content in the M1 site automatically leads to an increase in the Al^iv content (Figure 9).