The Parinari macrophylla (PM) shells were ground and sieved. The active carbons were produced using the granulometric fraction ranging from 2.5 mm to 5 mm.

The mass yield ranges are presented, for the three activation modes, in Table 2. These results do not allow to prefer the physical activation over the other activation modes (mean yield of 25%). They show that the use of orthophosphoric acid as an activating agent increases the pyrolysis yield through a dehydration process. This delays the thermal decomposition of the raw material, thus limiting

the loss of volatile matters, and it leads to a rigid carbonaceous matrix, as stated by Zhao et al. [22]. In the case of the chemical activation, a decrease of the pyrolysis yields is logically observed when the temperatures get higher (from 45% at 300˚C to 38% at 400˚C and 34% at 500˚C, all other parameters being held at their reference values).

The A.C. adsorption capacity, including the microporosity, in a condensed phase (aqueous) is estimated through the iodine number. Figure 2 shows the iodine numbers for the A.C. produced by physical activation. The physical activations under a flow of N₂ lead to low values of iodine numbers (between 131 mg I₂/g and 193 mg I₂/g). Moreover, the iodine number values of 200 mg I₂/g and less are largely under the application range of the norm. The only conclusion that can be drawn is that the produced solids actually develop some porosity but cannot be referred to as proper active carbons. The addition of water vapors to the process does not change the adsorption capacity of the solids. The only value that significantly varies from the other ones is the iodine number corresponding to the solid obtained when the water is heated at 70˚C. Not much can be concluded from this fact, as it might be a simple variation related to the lack of accuracy going along with the determination of iodine numbers lower than 200 mg I₂/g. The physical activations under a flow of CO₂, on the other hand, lead to higher values of iodine numbers and the higher the water vapor content, depending on the temperature of the water flask, the lower the iodine number. Thus, water has a negative effect on the iodine number value, during the physical carbonic activation.

Concerning the chemical activation with H₃PO₄, the iodine number rises with the increase of the acid concentration, the impregnation time and the pyrolysis temperature and time. When these parameters are set at their lowest, the iodine number is 256 mg I₂/g and has been raised up to 457 mg I₂/g (Figure 3). The effect of the orthophosphoric acid concentration on the iodine number has been observed during the mixed activations as well (Figure 4).

These results show that, whatever the activation mode, the microporosity of the active carbons obtained from Parinari macrophylla is always more developed when the medium is oxidative (CO₂, H₃PO₄), which is confirmed by the lower iodine numbers in the presence of water vapors. Indeed, when the activation occurs under a mixed flow of CO₂/H₂O, the iodine numbers decrease proportionally

with the amount of water vapors present in the pyrolysis reactor (depending on the temperature of the water through which the CO₂ is flowing to moisturize it). This shows that water vapors interfere with the action of CO₂ during the creation process of the porosity. On the other hand, the presence of water, during the activations under N₂ flow, shows little to no impact on the iodine numbers. This leads to think that water vapors are involved, in the physical activation process under CO₂ flow, either by inhibiting the creation of pores or by boosting the creation of meso- and macropores, which would also decrease the iodine number values.

The influence of the orthophosphoric acid concentration has been reported about the activation of several biomasses [5] [23]. It has been shown that this acid increases the microporosity adsorption capacity (as shown in this work as well) up to a certain concentration or a certain H₃PO₄ mass ratio, above which the medium becomes too oxidative and leads to the destruction of the porous network. This concentration limit in H₃PO₄ depends, of course, on the nature of the source material.

Concerning the activations in the presence of H₃PO₄ (chemical and mixed), the increase of its concentration produces A.C. with higher iodine numbers (Figure 3 and Figure 4). The effect is particularly obvious in the case of mixed activations, as the increase of H₃PO₄ concentration, from 0.5 mol/L to 5 mol/L, leads to a rise of the iodine numbers of 300 mg/g when the pyrolysis occurs under N₂ and of 450 mg/g when the pyrolysis occurs under CO₂. It is worth to note that the mixed activations give the A.C. with the highest adsorption capacities towards iodine.

The nitrogen adsorption at 77 K has enabled to determine the BET specific surface area, BET microporous surface area, total and microporous volumes of the produced A.C. All these characteristics are listed in Table 3.

During physical activations, the specific surface area values are too low for proper active carbons (less than 450 m/g), which is consistent with the low iodine number values exhibited by those solids. Nevertheless, the analysis of the BET isotherms proves that an internal (thus porous) surface has been developed during the process. The specific surface areas globally decrease as water vapors are added to the medium, going down from 315 m²/g (entry 3) to 171 m²/g

(entry 4) under nitrogen and from 447 m²/g (entry 6) to 408 m²/g (entry 8) under carbon dioxide. When carbon dioxide is used on its own, this leads to a more porous solid, developing more surface area, than when nitrogen is used alone (see entry 5 vs. entry 1). This fact could be due to the oxidizing properties of CO₂, which are favorable for the creation of pores during the pyrolysis of the biomass. The microporous surface and microporous volume follow the same decreasing trend. The presence of water vapors seems to lead to a less microporous volume under N₂ (ratio V_micro/V_total is lower with water at 45˚C or 70˚C, see entries 3 and 4) and a more microporous volume under CO₂ (ratio V_micro/V_total is higher with at 45˚C or 70˚C, see entries 7 and 8). The pore diameters follow the same order of evolution.

When the chemical activation with H₃PO₄ is studied, the specific surface area ranges from 331 m²/g (entry 15) to 727 m²/g (entry 9) and rises with the increase of the H₃PO₄ concentration (entries 12, 13 and 14), the impregnation time (entries 11, 10 and 9) and the pyrolysis temperature (entries 15, 16 and 17) and time (entries 18, 13 and 14). The micropores remain the major porosity type of all the A.C. produced, even for the solids with the lowest specific surface areas (entries 12 and 15). Indeed, all the chemically activated carbons display a microporous surface area representing more than 74% of the total area, a microporous volume representing more than 63% (sample which has the lowest porosity is entry 12) of the total pore volume and a mean pore diameter lower than 20 Å.

The same observations can be made for the mixed activations when considering the phosphoric acid concentrations (entries 21, 22 and 23 as well as entries 24, 25 and 26). Excepted for the lowest phosphoric concentration, nitrogen seems to lead to lower specific surface areas than carbon dioxide (entries 22 vs. 25 as well as entries 23 vs. 26). Microporous surface and microporous volumes are often, higher than 80%. This activation mode has allowed to produce active carbons displaying specific surface area values similar to the values obtained for the A.C. produce by chemical activation, but while using lower concentrations in H₃PO₄. It is to be noted that the mixed activations, using CO₂, give A.C. with higher specific surface areas than when N₂ is used. Chemical and mixed activations generally lead to active carbons with similar porosities whereas the ones produced by physical activation rather develop a lower porosity. Indeed, the orthophosphoric acid, used during the chemical and mixed activations, plays a major role in the pore formation. It takes part in the depolymerization reactions involving the macromolecules of the biomass (degraded in CO, CO₂ and CH₄), as well as in the condensation and cyclization reactions of the polyaromatic structures [24], which lead to the formation of a porous and rigid carbonaceous material. On the contrary, during the physical activation, the reactions at high temperature (up to 800˚C), under flows of CO₂ or N₂ alone, do not allow to get the same level of pore formation [24]. Furthermore, it is to be noted that the best active carbons, produced from the pyrolysis of Parinari macrophylla shells, have characteristics similar to these of some active carbons produced from others agricultural wastes [1] [2] [12] [20] [25] and of some commercial active carbons, which have generally higher pore volumes though (frequently close to 0.4 cm³/g or higher, while the active carbons produced for this work reach 0.3 cm³/g at best). For instance, the active carbon supplied by Merck (ref. 2184) show a specific surface area of 469 m²/g and a pore volume of 0.38 cm³/g, or the active carbon Lurgi Hydrafin show a specific surface area of 762 m²/g and a pore volume of 0.40 cm³/g.

Nevertheless, the nature of the pyrolysis gas has no effect on the percentage of microporous surface area and volume. This may mean that the microporosity is essentially due to H₃PO₄. These results show an opposite effect of water and orthophosphoric acid on the microporosity of the final A.C. Water tends to decrease the microporosity of the A.C. while H₃PO₄ increases it. In the production of A.C. from Parinari macrophylla, the studied activating agents contribute to the creation of microporosity in the following order: N₂ < CO₂ < H₃PO₄. The increase of the amount of these activating agents in the medium leads to higher specific surface areas [5] [8] [25] [26] [27].

Globally, the iodine numbers change alongside the BET surface areas. Although the two measures do not characterize exactly the same property of the solids, they both give an indication of the adsorption capacity for small size molecules. Indeed, both analyses are based on the adsorption of small molecules, N₂ having a molecular radius of 2.1 Å and a cross-sectional area of 16.2 Ų, and I₂ having a molecular radius of 3.3 Å and a cross-sectional area of 22.6 Ų. These dimensions give them a good accessibility to the majority of the active carbon porosity, including the microporosity (pore diameters lower than 20 Å) despite the fact that the N₂ adsorption occurs in gas phase and the I₂ adsorption occurs in condensed aqueous phase.

It is thus legit and interesting to plot the graph of specific surface area vs iodine number and micropore surface area vs iodine number. These plots are presented in Figures 5-7 for the A.C. produced by physical, chemical and mixed activations, respectively.

A linear trend can be observed between the iodine number and the surface area values and is validated by the correlation coefficient (r²) of the linear regression, which usually ranges from 0.8740 to 0.9996. The r² values for the graphs “surface area (total or microporous) vs iodine number” are low (0.5395 to 0.5848) in the case of A.C. produced by CO₂ physical activation. These low r² are nevertheless obtained for regressions performed on only four points. All it takes for r² to drop is one-point diverging from the regression line, which is the case here, as two A.C. with similar iodine numbers exhibit different surface areas. For chemical activations, the same phenomenon of different surface areas corresponding to similar iodine numbers can be observed for a couple of A.C. (iodine number around 460 mg I₂/g) but as the regression curves are plotted on a dozen points, the r² are still about 0.8903.

For a given activation mode, the linear regressions “total surface area vs iodine number” and “micropore surface area vs iodine number” are basically parallel, which indicates that there is, as well, a good correlation between the specific surface area and the micropore surface area. It is thus possible to estimate, roughly, the specific surface area or the iodine number if the other one is known, and it is also possible to estimate the micropore surface area from either the total surface area or the iodine number.

The porosity of the A.C. is created simultaneously with the decomposition of the biomass, during the pyrolysis and the activation processes. In the case of Parinari macrophylla as the source material, the production of the A.C. goes along with a mass loss of 55% to 77%, corresponding to pyrolysis yields of 45% and 23% respectively (Table 2). However, the highest values of surface area or iodine number are not connected to the highest mass losses, and this observation is true for all the activation modes.

Moreover, the chemical activation with H₃PO₄ leads to higher values of specific surface areas and iodine numbers, albeit with a higher pyrolysis yield (thus less mass loss), than the physical activation. This confirms that the orthophosphoric acid, as an activating agent, delays the decomposition of the biomass while it contributes to the formation of a rigid carbonaceous network [22], which can be explained by the creation of P-O-C bonds [12] [28] [29]. H₃PO₄ hence takes part in the biomass decomposition and also in the pore creation process. Marsh and Rodriguez-Reinoso [24] have suggested a reaction mechanism for the effect of H₃PO₄ on lignocellulosic biomass. It implies the formation of phosphoric acid esters at temperatures lower than 450˚C, the elimination of the acid at temperatures above 450˚C and the creation of ink bottle shaped pores as H₃PO₄ is released. However, a fraction of the acid remains in the A.C., as proved by the presence of FTIR signals corresponding to P-O-C bonds.