Study of Analytical On-line Pyrolysis of Oils from Macauba Fruit ( Acrocomia sclerocarpa M ) via GC / MS

Estudos pirolíticos dos óleos vegetais das partes (endocarpo mais mesocarpo e epicarpo) do fruto da Macauba (Acrocomia sclerocarpa M.) foram realizados sob atmosfera de hélio usando-se um pirolisador de filamento modelo Girdel 75 -Py -1, conectado diretamente a um CG/EM. Os experimentos foram realizados em diferentes temperaturas e tempos de pirólise, para avaliar como estes parâmetros afetam o processo pirolítico de obtenção dos produtos bem como a sua reprodutibilidade. A pirólise dos óleos gerou grande número de compostos tais como hidrocarbonetos, aldeídos e ácidos carboxílicos, este último refletindo a composição original dos óleos. A identificação de alguns compostos foi feita com o uso de um banco de dados , dos padrões de fragmentação das classes de compostos e dos cromatogramas de massas de alguns íons específicos tais como m/z 83, 85 e 60. O modo de ionização química positiva foi utilizado para confirmar as massas molares dos constituintes gerados a partir das pirólises, tendo como gás reagente, o metano, (CH4-CI).


Introduction
The need for different alternative sources of energy and technology using Brazilian natural resources have motivated many studies of biomass as a renewable energy source 1 .In addition, biomass is the only renewable source of raw material to the chemical industry and represents the only long term source of carbon 2 .Therefore it is important, in order to preserve in some ways our environment, to direct our efforts towards developing new technology to increase the use of derivatives from biomass in the chemical industry.
Biomass is, however, generally poorly suited for direct energy use and pre-treatment is necessary to change its physical and chemical forms 1 .
Pyrolysis of biomass, in particular, vegetable oils, has been studied since World War II.At that time studies concerning the chemical composition of the volatile compounds obtained were not completed, owing to the limitation in instrumental and analytical methods 3 .The direct thermal cracking of fatty acids (FFAs) and triglycerides (TGs) in the presence or absence of a catalyst, such as bentonite 4 , lignin 5 , calcium oxide [6][7][8] and zeolites [9][10] has been reported.During the last decade, the research into new concepts and processes for higher oil yields and new analytical techniques to elucidate the mechanisms of formation and constitution of the resulting pyrolysates has undergone a technological renaissance of the investigations in the area of pyrolysis 1,11 .
A natural source of a composite mixture of triglycerides (TGs) is the oleaginous plants, which comprises a large part of the vast and diversified flora of Brazil.The plant of interest here is the Macauba tree (Acrocomia sclerocarpa M), in particular, the oils from the Macauba fruit.This tree is native to different regions in Brazil, which are located mainly in the centre of the country and, in particular, in the state of Minas Gerais 8 .
We report herein a study carried out with a filament pyrolyser coupled to a GC/MS, using a flash pyrolysis procedure, employing the oils from Macauba fruit to investigate the effects of pyrolysis time and temperature on the yields of a particular pyrolysate composition.Experiments were performed to obtain information on the influence of these parameters upon the mechanism of pyrolysis of these oils, which is still largely unknown.

Sampling
The Macauba fruit has a spherical slightly flattened shape with an external diameter in the range of 3 to 6 cm and colour of the fruit varying from brownish to yellowish green [12][13] .
The fruit comprises an epicarp or external shell, a mesocarp or pulp and an epicarp or kernel which envelopes one or two nuts [12][13] .
The Macauba fruit was divided into two parts.The outer part comprised a mixture of external shell and pulp and the inner part, the endocarp nut.Both were dried at 80 °C to constant weight, ground into powder, extracted with hexane under a Soxhlet reflux for 72 h and concentrated by distillation to give the vegetable oil extracts 8 .
The composition of each oil, obtained after their esterification with BF 3 in methanol is summarised in Table 1.

Conditions and methodology
Pyrolysis was performed using a Girdel Pyrolyser 75-Py-1, a filament pyrolyser connected to a GC/MS (Finnigan Mat 1020).A gas and power control unit, a cable and a probe comprise the Girdel Pyrolyser 75-Py-1.Helium was used as the carrier gas.The system was purged for a short time (30 s) with the carrier gas before each injection and the probe connected directly to the GC injector.The sample was dispensed, as a thin film onto the filament and the solvent was allowed to evaporate before starting the pyrolysis procedures.The carrier gas was diverted through the control unit and probe ensuring that the volatile pyrolysate wad transferred from this device to the GC/MS equipment without any loss.
Preliminary experiments were performed in triplicate using different temperatures (400 to 1000 °C) and pyrolysis times (10, 20 and 30 s).Further experiments were carried out at a given temperature (700-800 °C) and different pyrolysis times (10, 20 and 30 s) to study the behaviour , reproducibility of the pyrolytic process.

Electron impact (EI) mode
General profiles of the pyrolysates were carried out using EI mode.Analyses were conducted on a Finnigan Mat model 1020 automated GC/MS using a fused capillary column (14 m x 0.25 mm; d f = 0.15 µm; DB1, J & W Scientific) with dimethylsiloxane as the stationary phase.Helium was used, as the carrier gas, at a flow rate of 1 mL/min.Aliquots of 1 µL of the sample in hexane together with 1µL of a mixture of standard compounds (d 10biphenyl, d 10 -anthracene, d 10 -pyrene) at 100 µg/mL as internal standard were spread onto the filament for pyrolysis.This set was employed as a parameter to locate the molecular weights of the compounds when the library search was performed as well as to confirm the reproducibility of the pyrograms.
The injection mode was short splitless with a 1s delay.The column was held at 35 °C for 2 min and then heated to 300 °C, at 20 °C/min.The final temperature was held constant for 2 min.Analyses, in a PCI (positive chemical ionisation) mode, were carried out on a GC model HP 5890 series II and MS model VG Trio 1000, using a fused capillary column (11 m x 0.32 mm, d f = 0.25 µm, DB5, J &W Scientific), and helium as the carrier gas and methane as the reagent gas.The GC conditions used were the same as for EI mode.

Compounds identification
The presence of different classes of compound from pyrolysates was confirmed using the total ion chromatogram (TIC) and mass chromatogram (MC) analyses, in addition to fragmentation patterns [16][17][18][19] and library search (NIST).Identification was carried out according to pre-established criteria for the analysis of the data 17 .

Data analysis
To obtain the data from the different experiments the following steps were undertaken: • TIC of each experiment at a specific condition : mode of injection, temperature and pyrolysis time; • Preliminary identification of the main peaks using the computer library data.The peaks were considered identified when the library response for purity was > 900 and fit was > 975.Positive chemical ionization (CH 4 -CI) was also used in an attempt to confirm the molecular weight of compounds; • Identification of classes of compounds to which the main peaks belonged; • A specific m/z characteristic of each class of compound was selected, for instance, alkane (m/z 85), alkene (m/z 83), alkynes or dienes (m/z 81), fatty acids (m/z 60 and 73), derivatives of cyclohexene (m/z 54 and 67); • Identification of the target compounds (however, in some cases only the base peak was identified); • Quantification of the peaks in terms of relative percentage was done using a software from the data system: SR and QL-QR-MQ-EQL; • The resulting data from quantification were compared for all replicate experiments, by a Fortran programme, at a given condition such as mode of injection, temperature and pyrolysis time; • Answers are registered into new files called RE-SULTS which were transferred to a spreadsheet.Statistical parameters such as mean, standard deviation, variance and error were calculated; • Each class of compound was classified according to its characteristic m/z and after that, constituents which belong to each class were added up to give a global view of the yield of products obtained.

Results and Discussion
Studies were carried out to evaluate the influence of the temperature and pyrolysis time on the behaviour and reproducibility of the process as well as the changes in the yield of products of both oils from Macauba fruit.
It has been reported 7,14,15 that a mixture of triglycerides submitted to a thermal cracking process, should undergo to decarboxylation, disproportionation and successive elimination of ethylene molecules yielding to some classes of compounds such as hydrocarbons and carbonylic compounds.In addition, the sample can undergo many different reactions such as hydrogenation and oxidation, the latter one providing the greatest difference between the cracking of vegetable oils and petroleum as the latter does not generate carbonyl groups in products 14,15 .

Preliminary studies
It was observed that the samples underwent a desorption like process rather than pyrolysis at lower temperatures (400 °C to 600 °C) and fixed pyrolysis time.The real pyrolytic process took place in the temperature range between 700 °C and 1000 °C, which could be observed by the changes in the total ion chromatogram (TIC), which become more complex as the temperature was increased, as shown in Fig. 1.
The pulp and nut oils only underwent partial pyrolysis.The pulp was more resistant to this process probably due to higher degree of unsaturation.Both oils led to carboxylic acids compounds generated from triglycerides of which the original oil is composed (see Table 1).A major peak (m/z 44), at the beginning of the chromatographic run, indicated the presence of carbon dioxide, probably originated from decarboxylation of fatty acids 9 .
A general feature of the pyrolysates of the pulp and nut oils is summarised in Fig. 2 , which shows the presence of several compounds, some of which were identified as indicated in Tables 2 e 3.

Analysis by EI mode
The driving force in the cracking of the pulp oil under helium atmosphere was the formation of aldehydes, cycloalkanes, alkenes, and dienes.The paramount product generated was 2-propenal or acrolein, a known product when triglycerides are pyrolysed 15 .Some compounds classified as unknown were also detected, which exhibited different base peaks and are reffered to as unknown m/z 55 and unknown m/z 67.In addition, the presence of small amounts of alkanes, alkenynes, alkylbenzene and cycloalkene were also evident (see Table 4).
The pyrolysis of the nut oil leads to alkenes, aldehydes and carboxylic acids as major products, as indicated in Table 5.The pyrolysate also exhibits cycloalkane, alcohol, alkane and some components, with base peaks at m/z 54 or 67, which are here classified here as unknown.These compounds are considered to be associated with cyclohexene derivatives [20][21][22] .Other constituents with base peaks, m/z 55 and m/z 57 were also noticed.
Some classes of compounds are common to pyrolysates products such as aldehyde, cycloalkane, alkene , alkane and those reffered as unknown/55 but others are not.These differences should be associated to the composition of the original oils.
The products generated by the pyrolytic process from both oils when compared to their respective calcium soap pyrolysates, are quite different 8,14,15 .In the latter case the main products are comprised of homologue series of n-alkane, 1-alkene and some carbonylic compounds.

Analysis by CI Mode
The crucial problem with the analysis in EI mode is the lack of molecular weight data for many of the compounds, which made the identification more difficult.CH 4 -CI/MS was used to overcome this problem in an attempt to match the identification and the molecular weight of the compound identified using EI mode.To ensure that the compound in EI mode was the same in CI mode the relative  retention time to biphenyl (t R = 6,74 min) was used as a reference.

Pyrolysis behaviour
Studies of the behaviour of the pulp and nut oils at different temperatures and pyrolysis time have shown different trends in profile (see Figs. 3 and 4, and Tables 4 and 5).

Pulp oil
The products of the pyrolysate can be classified into three categories: main products (> 10%), secondary products (3%-10%) and tertiary products (< 3%).The overall trend was the same for both worked temperatures.At 700 °C the main products (aldehyde, cycloalkane, alkene and diene) correspond to 71-79% of the total products; the secondary products (unknowns/55, 67 and others and alkylbenzene) ranging between 19 to 26% and the remaining products (alkenyne, alkane, cycloalkene) from 3 to 4%.At higher temperature ,however, it was noticed a slight decrease in the main products (61-76%), and a slight increase in the secondary products were noticed (22-34%).The percentage of tertiary products remained constant.
According to Table 4 and Fig. 3. it can be observed that short a pyrolysis time (5 s) tends to favour cyclization and generation of multiple bonds whereas a longer pyrolysis time favours secondary reactions.These trends may be explained by the extent to which the sample underwent secondary reactions (see Table 4).

Nut Oil
In this sample, the composition of the whole pyrolysate was distributed in the following way: main products (aldehyde, alkene and carboxylic acid) corresponding to 59-74%; secondary products (cycloalkane, alkane and unknown/55) varying between 24-37% and tertiary prod-ucts (unknown/54 and 57 and alcohol) ranging between 5-7%.
The relative percentage of carboxylic acids and unknown/55 increased when the pyrolysis time was raised in both worked temperatures.Other products, such as alkene, aldehyde and alkane have a tendency to decrease (see Fig. 4 and Table 5).This suggests that the generation of these classes of compounds is dependent on the cracking of carboxylic acids whereas the formation of unknown/55 is not.Additional trend is that classes of compounds present in less quantities of the total pyrolysate (unknown/54 and 57 and alcohol) increased with temperature and pyrolysis time enhancement.This fact can be explained by secondary reactions, which should be easier to occur under these conditions.
Short pyrolysis time (10 s) at both temperatures leads to a high amount of alkanes, alkenes and aldehydes instead of carboxylic acids.On the other hand, higher temperature and long pyrolysis time did not favour an extensive pyro-  ----------------lysis of this material.In this case, the desorption like process becomes more likely than the pyrolytic process.

Reproducibility of the pyrolytic process
The critical problem with the PI/GC/MS technique is the reproducibility of the experiments.There are several parameters which should be controlled during the experiments such as filament homogeneity and temperature, quantitative transfer of the pyrolysate material, amount of the material dispensed on the filament etc.All these parameters could affect in some way the reproducibility, especially in the quantitative aspect.
To ensure the reproducibility aspect of the experiments perdormed a various temperatures and pyrolysis times a  To verify the first assumption a set of internal standards comprising biphenyl, anthracene and pyrene) was used which were pyrolysed together with the samples.The chromatographic features were quite reproducible as illustrated in Fig. 2.
The second assumption was studied after obtaining the semi quantitative data presented in Tables 4 and 5 and Figs. 3 e 4, in an analogous manner the previously described procedure.These data allowed the evaluation of the observed trend of the yield of products when temperature and pyrolysis time were varied.The pyrolysate composition at each set of conditions was found to be quite reproducible.This means that the parameters of the pyrolysis systematically affected the pyrolytic process.
The third assumption was verified by calculation of the relative errors from replicate analysis.In the less representative classes of compounds the estimated errors were > 20% and for the main classes these values were ranging between 4 and 20%.

Conclusion
The technique of PY/GC/MS has been demonstrated to be a good analytical tool for identifying the main constituents generated from pyrolysis of the original oils from macauba fruit using the direct analysis with capillary GC/MS in EI and PCI modes.It was shown that the technique could be used to investigate the associated mechanism of the pyrolysis process.Under the conditions employed both oils undergo a pyrolytic process to a limited extent, generating a considerable amount of carboxylic acid component from the original oils.
The process as a whole was shown to be reasonably reproducible.

Figure 1 .
Figure 1.Comparison between pyrogram profiles at different pyrolysis temperatures (700-1000 °C)and the same pyrolysis time(30 s) by PI/CG/MS for nut oil.

Figure 3 .
Figure 3.Comparison between profiles at a given pyrolysis temperature and different pyrolysis times for pulp oil.(A)700 °C; (B) 800 °C.

Figure 4 .
Figure 4. Comparison between profiles at a given pyrolysis temperature and different pyrolysis times for nut oil..(A)700 °C; (B) 800 °C.

Table 1 .
Composition in percentage of fatty acids present in inner and outer oils from Macauba Fruit (Acrocomia sclerocarpa M).

Table 2 .
Some identified constituents of pulp oil pyrolysate.

Table 3 .
Some identified constituents of nut oil pyrolysate.

Table 4 .
Composition of the Pulp Oil pyrolysate at a given pyrolysis temperature and different pyrolysis times.

Table 5 .
Composition of the Nut Oil pyrolysate at a given pyrolysis temperature and different pyrolysis times.