Chromatographic analysis of bio-oil formed in fast pyrolysis of lignocellulosic biomass

2.1 Pyrolysis-gas chromatography/mass spectrometry (Py-GC/MS)

Pyrolysis-gas chromatography/mass spectrometry (Py-GC/ MS) is one of the most popular methods used for the analysis of the behavior of biomass in fast pyrolysis process. It consists in decomposition of a small amount of the investigated sample (usually several milligrams) in a micro-reactor with a controlled heating rate. The use of a small sample weight allows the minimization of the temperature lag during the analysis and heating of biomass with the rate even tens of thousands °C/s, which corresponds to the conditions of fast pyrolysis. In the case of pyrolysis vapor upgrading conducted with the presence of the catalysts, the obtained primary products of thermal decomposition of biomass are passed through a catalyst bed and then, temporarily, collected in the adsorption column at low temperature (i.e. 30°C). In the next step, the absorption column is rapidly heated to about 300°C which results in desorption of the collected mixture of pyrolysis products. Subsequently, they are directed to chromatograph for analysis. As mentioned earlier, chromatograms obtained for the mixture of fast pyrolysis products are very complex. Therefore, gas chromatograph should be coupled with a mass spectrometer in order to identify a higher number of compounds formed during biomass decomposition. An interpretation of the obtained mass spectra is commonly performed according to NIST MS library. However, in some cases, the difficulties in data interpretation may be noticed due to low probability of the formation of the suggested compounds. That is why some number of the chromatographic peaks usually remains unidentified.

The information about types of pyrolyzers, chromatographic columns and conditions of GC/MS analysis of the products of fast pyrolysis of lignocellulosic feedstock is presented in Table 1 [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]. The comparison of the collected data shows that helium is applied as the carrier gas with a split ratio in the range between 1:7 and 1:100 (with 1:50 and 1:80 the most often used). The injector temperature is usually maintained at the same level as temperature of desorption of reaction products from adsorption column (close to 300°C). The temperature of GC/MS interface is slightly lower (about 250°C), while the oven is heated from 30°C to 350°C (with 280°C as the most often used temperature). Mass spectra are collected in m/z range from 16 to 800.

The first limitation of the use of pyrolysis-gas chromatography/mass spectrometry to study fast pyrolysis of biomass can be related to the fact that Py-GC/ MS method does not allow conduction of a continuous process. However, literature shows that the composition of obtained products is in close agreement with that noticed for other bench-scale pyrolyzers [48]. It confirms that this method can be successfully applied to the studies of high temperature decomposition of lignocellulosic feedstock. Additionally, Py-GC/MS does not allow for the collection of formed mixtures of reaction products, which are only temporarily collected in the adsorption column and then passed to gas chromatograph for analysis. That is why determination of mass balance is practically impossible in this case.

It should be also noted that an application of Py-GC/ MS to the analysis of products of fast pyrolysis of biomass cannot guarantee fully quantitative determination of the chemical compounds present in the reaction mixture. However, some quantitative as well as qualitative information can be obtained. It is assumed during Py-GC/MS experiments that there is a linear relationship between change of intensity of chromatographic peaks ascribed to particular reaction products and their amounts. Therefore, changes of peak area percentages observed in the chromatograms obtained for various bio-oil samples can be considered linear with the concentration of formed substances. When the same mass of the feedstock is used in each experiment, the peak intensities observed for particular compounds present in different samples can be compared. This allows for determination of their relative content in the mixture of products. Sun et al. [21] suggested also that direct quantitative analysis with the use of Py-GC/MS can be very difficult due to a large number of identified substances and the lack or high cost of available standards.

On the other hand, Lu et al. [23] reported that Py-GC/ MS can be useful in quantitative analysis when the measurements are only focused on a small group of selected compounds. In this case, external calibration method was applied. It allowed the determination of the content of 14 different chemical compounds (for example 2-methoxy-4-methyl phenol as a representative of phenolics). The use of external calibration method for the analysis of the products of catalytic fast pyrolysis of biomass was also described in Ref. [31]. These studies were devoted to the production of aromatic hydrocarbons. That is why the concentration of 5 major aromatic compounds, such as benzene, toluene, xylene, naphthalene and 2-methylnaphthalene, was determined. Moreover, Harman-Ware et al. [33] applied Py-GC/MS to a determination of the ratio of sinapyl and coniferyl alcohol in pyrolyzed lignin. It was possible due to the selection of several phenolic compounds used as markers of mentioned lignin components.

Due to the problems with quantitative determination of the products of fast pyrolysis of lignocellulosic biomass, the Py-GC/MS studies are mainly focused on the qualitative aspects of the composition of bio-oil and comparison of the contribution of its components. In spite of that, the interpretation of the obtained results is not an easy task. The large number of pyrolysis products generates difficulties related to description of the composition of analyzed mixture.

Generally, in the first step of Py-GC/MS experiment, the total intensity of chromatographic peaks is determined. It allows for comparison of the efficiency of the production of a liquid fraction. Then, identified compounds are divided into several groups based on differences in their chemical structure. Different ways of the mentioned division are presented in Table 2. It is demonstrated that hydrocarbons, phenols, carboxylic acids, aldehydes, alcohols and ketones are the most popular groups of analyzed products. Some researchers also include carbohydrates, esters, ethers, N-containing compounds, etc. It is worth noticing that some part of the characterized substances are not assigned to any of the groups and collected as “others”. This may result from their complex structure (presence of different functional groups) or impossibility of identification.

In the next step, an interpretation of the results is focused on the analysis of the contribution of particular compounds interesting from scientific or industrial points of view. The main groups of products can be divided into more specific ones. For example, Zhang et al. [34] divided hydrocarbons into light hydrocarbon and aromatics. On the other hand, Vichaphund et al. [40] focused on

aliphatic and aromatic compounds, while Grams et al. [47] identified olefins, paraffins, aromatics and cyclic hydrocarbons.

Subsequently, an analysis of the particular substances within the most interesting subgroups takes place. Liu et al. [26] in their work devoted to hydro-pyrolysis of biomass discussed the effect of reaction conditions, presence of catalyst and type of feedstock on the content of phenolic compounds (including phenol and its derivatives, such as: 2-methyl-phenol, 4-methyl-phenol, 2-methoxy-phenol, 2-methoxy-4-methyl-phenol, 2-methoxy-4-vinyl-phenol and 2,6-dimethoxyphenol), aromatic hydrocarbons (benzene, toluene, p-xylene and 1-ethyl-methyl-benzene), cyclo-alkanes (i.e. cyclohexane, methyl-cyclohexane, 1,2-dimethyl-cyclohexane, ethyl-cyclohexane and propyl-cyclohexane), furans (2,3-dihydrobenzofuran, furfural, 2-ethyl-5-methyl-furan, furan and 2-methyl-furan) and alcohols (butanol and pentanol) (Table 3). Xue at al. [24] studied catalytic fast pyrolysis of bamboo residue and focused on the identification of phenols (phenol, xylenol, ethyl phenol, benzenetriol, syringol, metoxyphenol, vinyl-methoxy phenol, methyl methoxyphenol and ethyl methoxyphenol), while Zhang et al. [30] determined contribution of not only phenols, but also hydrocarbons (such as: benzene, toluene, xylene, indene and polycyclic aromatic hydrocarbons (PAHs)).

The other works also underlined complex composition of bio-oil formed in fast pyrolysis of biomass. Lu et al. [29] compared the intensity of peaks corresponding to the presence of 11 phenolic compounds and 4 hydrocarbons, among others. On the other hand, Zhang et al. [34] distinguished 7 aldehydes, 9 ketones, 3 furans, 6 aromatic and 12 light hydrocarbons (Table 3). The presented data suggests that detailed interpretation of the results obtained by Py-GC/MS is very difficult and probably not necessary. Therefore, researchers focused rather on the identification of the major bio-oil components which are the most valuable from the industrial point of view or undesirable ones (taking into account their toxic character or instability).

2.2 Types of reactors coupled with GC/MS analysis

The most popular way of the analysis of bio-oil formed in the fast pyrolysis of biomass is related with the use of setup consisted of Pyroprobe pyrolyzer coupled with GC/ MS as an analytical tool. However, literature shows that thermal decomposition of lignocellulosic feedstock can also be performed with the use of various types of chemical reactors, for example fixed bed, fluidized bed, moving bed, conical spouted bed, multi-zone, etc. (Table 4) [14, 18, 49, 50, 51, 52, 53, 54, 55]. The majority of them operate in continuous mode, which is widely applied in the industrial processes of biomass conversion. This creates a need for development of methods allowing transport of produced bio-oil from the reaction system to the chromatograph. It appears that the reaction mixture can be passed directly (on-line) to GC/MS system or condensed by condenser units in order to transform the pyrolysis products to the liquid phase. Then, the obtained samples of bio-oil are directed to the chromatograph by separate injections.

For example, Amutio et al. [49] condensed bio-oil formed in a conical spouted bed reactor by double-shell tube condenser cooled by tap water. They also used two coalescence filters in order to ensure the recovery of heavy molecules. Similarly, the use of condensers for transformation of pyrolysis products into liquid phase was reported in Refs. [52,54].

Generally, an identification of the collected products is performed according to the same procedure as that used during Py-GC/MS analysis described in the previous chapter. However, slight differences in the measurement parameters can be observed in this case (Table 4). The temperature of GC/MS oven varied from 40°C to 330°C, while split ratio ranged from 1:4 to 1:100. Mass spectra were collected in m/z range from 0-1000.

2.3 Novel chromatographic methods of bio-oil analysis

Complex nature and large variety of chemical compounds present in the mixture of products of fast pyrolysis of lignocellulosic biomass result in the problems with identification of bio-oil components. Despite high versatility of GC/MS, this technique does not allow for identification of all substances formed during decomposition of lignocellulosic feedstock (for example those which possess high polarity, low volatility or poor thermal stability). That is why, there is a need to design new complementary methods that would fill the gaps in this field. Taking that into account, researchers focused on the development of novel chromatographic methods devoted to determination of the composition of bio-oil formed in high temperature conversion of biomass. Previously performed studies were described in several review papers [19,56]. Moreover, the examples of the conducted investigations are presented in Table 5 [57, 58, 59, 60, 61, 62, 63, 64, 65, 66, 67].

Due to complex composition of bio-oil and difficulties in the identification of its components, the various methods of the pretreatment of samples before the main part of the analysis can be applied. According to Staš et al. [56], they can be divided into solvent and solvent-free methods. In the first case, bio-oil is dissolved in one of the solvents (for example – acetone, hexane, diethyl ether or dichloromethane). Owing to that several fractions of compounds differing in chemical properties can be obtained and directed separately to analysis, which makes an interpretation of the chromatographic data easier and more efficient.

On the other hand, fractionation of the analyzed bio-oil can be performed with the use of adsorption chromatography (Liquid-Solid Chromatography - LCS), gel permeation chromatography (GPC) or centrifugal partition chromatography (CPC) [56,58]. In the case of LCS, bio-oil components introduced to chromatographic column are eluted using solvents with increasing polarity. This results in the formation of several fractions having different polar character. GPC allows for the separation of analyzed compounds based on the size of their molecules, while CPC is based on the separation of solutes from the mixture of products of fast pyrolysis of biomass according to the differences in their partitioning coefficients between the mobile and stationary phases [59].

Moreover, solid phase extraction (SPE), molecular distillation (MD) or sample derivatization (SD) can be applied before chromatographic analysis of the products of high temperature decomposition of lignocellulosic feedstock. The latter technique is often used before GC or HPLC analysis. It is based on derivatization reaction (for example – acetylation or trimethylsilylation) which allows for transformation of the analyzed compounds into detectable ones. It is connected with the enhancement of their elution properties, improvement of detector response, change in the volatility, thermal stability or reduction of the strength of adsorption, among others.

Going back to the chromatographic analysis of products of fast pyrolysis of biomass, it should be noted that gas chromatography equipped with mass spectrometric or flame ionization detector was the most commonly used method. However, this method suffers from unsatisfactory resolution of chromatograms, peak co-elution, lack of analytical standards, available mass spectra of all identified substances in MS libraries or the presence of nonvolatile compounds in the analyzed mixture and, due to that, difficulties in qualitative and quantitative analysis. In spite of that, Py-GC/MS allows for on-line characterization of formed bio-oil and direct introduction of the sample to the chromatograph, it is not possible to detect high mass compounds due to their condensation in the transfer lines. Moreover, nonvolatile substances or the presence of water may be a reason for deterioration of chromatographic columns. It appears, that mentioned difficulties can be partly overcome by application of more sophisticated methods of the bio-oil analysis [19].

Two-dimensional gas chromatography (GC x GC) is based on the separation of the analyzed mixture on two capillary columns (the first nonpolar and the second with high or medium polarity) connected with a modulator. In the first step, the bio-oil components are separated according to their boiling points. Then modulator collects substances leaving the first column and directs them to the second one where further analysis is performed. This results in the increase in peak capacity and enhancement of the resolution of chromatographic peaks connected with limitation of coelution. It allows for the increase in the number of identified compounds in comparison to conventional GC-MS measurements. However, detection of the substances having boiling point above 400°C is still questionable [56]. Schneider et al. [64] applied two dimensional gas chromatography with fast-quadrupole mass spectrometry detector (GC x GC/qMS) to the analysis of polar compounds which were extracted from the bio-oil formed during fast pyrolysis of sawdust. They confirmed that the use of GC x GC/qMS allowed the identification of about 130 products based on their retention indexes and proved the considerable increase in peak capacity and resolution of chromatograms in comparison with GC/qMS. GC x GC connected with TOFMS detector has been also applied by Torri et al [60] to the characterization of bio-oil formed during fast and intermediate pyrolysis of softwood and hardwood forest industry residues.

The next group of methods of the analysis of bio-oil composition is based on the application of liquid chromatography (LC). One of the drawbacks of this technique is connected with lower separation ability in comparison to that achieved in the case of gas chromatography measurements. However, owing to the application of liquid chromatography, it is possible to detect heavier compounds present in bio-oil, which are not detectable during GC analysis. Moreover, LC allows for analysis of the compounds having higher polarity and gives opportunity to use a wide group of stationary phases possessing various selectivity [65].

Tomasini et al. [57] applied nanoLC coupled with mass spectrometer with mass ionization (EIMS) to determination of the composition of aqueous phase formed in fast pyrolysis of coconut fibers, sugar cane straw, and sugar cane bagasse. The obtained results suggested that owing to the reduction of the volume of eluate achieved by an application of nanoLC column, it was possible to directly introduce the liquid sample to the mass spectrometer. It was connected with considerable reduction in the volume of analyzed material which enabled the generation of vaporized mobile phase volume being compatible with that which could be analyzed by MS detector. On the other hand, the authors proved that the use of two-dimensional liquid chromatography (LC x LC), due to different retention mechanisms in each of the used chromatographic columns, allowed for an increase in peak capacity being responsible for better separation of the analyzed compounds and increase in the number of identified chemicals. Those findings were also confirmed by Le Masle et al. [63].

Literature shows that performance of liquid chromatography in determination of bio-oil composition can be enhanced by the application of centrifugal partition chromatography (CPC) or supercritical fluid chromatography (SFC) [58,59,65]. The use of CPC allowed for initial separation of bio-oil into fractions characterized by different solubility without sample loss at moderate temperature which increased overall efficiency of the analysis. On the other hand, SFC can combine the advantages of both gas (for example – low fluid viscosity or easy diffusion of solutes) and liquid chromatography (i.e. – separation of polar compounds, analysis of low volatile molecules) increasing the number of bio-oil components which can be detected with the use of this method.

2.4 Application of multivariate analysis

The results presented in the previous chapters confirmed that the interpretation of chromatograms obtained for bio-oil formed in fast pyrolysis of biomass is very difficult. The large amount of data hinders finding valuable information and makes analysis time consuming. It is especially important during investigation of a large number of variables (for example – temperature and time of pyrolysis, type of feedstock and its pretreatment method or effect of catalysts).

Py-GC/MS, the most common method for the analysis of products of fast pyrolysis of biomass, allows for detection of hundreds of chemical compounds, but only part of them can be unambiguously identified using the MS detector. This may be connected with co-elution of pyrolysis products on the gas chromatography column, lack of data in MS database or low concentration of considerable group of formed substances [68]. Moreover, the changes in the intensity of signal given by a MS detector may by characteristic for individual components of bio-oil. Therefore, this method is usually used for indication of trends but not for fully quantitative measurements. However, an interpretation of the results obtained with the use of Py-GC/MS can be simplified by the application of multivariate analysis [69,70].

It is known that principal component analysis (PCA) is one of the most popular statistical methods which can be useful for the characterization of the changes in composition of bio-oil formed during high temperature decomposition of lignocellulosic feedstock. The use of PCA allows for the extraction of information from very complex data sets due to the possibility of the transformation of a large number of possibly correlated variables into uncorrelated ones. The number of the latter is much smaller, and they are called principal components.

Xin et al. [68] used PCA for the interpretation of the results of Py-GC/MS study focused on the influence of biomass pretreatment (acid-leaching and torrefaction) on the distribution of products of fast pyrolysis of pinewood. The authors identified 45 chemical compounds, which were subsequently subjected to principal component analysis. It was observed that acid-leaching favored formation of levoglucosan and decreased the concentration of ketones in formed bio-oil. On the other hand, torrefaction led to a change in catechols and guaiacols contribution.

PCA was also used for monitoring the composition of bio-oil formed in pyrolysis of cassava rhizome conducted in the presence of various catalysts (zeolites, metal oxides or commercial materials) [71]. In this case, the performed experiments enabled observation of changes in the distribution of aromatic hydrocarbons, phenols, carbonyl products and carboxylic acids, among others.

On the other hand, Reyes-Rivera et al. [72] applied multivariate analysis for the results obtained in Py-GC/ MS studies of spines in Cactaceae. Principal component analysis (PCA), hierarchical clustering analysis (HCA) and hierarchical clustering on the principal components with k-means partition (HCPC) were applied in this case. The combined analysis allowed for identification of a large number of compounds formed during decomposition of the feedstock (derivatives of carbohydrates and lignin or N-compounds) and determination of their abundance patterns. Due to that the classification of lignocellulosic matrix originating from various species was possible.