**2. Material and methods**

of the interaction between the RF waves and the ions/molecules of the product [11]. Therefore, heat flows from inside the product to the outside, unlike conventional heating methods in which heat is transferred from the heating medium to the product via conduction or convection [8]. RF heating has the following advantages: 1) uniform electric field strength inside the application chambers, therefore preventing uncontrolled heating, overheating, local hot spots, and product degradation; 2) large penetration depth (10-30 m); and 3) higher energy efficiency [11-13]. RF has been successfully applied to leather drying [13], quantification of hydroxycin‐ namic acids and lignin in perennial forage and energy grasses [14], thermal therapy [15], and in other research fields such as food processing (blanching, tempering, pasteurisation, sterilisation) and medicine [11]. Recently, Izadifar et al. (2009) [12] demonstrated that RF can

be used for the extraction of podophyllotoxin from rhizomes of *Podophyllum peltatum*.

Fir (*Pseudotsuga menziesii*) at a reaction temperature of 200-220o

pretreatment of biomass.

304 Biofuels - Status and Perspective

molecular

Steam explosion is operated by introducing the feedstock into the reactor and heating under steam pressure (2000-5000 kPa; 200-260°C) for a few minutes [16]. Steam explosion induces chemical effects because "water itself acts as an acid at high temperatures" [17]. The sudden thermal expansion involved in the termination of the reaction causes the particulate structure of the biomass to open up [17]. Lam et al. 2011 [18] investigated the steam explosion of Douglas

min. Excoffier et al. (1991) [19] and Ferro et al. [20] (2004) have applied steam explosion on lignocellulosic biomass. These authors reported that steam explosion is effective in the

The traditional methods of determining the chemical compositions of biomass involve gravimetric and analytical procedures. These procedures are time-consuming, laborious and expensive to perform, with low sample throughput and often results in a corresponding degradation of natural polymers [1, 21-22]. In contrast, procedures involving infrared (IR) spectroscopy are useful tools in rapidly extracting information about the structure of biomass constituents and the associated chemical changes resulting from various biomass treatments [1]. Infrared spectroscopy offers researchers an alternative method that is easier, robust and rapid. Fourier transform infrared (FTIR) spectroscopy has been successfully applied in a variety of species in wood surface characterization, for estimating the carbohydrate and lignin components [1, 22]. The majority of the carbon based molecules in plants and animals alike are highly active in the IR. FTIR is associated with superior spectral resolution and provides information on the fundamental molecular vibrations [22]. This permits better discrimination of structural and compositional differences, and often better structural interpretation [22]. Furthermore, FTIR analysis requires only small amounts of biomass plant material, which helps when screening samples available in limited quantities [22]. Applications of FTIR have also been found in biomedical weld and medicine research such as cancer and bone [23-24].

Mid-infrared spectroscopy and near infrared reflectance spectroscopy (NIRS) are the two types of IR spectroscopy that have found application for the measurement of chemical composition in lignocellulosic biomass. NIRS has been used in the prediction of the chemical composition in bulk plant samples. While mid-IR spectroscopy in contrast to NIRS predicts the fundamental

C and a retention time of 5-10

### **2.1. Material procurement and preparation**

Barley straw of the "Xena" variety was grown in Maymont, SK (56.667°N, 107.794°W) and obtained from RAW Ag Ventures Limited (Maymont, SK) in October 2009. To increase the surface area of the biomass, the straw was ground using a hammer mill (Model No. GM13688, Glen Mills Inc., Maywood, NJ) with screen size of 1.6 mm. A dust collector (House of Tools, Model no. DC-202B, Saskatoon, SK) was connected to the outlet of the hammer mill to control dust during operation and to provide flow of the biomass in and out of the hammer mill. The initial moisture content of the straw was 8.09% (wet basis). The moisture content was measured based on ASABE standard method, ASAE S358.2 (2008). As a comparison between pretreat‐ ment methods, the 1.6 mm biomass grind was subjected to two different pretreatment methods: radio frequency (RF)-alkaline pretreatment using a RF machine (1.5 kW & 27.12 MHz laboratory dryer, Strayfield, Theale, Reading) in a blown glass reactor (volume 4.25 liters) stationed in University of Saskatchewan, Saskatoon, SK, and steam explosion (SE) pretreat‐ ment located at the Clean Energy Research Center, University of British of Columbia, Depart‐ ment of Chemical and Biological Engineering, Vancouver, BC. The material and operating variables considered in both methods of pretreatment are shown in Table 1 (a-b). Each pretreatment was performed in two replicates. For more details on the RF-alkaline and SE pretreatment, see [8-9].


**Table 1.** a) RF-alkaline material and operating variables using blown glass reactor; (b) SE material and operating variables with corresponding levels

### **2.2. Chemical composition analysis of lignocellulosic biomass**

The chemical composition analysis of the RF-alkaline and SE pretreated biomass grind was performed using the National Renewable Energy Laboratory standard (NREL) [30] at a laboratory facility at the Agriculture and Agri-Food Canada, Saskatoon, SK. Each sample was replicated twice. The NREL standard uses a two-step acid hydrolysis to fractionate the biomass into forms that are more easily quantified. The first step uses 72% H2SO4, while the second step uses 4% H2SO4. The lignin fractionates into acid insoluble and acid soluble material. The acid insoluble lignin is the residue (remaining solids) from the hydrolysis suspension. Acid-soluble lignin moieties were quantified using the Waters Acquity Ultra Performance Liquid Chroma‐ tography–MS system (Acquity 2004-2010, Waters Corp., Milford, MA), which has the capa‐ bility of separating and quantifying the various lignin components. The complex carbohydrates are hydrolyzed into monomeric forms (xylose, arabinose, mannose, glucose, and galactose) and subsequently quantified using UPLC-MS. The percentage hemicellulose was obtained by adding up the percentage xylose, arabinose, mannose, and galactose, while the percentage glucose was assigned to percentage cellulose. Further details on the material preparation, physical characteristics of the biomass grind, radio frequency alkaline technique, steam explosion process, and the chemical composition analysis using the NREL standard can be obtained in the research studies of [8-9]. The pretreated and non-treated samples were further ground to screen size of 0.354 mm using a precision grinder (Falling Number, Model No. 111739, Huddinge, Sweden).

### **2.3. Preparation of reference materials of known concentration**

laboratory dryer, Strayfield, Theale, Reading) in a blown glass reactor (volume 4.25 liters) stationed in University of Saskatchewan, Saskatoon, SK, and steam explosion (SE) pretreat‐ ment located at the Clean Energy Research Center, University of British of Columbia, Depart‐ ment of Chemical and Biological Engineering, Vancouver, BC. The material and operating variables considered in both methods of pretreatment are shown in Table 1 (a-b). Each pretreatment was performed in two replicates. For more details on the RF-alkaline and SE

> 1:5 - 110 g biomass and 550 g NaOH solution 1:6 - 100 g biomass and 600 g NaOH solution 1:7 - 100 g biomass and 700 g NaOH solution 1:8 - 90 g biomass and 720 g NaOH solution

> > C

(a)

(b)

The chemical composition analysis of the RF-alkaline and SE pretreated biomass grind was performed using the National Renewable Energy Laboratory standard (NREL) [30] at a laboratory facility at the Agriculture and Agri-Food Canada, Saskatoon, SK. Each sample was replicated twice. The NREL standard uses a two-step acid hydrolysis to fractionate the biomass into forms that are more easily quantified. The first step uses 72% H2SO4, while the second step uses 4% H2SO4. The lignin fractionates into acid insoluble and acid soluble material. The acid insoluble lignin is the residue (remaining solids) from the hydrolysis suspension. Acid-soluble lignin moieties were quantified using the Waters Acquity Ultra Performance Liquid Chroma‐ tography–MS system (Acquity 2004-2010, Waters Corp., Milford, MA), which has the capa‐ bility of separating and quantifying the various lignin components. The complex

**Table 1.** a) RF-alkaline material and operating variables using blown glass reactor; (b) SE material and operating

pretreatment, see [8-9].

306 Biofuels - Status and Perspective

Temperature (o

variables with corresponding levels

**Variables Levels** Hammer screen size 1.6 mm NaOH solution concentration 1% w/v

Soaking time 1 h

Temperature 70, 80, and 90o

C) 140, 160, and 180

**2.2. Chemical composition analysis of lignocellulosic biomass**

Residence time 20 minutes

Moisture Content (% mass fraction of water) 8, 30, and 50 Retention Time (min.) 5 and 10

**Variables Levels**

Biomass: NaOH solution ratio 1:4 - 110 g biomass and 440 g NaOH solution

In order to develop a predictive model that will be rapidly used for the quantitative prediction of the chemical composition contained in the RF-alkaline and SE pretreated biomass, pure cellulose (microcrystalline powder), hemicellulose (xylan from birch wood), and lignin (hydrolytic) powders (Sigma-Aldrich Canada Ltd., St. Louis, MO) were mixed in different proportions (Table 2). These were used as reference spectra to determine the relationship between the respective quantity in the mixture and the representative sample FTIR spectra. The FTIR wavenumber-dependent instrumental effects were corrected by using carbon black reference spectrum.


C, H, and L represent Cellulose, Hemicellulose, and Lignin, respectively

**Table 2.** Reference materials: Pure cellulose, hemicellulose, and lignin mixtures used to obtained the reference spectra (adapted from [21])

### **2.4. Fourier Transformed Infrared Photoacoustic Spectroscopy (FTIR-PAS)**

Intensity of spectra generally increases as the particle size decreases [30]. To avoid moisture interference, the biomass samples were further dried using the forced-air convection dryer [31] (Shaw et al. 2007) set at 40o C for 48 h. Photoacoustic intensities are lower for samples with high moisture content. This might be due to lower efficiency of heat transfer between the moist cellulose surface and the carrier gas [32]. The Infrared data/spectra of the reference materials and biomass samples were collected using Mid-IR beamline (01B1-1) with energy range of 4000-400 cm-1, at the Canadian Light Source Inc. (CLS, University of Saskatchewan, Saskatoon, SK). The beamline has a MTEC Model 300 photoacoustic cell (MTEC Photoacoustic Inc., Ames, IA) for FTIR-PAS of bulk samples. The sample cup was filled with reference biomass sample (52-75 mg, depending on the pretreatment type and combination) and purged with helium gas to remove water vapor and CO2. Helium gas is also needed in the medium because of it sound propagating properties. The collected FTIR spectra of the reference materials and biomass samples were recorded using Globar source (silicon carbide rod). When the radiation is incident on the sample, the energy of the Infrared beam is being absorbed by the sample layer. The photoacoustic signal is generated by thermal expansion of the gas caused by heat associ‐ ated with the thermal wave emanating from the sample. The photoacoustic signal is carried by a carrier gas (Helium) to a microphone which is transferred to the FTIR electronics (detector) for processing; this ultimately produces the needed spectrum [33-34]. The spectrum for each reference material and biomass samples were recorded separately averaging 64 interferograms (number of scans) collected from wavenumbers of 2000-400 cm-1 at a resolution of 4 cm-1. The higher the number of scans the better the signal (lesser noise). Stuart (1997) [36] reported that the signal-to-noise ratio (SNR) is proportional to the square root of the number of scans, n (SNR α n0.5). Therefore, the higher the number of scans, the higher is the SNR. Resolution of 4 cm-1 was used to be able to discriminate between too close overlapping peaks, help increase the SNR and subsequently obtain higher resolution. The OPUS 6.5 (Brucker Optics Inc. Billerica, MA) software was used for the collection of the FTIR-PAS spectra. Three replicates were performed for each reference and biomass samples.

### **2.5. Determination of concentration**

Beer-Lambert discovered that the amount of light transmitted by a solid sample was dependent on the thickness of that sample [35]. The Beer-Lambert law which can be applied to all electromagnetic radiation, states that the absorbance of a material is directly proportional to the thickness and concentration of the sample as shown: A=εCL. A=absorbance of the material, C=concentration, L=pathlength of the sample, and ε=constant of proportionality, which is referred to as the molar absorptivity [35]. Infrared spectra, particularly in the spectra of solid samples are often associated with the presence of asymmetric bands. As such, peak height cannot be used for the quantitative analysis of the spectra, because the baseline will vary from sample to sample. Instead, peak-area measurements should be used [35].

### **2.6. Quantitative and qualitative analysis of the FTIR-PAS spectra**

The two quantities of greatest interest in virtually any type of spectroscopy are, of course, band positions (wavenumbers) and intensities, the former generally conveying qualitative infor‐ mation, the latter quantitative [30]. Therefore, these two variables were used for the FTIR data analysis.

### *2.6.1. Spectrum manipulation*

**2.4. Fourier Transformed Infrared Photoacoustic Spectroscopy (FTIR-PAS)**

(Shaw et al. 2007) set at 40o

308 Biofuels - Status and Perspective

performed for each reference and biomass samples.

**2.5. Determination of concentration**

Intensity of spectra generally increases as the particle size decreases [30]. To avoid moisture interference, the biomass samples were further dried using the forced-air convection dryer [31]

moisture content. This might be due to lower efficiency of heat transfer between the moist cellulose surface and the carrier gas [32]. The Infrared data/spectra of the reference materials and biomass samples were collected using Mid-IR beamline (01B1-1) with energy range of 4000-400 cm-1, at the Canadian Light Source Inc. (CLS, University of Saskatchewan, Saskatoon, SK). The beamline has a MTEC Model 300 photoacoustic cell (MTEC Photoacoustic Inc., Ames, IA) for FTIR-PAS of bulk samples. The sample cup was filled with reference biomass sample (52-75 mg, depending on the pretreatment type and combination) and purged with helium gas to remove water vapor and CO2. Helium gas is also needed in the medium because of it sound propagating properties. The collected FTIR spectra of the reference materials and biomass samples were recorded using Globar source (silicon carbide rod). When the radiation is incident on the sample, the energy of the Infrared beam is being absorbed by the sample layer. The photoacoustic signal is generated by thermal expansion of the gas caused by heat associ‐ ated with the thermal wave emanating from the sample. The photoacoustic signal is carried by a carrier gas (Helium) to a microphone which is transferred to the FTIR electronics (detector) for processing; this ultimately produces the needed spectrum [33-34]. The spectrum for each reference material and biomass samples were recorded separately averaging 64 interferograms (number of scans) collected from wavenumbers of 2000-400 cm-1 at a resolution of 4 cm-1. The higher the number of scans the better the signal (lesser noise). Stuart (1997) [36] reported that the signal-to-noise ratio (SNR) is proportional to the square root of the number of scans, n (SNR α n0.5). Therefore, the higher the number of scans, the higher is the SNR. Resolution of 4 cm-1 was used to be able to discriminate between too close overlapping peaks, help increase the SNR and subsequently obtain higher resolution. The OPUS 6.5 (Brucker Optics Inc. Billerica, MA) software was used for the collection of the FTIR-PAS spectra. Three replicates were

Beer-Lambert discovered that the amount of light transmitted by a solid sample was dependent on the thickness of that sample [35]. The Beer-Lambert law which can be applied to all electromagnetic radiation, states that the absorbance of a material is directly proportional to the thickness and concentration of the sample as shown: A=εCL. A=absorbance of the material, C=concentration, L=pathlength of the sample, and ε=constant of proportionality, which is referred to as the molar absorptivity [35]. Infrared spectra, particularly in the spectra of solid samples are often associated with the presence of asymmetric bands. As such, peak height cannot be used for the quantitative analysis of the spectra, because the baseline will vary from

sample to sample. Instead, peak-area measurements should be used [35].

C for 48 h. Photoacoustic intensities are lower for samples with high

There are techniques that assist in both qualitative and quantitative interpretation of spectra. OriginPro software (Data analysis and graphing Version 8.6, OriginLab Corporation North‐ ampton, MA) was used for the spectrum manipulation, quantitative, and qualitative analysis of the FTIR-PAS Spectra:

### *2.6.2. Baseline correction, subtraction, and rescaling*

The Baseline Mode is a tool for choosing a baseline mode and creates the baseline. A userdefined baseline treatment was applied in this analysis. A common flat baseline of 0.012, joining the points of lowest absorbance (via fitting Pro) on the peak was selected and applies across all the reference materials and biomass sample spectra. Thereafter, baseline subtraction was performed from the input data, such that the absorbance difference between the selected baseline and the top of the band is then used. This helps to improve the accuracy of the peak finding. Rescaling of the baseline to zero was subsequently performed, so that all the spectra will have a common origin of zero.

### *2.6.3. Smoothing*

Smoothing is a signal processing technique typically used to remove or diminish noise from signals/spectrum. After a spectrum is smoothed, it becomes similar to the result of an experi‐ ment obtained at a lower resolution [35, OriginLab Corporation manual]. The features are blended into each other and the noise level decreased. A smoothing function is basically a convolution between the spectrum and a vector whose points are determined by the degree of smoothing one wish to apply [35]. There are multiple smoothing methods that work differently depending on the nature of the signal and the noise contained in the signal. Each method offers a different performance. In this present analysis, Savitzky-Golay was used. The Savitzky-Golay filter method performs a local polynomial (order of 2) regression around each point, and creates a new, smoothed value for each data point. This method is superior to other methods (such as adjacent averaging) because it tends to preserve features of the data, such as peak height and width, which can be "washed out" by adjacent averaging (OriginLab Corpo‐ ration, Northampton, MA, 2012). To increase the smoothness of the result, one can increase the "window size," used in each local regression (OriginLab Corporation, Northampton, MA, 2012). As such, the window size in this analysis was increased from one to two. But for very large window size, adjacent averaging may depart from the input signal too much, whereas Savitzky-Golay can still preserve the overall profile (OriginLab Corporation, Northampton, MA, 2012).

### *2.6.4. Peak finding settings*

Manual peak editing was performed so as to effectively select the required respective peaks. The second (2nd) derivative was used to search for all the hidden peaks and heavily overlapped bands included in the spectrum data. Differentiation is used to resolve and locate peaks in an envelope. Sharp bands are enhanced at the expense of broad ones, and this may allow easier selection of a peak, even when there is a broad band beneath it [35].

### *2.6.5. Characteristic peak assignment*

The characteristic wavenumbers for pure cellulose, hemicellulose, and lignin listed in Table 3, were used for the peak assignment. Five characteristic peaks were identified for the pure cellulose, six characteristic peaks for pure hemicellulose, and six characteristic peaks for pure lignin (Figure 1a-c). The number of the peaks identified for the respective reference materials depends on the mixture of the reference materials. All seventeen peaks were identified for the treated and non-treated biomass samples.

### *2.6.6. Peak integration*

To obtain quantitative values from the area under the manipulated spectrum/peaks, the area under the respective peaks were integrated and output into excel file.

### *2.6.7. Normalization of photoacoustic infrared spectra*

Prior to the spectrum manipulation, the FTIR-PAS biomass sample spectra were corrected for wavenumber-dependent instrumental effects by dividing the reference carbon black ("back‐ ground") spectrum intensity. This strategy implicitly assumes that the stability of the instru‐ mentation used is adequate to ensure reliable results, even though the sample and reference spectra were collected at different times [30]. Carbon black is featureless, in the sense that it does not show any major characteristics peak [30] Photoacoustic (PA) cell intensities varied with sample packing in the PA cell [36]. Stuart (1997) [35], also reported that absorbance varies linearly with the sample thickness. Therefore, the effect of bulk density of the reference materials and biomass samples was corrected by dividing the integrated areas with respective mass of the reference materials and biomass samples contained in the PA sample cup.

The model was further standardized by normalizing the corrected integrated area data from 0 to 1. This was performed by dividing individual reference materials and biomass samples corrected integrated area data by corresponding maximum corrected integrated area data value. The aforementioned steps were performed for the three major components of lignocel‐ lulosic biomass (cellulose, hemicellulose, and lignin). Therefore, this normalization approach ensures that the predictive model is adaptable for quantitative analysis of FTIR-PAS spectra obtained for any lignocellulosic biomass.
