Thermal combustion and oxygen chemisorption of wood exposed to low temperature long term heating 5

This study is concerned with the interpretation of such oxygen chemisorption experiments for cellulosic chars, through which the energetics of chemisorption process is elucidated in orde

Chapter Five: Gas-solid Reaction

In the study of spontaneous heating and ignition of wood, the role of gas-solid reaction plays a critical role in making a material oxidizable and hence liable to self-heating It is generally agreed that carbon gasification involves, first, chemisorption on the carbon to form carbon surface complexes, followed by desorption of these surface complexes as carbon monoxide or dioxide (Teng and Hsieh 1999) Characterization of either chemisorption or desorption process is important in the overall understanding of the role of gas-solid reaction or oxidation in the materials (Degroot, Osterheld and Richards 1991, Wang, Dlugogorski and Kennedy 1999, Wang, Dlugogorski and Kennedy 2003a)

This study is interested in the investigation of chemisorption as the initial step leading to gasification, and it being a significant factor in controlling chemical reactivity and heat release in the overall context of spontaneous heating and ignition This study is concerned with the interpretation of such oxygen chemisorption experiments for cellulosic chars, through which the energetics of chemisorption process is elucidated in order to characterize the reactivity of these cellulosic chars that have been created at low temperatures This chapter presents an analytical model to compute the energetics of chemisorption, in order to characterize reactivity of wood chars Experimental

preparation and procedures for chemisorption of cellulosic wood chars also warrant special consideration as chemisorption on cellulosic chars differ from the graphitic carbon such as coal in that these cellulosic chars are far more reactive In addition to the energetics of chemisorption, this study also investigates the changes in functional groups arising from oxidation via Fourier-transform Infrared (FTIR) Spectroscopy Experimental procedures using FTIR for wood chars are discussed in this chapter

5.1 Theoretical and experimental studies of chemisorption

In the description of the heterogeneous gas-solid reactions, including carbon-oxygen reaction, the reaction rate is generally assumed to be proportional to the accessible surface area (Khan, Everitt and Lui 1990, Teng and Hsieh 1999) One of the main interests in chemisorption study is to use oxygen chemisorption capacity to characterise the active sites in carbon, by way of activation energy, in the attempt to quantify reactivity of a given carbonized substrate (Teng et al 1999) Correlation is made possible on assumptions that the substance possesses a constant a number of reactive sites per unit surface area, and the area have the same distribution activity (Carpenter and Giddings 1964, Carpenter and Sergeant 1966, Bansal, Vastola and Walker 1970)

The fact that the activation energy for chemisorption is not constant but varies with surface coverage have since proliferated a large number of studies in making different attempts to quantify the possibility of a variety of sites so as to determine the overall kinetics of the chemisorption process For instance, Bansal, Vastola and Parker (1970)

have proposed that there were discrete types of active sites on carbon surface, relating to the number of breaks observed in a single Elovich plot in a single chemisorption run They have suggested ways to identify the different regions of active sites in order to calculate the activation energies

More efforts have also been devoted to develop very complex kinetic models to compute activation energy as a distribution, rather than discrete values Waters Squires and Laurendeau (1986) have developed a kinetic model using Collision theory and the Freundlich isotherm to characterise activation energy by a continuous distribution for turbostratic amorphous carbon Floess, Lee and Oleksy (1991) also have developed a distributed activation energy site model for microporous carbon Suuberg (1983) proposed the idea of Gaussian distributed activation energy models via approximate solution for carbon char Du, Sarofim and Longwell (1991) then continued to improve the Gaussian form of activation energy distribution for oxygen adsorption on carbons, by the use of an approximation technique suggested by Suuberg (1983), to determine the activation energy distribution of surface complexes Teng et al (1999) also purported a distributed kinetic parameter model similar to that of Floess et al (1991), except that their distribution function did not have a flat end, but resembled the tail end of a Gaussian form These distributed activation energy models generally hold a basic assumption that the energy distribution for oxygen chemisorption is continuous, and in the main seek to maintain generality in the allowable forms of the distribution function

Unfortunately the use of complex kinetic models such as those described above may be unattractive to those whose main concern is the modeling of combustion, rather than pyrolytic processes Even in these complex distributed models, the precise nature of the pathways to final surface oxides is not considered in the analysis, rendering the treatment pseudo-analytical (Wang, Dlugogorski and Kennedy 2003c) The creation and destruction of the rate-controlling intermediate steps and active sites are also ignored in these models in order to make them mathematically tractable The complicated models, albeit its complexity, thus examine only the irreversible oxygen adsorption The use of Gaussian distributed activation energy models inevitably results in more cumbersome comput ational models; they certainly require more comput ational resources to handle than simple first-order models

The chemisorption kinetics has by far defied any plausible theory, without being pseudo-analytical The empirical methods are still after all the most probable way to correlate the experimental data in order to calculate the instantaneous rates of adsorption which can be used in a kinetic analysis Several empirical methods have been put forth, such as parabolic diffusion equation (Allardice 1966), power-law form rate equation (Khan et al 1990) and exponential function (Wang, Dlugogorski and Kennedy 2002), in addition to the Elovich equation Both the parabolic diffusion equation and simple power-law model have failed to account for the chemisorption kinetics, except the exponential function proposed by Wang et al (2002) is able to provide a very good fit to the experimental data

of bituminous coal, where the rate of oxygen consumption is described as the sum of one constant and two exponential functions of time Khan et al (1990) other attempts at using

empirical models such as shrinking area model and diffusion model either was not able to predict data over the entire range with one constant or encountered significant errors at low surface coverage They concluded that Elovich equation still offers the best empirical method to describe the chemisorption kinetics

The Elovich kinetic law is an empirical model that has found wide application in kinetic studies of sorption process, particularly in the field of chemisorption (Laine, Vastola and Walker 1960) This kinetic model is able to account for the variation of the energetics of chemisorption with the extent of surface coverage (Allardice 1966)

Though first developed as an empirical model, the application of such empirical kinetic law has been thoroughly examined on theoretical grounds Taylor and Thon (1952) have mathematically shown that adsorption process conforms to the Elovich equation Landsberg (1955) on the other hand developed very elaborate chemical models to justify its form Low (1960) presented a comprehensive discussion of the use of Elovich equation in describing the chemisorption kinetics of gases on solids Elovich equation has continuously been applied to many experimental studies of chemisorption for carbonized char (Carpenter and Giddings 1964, Carpenter and Sergeant 1966, Harris and Evans 1975, Sevenster 1961, Suuberg, Wojtowicz and Calo 1989, Khan et al 1990)

In this study, a distributed activation energy model has been developed based on Elovich kinetic law to elucidate the energetics of chemisorption for low-temperature carbonized

wood char, in order to characterize the reactivity of these low-temperature, preheated wood chars

5.1.1 Analytical method to derive Elovich kinetics

Distributed Activation Energy Model

For a chemisorption process that conforms to Elovich kinetics, the rate of oxygen uptake with respect to time can be described as

( )1 ln ( )1 ln( 0)

where q is the amount of oxygen adsorbed; a and b are Elovich constants and t is an 0 integration constants equal to 1 ab Substituting t0 =1 abinto Equation (5.1) yields

( )1 ln ( )1 ln 1( )

If abtis >>1, a plot of vs lnq t should be a straight line, with a slope of 1

b At each isothermal run, values of a and b can be calculated from Equation (5.2)

To derive the activation energy, the instantaneous rates are to be calculated for adsorption

at different amounts of surface coverage The instantaneous rate is obtained by

substituting the Elovich constants a and b into the following differentiated form of Elovich equation

exp( )

dq

Taking logarithm of both sides of Equation (5.3), and let r=dq dt yields

Differentiating Equation (5.4) with respect to 1 T gives

(1 ) (1 ) (1 )

q

It has been suggested that lnd a d(1T)= −E c/R, where E is the apparent activation c

energy and R is the universal gas constant (Bansal, Vastola and Walker 1970) Rearranging Equation (5.5) yields

( )ln ( )

(1 ) (1 )

q

Equation (5.6) suggests a distribution of activation energies for various surface coverages

with associated values of constants a and b

5.1.2 Chemisorption experiments

Chemisorption experiments of wood chars warranted special attention as wood chars were much reactive than coal chars, requiring different preparation and testing procedures Physical adsorption and chemisorption of oxygen on char samples were measured thermogravimetrically in SDT 2960 simultaneous DTA-TGA Thermogravimetric Analyzer © TA Instruments

5.1.2.1 Materials and sample preparation

Kapor, a widely used hardwood species was used to create wood chars under low temperature, in both long-term preheating in air and rapid pyrolysis in nitrogen in thermogravimtric analyzer The untreated Kapur is of empirical formula C3.9H6.5O2.9 Two types of wood chars were used in the experiments: air-preheated wood chars, or aerobically produced chars that have been heated in air for an extended period in oven, and inert wood chars that were anaerobically created in nitrogen atmospheres

5.1.2.2 Air-preheated chars

To prepare air-preheated chars, wood blocks measuring 32 x 32 x 102mm were heated isothermally in air at heat treatment temperature (HTT) of 140°C for 50 days and HTT of

obtained by grinding the wood discs sliced from the wood blocks and sieved to the desired particle size of 200 to 300 μm The wood powders were stored at 50%RH at 23°C

in a desiccator These wood chars were designated as K140 and K150 respectively

5.1.2.3 Inert-heated chars

Inert-heated samples were prepared by heating fresh wood powder at heat treatment temperature (HTT) 300°C for 1.5 minutes in nitrogen at atmospheric pressure using thermogravimetric analyser (TGA) The 1.5 minutes of pyrolysis time was selected to approximate the condition of char formation by a smouldering combustion front (Shafizadeh and Bradbury 1979) These nascent wood chars were immediately subjected

to chemisorption after a few minutes of holding time in TGA to allow the respective chemisorption temperatures to equilibrate This process eliminated the formation of surface oxides arising from separate char preparation and chemisorption runs These inert-heated wood chars were designated as K300

5.1.2.4 Elemental analysis

The elemental composition of the respective wood chars are analysed and shown in Tables 5.1 to 5.4

Table 5.1 Elemental Analysis of Untreated Wood

* composition obtained by difference

Table 5.2 Elemental Analysis of Inert Wood Chars (K300-1.5mins-N 2 )

* composition obtained by difference

Table 5.3 Elemental Analysis of Preheated Wood Chars in Air for 30 days at 150°C)

(K150-30days-Air)

* composition obtained by difference

Table 5.4 Elemental Analysis of Preheated Wood Chars in Air for 50 days at 140°C)

(K140-50days-Air)

* composition obtained by difference

5.1.2.5 Argon purge

Argon was used to purge the balance enclosure at 750 sccm Displacing the undesired gases by introducing argon from the top produced a favourable density gradient in the reactor during the purging process, since argon is the lighter gas and is therefore less likely to mix with the displaced gas Argon purging was carried out for approximately five minutes to allow the sample weight reading to stabilise in a new buoyancy environment

5.1.2.6 Nitrogen purge

The initial argon purge rate was adjusted to 300 sccm Nitrogen was introduced into the reaction chamber through the reaction gas inlet at 250 sccm A total reaction gas flow rate was maintained at 250 sccm and 300 sccm respectively The nitrogen gas used has less

than 20 ppm impurities and less than 10 ppm oxygenated species, i.e O2, H2O, CO2 and

CO These low concentrations of oxygenated species were necessary to minimize char oxidation during the nitrogen purge and the subsequent heat treatment Nitrogen purge was maintained for 60 minutes at 25°C During this period, the reaction chamber would

be completely purged, the sample would lose some moisture and the gas-concentration sensors would be completely flushed

5.1.2.7 Empty pan correction

Because there would be small but measurable weight changes when a platinum empty pan was subjected to chemisorption conditions, all weight data were corrected for weight changes by running a blank experiment with an empty pan The baseline weight data would be used to subtract from each data point in the subsequent chemisorption experiments

5.1.2.8 Experimental Procedures

In a typical chemisorption experiment, the system was first purged with inert gases after the loading of sample to remove any air introduced The sample was pyrolysed to the required heat treatment temperatures (HTT) in the case of inert chars; for air-preheated chars, the chamber temperature was ramped directly from 25°C to the chemisorption temperatures In both cases, after the ramping and/or heat treatment segment, holding time was allowed to stabilize the furnace temperature to the required chemisorption

temperatures (CST) before the reaction gas was valved in to start the chemisorption experiment The chemisorption temperatures (CST) used in the experiment were: 74°C, 109°C, 139°C, 168°C, 185°C and 207°C The upper limit of the chemisorption temperature was limited by the possibility of simultaneous desorption of oxides of carbon, which normally occurs around 200°C (Allardice 1966)

For inert-heated samples, as soon as the chars were pyrolysed for 1.5 minutes at 300°C in TGA, the furnace temperature was immediately cooled to the chemisorption temperatures (CST) In each run, a few minutes of holding time were allowed for the chemisorption temperature to equilibrate and to establish the weight of the sample prior to chemisorption Chemisorption was initiated by valving air with a flow rate of 60 ml/min into the furnace that was maintained at isothermal chemisorption temperature The weight gain during chemisorption was monitored for 15 hours

For air-preheated samples, the same procedure and condition were adopted, where the air-preheated wood powders were heated from 25°C to the chemisorption temperature and the weight gain was observed for 15 hours However, there was no intercepting heating segment to clean the surface oxides before chemisorption experiment This was because the removal of surface oxides would require the heating of these wood chars to high temperatures Unlike high-temperature carbonized chars (Bradbury and Shafizadeh 1980b, Hshieh and Richards 1991) that were created at heat treatment temperatures (HTT) ranging from 375°C to 800°C, the low HTT of 140°C and 150°C for air-preheated chars used in this study were less stable and these low temperature chars remained very