56 KINETIC MODELS: KERR describe the dependence on both temperature and concentration of the reaction rates of all major wood components. Some of these are discussed below. 4. The Kerr model ‘One of the earlier investigations conducted to derive a kinetics-based model of kraft pulping was that of Kerr [117]. He rearranged equation (21) to the form u t a J kat (22) LAL) u ° in which {(L) expresses the relationship between the alkali concentration and the concentration of lignin remaining in the wood. This relationship has been observed to be well represented by two intersecting straight lines over much of the cook, shown in Fig. 43. Also, itis independent of the initial effective alkali concentration (except that the intercept is vertically displaced in accord- ance with the change in initial effective alkali concentration), as shown by the parallelism of the lines in the figure. Kerr substituted the ‘equations of these lines for f(L) in equation (22) and the defining equation for H-factor, equation (19), for the right-hand side. Integration then gave Ly, { i — S| by L + bya hy 1 L + ise = aH + ay (23) in which 4; = slope of alkali concentration vs lignin yield line for period of high alkali consump- tion, , = intercept of the same line, 4) = slope of alkali concentration vs lignin yield line for period of low alkali consump- tion, by, = intercept of the same line, 4; = constant related to the delignification rate constant in equation (21), and 4, = a constant added to correct for the failure of a plot of the left-hand side of equation (23) vs. H-factor to pass through the origin, The effect of changing the initial effective alkali concentration was incorporated into the 3 a g g (025% inital attective alkats X 16.75% intial effective aikati & EFFECTIVE ALKALI (gms/litre NaOH) x 8 8.8 3 o 5 10 15 2025 a LIGNIN (% of 0.0. woos) Fig. 43. Relation between effective alkall concen- tration and lignin removed [16,117]. model by noting that the separation between the intercepts of the two alkali concentration vs. lignin yield lines is independent of initial alkali concentration: by = by + as (24) and that the intercept of the first is linearly related to the initial effective alkali concentra tion, ¢;: by = age, + ay (25) This model predicts lignin yield (as a percent- age of the original wood weight) but not kappa number, which is related to lignin content based ‘on the weight of the pulp, and which therefore depends on pulp yield. The prediction of pulp yield requires thai the kinetics of carbohydrate dissolution be considered in addition to the kinetics of delignification. Kerr empirically circumvented this difficulty by using an earlier observation by Hartler [102] that, over a limited range of pulping variables, there exists a good correlation between pulp yield and lignin-to-car- ‘bohydrate ratio, Kerr conducted a series of laboratory cooks of Pinus radiata to evaluate the constants in the model and the constants defining the relation- ship between yield and lignin-to-carbohydrate ratio. Kappa numbers calculated from the model agreed with the observed values to within 1.5 units.Although apparently quite successful in pre~ dicting the outcome of laboratory cooks, this model was only applicable under conditions of fixed sulfidity, liquor-to-wood ratio and chip size. In a later paper, Kerr and Uprichard [120] described a useful simplification of the basic model as well as modifications allowing it to predict the effects of changes in the above three variables. The simplification resulted from the realization that the first term in equation (23) was always small in comparison to the second, suggesting that the alkali concentration-lignin yield relationship could be approximated by the bbulk delignification line over the complete range of delignification. Under this assumption the model becomes eg ay 26) 2 LF balay Sa 1 which was found to provide a very good fit for the experimental data. Varying the sulfidity of the white liquor had no significant effect on the relationship between effective alkali concentration and extent of delignification. Only a3, the coefficient of the H-factor, was affected. An approximately linear relationship between a; and $/(200-S) was observed, where S = sulfidity. Such a relation- ship had earlier been suggested by Edwards and Norberg [256] Changes in the liquor-to-wood ratio of the kind associated with variations in chip moisture content affected only one parameter — bo, the intercept of the alkali concentration vs. lignin yield line. Thus, the only effect of increasing the liquor-to-wood ratio at constant alkali charge is to slow down the cook by virtue of the decreased alkali concentration. This effect was easily incorporated into the model asa correlation with by. ‘The last and most empirical modification to the model was the inclusion of the effect of chip size. This variable, like sulfidity, had no effect on the alkali concentration-extent of delignification relationship but did affect the apparent rate constant, a;, The effect was incorporated into the ‘model as a linear relationship between chip size and a5, 5. The Edwards /Norberg model Edwards and Norberg (256) also produced a useful result by assuming that the rate of delignification is first order in both lignin and effective alkali. They recast equation (21) in dimensionless form to obtain KINETIC MODELS: EDWARDS/NORBERG 57 —(L*XC) Qn L — Lf(Lo — LA, Lf = content of “unremovable lignin” or “floor level” of lignin, based on wood, Lo = initial lignin content, Ct = C/Co, and the dimensionless time, @ is given by 8 = (kXCoXt) (28) in which k is the rate constant. Edwards defined the “tau factor”, which is proportional to the dimensionless time, 8. The tau factor can be written in terms of effective alkali, sulfidity and liquor-to-wood ratio as follows EA? eas 29) () (iw) o or in modified form for use at constant sulfidity as EA 30) % (t Jen £ A companion to equation (27) is {B] (dC*/d8)] = —(L*\C*) ey) which, together with equation (27), shows that L depends on only two dimensionless variables, 6 and B, the ratio of the amount of alkali present initially to that required for delignification. By this reasoning, the lignin yield at any time should betaine Deel ee a plot of experimental pulping data (as the product of kappa number and yield, which is Proportional to lignin yield) vs, the tau factor, The expected good correlation is observed, but no effect of alkali charge can be seen. It is apparently sufficiently in excess of the demand that f, the charge-to-demand ratio, does not have a significant effect. At lower alkali charges, the points were observed to lie significantly above the line. The plot can be linearized by plotting the reciprocal of the ordinate, as shown in Fig. 45. This model is apparently useful for correlating pulping data and for predicting the effects of changes in effective alkali charge or liquor to-wood ratio, but only within limits on effective alkali charge and sulfidity.(TOTAL PULP YIELD) 3 8 K+Y = (KAPPA NUMBER) * 100 200 300 400 500 600 "4/100 = (EA/L:W} + H/100 Fig. 44. Lignin yield data for kraft pulping ot ‘western hemlock plotted according to the model [286]. 0 100 200 300 400 500 600 "e/100 = (EA/L:Wy + H/100 Fig. 45. Lignin yleld data from Fig. 44, final correlation [256]. 6. The initial phase of delignification ‘The Kerr and Edwards and Norberg models consider pulping a simple reaction. However, as, mentioned earlier, the lignin in wood is structur- ally complex, and often does not behave as a single chemical compound. A manifestation of this in kinetic data for kraft pulping is the appearance of more or less sharp rate transitions as pulping proceeds. One such transition often occurs very early in the process and defines the end of a rapid initial delignification phase. In cooks where the temperature of the digester was rapidly increased to 150°C, for example, Wilder and Daleski [231] found that 20% of the lignin in loblolly pine shavings was removed within the first 5 min of the cook, after which the delignification reaction slowed down considera- 1D; Initial detigntication ‘8D: Bulk delignification {tn (lignin) o 2 3 3 ‘Cooking time Ly Fig. 46. Logarithmic rate plots for the Initial kraft delignification of western hemlock wood meal in the temperature range 100° to 150°C. The ‘ordinate Is given In natural logarithm of percent lignin on wood [123]. bly. In_a study of the kinetics of the initial delignification of industrial pine chips, Olm and Tistad [174] concluded that delignification during this part of the cook is first order in lignin and zero order in both hydroxide ion and hydrosulfide ion. In addition, they interpreted the relatively low activation energy of 40 kJ/mol tomean that the delignification rate is controlled by diffusion and not by the chemical reaction. Kondo and Sarkanen {123] showed that initial delignification of western hemlock woodmeal can be resolved into two kinetically distinguish- able phases (IDI and ID2), resulting in the dissolution of 13 and 11% of the original lignin, respectively. The second of these was shown to obey first-order kinetics with an activation energy of 73 kJ/mol, a value large enough to preclude the possibility of diffusion control. The data of Kondo and Sarkanen are shown in Fig. 46 and serve to illustrate the rapidity of initial delignification in comparison to the bulk phase. With regard to the mechanism of the rapid initial delignification, it has been postulated thats H3Cco 3 ss (CH}2804 ie a, > Hoe Pa Ms = darting structure Nysta'= mthyated structures Magoo. yt ga nn Me arom groupe HGocHs OCHS Masa yn ton Eymemyaten of Boneyle q myaromy groupe H3co” (OCH) sean se gn a Ma: As Ha, bulk phase further retarded by ‘athyiation of ¥ — hydroxy! groups Methylations of aryl ether structures (SS) in native lignins ‘Suttate pulping of wood shavings: 1 ignin dissolved 888 8 8 a8 8 intial phase ‘bak phase Fp #7: Course oF senetnnee ot Stale et presented wed Gherhioy ering IreRtuying60 BULK DELIGNIFICATION the reactive structures are a- and f-ethers containing free phenolic hydroxyl groups. In support of this, Ljunggren [137] has identified similarities in the kinetics of degradation of the corresponding model compounds and of the initial phase of delignification. Perhaps more ‘convincing is the evidence presented by Gierer and Noren [74]. These workers, as illustrated in Fig. 47, showed that methylation of wood with diazomethane, a reagent known to specifically methylate phenolic hydroxyl groups, eliminates the initial phase. Methanolic hydrochloric acid, ‘on the other hand, is specific for benzylic hydroxyl groups, the ionization of which is believed to be essential for the cleavage of f-aryl ether groups in non-phenolic units. Treatment of wood with this reagent slows bulk delignification but has little effect on the initial phase. Finally, methylation with dimethyl sulfate blocks all types of hydroxyl groups, thereby retarding both the initial and bulk phases. Although this evidence provides strong indi- cations for the central involvement of free phenolic hydroxyl groups in initial delignifica- tion, Obst [171] reminds us of the inherent complexity of the system and points out that parallels between the behavior of model com- pounds and lignin may be coincidental. He also ‘makes the point that the physical effects of wood methylation (restricted swelling and concomi- tant decreased accessibility) might be confused with chemical effects. 7. Bulk delignification ‘The bulk phase of delignification is that period during which most of the lignin is removed with a selectivity that is high compared to that prevail- ing during the initial phase it follows. The point of transition from initial to bulk delignification can be defined according to several different criteria, as illustrated in Figs. 48 to 50 {186}. The lignin yield at which the transition occurs may be defined as the intercept of a plot of lignin yield vs. H-factor, as exemplified in Fig. 48, as the abscissa of the intersection between two straight line portions of a plot of alkali concentration vs. lignin yield (Fig. 49), or as the abscissa of the corresponding intersection on a plot of carbohy- rate yield vs. lignin yield (Fig. 50). All three appear to give about the same result, although the degree of agreement and its significance does not appear to have been specifically studied. The kinetics of bulk delignification of pine has been investigated by LeMon and Teder (132), who used chips which had been partially predelignified to eliminate the initial stage. Interpretation of their results was complicated by what has been referred to as a lignin “Yield of nin (34) Fig. 48. The lignin yield as a function of the H-factor with different temperature rises [186]. roo} 7 i se g : | Chips amount 128 g od. prose | Non me 98 min Suk Phase Cigui fo wood {rato ae Fesidual phase | i 0 026 90 ‘loi of gin (%) Fig. 49. The effective alkall concentration as a function of the lignin yleld in pine sulfate cooking (186). “memory” effect — bulk phase behavior depends on initial phase conditions. This phenomenon, which had been previously observed by Wilder and Daleski [231] is illustrated by Fig. 51, which. shows that the first order delignification rate constant in the bulk phase increases with increased initial phase sulfidity, especially at lowYield of carbohydrates (%) Tb ky [S+ll) in stage 1 0.4 o2 M Fig. 51. The effect of sulfide in the pretreatment stage. The delignification rate in the bulk delig- nification stage plotted vs. sulfide sulfur concen- tration In the pretreatment stage at 145°C. In the bulk delignification stage at 170°C, [OH], = 0.3M In all experiments [S(—I1)] OM (0), 0.1M (©) and 0.2M (@) [132]. * Active alkali Sufi Chips amount Impregn. time Heating time | Liquid to wood 10 21.8 Yield of lignin (%) Fig. 50. The dissolution of carbohydrates and lignin during pine sulfate cooking [186]. BULK DELIGNIFICATION 61 18% as NagO 30% 125 g od. 15 min 90 min ratio 93.8:1 30 bulk phase sulfidity. In spite of this complica- tion, the authors were able to derive useful kinetic expressions, applicable over a range of pretreatment conditions. An equation that describes the progress of bulk deligni- fication after pretreatment with a sulfide-con- taining liquor is —dL/dt = [1.6 (OH~) + 2.3 (OH) x (HS) (L) G2 A simpler equation, applicable over a narrower range of sulfidities, can be derived from this by dropping the term containing only the concen- tration of hydroxide ion, which corresponds to the soda pulping contribution, and refitting to obtain new values of the remaining constants {173}, This results in values of the exponent on (OH) in the range 0.7-0.8 and values of the exponent on (HS~) in the range 0.1 to 0.4, depending on conditions. ‘A similar study of the kinetics of bulk delignification of birch by Edwards et al. [52] showed, as expected, that the hardwood pulped at a much higher rate. However, the dependence of the rate on liquor concentrations was somewhat different. These authors arrived at the following equation: —dL/dt = [282 + 11.1(0HS~/% * (OH) — 1.1) (33)62 RESIDUAL DELIGNIFICATION 8B 8ss Lignin %% on wood os oO og o2 on 10 30 $9 «6-7 910190 Time, min Fig. 52. Cooking of Picea exceisa sawdust in 4% ‘sodium hydroxide solution with rapid heating to reaction temperature [121]. a,a’ = 180°C; b,b’ = 175°C; cc’ = 170°C; dd’ = 165°C. delignification rate much slower than in the preceding bulk phase and, as shown in Figs. 49 and 50, by increased yield loss and alkali consumption per unit of lignin removed. Kleinert [121] has shown that, like the bulk phase, it is first order in undissolved lignin (Fig. 52), He also determined that the transition from bulk to residual delignification occurs at a lignin yield that decreases with increasing temperature and that the activation energy is about two thirds that of the bulk phase (90 vs. 134 kJ/mol). The former observation suggests that the “residual” or difficultly removable lignin is at least partly formed during the cook and is not simply a naturally intractable fraction of native lignin. 9. Other models To complete this discussion of the chemical aspects of delignification kinetics, it should be pointed out that approaches quite different from those described above have been suggested. For example, recognizing the polymeric nature of lignin, Szabo and Goring [190] have interpreted experimental delignification data in terms of the reverse of a gelation process, applying the Flory theory of trifunctional polymerization in reverse. Since these workers introduced the idea, others have explored its applications, particularly to delignification kinetics. These efforts have been recently summarized and extended by Yan (188). Thus far no immediately applicable practical model appears to have resulted. 10. Kinetics of carbohydrate removal The kinetics of carbohydrate removal during kraft pulping has received relatively little atten- tion. Data on the carbohydrate composition of the wood residue as a function of pulping time have been obtained by several authors but usually for purposes other than to derive rate laws for the removal processes. Examples are provided by Figs. 36-38 [15]. These data were sought out by Christensen et al. [46-48] who used them together with a parameter identification algorithm to obtain the best fitting values of the constants in the rate law —dC/dt = [k(OH™) + k"(OH~)(HS~)} X [(C-Co)] G4) More recently, Kondo and Sarkanen [123] measured carbohydrate loss during an investiga tion of initial phase pulping kinetics. By doing control experiments on wood which had been stabilized toward the peeling reaction by reduc- tion with borohydride, they were able to separate the observed losses into those resulting from the peeling reaction and those resulting from simple dissolution. Data giving peeling loss, A, as a function of time, , were then correlated by means of the equation = Xn = Infd./(4,.-A)] (5) to obtain the rate constant, k;, as a function of temperature and the activation energy, assuming values of 100 for Xn, the number average degradable chain length, and 18.5 for A.. the amount of carbohydrate degraded at infinite time. Their final correlation, shown in Fig. 53, gave an activation energy estimate of 93 kJ/mol. k B, MASS TRANSFER ASPECTS The previous discussion relates only to the Burely chemical aspects of the rate phenomena operative during pulping. It therefore assumes that all parts of the wood are freely accessible to the pulping liquor and that there are no concentration gradients within the chips. There are few practical situations where such models are directly applicable, since industrial wood chips are nearly always sufficiently thick that