Carbon 3. pp. 343-350. Rinted in Great Britain.
0008-6223187 $3.00 + .CKl 0 1987 Pergamon Journals Ltd.
THE ADSORPTION OF METHYLENE ACTIVE CARBON
STUART S. BARTON Department of Chemistry and Chemical Engineering, Royal Military College of Canada, Kingston, Ontario, Canada. (Received 1 February Abstract-The
1986; in revised form 15 September 1986)
extent of adsorption of methylene blue from an aqueous solution is a convenient indicator
in the evaluation of active carbons. The adsorption process is, however, complicated by factors inherent in the structures of both the methylene blue solution and the active carbon. These factors include the tendency of methylene blue to form molecular aggregates in solution, molecular sieving imposed by the pore size distribution of the carbon and the heterogeneous nature of the energies of adsorption sites. Temperature variation of adsorption and calorimetric experiments show, however, that the results obtained can be rationalized on the basis of these structural factors and the kinetic effects which result from them. Key words-Active
carbon, dye adsorption, pore structure, molecular aggregates, enthalpy of immer-
1. INTRODUCTION The
adsorption of the cationic dye, methylene blue, has been used for a long time for the evaluation of the adsorption properties of active carbons, particularly liquid phase carbons. The procedure was first suggested in 1924 and is still described in standard texts intended for industrial use and used in industry. For example, methylene blue adsorption was used extensively for monitoring the production and quality of active carbon from coal, in a fluidized bed. It has, however, been pointed out that when a large molecule, like methylene blue, and smaller, or “solvent” molecules, like water, are adsorbed on a porous solid with a distribution of pore sizes the effective adsorbent surface available to the larger molecule is limited by pore screening or molecular sieving[4,5]. Thus, for example, the surface of a certain active carbon (ACTIBON C) available to pnitrophenol was found to be about twice that available to methylene blue and dye adsorption, in general, was proposed as a method for the investigation of mesopore structure. The terms “macro, meso and micropore” will be used here as defined by IUPAC. The mechanism of dye adsorption by porous solids is however, complicated by the structure of the dye solution itself. This structure is usually discussed in terms of a monomer-dimer equilibrium although the presence of higher micelles or aggregates in concentrated solutions cannot be excluded[8,9]. A further complicating factor is the energetically heterogeneous nature of the solid surface so that the surface available, in the geometrical sense, to the large dye molecule may contain a range of adsorption sites of different adsorptive power. These factors operate so as to obscure the meaning 343
of the isotherm describing the adsorption and complicate a thermodynamical analysis of the temperature dependence of the adsorption. Finally the kinetics of porous adsorption from solution are usually slow so that the “equilibrium” state finally achieved is based on some subjective choice involving the decision that no further adsorption is occurring in the system under consideration. In the work reported here the interaction of aqueous and ethanolic methylene blue solutions and a commercial active carbon having both a large mesopore and large micropore volume is examined by conventional adsorption techniques and calorimetric determinations.
Adsorption measurements were carried out as follows. Weighed amounts of carbon were added to weighted amounts of aqueous or ethanolic methylene blue solution in 30 ml Hypo-vials which were then closed by means of teflon-silicone septa discs and aluminum seals. The methylene blue used was Fisher Laboratory grade which lost 13.9% by weight after heating for two hours at 110°C. This weight loss corresponds to 2.86 mole water per mole of hydrate. The aqueous solutions were prepared by weight from the hydrated dye and the ethanolic solutions from the dehydrated material. All the results given below were calculated on the basis of anhydrous methylene blue. The concentrations used ranged from about 2g . kg-’ to 21g * kg-‘. The amount of adsorption was controlled by using a fixed quantity of solution (ca 28 g) of a chosen concentration and changing the amounts of carbon added. These amounts ranged from 0.1 to 3.5 g,
depending on the concentration of the solution. The carbons, after being dried at 110°C overnight were stored over Drierite. The filled vials were loaded into holes drilled along the length of an aluminum bar which was supported axially in the central hole of a second large aluminum block. This second block was maintained at temperature by means of a flow of constant temperature water which passed through a copper coil wound around the block. The ends of the block were sealed with aluminum discs and the inner central bar, containing the samples, was revolved by a geared-down electric motor at about 2 r.p.m. so that the methylene blue-carbon systems were slowly and continuously tumbled. At the end of the reaction period, samples of the methylene blue solution were removed by a hypodermic syringe via the septum while the system was at the specified temperature. For measurements carried out for short reaction times the vials containing the methylene blue solution were pre-heated in the tumbler before weighted amounts of carbon were added. The closed vials were then rotated for twenty minutes and samples for analyses were taken as rapidly as possible at the end of this period. Analyses were performed by accurate (wt/volume) dilution with distilled water, to the Beer’s Law region, followed by an optical absorbance determination at 25°C in 10 mm matched cells in a Beckmann D.U. Spectra-photometer at 665 nm. Conversion of absorbance to concentration units (g . kg-‘) were made using a previously established Beer’s Law relation. In all experiments, a blank methylene blue solution of the initial concentration was also placed in the temperature controlled block and analysed at the end of the run. Separate loadings were made for all the adsorption determination, at the various temperatures used, except in the case where the reaction system was maintained at 76” for 72 hours, analysed, re-weighed and then kept at 25” for an additional 72 hours. For the separate loading experiments the apparent adsorption, (a), was obtained using the expression: mAC = a/g-methylene
When the same loading was used at two different temperatures, the second adsorption was calculated from a simple mass balance treatment expressed by: m,(F, - &) + Mr(F, - Fz) + NV*
- FJ 2:
where: m, = initial mass of solution/kg F, = methylene blue concentration on second analysis/g-kg-* M, = initial total mass of Hypo-vial plus contents/ kg M: = total mass of Hypo-vial plus contents after first analysis/kg The enthalpy of immersion measurements were carried out as previously described[ll]. The carbon samples were evacuated to low5 mmHg at room temperature. In some calorimetric experiments, previously wetted carbon was allowed to react with a methylene blue solution in an effort to measure the enthalpy of exchange of methylene blue with adsorbed water or ethanol. The technique is similar to that used by other workers[l2,13]. For these experiments, the carbon was weighed into thin walled glass bulbs and wetted with a weighed amount of water or ethanol before sealing. The amount of free liquid in the bulbs was estimated from the known imbibition volume of the carbon[l4], and the initial concentration of the dye solution was corrected for the small amount of dilution which occurred when the bulb was shattered in the calorimeter. Mercury porosimetry measurements were made using standard methods and an Aminco Porosimeter. The porous carbons used were, Type BPL Activated Carbon (Calgon Corp.) and a completely microporous carbon made by the slow thermal decomposition of polyvinylidene chloride (PVDC carbon) as previously described[ 151. 3.RESULTS AND DISCUSSION
blue * g-l carbon
where: AC = (F, - F,) F, = blank methylene (g-kg’ ) F2 = final methylene (g-kg-‘) m = mass of solution/kg w = mass of carbon/g
At low concentrations the above approximation comes more and more exact[lO].
Figure 1 shows the dependence of apparent adsorption of methylene blue on BPL carbon, from water solution, after 72 hours at 12”, 25”, 50” and 76°C. The adsorption increases with increasing temperature but is not proportional to the temperature rise. At 12” and 25°C the amount of reaction appears to be the same within the experimental error. A significant increase takes place at 50°C while at 76°C the amount of methylene blue adsorbed has gone up by a factor of 1.4. This phenomenon has been reported by other workers[ld] who described the effect to the fading of methylene blue solutions on storage at elevated temperatures. Since the results, reported here, were calculated on the basis of “blank” initial
Adsorption of methylene blue
F2 /g-kg-’ Fig. 1. Apparent adsorption of methylene blue. Cl PVDC carbon-aqueous-25” 7 BPL ” ” -12” I, I, -25 ,, ; II I, ,! -50 A ‘I ‘I I’ -76
concentrations, the increased apparent adsorption cannot be attributed to the disappearance of dye by processes other than reaction with the carbon surface. In any event, it was observed that no appreciable change in concentration took place as a result of exposure to elevated temperatures. Typically the “blank” concentration changed by only 51.5% of the value at 25” after extended periods of time at 50” and 76°C. Included in Fig. 1 is the data obtained when PVDC carbon was used as the adsorbent. This carbon does not adsorb methylene blue. Mercury porosimetry of the PVDC carbon showed that this carbon has no pore diameter greater than 3.0 nm. Indeed, molecular probe studies have shown that the pore diameter in PVDC carbon is about OS-O.6 nm. Therefore, the pore volume of this carbon, 0.40 ml-g-I[181 is not accessible to methylene blue. By way of contrast, for BPL, there is considerable volume with pore radius greater than 3.0 nm (0.46 ml-g-I)[141 and a Dubinin micropore volume (by cyclohexane adsorption) of 0.38 ml-g -I[ 191. A typical pore diameter distribution diagram for BPL, derived from mercury intrusion data, is shown in Fig. 2. In addition to very large pores, which probably represent interparticle voids equivalent to open surface, the macropore volume is contained in pores with diameter between 50 and 4800 nm. There is also considerable mesopore volume with pore diameters between 50 and 3.6 nm. The surface contained in these pores is presumably all available to
0 BPL carbon-aqueous -76 ; 1: I’ 0 -25” after 76 II ethanolic-26 0 ” ” ‘I -40 v It ‘I I’ -50
methylene blue since this dye molecule has a minimum diameter of about 0.8 nm and the limiting diameter of pores which will admit the molecule has been estimated to be about 1.3 nm. For the BPL carbon, therefore, the total pore volume is accessible to small molecules such as water and ethanol while only the macro and mesopore volume is available for the adsorption of the methylene blue. During the adsorption process, accordingly, molecular screening will ensure that there will always be a great deal of adsorbed solvent, independent of the dye concentration of the ambient solution. Adsorption of
PORE DIAMETER/ nm
LN PRESSURE /lb -
Fig. 2. Pore diameter distribution BPL carbon.
methylene blue involves a competition between the solvent and the dye for accessible adsorption sites. In Fig. 3 is shown the data obtained when adsorption experiments were carried out in very highly concentrated solutions. It appears that a decided adsorption plateau is reached which continues up to a dye concentration of 14 g-kg-‘. This concentration represents a relative concentration 0.06 based on the soiubility of methylene blue at 25”C. The plateau value is, however, reached at about l/6 of this concentration. If the system is at equilibrium after 72 hours reaction time then increased adsorption at higher temperatures must be explained (in the absence of an entropy increase on adsorption) by an overall endothermic enthalpy of adsorption. Similar results have been obtained by other workers for the adsorption of both anionic and cationic dyes, from aqueous solution, onto solid adsorbents[21,22]. The endothermicity of the adsorption process is explained by the necessity of an endothermic dissociation of dye dimers, micelles or aggregates prior to the exothermic adsorption of the dye monomer. In support of this mechanism these authors point out that adsorption from the non-associating solvent, methanol, is exothermic because the dissociation step is not required for adsorption. In Fig. 1 are also shown the adsorption results obtained when methylene blue-ethanol solutions were used. There appears to be little if any, temperature dependence and the adsorption does not increase as steeply at low concentration as does the isotherm for water solution. The results of Fig. 7 may be interpreted as indicating that the molecular size of the dye species in ethanol is small enough to insure that adsorption occurs readily. A possible, rapid test of the correction of the explanation, given above, is a comparison of the enthalpy of immersion (h,) of the carbon in the two solvents and the corresponding dye solution. The endothermic dissociation step in water should cause h, to be lower for the solution than for water alone. On the other hand for the ethanol system h, in the solution should be higher. Admittedly, the time involved in a calorimetric measurement (cu. 0.5 hr) is not long enough for “equilibrium” to be established. However, since the
0.3 b ,”’
process involves the immersion of evaculated carbon, a significant amount of dye will probably be adsorbed during the experiment. The calorimetric results are shown in Fig. 4. In all cases the heat was evolved rapidly (i.e., in less than 1 min.) and the temperature-time trace returned to a constant small slope. Further slow evolution or absorption of heat, if present, could not be detected by the calorimeter. It is apparent that, for both solvents the specific enthaipy of immersions (h,, given by the slopes of the heat vs mass carbon, straight line plots) do not depend on methylene blue concentration and are identical to the h, values for the pure solvents. The experimental uncertainty, indicated by the size of the data points, was obtained from a linear regression analyses of the data. During the calorimetric measurements the dye concentration decreased showing that dye adsorption took place. The initial dye/concentrations are given next to the points of Fig. 4. From the water solutions the adsorptions were 0.13 -C .006/g-g and from ethanol solution 0.12 + 0.1/g-g-‘. It must be concluded from Fig. 4 that the initial adsorption of methylene blue from the two solvents involves the same enthalpy change as the adsorption of the solvents themselves. Whatever the numbers of solvent molecules equivalent to a methylene blue molecule are, in the two cases, the dye adsorption is exothermic. Probably, during the enthalpy measurements, the methylene blue species, already present in the solution, which are small enough to be adsorbed are rapidly taken up. For the water system these species are replaced vin the endothermic dissociation step and further adsorption takes place by a slow diffusion of the adsorbable species into the completely wetted pore system. The same process must occur
Fig. 4. Enthalpy of immersion of BPL carbon (t = 25°C). Fig. 3. Apparent adsorption of methylene blue at high concentration
(t = 25°C).
A water A ethanol 0 aqueous methylene blue 0 ethanolic methylene blue
Adsorption of methylene blue in ethanol except that that endothermic dissociation is absent. It must be concluded that the enthalpy of adsorption step, from water solution, is sufficiently large to compensate for the endothermic dissociation. It is possible, from the h, values, in water, for the present sample of BPL and previously published data to estimate the micropore volume and hence the specific enthalpy of interaction of BPL with the two solvents[l9]. The calculated values are 350 J-g-’ or 16.1 kJ-mall’ for ethanol and 95.2 J-g-’ or 1.71 kJ-mall’ for water. If the adsorption of methylene blue involves a competition with solvent molecules for adsorption sites, these figures imply that less dye will be adsorbed from ethanol than from water and the isotherm will rise less steeply from the origin (Fig. 1). If it is assumed that the adsorption during the calorimetric measurement took place only on the easily accessible, high energy sites and that a one to one exchange of a dye molecule for a solvent molecule occurred at these sites, then a simple calculation can be made. The enthalpy of interacton must be at least, -16.1 kJ-mol-’ and so for the water system it can be concluded by Hess’ Law that the sum 16.1 kJ - 1.71 kJ = 14.4 kJ is an approximation of the endothermic enthalpy involved in the production of a mole of dye monomer. Values for this enthalpy change have been reported from temperature variation of equilibrium constant studies, by various methods[23,24,25]. The results are not in good agreement. In particular, the kinetic results differ markedly from results obtained by other methods. If the results of reference 25 are excluded, the enthalpy of dissociation of the methylene blue dimer is seen to be about 20 kJ-mall’ (dimer) or about 10 kJ-mall’ (monomer). In view of the assumption made above the agreement with the value calculated here is satisfactory. In Fig. 5, the apparent adsorption at 40°C and 64°C after 20 minutes reaction time, are plotted against final dye concentration. The adsorption increases with temperature by 1.3 + .04 times, over the range of concentrations considered. If this increase in initial adsorption is mainly due to the increase in the monomer concentration at the higher temperature then a correlation with the increased monomer concentration should be possible. Estimates of the dimer dissociation constant (K,) at the two temperatures can be made[23,26]. A simple calculation shows that the ratio of the monomer concentrations at two tempertures is given approximately by:
al.  is 1.39. These values agree very well with the ratio derived from the adsorption results. If the analyses of the calorimetric data and the adsorption data are meaningful then the conclusion must be drawn that it is the monomeric species which is mainly involved in the adsorption. The results of the calorimetric measurements of methylene blue-water exchange are shown in Fig. 6. Here, the amount of heat evolved during the initial, fairly rapid, reaction is divided by the apparent adsorption of methylene blue (to give the net integral specific enthalpy of the exchange reaction to that time) (Aa) and plotted against the apparent methylene blue adsorption. Experiments were carried out at two temperatures. If the enthalpy of dilution of the dye solution is negligible, the results shown in Fig. 6 represent the enthalpy of exchange of methylene blue for adsorbed water minus the enthalpy of dissociation of the dye dimer. The values range from about 50 kJ-mol-’ to 22 kJ-mol and fall with increasing adsorption. These large values must represent exchange at strong adsorbing sites, formed by the presence of surface oxide, inorganic impurities and energetically favoured pore radii, and reflect the heterogeneous nature of the surface. The increase in AH, at the higher temperature is believed to result from the fact that there is more dye monomer present, when the exchange reaction starts at the higher temperature, so that more reaction takes place as the monomer diffuses into the pore system. It is also possible to extract some kinetic information from these experiments. In Fig. 7 is shown a typical result obtained by transforming the calo-
b . 0
for the same analytical concentration of dye. The value of the ratio from the data of Dunkel et al. is 1.29 and the value from the data of Mukerjee et
F2 /g-kg-’ Fig. 5. Fast apparent adsorption of methylene blue from water onto BPL. A 64” C
0 40” C
a /g-g-’ Fig. 6. Enthalpy of exchange, methylene blue-water. A 47°C 0 25°C
rimeter recorder trace into digital form by means of a desk-top digitizer (CFM Mode H-241-1E). The response of the calorimeter to heater power was measured so that it could be certain that the rate of temperature increase observed during the exchange experiment was a measure of reaction rate and not an artifact of the calorimeter. It was found that the rate of increase of temperature in the calorimeter was linear with input wattage from 0 to 1.17 watts. The rates of heat evolution in the exchange experiments were all less than 0.08 watts so that it may be concluded that the observed rate of heat evolution was, indeed, a measure of the rate of exchange at the carbon surface. In Fig. 7 the amount of heat evolved is plotted both against time and (time)“*. After about 12 minutes, the heat evolution becomes so small that it could not be detected by the calorimeter. In all the experiments performed, the rapid heat production appeared to be complete after about 15 minutes, when the time-temperature trace became constant with a slope close to initial slope. The (time)“* plot is moderately curved and appears to be the type II of Giles’ classification. Fig. 7 implies that the reaction follows broadly a parabolic rate law and is diffusion controlled. The mechanism is, however, not simple and two diffusion controlled regimes can be noted. Integral diffusion constants may be calculated from the slopes of the initial straight line portions
using the equation given by Adamson et al. for intra-particle diffusion. D = (slope)*R%r AH*a* c 0 36
where: AH, = integral exchange heat/J-g-’ (dye) a, = adsorption/g-g-’ R = average radius of carbon particles assume spherical, (from mesh size) = 0.071 cm. For 6 experiments at 25°C the value of D was found to be 1 ? 0.2 cm’ . see-’ and for 4 experiments at 47°C 1 2 0.3 cm* . set-I. The diffusion of methylene blue, in this case, appears to be considerably more rapid than the diffusion of iodine into porous carbon recently reported. The insensitivity of the calculated diffusion constant to temperature indicates that the global reaction is under diffusion control and that the adsorption step itself is very rapid. It may be demonstrated that adsorption at a high temperature pertains more closely to the equilibrium situation than the apparent equilibrium reached at lower temperatures. In Fig. 1 are also plotted the results of experiments in which the adsorption reaction was allowed to proceed at 76°C for 72 hours and then at 25°C for an additional 72 hours. It is at
Adsorption of methylene blue q%z
TIME / MIN Fig. 7. Heat evolution during methylene blue-water exchange of BPL. 0 heat evolution vs time * ,, I, u ” vzz
once seen that the adsorption does not change on lowering the temperature. If these isotherm points represent a true equilibrium then it must be concluded that the average net enthalpy of the adsQrption, from water, must be nearly zero. The observed increase in adsorption with increasing temperature is then a kinetic effect resulting from the increased monomer concentration in the ambient solution. If it is assumed that the plateau of the high temperature isotherm from water represents the complete coverage of the macro and mesopore surface with methylene blue then an estimate of the area of an adsorbed dye molecule can be made. If the macro and mesopore volume surface area is given by the mercury porosimeter data, 91 m*-g-‘, and the “equilibrium” adsorption is 0.42 g dye-g-’ then the area of the adsorbed molecule is 0.35 nm*-molecule-‘. This value is close to the end-on molecular area 0.39 nm*[ 161.
Consequently, the exchange reaction between the dye and the adsorbed solvent molecules is initially exothermic. The enthalpy change however decreases rapidly with increasing adsorption and finally becomes nearly equal to the enthalpy of dimerization of the dye. The overall reaction is then nearly atherma1 or perhaps endothermic. Diffusion into the easily available part of the pore system is fairly rapid but subsequent diffusion into smaller pores is very slow. Because of the ease of analysis, the adsorption of methylene blue from aqueous solution continues to be a useful tool for product control in the manufacture of active carbon. Test adsorptions should, however, be carried out at high temperatures and interpreted as a measure of the macro and mesopore areas.
Acknowledgement-The author wishes to thank Maureen Good for skillfully carrying out much of the experimental work.
The adsorption of methylene blue is complicated by the structure of the dye solution, molecular screening effects, and energetically hetergeneous nature of the adsorbent. Initial enthalpies are quite large because of adsorption on the high energy sites.
1. F. Paneth and A. Radu, Ber. Dtsch. Chem. Ges. B57, 1221 (1924). 2. J. W. Hassler, Purification with Active Carbon, p. 326, Chemical Publishing Co. Inc., New York (1974).
STUART S. BARTON
3. K. Kudo, H. Hosada, S. Honma, and M. Komatsu, J. Fuel Sot. Japan 52, 335 (1973). 4. D. Graham, J. Phys. Chem. 59, 896 (1955) and references given there. 5. M. M. Dubinin and E. D. Zaverina. Acta Phvs-Chins. U.S.S.R. 4, 647 (1936). 6. C. H. Giles and A. P. D’Silva, Trans. Far. Sot. 65, 1943 (1969).
7. D. H. Everett, Manual of Symbols and Terminology for Physicochemical Quantities and Units. Appendix II, Part I, p. 585. Butterworths, London (1972). 8. R. W. Chambers, D. R. Kajiwara, and J. Kearns, J. Phys. Chem. 78, 380 (1974). 9. D. R. Lemin and T. Vikerstaff, Trans. Far. Sot. 43, 491 (1947).
10. J. J. Kipling, Adsorption from Solutions of NonElectrolytes, p. 31. Academic Press, New York (1965). 11. S. S. Barton, J. R. Dacey, B. H. Harrison, and J. R. Sellors, J. Coil. Interface Sci. 71, 367 (1979). 12. J. M. Lamarche, A.. Foissy, and G. Robert, in Adsorvtion at the Gas-Solid and Liauid-Solid Interface (J. Rouquerol and K. S. W. Siny, Eds.), p. 117. Elsevier, Amsterdam (1982). 13. J. Rouguerol, et al. in Journees de Calorimetrie et dilnalyse Thermique, Turin 1978 and Nancy 1981. 14. M. P. McDaniel and T. D. Hottovy, J. Coil. Interface Sci. 78, 31 (1980).
15. S. S. Barton, M. J. B. Evans and B. H. Harrison, J. Coil. Interface Sci. 49, 462 (1974).
16. J. J. Kipling and R. B. Wilson, J. Appl. Chem. 10, 109 (1960). 17. S. S: Barton, G. L. Boulton, J. R. Dacey, M. J. B. Evans, and B. H. Harrison, J. Coil. Interface Sci. 44, 50 (1973). 18. S. S. Barton, M. J. B. Evans, and B. H. Harrison, J. Coil. and Interface Sci. 49, 462 (1974). 19. S. S. Barton, M. J. B. Evans, J. Holland, and J. E. Koresh, Carbon 22, 265 (1984). 20. W. J. Schnellbach and J. J. Rosin, Amer. Pharm. Ass. 20, 227 (1931). 21. M. M. Allingham, J. M. Cullen, C. H. Giles, S. K. Jaun, and J. S. Woods, J. Appl. Chem. 8,108 (1958). 22. C. H. Giles. J. J. Greczek. and S. N. Nakhowa. J. Chem. Soi. 93, (1961). 23. H. Dunken, D. Schmidt, and K. Palm, Z. fur chemie 2, 349 (1962). 24. R. E. Ballard and C. H. Park, J. Chem. Sot. A, 1340
(1970). 25. W. Spencer and J. R. Butler, J. Phys. Chem. 83,1573 (1979). 26. P. Mukeriee and A. K. Ghosh, J. Am. Chem. Sot. 92:22 6419 (1970). 21. C. H. Giles. in Adsorvtion from Solution at the Solid Liquid Interface (G. b. Paifitt and C. H. Rochester,
Eds.), p. 369. Academic Press, London (1983). 28. R. M. Barrer, Trans. Far. Sot. 45, 358 (i949). 29. A. N. Adamson. G. E. Bovd. and L. L. Mvers. J. < Adm. Chem. Socl 69, 2836 0947). 30. Y. Baba, T. Nakao, K. Inoue, and I. Nakamori, Sep. Science and Technol. 20,21 (1985).