MOLECULAR PROPERTIES OF LIGNIN SOLUTIONS'
FROM VISCOSITY,OSMOTICPREBSURE, BOILING-POINT RAISING, DIFFUSION, AND SPREADING MEASUREMENTS D. L. LOUGHBOROUGH AND ALFRED J. STAMM Forest Products Laboratory,a Forest Service, U . S. Department of Agriculture, Madison, Wisconsin
Received June 11,19$6
Recent chemical researches (2, 12) indicate that lignin, as isolated from previously extracted wood by several different methods, is a reproducible material with rather definite properties. It has ten polar groups, methoxyl and hydroxyl, for each empirical molecular weight unit of approximately 900, the proportion of each varying somewhat with the source of the lignin and the means of its isolation (12). The presence of unsaturated groups has also been demonstrated. An uncertainty still exists as to whether the basic part of the molecule is made up of aromatic or furan units. The work reported here was undertaken with the hope that physical data on lignin solutions would aid in the further characterization of lignin. Previous measurements of the molecular weight of lignin are meager and have been made chiefly on lignin derivatives which, because of the nature of the preparations and the drastic solvents used, must have been degraded. These values range from 200 to 2000 (1, 6, 7, 9, ll, 17, 26, 27, 28, 29, 36). The most reliable value seems to be that of Samec (30) who, using a static osmotic pressure method, reported a value of 4000 for the molecular weight of Willstatter lignin dissolved in ammonium hydroxide. Spreading measurements have also been made to determine the thickness of lignin films on a water surface. Wedekind and Katz (37) obtained 10 A. U. for phenol lignin and Freudenberg (5) reports 20 A. U. for the sodium salt of the diazobenzenesulfonic acid derivative of cuprammonium lignin. 1 Presented before the Thirteenth Colloid Symposium, held a t St. Louis, Missouri, June 11-13, 1936. Abstracted from a thesis submitted by D. L. Loughborough to the Faculty of the University of Wisconsin in partial fulfillment of the requirements for the degree of Doctor of Philosophy, June, 1936. 2 Maintained at Madison, Wisconsin, in cooperation with the University of Wisconsin.
1113
1114
D. L . LOUGHBOROUGH AND ALFRED J. STAMM
PREPARATION OF MATERIAL
The lignin used in this investigation3 wa3 prepared from maple and spruce wood. The sawdust wab extracted 11 ith alcohol-benzene mixture for two or three days to remove resins and fatty substances. This was follo%ed in consecutive order by a thorough extraction in hot mater, cold water, and ether. The lignin was isolated from the extracted sawdust in three different ways: (a) by the modified sulfuric acid method of the Forest Products Laboratory (31) ; (b) by extracting the lignin directly with methyl alcohol acidified with hydrochloric acid (6); and (c) by extracting in a sodium hydroxide-ethyl alcohbl mixture (25). The soluble portions of the isolated lignin preparations were purified by alternate solution in alcohol and precipitation in water three or four times. I n some cases the product was also precipitated in ether. The measurements reported here were made on the product after different numbers of precipitations, and in some cases the final product was electrodialyzed for forty-eight hours a t 230 volts. The material was dried in a vacuum oven at 50OC. for more than forty-eight hours and stored over phosphorus pentoxide until needed. The solvents were of reagent quality, dried by standard drying methods. The completely nitrated and methylated lignins were prepared from methyl alcohol-hydrochloric acid lignin. The nitrated lignin was prepared according to the method of Gilman (10) for nitrating furfural; the methylated lignin was prepared by the ordinary methylation process with dimethyl sulfate. SOLUBILITY O F LIGNIN
The solubility of lignin in organic solvents depends to a large extent on the method used for isolating the lignin. When isolated by the sulfuric acid method, only about 12 per cent of maple lignin and less than 2 per cent of spruce lignin is soluble in methyl alcohol. Since the sulfuric acid method removes the lignin quantitatively, these fractions represent the fraction of the total lignin that may be brought into solution b y this method. Both the alkali and methyl alcohol-hydrochloric acid methods yield from one-third to one-half of the total lignin content of the wood. Virtually all the yield is soluble in methyl alcohol after precipitation and washing free of alkali and acid. No chemical difference, however, has been found between the soluble and insoluble fractions of the isolated lignins. Thus the solubility differences are probably due to polymerization. Since all the physical measurements considered in this study were made 3 Acknoirledgment is made t o E. E Harris of the Forest Products Laboratory for the preparation of a number of the lignin samples and lignin derivatives used in this investigation. To hmi also are due thanks for many suggestions and advice on the chemical aspects of this ptoblem
MOLECULAR PROPERTIES OF LIGNIN SOLUTIONS
1115
on solutions of lignin, no definite conclusion can be drawn regarding the total isolated lignin, or the lignin as it occurs in the tree. Lignin is somewhat soluble in many solvents, among which are the aliphatic alcohols and acids, methyl and ethyl acetate, acetone, and chloroform. Lignin that is air-dried after precipitation in water is insoluble in water, ether, benzene, or carbon tetrachloride. Although the solubility is not very great in most of these solvents, the amount that can be put into solution can be greatly increased by evaporation of the solutions. It is almost impossible to dissolve all the dry material added to the solvent, since some of the lignin almost invariably forms a tarry insoluble mass. The solubility in chloroform is greatly increased by removing every trace of water. I n the alcohols, on the other hand, the solubility is increased considerably by the addition of small amounts of water. For example, the direct solubility of alkali lignin in methyl alcohol was found to increase to a waher content of approximately 18 per cent by volume and then to decrease with a further increase in water content. All the solutions of dry lignin were clear. They showed no Tyndall cone unless treated in a way that causes precipitation, as when appreciable amounts of water are added. Solutions have been kept as long as two years without showing any cloudiness. Lignin dissolved in the foregoing solvents can thus be considered as forming true solutions. VISCOSITY
A Bingham viscometer, flow time 1,164 sec. for water under 6 cm. of mercury, was rigidly mounted in a thermostated water bath held a t 25' f 0.02"C. The applied pressure was controlled from an auxiliary tank of 70 liters capacity. This tank could be filled to the desired pressure, closed from the air line, and held at this pressure for long periods of time. The pressure was read with a cathetometer to a tenth of a millimeter of mercury. Several measurements of the efflux time for flow in both directions Rere made at each pressure. Poiseuille's law, in which the reciprocal of the efflux time is proportional to the pressure, was obeyed over the complete pressure range used. Over the concentration range used (0.2 to 8.0 per cent) there was no great deviation from the modified Einstein law
where qsp is the specific viscosity (increase in relative viscosity of the solution over that of the solvent), C is the concentration in grams per 100 cc. of solution, V is the specific volume of the solute, and B is a constant, which
1116
D. L. LOUGHBOROUGH AND ALFRED J. STAMM
is unity for the ordinary Einstein equation. The expression
' 00 has ' CV
been designated by Kraemer (15) as the specific hydrodynamic volume. V was determined pycnometrically a t 25°C. for sulfuric acid lignin and methyl alcohol-hydrochloric acid lignin dissolved in methyl alcohol and acetic acid. The values were in fairly good agreement, and average 0.75. The change in the specific viscosity per unit concentration with concentration for methyl alcohol-hydrochloric acid lignin in different solvents is given in figure 1. The curves are practically horizontal over an appreciable concentration range, with the expected upward curvature a t higher concentrations in the case where higher concentrations were obtainable. It is uncertain if the curvatures in either direction a t extremely low concentrations have any significance, though the deviations seem to be greater than the experimental error. The linear portions of the curves have been
FIQ.1. Change in the specific viscosity per unit concentration with concentration for methyl alcohol-hydrochloric acid lignin. A , chloroform; A,acetone; 0 , methyl alcohol; 0, ethyl alcohol. extrapolated to zero concentration in order to calculate the limiting values of e, without regard to the curvsture a t low concentrations, as has been done by other investigators (16, 34). These values, as shown in table 1, are practically constant for all the solvents. Values of e for sulfuric acid and alkali lignins are also given. These measurements were not made over a sufficient concentration range to make extrapolation advisable, so that they represent the directly calculated values. They are not in so good agreement as the values for methyl alcohol-hydrochloric acid lignin, but they do represent the order of magnitude of 8. Three explanations are possible for the deviation of the data from the simple Einstein equation in which e is unity. The first is that the departure may be due to solvation of the solute particle. The method of measuring the specific volume used does not give a true measure of the volume of the solvzted particle. The constancy of 0, however, for quite
1117
MOLECULAR PROPERTIES O F LIGNIN SOLUTIONS
different solvents makes this factor, in all probability, of minor importance. The second explanation is that the equation is derived on the basis of no interaction between particles, a condition which would ocour only at infinite dilution. Unfortunately, the extrapolation to zero concentration can not be definitely made until more accurate measurements a t extremely low concentrations are available. The third explanation is that if the particles were not spherical, they would not be expected to obey the original Einstein equation derived on that basis. The equation has been extended to include differently shaped particles by Eisenschitz (4), Jeffery (13), and Kuhn (17, 18). The Eisenschitz equation for appreciably elongated elliptical particles when substituted in the modified TABLE I Multiples of Einstein constant obtained for various lignin preparations from viscosity data at 96°C. V A L U l S OF
2.5
* loanap
cv
BOLVBNT
Alkali lignint
Methyl alcohol.. ........................... Ethyl alcohol. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Ethyl alcohol (few per cent water). . . . . . . . . . n-Butyl alcohol. ........................... n-Propyl alcohol. .......................... Acetone.. .................................. Methyl acetate.. ........................... Chloroform. ................................
3.36
3.00 4.12
4.90
3.26 3.30
4.50 4.20 2.08 2.60
3.46
* By
extrapolation to zero concentration. single measurements below 1 per cent concentration, by assuming qap/C as constant.
t From
Einstein equation gives the shape factor l/a of the elliptical particles in terms of the multiple of the Einstein constant e, thus I 1 0.159 - -= 2.5 6 a 1 log 2 a This gives a shape factor for lignin of approximately 7.5. Although this shape factor is in a range for which there is no apparent reason for the validity of the equation, Theis and Bull (34) have found the relationship to be obeyed by stearic acid which, from x-ray and spreading measurements, has been shown to have a shape factor of this order. If the specific viscosity is a function of the shape of the molecule, it
1118
D. L. LOUGHBOROUGH AND ALFRED J. STAMM
should vary with the temperature. A t the higher temperatures the frequency of rotation about even the longest axes will be large in terms of the collision frequency, and a molecule will offer nearly spherical collision volume. The values of the sperific viscosity should tend more and more toward that predicted by the Einstein equation. The plot of 0 as a function of temperature over a small range is shown in figure 2. The extension of this curve could readily give a value of unity for 0 a t high temperatures. Direct application of the Staudinger method (33) of determining the molecular weight from viscosity to these data is impossible, since no information is available on the lower molecular-weight homologues of lignin, if there are any. An idea of the magnitude of the molecular weight
.. ,O
U
20
25
30
7 r N P E I A rURE
35
do
.I
(*e)
FIQ.2. Variation of the specific hydrodynamic volume of a sulfuric arid lignin in methyl alcohol x i t h temperature
rn may be obtained by choosing a representative value for the constant in the equation
in which C is the concentration in terms of the weight of the repeating basic group and K is a constant. Taking the basic chemical molecular weight of lignin (900) as the repeating group and a value of K equal to that for rubber, 3 X (33), the molecular weight obtained is 18,000. If the value of K of 10 X lo-* for cellulose (33) is used, the molecular weight is 6000. These values can be thought of merely as an interesting speculation. The viscosity data on lignin show conclusively that lignin is not a linear polymer surh as cellulose, which has a value of e of 280 to 400 (21). The lignin molecule, however, deviates definitely from spherical, a probable
MOLECULAR PROPERTIES O F LIGNIN SOLUTIONS
1119
shape factor being about 7.5. The approximate constancy of 0 for different solvents indicates that lignin is probably not differently associated in the different solvents. Differences in solvation in the different solvents also appear to be small. OSMOTIC PRESSURE
The measurement of molecular weight by osmotic pressure provides a method that is less affected by impurities than any of the other thermodynamic methods, such as freezing-point lowering or boiling-point raising. The method used was a modification of the dynamic method of S$rensen
FIG.3. The osmometer
(3, 32, 35), altered to give a better support to the membrane and to make possible the use of less material. The osmometer (figure 3, side view) consists of two brass cells, B and B’, 4 in. in diameter and 1 in. thick, with a hole 3 in. in diameter turned to a depth of 3 in. in each. The cavities between the cells are separated from each other by two perforated brass disks, E and E’, which rest on the circular flanges. The membrane D is held securely between these two disks. The rest of the apparatus serves in filling, applying the pressure, and measuring the flow. The osmometer is filled through F and F’ (the primed letters indicate solution and the unprimed letters the solvent) by suction applied at H and H’, the level of both solution and solvent being raised, free from bubbles, to the horizontal part of H. F and F’ are TED JOURNAL OF PHYSICAL CHDMISTBY,
voc. 40,
NO.
9
1120
D. L. LOUGHBOROUGH AND ALFRED J. STAMM
securely closed with brass screw caps. All stopcocks are closed and the system allowed to stand for eight or ten hours. This procedure allows any material of very small molecular weight to come to equilibrium before the measurements are made. The whole apparatus, with the stopcocks open, is then placed in a constant-temperature bath held a t 25" =t0.05"C. When thermal equilibrium has been established, stopcocks K' and L are closed, the pressure applied at H', and the flow measured in the capillary tube I by means of a cathetometer. If the applied pressure is greater than the osmotic pressure, the liquid will be forced across the membrane and up the capillary tube I. The rate of flow is proportional to the pressure in excess of the osmotic pressure, so that when the flow is zero, the applied pressure equals the osmotic pressure. Readings of the height of the men-
I
I
l
/
I
/
FLOW IN
I
l
1
1
1
1
1
1
I
1
1
C C P b R SEC X ( 3 x l O . s )
FIG.4. Dynamic means of determining the osmotic pressure from the rate of flow under different hydrostatic pressures for methyl alcohol-hydrochloric acid lignin (0.00142 g per cubic centimeter) in methyl alcohol.
iscus in I were measured every half-minute for 20 minutes, and these values were plotted against the time to determine the rate of flow. The rates of flow obtained under different applied pressures were plotted against the pressure and the curyes extrapolated t o zero flow, the intercept being the osmotic pressure. Rubber dam and cellophane membranes were used. The cellophane membranes were of 600 and 300 gage. The cellophane was swollen in water, and the water replaced by the solvent. The fact that the osmotic pressure is independent of the membrane may be noted in figure 4, in which a typical plot of flow against pressure for a solution of methyl alcohol-hydrochloric acid lignin in methyl alcohol is given for all three membranes. The molecular weights, as tabulated in table 2, were calculated from the
MOLECULAR PROPERTIES O F LIGNIN SOLUTIONS
1121
osmotic pressure on the basis of the van’t Hoff law. The average molecular weight is, within experimental error, unchanged by the solvent, the method by which the lignin was isolated, the number of reprecipitations after the second in the purification, the fractionation, the temperature a t which the measurement was made, the presence of small amounts of water or small amounts of salt, or the concentration. The fact that elevation of the temperature does not affect the molecular weight indicates that the lignin molecule is not a loose aggregate of smaller molecules subject to thermal dissociation. No dissociation or polymerization occurs when small amounts of water are added. When sufficient salt is ttdded to overcome any Donnan effect that might be present, the molecular weight is unchanged, indicating that there can not be any appreciable Qonnan effect in the case of the electrolyte-free systems. Acetone and ethyl alcohol solutions of methyl alcohol-hydrochloric acid lignin were poured through a filter containing a small amount of activated charcoal. When this process was carried out rapidly, it was possible to halve the concentration. Molecular weights calculated from the osmotic pressure exerted by these solutions were definitely of the same order of magnitude as those calculated from the total fraction. Both the difficulty of performing the fractionation and the similar molecular weights measured indicate strongly that lignin forms monodisperse solutions. This is not in agreement with the results of Fuchs (8) for phenol lignins. The most probable molecular weight from the measurements on maple lignin prepared by the three methods previously mentioned is 3900 It 300. This is very close to the value reported by Samec (30), and also approximately four times the empirical molecular weight (4 X 900). Completely methylated and nitrated lignins show increases in molecular weight that can be accounted for on the chemical basis. The fact that the different lignin preparations, which represent different portions of the total lignin, give similar molecular weights is a strong indication that the lignin is not a mixture of molecular species with appreciable variation in molecular weight, for it is highly improbable that each of the preparations should contain the same molecular distribution. BOILING-POINT RAISING
A few measurements of the boiling-point raising of lignin in chloroform and acetone were made. I n this work two sets of the modified4 Cottrell boiling-point raising apparatus were used so as to require only 10 cc. of solution. They were heated on an electric hot plate by inserting theend of each into a form-fitting brass thimble so that only the part beneath the inverted funnel was heated. The tubes were inserted through holes in an This was done by H. D. Tyner of the Forest Products Laboratory.
1122
D. L. LOUGHBOROUGH AND ALFRED.J. STAMM
MOLECULAR PROPERTIES OF JGNIN SOLUTIONS
0
B
__
1123
1124
D. L. LOUGHBOROUGH AND ALFRED J. STAMM
asbestos board up to the liquid level to eliminate the heating of the upper part of the apparatus by other than the boiling liquid and vapor. Measurements of the temperature were made with Beckmann thermometers when solvent was in both tubes and when solvent was in one and solution in the other. Thermometer readings were taken after different lengths of time, and the values obtained after experimental constancy was attained were averaged. The molecular weights calculated from the boiling-point raising, together with the probable error due to fluctuations of the temperature, are given in table 3. Considerable difficulty was encountered TABLE 3 Molecular weight of various l i g n i n prepayations .front bozling-point raising measurements TYPE O F LIGNIN
SOLVENT
CONCENTRATIOK
MOLECTjLAR VYEIGHT
ERROR
gTams p e r
gram of solvent
.)
Chloroform Chloroform Chloroform Chloroform Acetone Acetone
0 0 0 0 0 0
0672 0336 0308 0154 0952 0472
3780 3620 3760 3670 3180 3460
flO0 f400 1200 f500 1300 A500
Maple lignin (sulfuric acid). . . . . . . . . .
Chloroform
0 00605
3160
fl000
Chloroform Chloroform Chloroform
0 0432 0 0216 0 0093
3510 3810 3410
1400 1300 1500
Fullymethylated maple lignin (methyl alcohol-hydrochloric acid) . . . . . . . . .
Chloroform
0 0101
4220
f700
Spruce lignin (methyl alcohol-hydrochloric acid). . . . . . . . . . . . . . . . . . . . . . .
Chloroform
0 00598
3640
11000
Maple lignin (methyl alcohol-hydrochloric acid) . . . . . . . . . . . . . . . . . . . .
I ;
I
Maple lignin (alkali). . . . . . . . . . . . . . . .
b
$ 4or gZ30r
IO
30
40
30
60
70
BO 30 /W O ! SCL7IE o , n s , 0 N s
120 /30 / 4 G 'WO /60 '70
FIG.7 . Diffusion curve for methyl alcohol-hydrochloric acid lignin in methyl MS a function of thc height in the diffusion cell
s l ( : o h i ~ l . Scale displaccmcnt
MOLECULAR PROPERTIES O F LIGNIN SOLUTIONS
1127
thermal equilibrium, for which two hours usually was considered sufficient, the air being stirred constantly. Then the upper cell was pushed over the lower by a steadily moving screw. The boundary formed between them was very sharp, as may be noted by the fact that the diffusion constant measured was independent of the time, even for times as short as half an hour The displacement of each line on the photographs was measured with a comparator to 0.002 cm., and the displacement plotted against the height JOO
,
/t
FIG.8. Diffusion data in terms of the relationship between the square of the standard deviation and the time for sugar in water solutions and various lignin preparations in several organic solvents. 0, sulfuric acid lignin (0.2-0.4 per cent); 0 , methyl alcohol-hydrochloric acid ligin (0.1-0.6 per cent); A, alkali lignin (0.3 per cent); A,sugar (0.5 per cent). in the cell. Representative plots for sugar and for lignin are shown in figures 6 and 7. On the basis of the index of refraction being proportional to the concentration, these graphs are, except for a scale factor, a plot of the rate of change of concentration against the height in the cell. The solution of Fick’s law for diffusion taking place under these conditions is a normal probability curve, the standard deviation u of which is related to the diffusion constant D through the following equation:
-
Ku2 2t
(4)
1128
D. L. LOUGHBOROUGH AND ALFRED J. STAMhI
where t is the time. The standard deviation was determined by finding the normal curve of known standard deviation that most closely matched the experimental points. The lines in figures 6 and 7 represent the iiormal curves chosen t o represent those data. The squares of the standard deviation were plotted against the time, and the diffusion constant was detrrmined from the slope. When the concentrations were of the order of one-half of 1 per cent or more, the displacement of the lines was so great as t o make accurate determinations impossible. Only those curves were used to determine the standard deviation for which most of the scale lines could be read with certainty. -1plot for sugar and the various lignin preparations in several of the solvents ib shown in figure 8. The diffusion sq. cm. per second, which is in constant obtained for sugar i- 4.58 X good agreement with the accepted values found in the literature. The determination of the molecular weight of lignin was made from the diffusion constant by means of Einsteink diffusion equation
The molecular weights so determined are shown in table 4. A single run with a AlcBain-Northrup diffusion cell (22, 23) was made on a solution of methyl alcohol-hydrochloric acid lignin in methyl alcohol at a concentration of 0.00248 g. per cubic centimeter. The cell constant was determined with sugar. A molecular weight of 9500 was obtained. The molecular weights for all the lignins in all the solvents are nearly identical, but they are considerably higher than the molecular weights determined by osmotic-pressure data. There are two reasons that might account for this difference. First, the lignin solution might be a mixture of different molecular species. The osmotic pressure gives a molecular weight that depends upon the number of particles in the solution, whereas the diffusion constant gives a molecular weight that depends upon a weight average of the particles. Any appreciable deviation from monodispersion would result in higher calculated molecular weights from diffusion than from osmotic pressure, because the larger particles would have the dominating effect in the former measurements and the smaller particles in the latter (21). Several reasons have already been given t o indicate that the lignin solutions approach monodispersion. The diffusion data also give further evidence of monodispersion. If the solutions were polydisperse, each species would diffuse at its own rate and its data would fall on a normal curve whose shape would depend on the weight and the number of particles of that particular species. The measured curve would be a weighted sum of these separate curves. This sum would probably not be normal, though it might approach normal if there were a great many particles of nearly equal molecular weights. The fitting of the data to a normal curve
MOLECULAR PROPERTIES OF LIGNIN SOLUTIONS
1129
can not be done with enough accuracy t o show small departures from normal, but the agreement, as shown in figures 6 and 7, is very good. Also the plot of the square of the standard deviation against the time is linear over long periods. If the systefi were polydisperse this would not be true. Thus appreciable deviations from monodispersity are highly improbable. If the molecules were not spherical, the use of the Einstein equation would be unjustified. Though the departure from spherical, as indicated by viscosity data, is small for the lignin molecule, the difference between TABLE 4 Molecular weight of various lignin preparations f r o m diffusion measurements at W C .
I
TYPE OF LIQNIN
Maple lignin (methyl alcohol-hydrochloric acid). . . . . . . . . . . . . . . . . . . . . .
,
I
MOLECUSOLVENT
*200 f200
11,000 11,000
*loo0 f500
10,000
fS00
12,000 12,000
11000 1600
Methyl alcohol Ethyl alcohol
10,000 9,000
1500
Methyl alcohol
7,350
f300
9,500
f5OO
Methyl alcohol Methyl alcohol (0.045 N with NaC1) Maple lignin (sulfuric acid). . . . . . . . . . < Methyl alcohol ( f l l pe cent water) Ethyl alcohol Acetone
I
Spruce lignin (methyl alcohol-hydrochloric acid). . . . . . . . . . . . . . . . . . . . . . .
ERROR
10,000 10,000 9,500 10,500
Methyl alcohol Ethyl alcohol Acetic acid Chloroform
'
Maple lignin (alkali) . . . . . . . . . . . . . . . .
LAR WEIQHT
Fully methylated maple lignin (methyl alcohol-hydrochloric acid lignin). . . Acetone
1500
f500
1500
the molecular weights obtained by osmotic pressure and by diffusion increases rapidly with increasing ellipticity. Perrin (24) has developed a theory for the rotary diffusion of elliptical particles. The result is expressed in terms of the ratio of the time of relaxation for an elliptical particle (TI)and for a sphere (TO). The ratio of T ~ / Tfor ~ a prolate spheroid rotating about its minor axis is
1130
D. L. LOUGHBOROUGH AND ALFRED J. STAMM
where p is the ratio of minor to major axis. If a somewhat Pimilar relationship holds for linear diffusion, it should be possible to determine a shape factor for the lignin molecule. If the true molecular weight of lignin is taken as 3900, the diffusion constant, considering the molecule t o be spherical, can be calculated. This diffusion constant is three-quarters of the measured values and, acrording to Perrin’s theory, gives a shape factor of approximately 8. This is in as good agreement with the shape factor found from viscosity data as could be expected. It is thus quite definite that the deviations of the molecular weight obtained by diffusion from the values obtained from osmotic pressure arc due to a deviation of the molecule from a spherical shape rather than to the system being polydisperse. Measurements, to be published later, made on the dispersion of the dielectric constant of chloroform solutions of lignin gave molecular weight too BO
0
2
4
CWOLM7R4TION
-
6 8 /O V 14 16 I8 (CRAM8 PCP I C O C C OF sOLuT/o,V)
FIG.9. Area covered per unit weight of methyl alcohol-hydrochloric acid lignin in chloroform a8 a function of the concentration.
values that were in good agreement with the values calculated from the diffusion constant. This was to be expected, since the dispersion of the dielectric constant is a measure of the rotary diffusion constant. SPREADING MEASUREMESTS
Some concept of the dimensions of large organic molecules can be obtained from the measurement of the area covered by a material when spread on a liquid in which it is insoluble. If the molecular weight and density are known, the area per molecule as well as the thickness of the film can be determined. Lignin was dissolved in chloroform and spread on water against the pressure of a thin film of lycopodium powder. When the area covered per gram of lignin was plotted as a function of the concentration, a limiting concentration of 0.005 g. per cubic centimeter was obtained, below which
MOLECULAR PROPERTIES O F LIGNIN SOLUTIONS
1131
the area covered per gram mas a constant. This constant value corresponds to a film thickness of 16 A. U. (figure 9). If the molecular weight is taken as 3900, the area per molecule is 318 sq. A. U. Spreading measurements were also made with a modified Langmuir film balance. The pressure exerted was practically zero until an area of 308 sq. A. U. per molecule, using several different concentrations, was reached. If the area was decreased slightly below this value the pressure rose very rapidly, the film collapsed, and the measurements were no longer reversible. Measurements upon 0.1 N hydrochloric acid solution gave similar results, and also measurements with nitrated lignin. If the two surface dimensions were equal, they would be equal to 17.6 A. U. and the molecule would be practically spherical. This is in disagreement with the previous findings reported here, which indicate a length from six to eight times the effective cross-sectional diameter. If the dimensions are taken as 16 A. U. by approximately 100 A. U. by 3 A. U., all the known data may be correlated. SUMMARY AND CONCLUSIONS
Lignin isolated by three different methods and dissolved in a number of different organic solvents gives molecular weights of two magnitudes, one of which is 3900 f 300 and the other 10,000 f 1,000. The lower value was obtained by the osmotic-pressure and boiling-point raising methods, both of which give number average values and are independent of shape. The higher value was obtained by diffusion measurements, which give weight average values and are affected by shape. Two explanations for this difference are offered, one depending on the concept of polydispersity and the other depending on the departure of the molecules from a spherical shape. A number of reasons are given to show that the latter explanation is probably correct. The viscosity data indicate that the lignin dispersion approaches monodispersion, because of the practical constancy of the multiples of the Einstein constant a t infinite dilution in the different solvents. The constancy of the molecular weight of all the lignin preparations in all the solvents, the nearness with which the actual diffusion data approach the theoretical standard deviation curves, and the nonvariation of the diffusion constant with time are also given as evidence that the isolated lignin is a single molecular species. On the other hand, the magnitude of the multiple of the Einstein constant obtained from viscosity data and the deviation of the molecular weight calculated from the diffusion constant from the molecular weight from osmotic pressure both indicate that lignin has a shape factor of the order of 6 to 8. It thus appears that the true molecular weight is of the order of 3900 f 300, which is four times the empirical molecular weight. A possible set of molecular dimensions based on the film thickness obtained from spreading measurements and the shape factor is of the order of 3 X 16 X 100 A. U.
1132
D. L. LOUGHBOROUGH AND ALFRED J. STAhlJl
The lignin preparations prepared by methods which deviate so appreciably and which give soluble fractions representing such different portions of the total lignin all approached monodispersion and gave identical molecular weights within experimental error. These phenomena certainly indicate that the soluble fractions of isolated lignin are similarly associated, namely, four times the empirical molecular weight, of 900. REFERENCES (1) BECHMAXN! E., LIESCHE,O.,
~ N LEHMANS, D F . : Z. angew. Chem. 34, Bufsatzteil, 285 (1921). (2) BRIUSS, O., A K D HIBBERT,H . : Pulp 8r Paper LIag. Can. 34, 187 (1933). (3) B~-CHPER, E. H . , . ~ S D S.AMWEL, P.'J. P.: Trans. Faraday Soc. 29, 32 (1933). R.: Z. physik. Chem. MEA, 78 (1931). (4) EISEKSCHITZ, K. : Tannin, Cellulose, Lignin, p. 131. Julius Springer, Berlin (5) FREUDENBERG, (1933). (6) FRIEDRICH, h.,. ~ S DDIFALD,J.: Monatsh. 46, 31 (1925). (7) FUCHS, W.:Brennstoff-Chem. 9, 298,'363 (1928). W.:J. Am. Chem. SOC.68, 673 (1936). (8) FCCHS, IT., A N D DAER,R.: Cellulosechem. 12, 103 (1931). (9) FUCHS, (10) G I L M . ~ ~H.: ; , J. Am. Chem. SOC. 56, 464 (1934). (11) HBGGLUND, E., A N D URBAN,H . : Cellulosechem. 9, 49 (1928). (12) HARRIS,E.E., SHERR.4RDj E. C., A S D hlITCHELL, R . L. : J. Am. Chem. SOC. 66, 889 (1934). (13) JEFFERT,G. B. : Proc. Roy. SOC.London 102A, 161 (1931). (14) KL.ASOS,P. : Ber. 65B, 448 (1922). (16) KRAEXER, E. O., APD SEARS,G.: J. Rheol. 1 , 231 (1930). (16) KRAEMER, E. o., A N D \ ' A N ~ A T T A , F. J.: J . Phys. Chem. 36, 3175 (1932). (17) KUHN,IT.: Z. physik. Cheni. 161A, 1 (1932). (18) KUHX,W.: Kolloid-Z. 62, 269 (1933). (19) LAMM,OLE: Z. physik. Chem. 138, 313 (1928). (20) LAMU,OLE: Z.physik. Chem. 143, 177 (1929). E. 0 . : J. Am. Chem. SOC.67, 1369 (1935). (21) LAXSING,W.D . , A N D KRAEMER, (22) LICBAIN,J . W.,A N D LIU, T . H.: J. Am. Chem. SOC.63, 59 (1931). J. H., A K D Assox, hl. L.: J. Gen. Physiol. 12, 543 (1929). (23) NORTHRUP, F. : J. phys. radium 6, 497 (1934). (24) PERRIN, (25) PHILLIPS, M.:J. .4m. Chem. Soc. 49, 2037 (1927). , (26) PHILLIPS, M.: J. Am. Chem. SOC.60, 1986 (1928). (27) RASSOW,B., ASD GABRIEL,H.: Cellulosechem. 12, 227, 249, 290, 318 (1931). R , Wochbl. Papierfabr. 63, 103, 161, 243, 303, 342 (28) R ~ s s o w B., , A N D ~ T A G X EK.: (1932). (29) ROUTALA, O., A K D SEVOS,J.: Ann. Acad. Sei. Fennicae 29A,No. 11, 50 (1927). (30) SAMEC, hf., AND P I R K M A I E R , B.: Kolloid-Z. 61, 96 (1930). (31) SHERRARD, E. C., A N D HARRIS,E . E . : Ind. Eng. Chem. 24, 103 (1932). S. P. L.: 2 . physik. Chem. 106, 1 (1919). (32) SFJREKSEX, (33) STAUDINGER, H . : Trans. Faraday SOC.29, 18 (1933). (34) THEIS,R. M., A N D BELL,H. B.: J. Phys. Chem. 40, 12.5 (1936). P. : Rec. trav. chim. 60, 915 (1931). (36) VAFCAUPEN, (36) WAEKTIG,P. : Z. angew. Chem. 41, 493, 977, 1001 (1928). (37) WEDEKISD,E., . ~ N DKATZ,J. R.: Ber. 62B, 1172 (1929).