KITROCELLCLOSE DIFFCSIOS EXPERIMENTS* BY R. 0 . HERZOG ASD D. K R ~ G E R
The following communication deals with experimental work which was performed several years ago and which is not yet concluded. If cellulose is nitrated with so-called mixed acid, nitrocellulose products are obtained which differ according to the conditions employed (concentration of the mixed acid, time and temperat'ure of the action!. We have dissolved the nitrocelluloses made from cellulose of various botanical origins in suitable liquids and determined their diffusion coefficients, finding as was expected that the diffusion coefficient is dependent upon the raw material employed as well as upon the conditions of nitration. .in essential difference revealed itself however in the course of the investigation, which consisted of several hundred single experiments, namely, only in a relatively few number of experiments was the diffusion coefficient constant within a single experiment, i.e. for the four different heights in one diffusion cylinder; by far the majority of determinations displayed a deviation from Fick's law of diffusion. If, on the assumption that this law holds, our values for the concentration in the various layers of the diffusion cylinder are calculated by the use of Stefan's derived tables for the Graham determinations, one finds in these anomalous cases always the same trend: the coefficients for the topmost layer and, for the second layer from the bottom is too great; the coefficient of thp lowest and of the third layer from the bottom are approximately equal. (An indication of this trend is present also in the cases where the coefficients of all four layers may be considered as equal.) This trend in the values of the coefficients of diffusion indicates that the conditions on which the validit'y of Fick's law is based, i.e. that the solution is homodisperse and that the coefficient of diffusion does not depend upon the concentration, are not fulfilled in the present case. I n addit'ion to heterodispersit,y, the extent of which will be determined by the nature and pretreatment of the raw material, the conditions of nitration and dispersion etc., a splitting off of larger particles into smaller ones in the course of the diffusion process must be taken into consideration. This tendency to disaggregate on dilution will also vary with the kind of the cellulose material and other factors. Xs a matter of fact, certain experiments which one of us (H) has performed together with Mr. Lange, namely the investigation of nitrocellulose solutions by means of depolarized Tyndall-light point to such a disaggregation on dilution. I n this communication only those experiments are recorded in which the "heterodispersity" of the nitrocellulose solution may be disregarded. *From the Kaiser Wilhelm Institut fur Faserstoffchemie, Berlin-Dahlem.
I80
R. 0 . HERZOG AND D. KRUGER
Experimental Procedure Nitration The conditions of nitration were so chosen that the destructive action of the nitrating acid upon the raw material was as small as possible, this is thr case where the reaction is carried out a t a low temperature and with acid containing but little water. An acid consisting of 71,jo:T H2S04,18.6jc: HSOa and 9.63y0 water was used a t o°C. The ratio by weight of fibre to mixed acid was I :IOO. The fibres were in part scoured by a simple boil in 1‘; soda solution and in part especially purified in the manner described below. The “purified” materials were heated as follows: 2 0 grams of the cellulosr material, previously freed from impurities, was introduced into 400 cc of soda solution and boiled for two hours with the cellulose submerged boiling 17~ below the surface of the liquid. The hot liquor was then gradually displaced by cold water. The expressed cellulose was next treated at 3 j°C for two hours with 400 cc of a bleaching powder solution containing 2 . 7 gm C1 of per litrr, soured with hydrochloric acid and washed n i t h water until acid frer. Then followed in the same manner a second boil with 15% soda solution, a second bleaching with chloride of lime solution containing z grams of C1 per liter, a third boil in 1 7soda ~ and a third bleach in 1.3 gram per liter C1 and finally one more boil in 1 7soda. ~ The cellulose after washing in running water until free of alkali was dried in the air. Experiments were also carried out with only one 13 hour boil in 1ycsoda solution. Experiments on different materials (hemp and alkali cellulose) show that within rather wide limits ( I O minutes to 3 hours treatment) the diffusion coefficient is independent of the duration of the nitration. In all further esperiments the nitration period was three hours unless otherwise noted. Determination of the velocity of nitration under the same conditions indicated t’hat after I j minutes some 7% K had been taken up, while the maximum S content was reached after about I + hours. Those products which werr nitrated for shorter periods were only partly soluble in acetone. The X content of the product produced by a three-hour nitration with a mixed acid containing 7 1 . 7 7 ~H2S04,18.657~ HSO, and 9.63Yc water at ocC amounted in the case of nitrated cotton, regardless of its preliminary treatment (mercerization, action of acids) to 12.8 - I Z 9Yc; in the case of nitrated wood cellulose the 7S content was lower (11.8 - 1 2 . 3 7 ~ which ) is in accord with previous observations. The nitrocelluloses produced in this manner are easily soluble in aceton(,, methylethylketone, amylacetate, nitrobenzol, pyridine and chloroform. The solutions produced in the cold are highly viscous or jelly-like even with concentration of less than 1 7nitrocellulose. ~ The solutions in hot acetone or methylethyl ketone also show high viscosities. The viscosity varies considerably according to the nature and preparatory treatment of the raw material. In ether-alcohol or methyl alcohol these nitrocelluloses are insoluble,
+cc
181
NITROCELLCLOSE DIFFSUION EXPERIMENTS
Unstabilizcd nitrocellulose decomposes gradually when kept at ordinary temperature with the splitting off of nitrogen oxide, In order to determine to what extent the properties of the above described nitration products are altered in storage, the following experiments were carried out. The nitrates were first washed thoroughly with cold water. For nitrated hemp, after storage for seven months only a small increase in the diffusibility was observed, while in the case of nitrated pulp no change had occurred after four months. Since commercially produced nitrates, of the sort here in question, also remain (for our purpose) unchanged for a long time without any special stabilizing treatment,, this procedure was omitted in the experiments dealing with the comparison of different raw materials, and the nitrates were simply washed with cold running water, until a blue litmus paper when pressed against the fibres showed no tendency to turn pink. This method was also chosen because after-treatment of nitrocellulose may affect considerably its colloidchemical properties. By boiling nitrocellulose in water preferably in the presence of an acid or :zlcohol, not only the stability increases, but also the viscosity of the subsequent solution in acetone (a fact which is well known). The increase in viscosity by this method of stabilization was also observed in a few experiments on nitrated cellulose as follows: Cellulose was nitrated for I hour at zo°C with an acid mixbure of the com~ 7 ~ and 18.4j7, water. The nitrocellulose poeibion: 61. j 3 % H2S01, ~ 0 . 0"03 was first thoroughly washed in cold running water, a part of it was then further washed for I hour in water, twice boiled for periods of one hour in 1% soda solution, then boiled for an hour in water (the water from this first water boil gave a neutral reaction with litmus.) The nitrocellulose was then dissolved by heating with acetone for one hour under a reflux condenser and the period of flow of the solution a t zo°C was determined with the Ostwalti viscosimeter. Concentration acetone
Discharge Time (seconds)
1.74 1.63
31.8
g/100
Unstabilized nitrocellulose Stabilized nitrocellulose
cc
18.1
I n the industry no such energetic stabilization as boiling with 1 7soda ~ solution is employed. By this treatment a considerable quantity of substances which are regarded as nitro-oxycelluloses and nitro-hydrocelluloses and which form the less dispersible part of nitrocellulose are dissolved out. I n general the nitrocelluloses were used within a few days of their production. Solvents The choice of suitable solvents for diffusion experiments on nitrocellulose is limited by the following considerations: I. The solvent must be sufficiently pure and stable and must not react with the nitrocellulose.
R. 0. HERZOG ATD D. K R ~ ~ G E R
182
2. Solvents which are specifically heavier than nitrocellulose (chloroform for example) can not he used without altering the method. 3. Very low boiling solvents are inconvenient for practical reasons. We used principally acetone (Bp. j6.3') and methyl-ethylketone (Bp. 81")~ occasionally also other solvents such as amylacetate. The commercial ester was freed from acid, dried and distilled; in the case of the other solvents, the pure preparations of Kahlbaum were used directly. The presence of moisture within certain limits was without effect upon t'he results: therefore in general air-dried nitrocellulose was employed. The nitrocelluloses prepared as described gave even in small concentrations with cold acetone, amylacetate, etc. (insofar as the raw material had not received extensive chemical pretreatment), highly viscous gelatinous solutions which were difficult to maintain homogeneous. It was therefore found advisable to effect the solution by heating with the solvent concerned. Experiments with methyl-ethylketone solutions indicated that the duration of heating (2-4 hours) was without influence. The question as to what extent the different solvents cause aggregat'ion must be reserved for further investigation. The nitrocelluloses were produced therefore, when not expressly otherwise stated, by a three-hour nitration with a mixed acid whose composition was 7 1 . 7 0 7 ~ H2S04,1 8 . 6 5 7 ~HKOs and 9.63Yc H20, washed with cold water and dried in air. The air dry nitrocellulose was dissolved by treatment for four hours with the boiling solvent, usually acetone or methyl-ethylketone and the solution finally centrifuged.
Diffusion The diffusion was carried out in the apparatus of Oeholm'. The experiments were conducted in a cellar whose maximum temperature variation in the course of a day was 0.1' and in the course of the whole experimental period 0.j". Because of the volatilit,y of the solvents employed, special precaution were taken to limit as far as possible the evaporation during the long time required for t,he experiments. To the capillary tube on the upper part of the diffusion cylinder, a bent glass tube with a blown bulb in it was connected and this tube was drawn out to a long fine capillary. This dipped into a vessel which was filled with the solvent and which was connected to the outer air only through a capillary tube. The diffusion coefficients at to in organic solvents when calculated to the diffusion coefficient in water a t 20' (Dw20) give within the limits of experimental error, equal values for solutions in acetone, methyl-ethyl ketone, ethyl formate and methyl alcohol. After tapping off the xitrocellulose content of the four layers was determined by evaporation in vacuum to constant weight. From the amount of substance thus found the diffusion coefficient was ascertained by the use of Kawalki's tables and curves. 'L.FY. Oeholm: %. p h y s k Chern., 50, 309 lI904! ~
SITROCELLULOSE CIFFUSION EXPERIMESTS TABLE
1
Temp. Time in (to) days(s)
Content of layer
mg.
%
D.
Xitrocellulose from purified hemp, 12.8% K, dissolved in hot methylethyl ketone.
DWPO
0.018 0.018 0.017
0.019
0.018 13.6 The same nitrocellulose kept at room temperature seven months and then dissolved in hot acetone.
0.068
0.024
0.071
0.025
0.056 0.083 __
0.019 0.029
Mean
0.069
0.024
j4,8 31.8 8.9
0.032
0.015
0.029
0.014
0.026 0.043
0.012
B = 2 1 4 . 8 Mean
0.032
0.015
29
0.047
0.022
0.039
0.018 0.019
19
162.4 107.4 41.1 22.7
__
B = 333.6 Nitrated ramie, 12.8% K, dissolved in hot methylethyl ketone
12.8 29
117.8 68.3 19.2 9.5
Xitrated flax, 1 2 . 8 % K,dissolved in hot methyl-ethyl ketone
Nitrated cotton, IO.jyc K,dissolved in cold acetone
12.8
127.4 88.9 37.5 20.4
E = 274.2 38
2 14.2
=
38
=
0.040
0.027
Mean
0.046
0.021
98.5 77.7 43.8
39.9 31.5 17.8
0.054 0.045 0.046
0.019 0.016 0.016
26.6
10.8
0.054
0.019
- __
246.6
Mean
102.6
77.6 41.2
0.050
0.017
41.6 31.4 16.7
0.049
0.017
0.045
0.016
0.042
0.01j
25.2
10.2
0,OjI
0.018
246.6
Mean
0.04;
0,016
S
0.020
- __
o.oj9 __
The same, dissolved in hot acetone
4.4
46.5 32.4 13.7 7.4
14.2
48.7 32.2 12.3 6.8
-
R. 0. HERZOG .4SD D . KR6GER
TABLE I (Continued) Temp. Time in (to) dayslz)
Sitrocellulose from sulfite pulp, 12.3qCS , dissolved in hot acetone
13 3
Sitrated a-cellulose from sulfite pulp', ~ 2 . 5 S ~ ;, dissolved in hot acetone
46 j 31 'i I3 1
0.04j o.oj2
0.036
0.018 0.013
2 7 4
8;
o.oj7
0.020
316.3
Mean
0.04;
0.017
128 6
48 I 31 j
100
12
9
34
=
84
13 3
27
I
293
T O
2 j 2
9 4
34
0.019
o.oj2 0.029 o.oj9
0.024 0.014
0.045
0.021
0.02;
0.98
0.034
1.02
j
0.90
3
1.04
0.036 0.031 0.036
j
Mean
0.9;
0.034
5 88 o 43 9
44 0 30 8
0.056
0.026
O.Oj0
0,033
4
o.oj0
0.023
281
9 8
0.067
0.031
= 78j j
3Iean
0.061
o
43
9
0.041
;
8
88 ; 58 9
12
0.016
4
IIO
I : = 301
Sitrated viscose .ilk, 1 2 . 7 5 S , dissolved in hot met hyl-et hyl ketone
D.
3 41 5
14; I
28
S
Kitrocellulose from German sulfite pulp, 12.3cIC S , dissolved in hot methylethyl ketone
Content of lazer mg
1
125
36 29 19 14
15
-
Kitrated alkali-cellulose? which has been ripened 28 days. Dissolved in hot acetone. Kitration time I O nun.
13 3
2;
028
8
41 3
0.069
0 024
81 4 46 8
29 8
0.091
0.032
I;
2
0.064
0 022
31 9
11
7
0.082
0.029
0.0;6
0.02;
TIZ
__
s = 272
9
Mean
'.The a-cellulose from sulfite pulp was obtained from artificial silk wood pulp accordjng to the instructions of Schwalbe, the other a: celluloses as well as the @ cellulose according to the process of Lenae. Pleuss and Nueller. According t o their instructions, a three-fold treatment with alkali is nzcessary to remove the hemicelluloges. d lengthening of the alkali treatment is however useless, since hv each increase in time only small and approximatelv constant quantities go into soluhon and these consist of the soluble products formed by"the action of the alkali itself. 2 By alkali cellulose is meant here as in the artificial silkindustrv-woodpulp which ha. been soaked in approximately 1 6 S.%OH ~ ~ liquor, ~ pressed until i t weighed about three times as much as the original pulp and stored for a certain period.
18;
SITROCELLULOSE DIFFUSION EXPERIMESTS
(TABLE I Continued) Temp. ‘Time in (to) days(zj
C,ontent of layer mg.
Ditto, nitration time 60 min.
13.3
0.1421
2;
0.1120
o.ojo8 0.0484
5
38.1 0.076 0.076 30,o 1 9 . u o.06j 0.08j 13.0 ___
~
E = 0.3733 Mean
Ditto, nitration time 3 hours
13.3
0.067 3 0 . 3 0.076 17.6 0.063 0.069 11.3
133.8 4 0 . 9
28
99.0 57.j 37.0 P I
0.076
- 327.3 Mean
3.069
0.027 0.02;
0.023 0.030 ~
0.027
0,023 0.02; 0.027 0.024
3.026
The diffusion time was reckoned from the beginning of the admission of the solut,ion into the cylinder t o the beginning of the drawing off of the first layer. The filling of the cylinder required a t least 30 minutes, the tapping off of each layer an average of 2 0 minutes. The duration of the diffusion n-as usually so chosen as to be most favorable for the evaluation (with regard to the slope of the concentration-time-curve) of the first, third and fourth layer; in the second layer, on the other hand, the quantity of material present was usually in the maximum region or in the flat part of the curve after the maximum; the D values of the second layer are for this reason less trustworthy than the rest. The concentration used seldom reached I % nitrocellulose. For every solution, two parallel experiments were made. I n the tabulated data, the result of one experiment only is given in each case.
Experiments In Table I the solvent, the diffusion time in days (z), the temperature (t), the content of the layer in milligrams and in percent of the total content are given. Cnder D stands the calculated diffusion coefficient for the layer and O coefficient calculated for water a t zo°C (for comparison of under D ~ Z the the solvents and of the numbers independent of temperature). \T7e add here another group of experiments, Table 11, which refer to the influence of the duration of nitration by a mixed acid considerably richer in water content. They are taken from the dissertation of E. Kausmann whose work was performed in our Institut’e. The composition of the acid was 20Cc H K 0 3 , 61.55: H2S04,18.5:; H.0: and the nitrating temperature 2o°C.
I 86
R. 0. HERZOG AXD D. KRUGER
TABLE I1 Temp.
Sitrated cotton dissolved in hot acetone. Time of nitration 3 j min.
Ditto, nitration time 60 min.
(to)
Time in days (2)
1 1 .j o
31.1
12.5
11.;
D
36.9 0.086 30.7 0.071 18.j 0.070 13.9 0.089 -B = 239.6 Mean 0.079
34.0
u v
Ditto, nitration time 65 m h .
Content of layer mg. 70
--
Dsza
70.0
0.031
58.2 8j.o 26.4
0.02j
51.7 35.1 41.9 28.4 28.6 19.4 25.3 17.2
147.j Mean
0.02j
0.032 0.028
0.080 0.029 0.089 0.032 0.065 0.023 0.093 0,033 - 0.082 0.029
0.028 0,077 0.070 0.025 19.0 0.066 0.024 14.1 0.079 0.028 - 0.026 B = 207.9 Mean 0,073
31.1
76.3 62.7 39’5 29,4
36.7
30.2
_c_
The following summary of the diffusion coefficients calculated for water a t zo°C. shows that the natural fibres yield values between 0.015 and 0.021, The values for alkali cellulose (see below) are not different from those for viscose, so that the diminution in particle size in the manufacture of artificial silk must occur chiefly in the “preripening” process. The last three series of experiments show that under other than the above specified conditions of nitration greater values are obtained for the same raw material: D,zo increases from 0.016to 0.028 (the particle diameter is inversely proportional to the diffusion coefficient). For our purposes it was most expedient to work under conditions which would lead t o the smallest practicable change in the particle size, Table 111, as against the cellulose crystallite. TABLE I11 Dria
Cotton Hemp Ramie Flax
0 .o r 6
0.018 0.OIj 0.021
Sulphite pulp1 “Alkali cellulose” a-Cellulose Viscose silk
0.019 0.026
0.034 0.028
1 Because of the impurities in sulfite pulp experiments with it cannot be considered positively dependable.
SITROCELLULOSE DIFFUSION EXPERIMENTS
TABLEI V Kitrated cotton (12.8% N) in acetone. Diffusion time days
1
Content of the layers mg. 70
Temp. (t"C)
(2
2.8
I3,j
80.2 16.0 7.1 4.4
-
z= 11.0
2 6.0
107.7
81.6 15.8 7.5 5.1
12.8
=
81.3 15.5 7.8 5.3
12.7
62.7 26.0 1 1 .j
9.9 __ =
74.1 14.5 6.8 4.6
110.0
12.8
z
74.5 14.9 6.6 4.1
73.9 14.2 7.1 4.8
57.0 23.6 10.4 9.0
110.1
Nitrated mercerized cotton in acetone. 7
14.0
2 I9
=
93.5 26.8 4.6 2.6 127.5 76.0 33.3
14.0
10.0
8.3
-?-
u
- 127.6
73.3 21.0
3.6 2.0
59.6 26. I 7.8 6.j
I88
R. 0 . HERZOG ASD D. KRUGER
It remains to give the results of further experiments which are not recorded in the foregoing tables. The content of so-called a-, P- and y-cellulose does not stand in any easily understood relation to the observed diffusion distribution. I n this regard the diffusion coefficient of the nitrate from a-cellulose is always smaller than that for nitrated P-cellulose. Mechanical treatment, for example in a Hollander mill, increased the diffusion coefficient. Two series of experiments remain for discussion. If as for example in the following experiments the diffusion coefficient be calculated for the lowest layer for different time periods on the assumption that Fick's law is valid, it decreases sharplv with time, whileunder the same assumption- the coefficient for the topmost layer increases. With increasing time the values for D in the different layers approach nearer t o equality. It appeared possible that disturbances which are FIG.I present a t the beginning, immediately after the two Concentration gradient after different time liquids brought in contact might be caused by the periods. (Photometric nitrocellulose solution swelling as it stands beneath thr determination of the Tyndall-cone on the pure solvent, in other words: that the true diffusion of surface between solu- the particles does not occur immediately but rather that tion and solvent). the solution a t first increases in volume bv the absorption of solvent, therefore the lowest layer is lifted, so to speak, into the second. Such a process has been observed by R. 0. Herzog and R. Gaebel' when viscose is covered by a layer of alkali. I n the experiments with nitrocelluloFe to be sure such a disturbance is not so probable because of the small concentration used. Its absence was in fact proven in the following manner. In a small bulb 1vc nitrocellulose solution (in acetone) Ras carefully pipetted under acetone with the aid of a fine capillary. At first every ten minutes, later every 30 minutes and finally every hour the Tyndall-cone on the intersurface of the solution and solvent was photographed and the photograph evaluated. If swelling occurred then the Tyndall-cone should have raised in relation to a mark on the bulb. This however was not the case, the photometer curves on the contrary, from which the concentration distribution a t various heights of the bulb could be read (since the Tyndall cone represents a relative measure of the concentration of colloid particles), showed a normal course. Immediately after mixing a normal concentration gradient began, which with time mounted even higher in the bulb. The following graph represents such photometer curves (somewhat smoothed out) curve I was obtained 5 minutes after the two liquids were brought in contact, I1 after 20 minutes more, I11 'Iiolloid-Z., 35, 193 (1924);39,252 (19261.
TITROCELLULObE DIFFUSIOS E X P E R N E S T S
I 89
one hour later, 11- 13 hours after 111. For the execution of such exprriinent.5 wr inuet thank N r . H. Kunze. The particle sizes, which were calculat,ed from the diffusion coefficients of the nitrocelluloses made by the mild process described above, lie approxinintely in the neighbolohood oj the cnlues .found from the width of the S-i.acl 2'rrIerjewnren for natural cellulose as well as for the nitrated fibres.' The cellulose crystallite therefore holds together under certain conditione of nitration I ) in fact the single crystallites disintegrate but little or not at all, even when the nitrated fibre is dissolved. The colloid particles in the nitrocellulose solution correspond to the cellulose crystallite in the fibre. This behavior is the more surprising since dilute (1-2$) solutions are stable for 6 weeks and longer. I t seems to us of particular interest to search farther for the causes which determine this cohesion and to seck an exacter understanding of the Ian-r of the di-integration of colloid particles,-whether they be static or controlled by a dynamic equilibrium?). One m w t expect here in the case of organic colloids quite different reactions, and chemical action also must be carefully considered. For these reasons we h a w begun to rxtend our investigation with the aid of an ultracentrifuge.
Summary Diffusion experiments with nitrocellulose in diff erent solvents are described. The diffusion coefficient's obtained when calculated to water a t zo°C as the dispersing medium, lie mostly between 0.01 j and 0 . 0 2 1 for the natural fibres. Methods for obtaining such solutions are described; these methods however do not suffice t o guarantee the production of a hornodisperse solution of the indicated particle size, still less to explain more exactly the true causes of the cohesion and dispersion. A few disturbances are pointed out and the line of future investigation is indicated. For example, in the case of ramie the particle diameter of the natural fibre is found t o be 128 X IO-* cm., and after nitration and subsequent denitration, 130 X 10-8 em., i.e. constant. R. 0. Herzog: Svensk Pappers Tidning, 1927, 2 3 1 (contains% misprint]. >I. Yolmer: Z . physik. Chem., 125, I j~ (1927).
