International Journal of Pharmaceutics, 83 (1992) 65 —72 65

1992 Elsevier Science Publishers B.V. All rights reserved 0378-5173/92/$05.00

UP 02752

Characterization of adsorption behavior by solid

dosage form excipients in formulation development

Richard I. Senderoff, Majid Mahjour and Galen W. Radebaugh

Parke Da r’is Pharmaceutical Research Division, Warner-Lambert Co., Morris Plains, NJ 07950 IUSA)
(Received 17 October 1991)
(Modified version received 6 January 1992)
(Accepted 7 January 1992)

Key words: Adsorption; Solid dosage form excipient; Microcrystalline cellulose; Croscarmellose sodium;
Sodium starch glycolate; Ionic strength effect; pH effect; Adsorption

The adsorption of drugs onto solid dosage form excipients may influence dissolution characteristics, analytical testing and
bioava ilability. CI-977 ([5ft-(5n,7a,8Q)]-N-methyl-N-[7-(pyrrolidinyl)-1-oxaspiro[4.5]dec-8-yl]-4-benzofuranacetamide monohydro-
ch loride), a x-opioid agonist analgesic compound, was found to adsorb onto microcrystalline cellulose, croscarmellose and sodium
starch glycolate. The extent of adsorption was affected by pH, ionic strength and ionic species. The absorptive capacity and
relative affinity for CI-977 adsorption to microcrystalline cellulose was characterized under various experimental conditions using
Langmuir isotherms. The results show that electrolytes inhibit adsorption by reducing both the affinity of the adsorbent for the
adsorbate and the adsorptive capacity of the adsorbent. Divalent cations reduced the adsorption to a greater extent than
monovalent cations. The effects of pH and electrolytes suggest that the predominant mechanism of CI-977 adsorption to
microcrystalline cellulose is electrostatic attractive forces.

Introduction cological screens, the projected oral human dose

is less than 1 mg. In the present study, the ad-
CI-977 (Fig. 1; [5ft-(5n,7n,8Q))-N-methyl-N-[7- sorption of CI-977 to solid dosage form excipients
(pyrrolidinyl)-1-oxaspiro[4.5]dec-8-yl]-4-benzofu- was studied in order to determine its significance
ranacetamide monohydrochloride) is a centrally in formulation development.
acting x-opioid agonist intended for use as an In the development of oral formulations many
analgesic in the treatment of mild to moderate factors must be considered. They include: (1) the
pain. The hydrochloride salt form of the com- drug shpuld be stable in the presence of formula-
pound ‹ pK a 8.8) was found to be soluble ( > 200 tion excipients; (2) the drug should be recovered
mg/ml) and thermally stable. Based on pharma- from the formulation excipients; (3) the drug
should exhibit acceptable dissolution characteris-
tics from the formulation; (4) processing should
Correspondence. R.I. Senderoff, Parke Davis Pharmaceutical allow for acceptable content uniformity; and (5)
Research Division, Warner-Lambert Co., Morris Plains, NJ the formulation should have acceptable physical
07950, U.S.A. characteristics. Many of these aspects are ad-
 O croscarmellose sodium, N.F. (Ac-Di-Sol) were ob-
tained from FMC Corp.; pregelatinized starch,
N.F. (National 1551) from National Starch and
Chemical Co.: corn starch N.F. from Husingcr
y”
Co.; dicalcium phosphate dihydrate USP from
Stauffer Chem. Co.; lactose, N.F. (Fast-Flo) from
Van Waters and Rogers, Inc.; magnesium stcaratc
N.F. from Mallinckrodt, Inc.; and sodium starch
Fig. 1. Structure of CI-977 ([S R-(.$ a,7r‹,8Q)]-N-methyl-N-[7-
glycolate, N.F. (Explotab) from Edward Mendcll
(pyrrolidinyl)-1-oxaspiro[4.5]dec-8-ylJ-4-benzofuranacetamide Co. Single lots of each excipient were used
monohydrochloride). throughout this study.

Analytical methods
dressed in preformulation through drug-excipient CI-977 was assayed by an HPLC method. The
compatibility studies (Carstensen, 1990). Chemi- HPLC system was comprised of a Hewlett
cal stability is often the primary concern with Packard HP1090 liquid chromatograph, a Kratos
drug-excipient compatibility evaluations. How- 783 variable-wavelength detector (equipped with
ever, the drug-excipient physical interactions can a 2.4 pl flow-cell, â = 250 nm), a Hewlett Packard
also affect the formulation performance and the HP3393A computing integrator, and a Beckman
development of analytical methodology. For in- ODS (250 x 2.0 mm, 5 p m) column. Two sets of
stance, although the interactions resulting in the chromatographic conditions were employed. The
adsorption of drugs onto solid dosage form excip- first consisted of a mobile phase containing 0.05
ients are generally of a weak type such as van der M triethylamine adjusted to pH 3 with phospho-
Waals forces and hydrogen bonding, they have ric acid (75 Vr), tetrahydrofuran (12.5%) and ace-
been shown to influence dissolution characteris- tonitrile (12.5%) with a flow rate of 0.2 ml/min.
tics (Aboutaleb et al., 1956; Aly and Udeala, An alternative mobile phase was used to separate
1987) and bioavailability (Calis et al., 1986; El chromatographic interference caused by pregela-
Gamal et al., 1986; Aly and Megwa, 1957). Ad- tinized starch, corn starch and dicalcium phos-
sorptive behavior can also affect content uniform- phate dihydrate. This mobile phase consisted of
ity following a wet granulation process. Zografi 0.05 M triethylamine/0.05 M ammonium dihy-
and Mattocks (1963) have shown that the migra- drogen phosphate/0.025 M sodium octanesul-
tion of water-soluble dyes during tablet manufac- fonate adjusted to pH 3 with phosphoric acid
ture can be related to adsorption with excipients. (70Hr) and acetonitrile (30%) with a flow rate of
Additionally, analytical methods must be devel- 0.6 ml/min.
oped taking drug-excipient adsorptive behavior
into account to ensure accurate assays (Pramar Ei aluation oJ’ potential for CI-977 to adsorb to
and Gupta, 1991). Subsequently, it is important excipients
to characterize drug-excipient adsorption behav- The potential for microcrystalline cellulose, di-
ior during formulation development. calcium phosphate dihydrate, croscarmellose
sodium, sodium starch glycolate, corn starch and
pregelatinized starch, to adsorb CI-977 was evalu-
Experimental ated using the following procedure. 50 mg of
excipient (10 mg of sodium starch glycolate was
Materials used due to its gelation properties) was placed
Cl-977 was supplied by Parke-Davis Pharma- into 1.5 ml polypropylene centrifuge tubes. 1 ml
ceutical Research (Ann Arbor, MI). Microcrys- of a stock solution composed of either water (pH
talline cellulose, N.F. (Avicel PH 102) and 5.8), phosphate buffer (0.05 M, pH 7.0), citrate
buffer (0.05 M, pH 5.0), 0.9% NaCl or 0.1 N HCI
 67

containing 1.0 g/ml of CI-977 was added to

each tube. The samples were shaken for 1 h
(ambient temperature) using a rotary shaker (lKA-
Vibrax-VXR, lKA-Works, Inc.). Controls (without
excipients) were treated identically. Fol- lowing
the 1 h equilibration period, the samples
were centrifuged (Eppendorf Centrifuge 54l5C)

% and the supernatants were assayed. The recovery

was determined as the peak response compared
with controls. Samples were prepared and evalu-
ated in duplicate.
TU (m n.I
Preparation of yowder blends and dissolution anal-
ysis Fig. 2. Typiciil plot indicating equilibrium adsorption occurred
Three powder blends containing 10 mg CI- within 15 min. Experimental conditions were: 1).5 g of micro-
crystalline cellulose dispersed in 10 ml of CI-977 stock solu-
977/200 mg blend were prepared by geometric tion (5 p g/m1, distilled water).
trituration in a mortar. The formulation composi-
tions were (blend A) microcrystalline cellulose /
croscarmellose (2%)/magnesium stearate
(().25%); (blend B) lactose (spray dried)/ prege- strength adjustments was made on the basis of
latinized starch (10Ho) magnesium ste a rate preliminary adsorption experiments, which indi-
(0•2 8); and (blend C) dicalcium phosphate dihy- cated that limited adsorption occurred at higher
drate/ pregelatinized starch (10*/r)/magnesium ionic strengths. The vials were equilibrated for 1
stearate (0.25Wc). The powder blends were hand- h at 25 + 0.5°C on a mechanical shaker water
filled into hard gelatin capsules (200 mg blend bath (Versa-Bath S, Fisher Scientific). Prelimi-
capsule). The capsule sizes used were nos 1, 3 nary experiments established that equilibrium ad-
and 2 for capsule blends A, B and C, respectively. sorption was reached within 15 min (Fig. 2). Fol-
Dissolution behavior was evaluated in distilled lowing the 1 h equilibration period, samples were
water using USP Method II. Samples were inter- centrifuged at 25 1.0°C (BHG Hermle Z 360 K
mittently withdrawn over a 2 h time period and centrifuge) and the supernatants were assayed by
assayed by HPLC. HPLC to determine the equilibrium concentra-
tion. Samples were prepared and evaluated in
Characterization of C/-977 adsorption to micro- duplicate.
cwstalline cellulose The effect of pH was evaluated using the fol-
The effects of ionic strength and pH on ad- lowing procedure. CI-977/ microcrystalline cellu-
sorption of CI-977 were evaluated using micro- lose suspensions were prepared and evaluated by
crystalline cellulose as the modcl adsorbent. The the method stated above. Stock solutions contain-
effect of electrolyte was determined using the ing 50 p g/ml CI-977 were used for the pH effect
following procedure. A series of CI-977/micro- studies. The pH of the stock solutions was ad-
crystalline cellulose suspensions were prepared justed using 0.1 N HCl or 0.1 N NaOH. An
by the addition of 10 ml of a CI-977 stock solu- appropriate amount of sodium chloride was added
tion (5, 10, 25, 50, and 100 p g/ml) to 20 ml glass to adjust the ionic strength of each stock solution
vials containing 0.5 g of microcrystalline cellulose. to 0.001. The pH of each suspension was deter-
The CI-977 stock solutions were prepared in dis- mined following the equilibration period. Ad-
tilled water without ionic strength adjustment, sorption was evaluated in the pH range 3— 10. CI-
and also in distilled water with the ionic strength 977 is stable over the experimental time pe- riod
adjusted to 0.0005 and 0.001 with either sodium under these conditions. Samples were pre- pared
chloride or calcium chloride. The degree of ionic and evaluated in duplicate.
Results and Discussion The results of the evaluation of the potential for
 CI-977 to adsorb to excipients are listed in Table
1. As can be seen, complete recovery of CI-977
was not achieved from microcrystalline cellulose,
croscarmellose or sodium starch glyco- late
suspensions in water. However, the adsorp- tion
was eliminated in the case of microcrystalline
cellulose and reduced in the cases of croscarmel-
lose and sodium starch glycolate from phosphate
buffer (0.05 M, pH 7.0), citrate buffer (0.05 M,
pH 5.0), 0.1 N HCI and 0.9% NaCl solutions. In
this way, it was demonstrated that analytical
methods and formulation development should
take the physical interactions between CI-977 and
solid dosage form excipients into account. Ad-
sorption behavior is often the result of hydropho- T it (min.)

bic interactions with lipophilic compounds. The Fig. 3. Dissolution of CI -977 from powder blends. ( ■ ) Blend
propensity for adsorption of a drug has been A, (•) blend B, ( z ) blend C.
related to the oil/water partition coefficient as a
measure of the lipophilicity (Fung, 1990). The
solubility of CI-977 suggests that mechanisms mellose (powder blend A) exhibits a rapid disso-
other than hydrophobic interactions contribute to lution phase followed by a slow, incomplete disso-
adsorption to these solid dosage form excipients. lution phase over 2 h. Capsule blends containing
The protocol in which the excipients were excipients without potential to adsorb CI-977
evaluated for potential to adsorb CI-977 repre- (powder blends B and C) exhibit rapid and com-
sented exaggerated experimental conditions in plete dissolution (within 15 min). The results with
terms of CI-977/ excipient and excipient/medium each powder blend are consistent with the previ-
volume ratios. The influence of excipient adsorp- ous results in that excipients with potential to
tion on dissolution analysis in water of powder adsorb CI-977 cause a delay in dissolution. The
blends is shown in Fig. 3. The powder blend dissolution behavior from powder blend A is un-
containing microcrystalline cellulose and croscar- acceptable, considering the high solubility of the
compound.
The results of the evaluation of excipients for
potential to adsorb CI-977 suggest a relationship
between electrolytes and pH with the extent of
TABLE 1
Summary of recot'ery data from the evaluation of CI-977 adsorption to excipients

Excipient ' Fraction of CI-977 recovered

Water Phosphate Citrate HCI NaCl
pH 6.0 pH 7.0 pH 5.0 pH 1.1 pH 5.9
Microcrystalline cellulose 0..34 0.99 1.00 1.01 0.99
Dicalcium phosphate
dihydrate 1.02
Croscarmellose 0.22 0.GB 0.68 0.74 0.72
Sodium starch glycolate 0.60 0.94 0.95 1.00 0.95
Pregelatinized starch 1.00
Corn starch 0.99
Fraction of CI-977 recovered from suspensions containing 1y g/ml of CI-977. Values represent the mean fractional peak
response (n = 2) compared with controls fwithout excipients).
' Suspensions were prepared using 50 mg of excipient except for sodium starch glycolate in which 10 mg was used.
 f›9

adsorption. In order to determine potential ad-

sorption mechanisms and the significance of CI-
977 adsorption to the development of solid dosage
forms, the adsorption of CI-977 to microcrys-
talline cellulose was further characterized. The
effect of added electrolyte on adsorption was
evaluated using Langmuir isotherms. The equilib-
rium concentration Ce , mg/100 ml) and the
amount of drug adsorbed in mg per g of micro-
crystalline cellulose (A/ñf) were determined un-
der various experimental conditions. The
adsorp- tion isotherms are shown in Fig. 4. The
linear form of the Langmuir adsorption isotherm
EQJlLlBFtlW CONCENTRAT IN (mo/ 1OC' mil
is given by:
Fig. S. Langmuir plots of CI-977 on microcrystalline cellulose
where ionic strength and ionic species was varied. ( a ) Dis-
(1) tilled water; (0) NaCl solution,p = 0.0005; ( G ) NaCl solution,
X/M k ,k 2 p = 0.001; (z) CaCl 2 solution, = 0.0005; (w) CaClt solution,
2
= 0.001.
where k and k 2 are constants. The constant k 1
is a measure of the relative affinity of the adsor- both reducing the affinity of the adsorbent for the
bate for the adsorbent. The constant k 2 is the adsorbate (k decreases with increasing ionic
adsorptive capacity of the adsorbent (mg of ad- strength) and by reducing the adsorptive capacity
sorbate which can be adsorbed by 1 g of adsor- of the adsorbent (k 2 decreases with increasing
bent). The linear plots of the Langmuir adsorp- ionic strength). The results also demonstrate that
tion isotherms obtained under varied ionic condi- the extent of adsorption (in terms of both k and
tions are shown in Fig. S. The adsorption con- k 2 ) wit s more sensitive to the divalent cation
stants obtained from these linear fits are summa- (Ca 2*) than the monovalent cation (Na +).
rized in Table 2. This analysis of the data indi-
Adsorption mechanisms have been previously
cates that added electrolyte inhibits adsorption by
attributed to electrostatic attractive forces be-
tween a drug and an adsorbent (Carstensen and
Su, 1971; Franz and Peck, 1982; Hollenbeck,
L2O -
1983). Carboxyl groups (px, — 4.0; Edelson and
1.OO
Hermans, 1963) can be formed on the surface of
microcrystalline cellulose by oxidation of the
0
.80 hy- droxy groups on individual anhydro-glucose
units

O.40
TABLE 2
Summary of constants obtained from linear plots of the Lang-
muir equation for CI-977 adsorption to microcrystalline cellu-
lose
0

0
Ionic strength Distilled
EDLJIL IBRILA4 COrJCENTRAT ION (mgY 100 ml)
0.001 0.0005 water
Fig. 4. Adsorption isotherms of CI-977 on microcrysta)line Ionic species: NaCl CaCl 2 NaC1 CaCl
cellulose where ionic strength and ionic species was varied.
I a ) Distilled water; (•) NaCl solution, p = 0.0005, ( 0 ) NaCl k 0.()709 0.0481 0.0936 0.0689 0.271
solution,p = 0.001; foJ CaCTt solution,p = 0.0005; (oJ CaCl 2 k 1.35 0.748 1.62 1.18 1.65
solution, p = 0.001. r2 IN.965 0.879 0.991 0.957 0.996
70
(McBurney, 1954; Edelson and Hermans, 1963;
Mark, 1965; Franz and Peck, 1982). Hence, at
pH > 4.0 the surface of microcrystalline cellulose 50 1.OO
becomes negatively charged due to ionization of g 0.90
O.8Oi
the carboxyl groups. An adsorption mechanism 4 0.70
based on ionic interactions can be described in its 0
simplest form (constant stoichiometry, i.e., 1 : 1) 0g 30
by the following relationship 0.50T O
0 *’ 0.40@
20
[X ]+[Y’] (X-Y] 1
0.30
5
where [X*], [Y +] and [X — Y] represent concen-
5
trations of the anionic, cationic and bound O'
species, respectively. This relationship defines an 0.OO

equilibrium adsorption constant, N„d , for a de-

fined set of conditions (i.e., concentration of sub- Fig. I›. Effect of pH on percent of CI-977 adsorbed (50 pg/ 10
strates): ml to P.5 g of microcrystalline cellulose). ( ■ ) *Zr CI-977
adsorbed; ( G ) calculated ionization fraction coefficient.

K.,d - (2)
cient as a function of pH, based on the ionized
where /g - and /g. represent the ionized frac- fraction of CI-977 ( l a 8.8) and the ionized
tion of [X] and (Y] at a given pH, respectively. fraction of anionic surface sites (assuming car-
Rearrangement leads to the following equation boxyl groups are present with px., —— 4.0) on mi-
for the concentration of bound species ([X—Y]), crocrystalline cellulose. The ionized fractions of
assuming the stoichiometry remains constant with CI-977 were calculated using the following equa-
tion:
varying fractions of ionized species:

[X — Y] = d la —la [ ][ ]' ad F[X] [Y] (4)

(3)
where p Na represents the pe a of the cationic
where F represents the ionization product coeffi- species (CI-977). The ionized fractions of anionic
cient. The ionization product coefficient ranges surface sites on microcrystalline cellulose were
from 0 to 1 and varies with pH depending on the calculated using the following equation:
pl, of the adsorbent and adsorbate. According
to Eqn 3, the maximum adsorption will occur
(when the total concentrations of adsorbent and (5)
adsorbate are kept constant) as F approaches
unity, since the adsorbent and adsorbate will be where x a Represents the pl, of the anionic
in their most favorable state of ionization for species (microcrystalline cellulose). As can be
adsorption. The contribution of electrostatic at- seen, as the pH increases from 3 to 6, the percent
tractive forces to CI-977 adsorption to microcrys- of CI-977 adsorbed increases most likely due to
talline cellulose was determined by evaluating the ionization of the carboxyl groups on the cellulose
effect of pH on adsorption. The results are de- surface. Likewise, as the pH increases from 8 to
picted in Fig. 6 in which the percentage of drug 10, the percent of adsorbed CI-977 decreases
bound is plotted vs pH. Also shown in Fig. 6 are since less ionized drug is available. Assuming a
calculated values of the ionization product coeffi- mechanism of adsorption based on electrostatic
interactions between the adsorbate and the ad-
sorbent, an estimate of the pH where maximum
the intersection of the ionized fraction curves
(maximum ionisation product coefficient) for drug
adsorption (pHq„) occurs can be calculated as and anionic surface sites. An equation to calcu- late
the intersection point can be derived by setting the 71
equations for the respective ionized fractions equal.
This yields the following equation to estimate pH q. versely affected, due to the pH and abundance of
„: cations (Ca 2 +, Mg 2 +, Na“ and K +) in gastric
fluid. However, adsorption to these excipients
(6) should be considered when developing in vitro
testing methods (i.e., selection of dissolution
medium) to ensure valid interpretation of dissolu-
where px„ and px, represent the pt., of the tion data and accuracy of analytical methods,
anionic and cationic species, rcspectively. The especially for low dose compounds such as CI-977.
calculated pH g„ value (6.4) is dependent on the Subsequently, screening excipients for adsorption
pl, estimates and does not take other adsorptive potential is useful in formulation development.
mechanisms into account, such as adsorption of
the nonprotonated free base. These factors may
account for a shift of the maximum adsorption
observed, from the calculated pH q„ value. The References
maximum adsorption found experimentally was
at pH 7.1.
Aboutaleb, A.E., Rahman, A. A.. Saleh, S.l. and Ahmed,
M.O., Factors affecting the release of diaze pam from directly
compressed t ublets: I. Interaction of diazepam with certain
Conclusions ‹lirect ly compressible veh icles. SI“P Pharma, 2 ( l98G) 23—
27.
The effects of pH and electrolytes suggest that a Aly, S.A.S. and Megwa, S.A., Drug excipient interaction. Effeci
of adsorption of oxytetracycline hydrochloride by some
predominant mechanism of adsorption of CI-977 to tablet excipients on the physiologic av'a ilab ility of the
microcrystalline cellulose ' is electrostatic at- tablets. STP Pharma, 3 (1987) f›52—fi57.
tractive forces between positively charged drug and Aly. S.A.S. and Udeala. O.K., Drug excipient interaction. Effect
the negatively charged surface of the micro- of absorption of oxytetracycline hydrochloride by some tublet
crystalline cellulose. The adsorption capacity of excipients on the dissolution prtifile of the tablets. .S TP
Plnirrnu, 3 ( 1957) 23—27.
microcrystalline cellulose is reduced due to in- Calis, S., Sum n u, M. and Hincal, A.A.. I n vitro ads‹arption of
creased competition between positively charged propranolol hydrochloride by various antacids. Drup Det .
drug molecules and other cations. The adsorption hid. Pltui nt., 12 (1986) 1833— 1 S4fi.
of counterions can cause a reduction or a reversal Carstensen, J.T., Preform ulation. In Banker. G.S. and Rhodes,
of net surface charge. This could account for the C.T. (Eds), modern Pluirma‹ critics. Dekker, New York,
19914, pp. 2fi 1 —2fi3.
reduced affinity of CI-977 for microcrystalline Carstensen, J.T. and Su, K.S.-E., Nature of bond in g in naont-
cellulose in the presence of additional cations morillonite acisorbates. I: Surface adsorption. J. Pli‹irnn
(especially cations of high charge number). Sci., hL1 (1971 ) 733 735.
Hollenbeck (1988) has shown that bioavailabil- Edelson. M.R. and Hermans, J., Flow' of gels of cellulose
ity of phenylpropanolamine HCI was not affected microcrystals. 11. Effect of added electrolyte. J. Polj'm. S‹ i. C,
2 (19b3) 145—152.
by ion-exchange interactions with croscarmellose El Gamal. S.S., Boraie, N.A. and Naggar. V.J., An in vitro
even though a significant drug-excipient interac- adsorption study of fluraze pam on some ant acids and excipie
tion was observed in distilled water. Likewise, oral nts. Pha i‘in. Ind„ 48 ( 1986) 1207— 1210.
bioavailability of CI-977 from formulations Franz, A.M. and Peck, G, E„ 1n-vitro adsorption-desorption ot
containing microcrystalline cellulose, croscarmel- fluphenazine dihydroch loride and promethazine bydro-
chltiride hy microcrystalline cellulose. 1. Pharm. Sci., 71
lose or sodium starch glycolate may not be ad- (1982) 1193— 1199.
Fung, H.-L., Chemical kinetics and d rug stability. 1n Banker,
G.S. and Rhodes, C.T. (Eds), 4 oderit Pharma‹ •uti‹ s,
Dekker, New York. 1990, p. 2 t7.
Hollenbeck, R.G., Bioavailability of phenylpropanolamine HCI
from tablet dosage forms containing croscarme Hose
sodium. hit. J. Ph mm., 47 ( 1988) 89—93.
Hollenbeck, R.G., Estimation of the extent of drug-excipient
72

inleractions involving croscarmellose sodium. J. Phariti. Pramar, Y. and Gupta, V.D., Quantitation of scopolamine
Sci.. 72 11963) 325—327. hydrobromide when adsorbed onto microcrystalline cellu- Mark,
H.F., Gaylord, N.G. and Bikales, N.M., The Encyclope- lose and sodium carboxymethyl cellulose. Druy Det . Ind.
dia of Polymer Science and Technology, , Vol. 3, Wiley. New Pharm„ 17 tl99l 2401 —2407.
York, 1965, pp. 170—178. Zografi, G. and Mattocks, A.M., Adsorption of certified dyes
McBurney, L.F., High Polymers, Vol. 5, Part 2: Cellulose anâ by starch. J. Pharm. Sci., 52 t1963) 1103—105.
Cellulose Derry atit e.s, Wiley, New born, 1954.