`
`parallelepiped having the dissolution rate
`Figure 2-Rectangular
`constants ki, nki, and bki.
`
`average dissolution rate constant, k,, ( M = 1.0, u = 0.5, p = 1.5, R = 1.0,
`and b = 10.01, and with isotropic behavior are presented in Table 111. Both
`profiles are almost the same. The ratio of the time necessary for 50%
`dissolution is L0.95. This result implies that the evaluation of the dis-
`solution for the acicular or flaky particles with nonisotropic behavior is
`roughly possible by means of Eq. 6 using the mean dissolution rate con-
`stant, k,,, which may be obtained experimentally as described.
`The basic assumptions behind the theory are that the constituent
`particles are rectangular parallelepipeds that are similar in shape and
`dissolve isotropically under sink conditions. In actual situations, more
`
`Autoxidation of Polysorbates
`
`factors may affect the overall dissolution profiles such as deviation from
`the log-normal law, irregularity in shape, and differences in the diffusion
`barrier for each particle, none of which is easily available. But these
`factors do not prevent an understanding of particle-shape effects on drug
`dissolution profiles, since they normally act independently of the ef-
`fects.
`In practical terms, the sizes of acicular or flaky particles measured
`microscopically or by an automated counter tend to be larger than those
`available for the evaluation of their dissolution profiles since the micro-
`scopic method does not always give the smallest side length but gives a
`larger one and the automated counter method gives volume diameter.
`
`REFERENCES
`(1) W. I. Higuchi and E. N. Hiestand, J . Pharm. Sci., 52,67 (1963).
`( 2 ) W. I. Higuchi, E. L. Rowe, and E. N. Hiestand, ibid., 52, 162
`(1963).
`(3) D. Brooke, ibid., 62,795 (1973).
`(4) Ibid., 63,344 (1974).
`(5) P. V. Pedersen and K. F. Brown, J . Pharm. Sci., 64, 1192
`(1975).
`(6) J. T. Carstensen and M. Patel, ibid., 64, 1770 (1975).
`(7) P. V. Pedersen and K. F. Brown, ibid., 65,1437 (1976).
`(8) “Handbook of Mathematical Functions with Formulas, Graphs,
`and Mathematical Tables,” M. Abramowitz and I. A. Stegun, Eds., Na-
`tional Bureau of Standards, Washington, D.C., 1965.
`
`M. DONBROWX, E. AZAZ, and A. PILLERSDORF
`Received September 6,1977, from the Pharmacy Department, School of Pharmacy, Hebrew Uniuersity, Jerusalem, Israel.
`publication March 31,1978.
`
`Accepted for
`
`Abstract 0 Aqueous solutions of polysorbate 20 undergo autoxidation
`on storage, with the peroxide number increasing and subsequently de-
`creasing again, the acidity increasing continuously, the pH and surface
`tension falling and tending to level off, and the cloud point dropping
`sharply until turbidity begins at room temperature. The changes are
`accelerated by light, elevation of temperature, and a copper sulfate cat-
`alyst. At the same time, hydrolysis occurs, liberating lauric acid. Analysis
`of the alterations in these properties leads to the conclusion that hy-
`drolysis has the major influence near room temperature and that oxy-
`ethylene undergoes chain shortening at temperatures above 40°. How-
`ever, evidence of degradation is detectable even in previously unopened
`commercial samples of polysorbates 20,40, and 60, warranting attention
`to the stability of and standards for these surfactants as compared with
`the solid alkyl ether type of nonionic surfactant.
`Keyphrases Polysorbates, various-autoxidation
`on storage, effect
`of light, temperature, and copper sulfate Oxidation-various poly-
`sorbates on storage, effect of light, temperature, and copper sulfate
`Stability-various polysorbates, autoxidation on storage, effect of light,
`temperature, and copper sulfate 0 Degradation-various polysorbates,
`autoxidation on storage, effect of light, temperature, and copper sulfate
`0 Surfactants-various polysorbates, autoxidation on storage, effect of
`light, temperature, and copper sulfate
`
`In view of accumulating evidence of the ease of autox-
`idation of polyethylene glycols and polyoxyethylene fatty
`alcohol ethers (1-4), it was suspected that other nonionic
`surfactants might undergo a similar process. Information
`about such reactions could increase the understanding of
`drug instability in aqueous solutions containing nonionic
`surfactants (1,3).
`The only systematic investigation of autoxidation in
`
`nonionic surfactants was carried out on cetomacrogol (1,
`2). Peroxides were formed and decomposed spontaneously
`at rates increasing with temperature and decreasing with
`surfactant concentration. Furthermore, the induction
`period for peroxide chain propagation was shortened by
`an increase of temperature, a reduction of pH, a copper
`sulfate catalyst. The period was also reduced by the ad-
`dition of chemical initiators, such as hydrogen peroxide
`or partially oxidized surfactant, and by free radical-ini-
`tiating processes, such as exposure to light or thermal
`treatment as in sterilization by autoclaving.
`During storage, the pH and cloud point fell and the acid
`content rose while the surface tension characteristics
`changed drastically. Polyglycols exhibited parallel changes
`in peroxide and acid content and in pH after autoclaving.
`These changes were interpreted as showing that degra-
`dation occurred in the hydrophilic chain with progressive
`reduction of the oxyethylene content until the hydro-
`philic-lipophilic balance fell below the critical value for
`solubility in water, when phase separation of the surfactant
`occurred at room temperature.
`Since sorbitan derivatives are used widely, knowledge
`of their stability is important. Their behavior relative to
`the fatty alcohol ether type of surfactant may govern the
`choice between these two agents in a formulation. In the
`present work, the decomposition of polysorbate 20 (poly-
`oxyethylene 20 sorbitan monolaurate) was studied sys-
`tematically at controlled temperatures. The results are also
`
`1676 I Journal of Pharmaceutical Sclences
`Vol. 67, No. 12, December 1978
`
`0022-35491 781 1200-1676$01.00/0
`@ 1978, American Pharmaceutical Association
`
`Page 1 of 6
`
`SENJU EXHIBIT 2097
`LUPIN v. SENJU
`IPR2015-01097
`
`
`
`relevant to the stability of other polysorbates. Some data
`on the purity of' polysorbates 40,60, and 80 indicative of
`the state of commercial samples also are reported.
`For comparison with cetomacrogol, the peroxide con-
`tent, pH, and total acidity were measured as parameters
`of chemical decomposition; cloud point and surface tension
`were used as criteria of physical changes bearing on both
`the decomposition process and the possible failure of the
`surfactant as a solubilizer or emulsifier in a formulation.
`
`THEORETICAL
`
`400
`
`300
`
`2
`
`200
`
`100
`
`Apart from peroxide formation, which might be expected to occur in
`all oxyelhyleiie-containing materials under suitable conditions, the
`possibility of hydrolysis also must be considered in the polysorbate ester
`surfactants. The first process is a chain reaction analogous to the per-
`oxidation of oils and ethers (5) and occurs by autoxidation reactions as
`shown in Schemes I and 11:
`
`RH + light or catalyst - R + H
`
`+
`
`Scheme I-Initiation
`I? + 0 2 + ROO
`ROO' + RH
`ROOH + R
`Scheme 11-Propagation
`Free radicals also may be formed by the processes in Scheme 111 and
`removed by those in Scheme IV:
`
`ROOH - ROO. + H.
`ROOH - RO. + .OH
`ROO + RO. + HzO
`2ROOH
`Scheme III
`-
`(a)
`2RO; --+ inactive products
`RO; + R -+ inactive products
`
`-+
`
`(b)
`
`(C)
`
`inactive products
`2 R
`Scheme IV-Termination
`On the basis that the peroxidation occurs in the hydrophilic chain in
`pcdyoxyethylene surfactants (1,2), the initiation and propagation steps
`(Schemes I and 11) yield hydroperoxides of the oxyethylene units ac-
`cording to Scheme V:
`
`OOH
`
`Scheme V
`The hydroperoxide concentration is measured iodometrically in oils
`and polyglycol solutions, and results are expressed as the hydroperoxide
`equivalent to the iodine liberated in niilliequivalents or millimoles per
`liter (5,6). Recalculation with respect to the weight of material under-
`going autoxidation gives the peroxide number (P.N.), expressed in mil-
`liequivalents per kilogram of surfactant (1,s).
`Peroxide Decomposition-Degradation
`of both hydroperoxides and
`peroxide free radicals may occur by a number of routes (7-10), as sum-
`marized elsewhere (1 1,12). The peroxide formation rate during the initial
`stage of propagation is normally faster than that of its decomposition.
`The rates subsequently become equal, giving rise to a short plateau
`representing a temporary steady state, following which decomposition
`is the faster reaction. The analytical data give the residual peroxide
`content, which is determined by the rates of the various simultaneous
`reactions involved (8,9); these reactions are temperature, concentration,
`and catalyst dependent in aqueous surfactant systems (1).
`The typical kinetic pattern, showing the rise to the maximum P.N.
`value and then the fall, is a clear indication of autoxidation with degra-
`dation. The presence of more than minimal quantities of peroxide (P.N.
`> 5) signifies that autoxidation is underway and has probably passed the
`lag phase, but a low P.N. value may also be obtained if the decomposition
`
`0
`0
`
`10
`
`20 0
`
`10
`DAYS
`Figure 1-Rate of peroxide formation in 3% aqueous polysorhate 20
`at 70,60,40, and 25' in daylight (a), darkness (b), or daylight with I X
`10-4 M copper sulfate catalyst (c). P.N. = peroxide number in millie-
`quivalents per kilogram.
`
`20 0
`
`10
`
`20
`
`rate is or has been high. Therefore, a low P.N. value is not in itself a cri-
`terion of nondegradation of the surfactant.
`chemical degradation
`Surfactant Chain Degradation-Various
`products have been detected, including carbonyl compounds and acidic
`products (12), but no stoichiometric relationship with the degree of de-
`composition of the surfactant has been established. Acid formation is
`readily measured by the change in pH and the total acid content, and it
`is an important indicator of the extent of degradation and the quality of
`the surfactant (1, 12). Increased degradation leads to larger amounts of
`low molecular weight acids and, thus, a lower pH. The terminal hydroxyl
`groups are relatively stable compared with hydroperoxide and free rad-
`icals, requiring highly acidic conditions with strong oxidizing agents for
`their oxidation to terminal aldehydic and carboxylic groups.
`However, substitution of a hydroperoxide radical in the n- or &position
`to the terminal hydroxyl leads to instability, which generally causes mi-
`gration and/or C-C or C-0 fission (9, l l , 12) with consequent formation
`of a two-carbon acid or formic acid, respectively. The latter has been
`identified as a degradation product (12).
`In view of the complexity of the degradation reactions and the difficulty
`of performing separations quantitatively in dilute aqueous surfactant
`solutions, physical-chemical methods have particular importance. Sur-
`face tension-concentration curves enable measurement of changes in the
`critical micelle concentrations (CMC); furthermore, the sub-CMC slope
`is determined by the surfactant area per molecule at the air-liquid in-
`terface if the molecules are close packed. In the oxyethylene type of
`nonionic surfactant, the area per molecule at close packing is determined
`by the number of oxyethylene groups present in the molecule and is rel-
`atively independent of the hydrophobic group due to coiling of the
`polyglycol groups (13). The surface tension value above the CMC is rel-
`atively constant and is also a characteristic property, becoming pro-
`gressively lower in a series based on the same hydrophobic group as the
`hydrophilic chain is shortened (2).
`The cloud point of nonionic surfactants is the temperature a t which
`turbidity appears on heating their aqueous solutions. It is related to the
`hydrophilic-lipophilic balance; in a series based on the same hydrophobic
`group, phase separation occurs at progressively lower temperature as t,he
`oxyethylene chain length is reduced (2), but the cloud point is also sen-
`sitive to the presence of additives (14).
`Changes in surface tension and cloud point properties were used to
`establish that degradation occurred in the hydrophilic chain in cetoma-
`crogol and to estimate the rate of loss of oxyethylene groups at 50"
`(2).
`For polysorbate 20, chain breakdown may be represented as shown in
`Scheme V1:
`- C~~HBCO(OR)
`CIIHZ~CWOR) (OCH2CH2),-, (OCHzCHz),OH
`(OCH2CH2),-x OH
`+ short chain degradation products
`Scheme VI
`where R is the sorbitan ring, n is the number of oxyethylene groups
`originally present, and x is the number of oxyethylene groups peroxidized
`and subsequently degraded.
`Hydrolysis of Polysorbate-Hydrolysis, which may be acid, base,
`
`Journal of Pharmaceutical Sciences f 1677
`Vol. 67, No. 12, December 1978
`
`Page 2 of 6
`
`
`
`Table I-Peroxide Formation in Polysorbate 20 a t Different
`Temperatures
`
`Time of'
`Induction Maximum' Maximum
`Period b,
`P.N.,
`P.N.,
`hr
`mEq/kg
`days
`
`1
`5
`2.5
`5.,i
`7
`7
`17
`
`Temper-
`ature
`
`700
`
`60"
`
`40'
`
`25"
`
`Conditionsn
`C + L
`<2
`50
`L
`5.5
`168
`D
`75
`-
`C t L
`130
`<2
`L
`10
`368
`D
`268
`-
`C t L
`<5
`440
`L
`-50
`0 7 0 )
`D
`0 7 0 )
`(>20)
`-
`C t L
`<24
`(>140)
`(>25)
`L
`144
`(>70)
`(>SO)
`D
`0 5 0 )
`(>50)
`-
`0 C = catalyst ( 1 X 10-4 M CuS04), L = light, and D = dark. * Based on P.N. of
`Numbers in parentheses indicate values at end of experiment before ter-
`5 (I).
`mination was reached.
`
`+
`
`or solvent catalyzed (15), would be expected to proceed at an increasing
`rate as the pH falls on formation of acidic degradation products, following
`the reaction in Scheme VII:
`CII H,,CO(OR) (OCHsCHZ), OH
`CIIH~:~COOH t H(OR) (OCH2CHz)"OH
`Scheme VII
`Of the products, lauric acid is stable and micelle soluble. The micelle
`saturation point would be determined by the increasing quantity of lauric
`acid and decreasing quantity of micelles, both functions of t.he hydrolysis
`rate, and also by changes in the micelle-solubilizing capacity as a result
`of chain shortening. The sorbitan polyglycol would pass out of the mi-
`cellar phase, raising the aqueous concentration of hydrophilic solute
`progressively and possibly reducing the cloud point of the surfactant. It
`would be expected to undergo peroxidation and degradation by the re-
`actions outlined for the parent surfactant; but since the process would
`occur outside the micellar phase, the rate constants and mechanism might
`not be identical with those of the polysorbate.
`Turbidity in the solution could be the result of either lauric acid sep-
`aration or reduction of the cloud point.
`EXPERIMENTAL
`Polysorbate 20 was neutralized before use to pH 6.00 with sodium
`hydroxide.
`Autoxidation was effected under the same conditions as used previ-
`ously in studies on cetomacrogol(1,2). The method was designed to en-
`sure adequate agitation and free access of air during storage, without loss
`of solvent by evaporation. Therefore, the oxygen concentration remained
`constant and was not rate limiting.
`The acid content, pH, cloud point (l), and surface tension (2) were
`determined as described previously.
`The peroxide number was determined using the spectrophotometric
`method developed for determining hydroperoxide in micellar solutions
`(6). Final readings were made at a concentration level of 1% polysorbate
`containing pH 6.00 buffer and potassium iodide at the same concentra-
`tions as described for cetomacrogol. Readings were taken at 360 nm i e
`= 11,400) and were time independent in this system. Dilutions, when
`necessary, were made with solutions of 1% polysorbate 20; readings were
`corrected by deducting the amount of iodine liberated by the polysorbate
`content of the diluent. The polysorbate used as the diluent was stored
`under nitrogen and refrigerated.
`
`RESULTS AND DISCUSSION
`three stages of autoxidation-uiz.,
`Development of Peroxides-The
`induction, propagation, and termination, were observed (Fig. 1). The
`peroxide number (P.N.) values at each st,age varied with conditions but
`were comparable with those of cetomacrogol under parallel conditions
`(1).
`Elevation of temperature from 25 to 70" reduced the induction period
`and raised the peroxide formation rate under all conditions. Copper
`sulfate and light had the expected catalytic effects, shortening induction
`and raising the peroxide formation rate relative to the dark uncatalyzed
`reaction. Pronounced catalysis of peroxide breakdown by the metal ions
`at 60 and 70" occurred (1,3, l l ) , with a shorter time required to reach the
`
`1678 I Journal of Pharmaceutical Sciences
`Vol. 67, No. 12, December 1978
`
`2.0 -
`
`1.5- .
`
`U w
`E
`>'
`Q 1.0-
`a 0
`
`0
`
`0-0
`0-0
`
`60°
`70'
`
`/
`
`I
`
`
`
`k 00038
`
`k 000061
`
`0
`
`10
`
`30
`
`1
`
`20
`DAYS
`Figure 2-Rate of acid formation in 3% aqueous polysorbate 20 at 70,
`60,40, and 25" in daylight with no catalyst. Broken lines represent lauric
`acid produced by hydrolysis and were calculated using the indicated
`ualues of k (hour-'), the first-order hydrdysis constant, on both acidity
`and percent hydrolysis scales.
`
`maximum P.N. value and a lower value of P.N. obtained (Table I). Ele-
`vation of temperature also catalyzed the decomposition of peroxide.
`Generally speaking, the greater the peroxide decomposition rate, the
`lower was the P.N. value at the termination stage, in which the degra-
`dation rate of the peroxides equaled or exceeded their formation rate.
`The enhancement of formation and decomposition rates by the tem-
`perature-catalyst combinations was such that termination was reached
`even at 40'; yet a t this temperature, the rate balance brought about the
`highest P.N. value. Even at 25' with the catalyst, the P.N. rose to 150
`within 25 days, and there was virtually no induction period (Fig. l c ) .
`pH and Acidity-As observed with cetomacrogol (l), the increase in
`acidity (Fig. 2) continued after the P.N. fell and was, therefore, the most
`reliable factor for following the degree of deterioration. The phenomenon
`of incessant increase in acidity in short chain polyglycols also was de-
`scribed by McKenzie (11). The rate of development of acidity had an
`inverse relation to the time of onset of propagation (Fig. l a ) and also to
`the initial rate of formation of peroxides.
`The pH value approached 4.0 at 25 and 40°, whereas at 60 and 70' it
`fell rapidly and continuously to 2.5 (Fig. 3a). The relation between pH
`and acid concentration, c (Fig. 3b), indicated that the acids developed
`at the lower temperatures contained weaker functions. Log acid con-
`centration-pH plots were linear at 40,60, and 70' and tended to converge
`at high acidity. By use of the equation pH = '/Z (pKa t log c), the inter-
`cepts at log c = 0 enabled estimation of apparent pKa values', which were
`3.6 f 0.5 at 60 and 70" and 4.9 f 0.8 at 40".
`These results suggest the presence of a larger fraction of stronger acids
`at the higher temperatures, which is consistent with greater rupture of
`the oxyethylene chains (Scheme VI). Indeed, in cetomacrogol, an ether
`for which hydrolysis is not to be expected, formic acid (pKa 3.75) con-
`stituted 50% or more of the acid formed during the initial stages of aut-
`oxidation under drastic conditions (12). The weaker acids present at 25'
`could be constituted of micelle-solubilized lauric acid (Scheme VII),
`carboxylated surfactant, or acetic acid (12).
`The hydrolysis rate of polysorbate 80 was reported to be relatively
`constant and lowest between pH 3 and 7.6, increasing rapidly as a func-
`tion of pH below 3 and above 7.6 (15). There was little difference between
`the hydrolysis rates of different polysorbate esters. To estimate the degree
`
`1 These pKa values represent mixed acid systems. The linearity over one order
`of concentration could indicate that the acid mixture has a relatively constant
`composition over this region in the aqueous phase. Lyophobic acids solubilized in
`the micelles would have a lesser influence on the experimental pH (16), and the pKa
`estimate would relate to their apparent pKa values in the micellar solutions and
`not to their aqueous pKa values.
`
`Page 3 of 6
`
`
`
`A-A
`0-0
`0-0
`0-0
`
`2 5 O
`L o ;
`60
`700
`
`7-
`b
`
`0
`I
`1 k \
`
`M'
`
`R,A
`
`C E T O M A C R O G O L 50°
`/
`
`O-0..
`
`35
`
`\ \ \
`
`A-
`0-
`0-
`0-
`
`A 25'
`0 100
`0 60°
`0 70°
`
`a
`
`7.0
`
`6.0
`
`5.0
`I,
`
`4.c
`
`3.c
`
`2
`
`0
`
`5
`
`5
`
`20
`
`10
`DAYS
`Figure 4-Rate of change of surface tension, y, of 3% aqueous poly-
`sorbate 20at 70,60,40, and 25" in daylight with no catalyst. (Cetoma-
`crogol, 3% aqueous solution at 50°, is shown for comparison.)
`
`
`
`2.c
`
`10
`
`30 0
`
`0.5
`1.5
`1.0
`20
`ACIDITY, mEqIg
`DAYS
`Figure 3-The pH change>f3% aqueous polysorbate 20 at 70,60,40,
`in daylight with no catalyst, d t h time (a), and as a function
`and 25'
`of total acidity at the corresponding time (b).
`
`of hydrolysis expected at the temperatures used in the present work,
`approximate rate constants, k , were calculated based on the rate constant
`reported for 0.02% polysorbate 80 at pH 3.95 and 80', utilizing the re-
`ported energy of activation (15). This calculation gave k values of 3.80,
`2.65,1.21, and 0.62 X 10-3 hr-1 at 70,60,40, and 25O, respectively, and
`these values were used to calculate the respective quantities of lauric acid
`that would be yielded at various times.
`Some of these results have been included in Fig. 2 for comparison. The
`values are of the same order as the observed acidities at 25 and 40°, al-
`though somewhat overestimatedz; the curves are also similar in form.
`However, at 70 and 60'. the quantity of acid formed on storage after 3
`and 6 days, respectively, greatly exceeded that expected theoretically
`from hydrolysis. Again, the upward curvature indicates a rate rising with
`time, as in cetomacrogol (I), characteristic of degradation processes of
`the chain-reaction type3.
`Surface Tension ( y ) Changes-The
`surface tension above theCMC
`decreased on storage and ultimately reached a constant value (Fig. 4).
`The rate of fall of y increased systematically with a temperature rise; the
`minimum values for 73% developed at 60 and 70' were lower than at 40
`and 25' (about 28 and 30 dynes/cm, respectively). The rate was much
`greater than observed under parallel conditions in cetomacrogol (1); a
`typical result is included in Fig. 4. There was a rank correlation among
`the rates of fall of y, increase of acidity, and fall in pH suggestive of an
`acidic reaction product increasing in quantity with temperature and in-
`fluencing surface tension.
`For the consideration of the hydrolysis of the fatty acid ester, tem-
`perature acceleration has already been discussed. The increasing quantity
`
`Hydrolysis rates are concentration dependent in acid solution, with the rate
`constant falling by some 60% at high surfactant concentratibn (15). This fact is the
`robable reason for the discrepancy; the 40 and 25" experimental acidities accord
`getter when the estimated k values are reduced by some 60%. as do the acidities
`at 60 and 70" during the first 2 days (Fig. 2). A similar reduction of the rate constant
`a t a high surfactant concentration also was observed for peroxide formation in ce-
`tomacrogol(1).
`,* The general picture and conclusions drawn would not be altered by the errors
`inherent in this treatment due to the approximated k values or to the possibility
`that the hydrolysis data on which they are based (15) are uncorrected for acids
`formed by autoxidation, which could be a serious source of error at 80'. particularly
`at low pH values.
`
`of hydrophobic lauric acid formed would be largely solubilized in the
`micelles. However, being in equilibrium with the surface, some acid would
`tend to be adsorbed there, forming a mixed surface film with the more
`hydrophilic surface-active monomers of the polysorbate. Mixed films are
`closer packed and are expected to give lower surface tension values than
`the separate amphiphiles (13). Confirmation that this was the probable
`explanation was obtained by measurement of the surface tensions of
`mixtures of lauric acid and polysorbate 20 (Fig. 5).
`The 71% value of the polysorbate fell steeply and almost linearly as the
`lauric acid concentration was raised to about 2.2% (w/w) in polysorbate
`
`[Ay/Sc: - -3.1 dynes/cm/l% (w/w) lauric acid in polysorbate or -61
`
`dynes/cm/mEq of lauric acid/g of polysorbate). Above 2.2% lauric acid
`(equivalent to about 10% hydrolysis of the polysorbate), the rate of fall
`decreased greatly to 0.32 dyne/cm/l% but remained approximately linear
`up to saturation concentration [9% (w/w) lauric acid, 0.45 mEq/g,
`
`A-A
`
`0-0
`0-0
`0-0
`
`2 5 O
`400
`600
`700
`
`25
`0.0
`
`1 .o
`0.5
`ACIDITY, mEqlg
`tension of 31'1 aqueous polysorbate 20 at 70,60,40,
`Figure 5-Surface
`and 25' as a function of total acidity at the corresponding time. Broken
`lines represent surface tension of lauric acid solutions in polysorbate
`20 ( I 7; w/u), expressed as milliequioalents of lauric acid per gram of
`polysorbate (V); and experimental surface tensions at 70" "corrected"
`for acidity due to autoxidation by plotting against hydrolyzed poly-
`sorbate at the corresponding time using k = 0.00211 h r - l ( + ) .
`
`1.5
`
`Journal of Pharmaceutical Sciences 1 1679
`Val. 67, No. 12, December 1978
`
`Page 4 of 6
`
`
`
`Table XI-Phase Separation Time as Measured by Surface
`Tension, 7, and Cloud Point
`
`Temperature
`
`25'
`40'
`60'
`700
`
`Phase Separation Time, days
`y Constant
`Cloud Point 59'
`12.5
`11.5
`15.5
`12-13
`10.5
`11
`6
`6.5
`
`equivalent to about 50% hydrolysis; Ay(tota1) = -9 dynedcm approxi-
`mately to ~aturation]~.
`The y-time curves of Fig. 4 are closely predictable by using selected
`values of the hydrolysis constants and estimating surface tensions from
`the data for the prepared mixtures corresponding to the amount of lauric
`acid formed at the time intervals, but they are highly sensitive t,o the k
`values selected. There was reasonable agreement on a y-acidity scale (Fig.
`5) between the lauric acid soluhilizate curve and the experimental curves
`at the three lower temperatures initially, hut t,he 70" curve deviated al-
`most, from the start and the 60" curve deviated at a later stage toward
`increased acidity at corresponding y values; both fell ultimately to lower
`y values than were accounted for by the lauric acid-polysorbate mix-
`tures.
`Degradation of the oxyethylene chains (Scheme V1) also would be
`expected to be temperature accelerated and, as noted, to give rise to short
`chain acids. The residual surfactant would have a lower hvdrophilic-
`lipophilic balance and, hence, be preferentially adsorbed on the surface
`or from a mixed surface film, as described for lauric acid. Consequently,
`lowering of surface tension should also occur during degradation and was,
`in fact, demonstrated in retomacrogol where a series of hexadecyl poly-
`oxyethylene ethers was used to determine the degree of degradation (2,
`1.3). One must assume that the y values at 70 and 60' in particular reflect
`the added effects of hydrolysis and chain shortening in view of the pattern
`of acid formation noted earlier.
`
`POLYSORBATE
`2 5 O
`A-A
`0-0
`G O O
`60°
`a-0
`70'
`0-0
`
`0
`\
`
`9 on
`
`80'
`
`k- E 12
`0 70'
`3
`3
`
`0
`
`60'
`
`50
`
`DAYS
`Figure 6-Rate of changf of cloud point of 3<,';> ayueous polysorbate 20
`at 70,60,-10, and 2 5 O in daylight with no catalyst, measured after ad-
`dition o/ I M NaC1. (Cctomacrogol is shoum for comparison.)
`
`4 Cetomacrogol 1000 gave a similar picture [A.r/Ar: - -1.9 dynes/cm/l% lauric
`
`acid up to 6% (w/w) lauric acid, after which the rate fell to -0.44 dyne/cm]. The total
`fall was ahout 12 dynes/cm because of higher lauric arid solubility [1S% (w/w)].
`
`1680 I Journal of Pharmaceutical Sciences
`Vol. 67, No. 12, December 1978
`
`This hypothesis was borne out by the y-total acid plot (Fig. 5). At the
`higher temperatures, the acid content was greatly in excess of the amount
`of lauric acid that could he solubilized in the micelles, this excess acid
`being constituted of short chain autoxidation products that were probably
`not very surface active. To test this proposition, the total acid formed at
`70" was "corrected," with y being replotted against the amount of lauric
`acid formed hypothetically by hydrolysis on the basis that k was about
`0.0021 hr-l. This approach brought the curve more closely into line with
`the data for the lower temperatures and with the curve for the lauric
`acid-polysorbate mixtures. Thus, at low acid concentration, most of the
`acid formed at 25,40, and 60' was accounted for by lauric acid; at 70' and
`at the higher acidities at FO", the acids formed by autoxidation became
`predominant. A t the same time, the sharper fall than expected of y at the
`lowest temperatures may indicate that oxyethylene chain degradation
`occurs here as well.
`The region of time independence of y (Fig. 4) is explained by phase
`separation at room temperature, as evidenced by the appearance of
`cloudiness, with the nature and concentration of water-soluble surface-
`active species at the surface remaining virtually constant. Phase sepa-
`rations occurred earlier as the temperature was raised because of the
`accumulation of larger quantities of the hydrophobic products. A t the
`same time, the higher terminal y values at 40 and 25", close to the lauric
`acid saturation y value, imply that the main reaction was that of lauric
`acid formation. The greater surface activity at 60 and 70" was compatible
`with the summated effect of the two reactions.
`Cloud Point-As
`in cetomacrogol(1,2), the cloud point of polysorbate
`20 solutions (3% containing 1 M NaCI) decreased during autoxidation
`(Fig. 6); but unlike in cetomacrogol, the rates of fall were not in rank order
`of temperature change. Results could not he determined accurately below
`60" because of cloudiness appearing at room temperature (the cloudy
`phase partially dissolved on heating but interfered with observations).
`The turbidity stage corresponded fairly well to the time at which y:3%
`reached a constant value at 70 and 60° but less closely at 40 and 25'
`(Table 11).
`The cloud point dropped unexpectedly rapidly at 25 and 40°, and room
`temperature turbidity appeared well within 15-20 days. Under compa-
`rable conditions, cetomacrogol exhibited little change in the cloud point
`at 50°, phase separation being evident only after 60 days of storage; even
`catalyzed solutions remained optically clear for some 30 days or more (2).
`At 60 and 70°, cetomacrogol ultimately exhibited a similar rate of fall of
`the cloud point to that of polysorhate, but there was a longer induction
`period (Fig. 6). These differences again'point to the hydrolysis reaction
`as a prohahle explanation. Lauric acid will be taken almost entirely into
`the micelles up to saturation concentration, above which turbidity will
`appear at room temperature; warming will clarify the solution due to a
`soluhility increase.
`The phenomenon was studied in some depth for benzoic acid and a
`series of substituted phenols (14). Except for certain aliphatic hydro-
`carbons, most solubilizates lower the cloud point. The effect is not
`straightforward, however, being dependent on factors such as concen-
`tration, type of polar group present, and site of solubilization in the mi-
`celle (see references in 14). In the present work, saturation with lauric
`acid (90 mg/g of polysorbate) lowered the cloud point of 3% polysorbate
`to 59" and that of 3% cetomacrogol to 64" (150 mg of lauric acid/g of ce-
`tomacrogol).
`The form of the 60" curve was deviant, showing a lag during the first
`6 days and then falling more steeply; i.e., the initial stage resembled 40°,
`and the later stage resembled 70'. T o the extent that the cloud point fall
`in polysorbate IS a function of polyglycol chain shortening, as demon-
`strated in cetomacrbgol (2). the chain breakdown is presumed to occur
`uia reactions of the hydroperoxides formed in the initial stages (Scheme
`V) and modified by acid catalysis (Figs. 7 and 8 in Ref. 1). The P.N. data
`(Fig. 1 and Table I) show that the peroxide stability was enhanced at 60'
`relative t.o 70", which could explain the initial lag in chain breakdown at
`60'. However, the absolute peroxide concentration increased greatly by
`the 7th day (P.N. 368), which would enhance the rate of reactions in-
`volving the peroxide species, and, indeed, the slope of the cloud point-
`time relation increased at the time corresponding to the peak P.N.
`value5.
`This time was also close to the time at which the pH drops below the