`
`Leon.-mo J. GORDON .-mo Rossar L. Scorn‘
`
`Vol. 7-l
`
`lCO:‘~'TRIBI.'TIO.‘~.' FROM THE DEPARTMENT or CHEMISTRX‘,
`
`'LT.~'IvERs1Tv or‘ CALl‘F(1R.‘x'Ie\, Los A.\'GE1..t-‘S!
`
`I. The System Phenanthrene-
`Enhanced Solubility in Solvent Mixtures.
`Cyclohexan-e—Methy1ene Iodidel
`
`Bv Lsomno J. Gonoon" AND Roses": L. Scorr
`RECIEIVED ?\Luzcn 8. 1.052
`
`The Hildebrand theory of regular solutions predicts that whenever the internal pressure of a solid (or rather the metastable
`supercooled liquid) lies between those for two solvents, there will be a range of solvent mixtures in which the solubility of
`the solid is greater than in either solvent alone. This enhanced solubility has been oliservetl in the system phenant|1rene—
`cyclol1e::ane—rnethylene iodide. allllotlgh the m;txi1I1In11 soliihilily of 1|};-_ fulyr-n;m1}]r9m_. in 1]“. um-i1m,m 1.-.01‘-enl A-njxmre fu]|.;
`sliorl of the ideal value prt-dictcrl by tho rlu-rnr_\'.
`
`Introduction
`
`Regular solution theory” when applied to the
`solubility of solid non-electrolytes gives solubility
`relationships in terms of the thermodynamic prop-
`erties of the pure substances, principally the melt-
`ing point and heat of fusion of the solute. and the
`so-callecl “solubility parameters" of
`the various
`components. The solubility parameter :5 is usually
`defined as the square root of the internal pressure
`or cohesive energy density, and may be calculated
`for a given substance in various ways." usually from
`the molar energy of vaporizatioxl and the molar vol-
`uine
`
`.5 = {.11=.",*\-'.1‘r'=
`
`{I}
`
`Where the solubility parameter of a solid solute
`(calculated for the supercooled liquid) is between
`those of two solvents, the theory predicts that there
`will be a range of solvent mixtures in which the
`solid is more soluble than in either pure solvent.
`As far as we know this effect has not heretofore been
`reported for three component systems of normal
`non-polar or slightly polar liquids, although an anal-
`ogous enhanccd solubility in solvent mixtures has
`previously been observed in the case of high poly-
`mcrs.‘-"'
`
`The system chosen for studying this enhanced
`solubility consisted of phenanthrene ('6 = 9.8. es-
`timatedl, cyclohexane (5 = 8.2) and CH2],
`(5 =
`11.8).
`In addition to satisfying the necessary solu-
`bility parameter conditions, this system is reason-
`ably free from the complications of hydrogen bond-
`ing and high polarity“ and the thermodynamic prop-
`erties of the components make the system a conve-
`nient one for experimental measurements.
`It was of interest to investigate the system fur-
`ther and, therefore, the two phase solvent—solver1t
`phase diagram was determined, and some aspects of
`the three-phase freezing diagram were determined.
`
`Theory
`The ideal solubility of a solid (expressed as mole
`fraction xi in its saturated solution} is _2‘i\'en by the
`equation
`
`RTIH 3*‘ = —.m"':_'I —-(1"‘.".",...Iu
`
`I.’2=
`
`(1) First presented at the Xllth International Congress of Pure
`and applied Chemistry. New York. N. "i'.. September 11. 1951.
`(la! Present address: Aerojet Engineering Corporation. Azusa.
`Calif.
`{2} _]'_ H_ Hildebrand and R. L. Scott_ ""I‘l1e Solubility 0|’ None|e.c-
`trolytes." Third Edition. AC5 Monograph HT. Reinllold. 1'35”
`(3) Reference 3. Chapter XXIII.
`r.1) G Giro. Tram‘. Fro-rm’-:_v Smz. I0. -1H-3 ll'J-HJ.
`E5! R. L. Scott. J. Cir.-nu.
`i"i:_1-3 . 1'1‘. 268 (1949).
`rm According to C P 5-=m_\-Lh and II. E. Rogers (Tms Joni.-'AI.. I2.
`'_-227 (19301). the dipole rnoment of Cf-Tel: is 1.1 De-by: units
`
`where A11” is its heat of fusion, and Tm its melting
`point.
`If the solution is not ideal. RTln x‘ is replaced by
`RT In rt. where n is the activity of the solute in the
`saturated solution. According to regular solution
`theory, the deviations from ideality are related to
`the partial niolal heat of mixing .-_\u; of the super-
`cooled liquid solute with the solvent
`R3"ln (.\-.i..=.\-.~ = rerun or. -.i-.. = in. = v.¢3(.s. — an
`(3)
`
`the ¢’s
`where the as are solubility parameters,
`volume fractions and the v’s molar volumes of
`components 0 (solvent) and i l solute).
`For three component systems the theory’ leads
`(with no further assumptions than those involved
`in the two-component case) to equations which are
`identical with Equation 3 if «.15.; is taken as the total
`volume fraction of solvents 2 and 3 and 5., is takcn
`as the volume fraction avera_t:t- of 59 and 53
`‘-.5--5~ -l‘ $53
`C»: ‘l’ $2
`= "53: "‘ Q
`
`.
`J
`
`‘
`
`n
`
`Consequently the thermodynamic behavior of
`the solute should not be affected by the [act that
`the solvent is a mixture * and its solubility should
`be equal to that in :1 hypothetical “single Iiquid" of
`average properties.
`It is easily seen that if 5, for
`the supercooled liquid solute lies between 5.», and
`53.. a proper choice of solvent ratios will make Bu ap-
`proach 5; and for the solvent mixture such that 5.; =
`5.. the ideal solubility should be observed.
`
`Experimental
`All Inatcrials were Eastman Kodak Co. "white label."
`and were not further purified.
`(Small traces of impurities
`should have no effect except
`to depress all temperatures
`slightly in a more or less uniform way.)
`Liquid mixtures were prepared having diflerent ratios of
`cyclollexatlc to methylene iodide. For each "single liquid”
`at series. of incasurcn1en1s of the solubility of phenarltllretie
`were made b_v the "synthetic" method: Solid and solvent
`were weighed and then sealer! in a glass ampule. The am-
`plllc was iwlrmc-‘l until the solid dissolved and then coolctl
`quit-l<l_v so that [he .-solid precipitated as finely divided crys-
`t.-ilr-'. Tlic alupule was again healed this time at a rate not
`exceerliiig oue~l1alf degree per minute and the temperature
`at which the last crystal of solid dissolved was noted.
`Figure 1 gives the experimental results plotted in the usual
`manner. The precision of the temperature measurement
`was within one-half degree. From the best straight line
`for each liquid mixture the solubility at a given temperature
`was determined. A series of solubility curvcs at different
`-'7'; Reference 3. Chapter XIII. p. 20].
`r8‘ Tllis com-lu.-.ion is valid’ only if the solid phase in eqtrilihriurn
`with the .~mlutirm i-. purc I'urn1\ont'nt
`I;
`solirl solution-a intrutlurc com-
`;JllI.'HllIIIflfi
`
`
`
`MYLAN PHARMS. INC. EXHIBIT 1048 PAGE 1
`
`MYLAN PHARMS. INC. EXHIBIT 1048 PAGE 1
`
`
`
`oooooorion"r;-ulruucgu
`
`
`not: FRICTION
`691.2 IH SOLVENT
`
`
`
`DOUGDGCFJ
`
`1').-(inm—~g
`
`553
`
`350
`
`31:‘
`
`J C
`320
`.330
`l'£InPEIt.n1‘uI1£ ‘I
`
`BOD
`
`2'0
`
`-
`
`2'0
`
`ITO
`
`0 Bl
`3“!
`
`Fig. 1.-—Solubility of pheiianthreno in mixtures of cyclo-
`hexane and methylene iodide. Dotted line represents ideal
`solubility.
`
`
`
`
`
`MoleFractionPhenonlnrene.
`
`0.35
`
`0.30
`
`0.25
`
`0.20
`
`0_l5
`
`0.00
`
`300
`
`§
`
`250 250
`
`rcueermune,-x.MI0«I(DOO
`
`Cali”
`X.
`CH-J:
`Fig. 3.—-Experimental phase diagram for the binary system
`methylene iodidthcyclohexanc.
`
`3|O
`
`ueennrune.-u.'32"’oo3
`1':
`
`I00
`
`290
`
`Cs.I‘I|Ig
`.Y.
`CIJI-,II2
`Fig. 3.A.—-Calculated phase diagram for the binary system
`methylene iodide—-cyclohexane (using thermodynamic par-
`ameters of Table 1).
`
`The solubility points were found to fall fairly well on
`straight lines on a log mole fraction rs.
`temperature plot
`(Fig. 1). Each straight line was drawn to best fit the points
`for that particular solvent mixture and no attempt was made
`to draw the lines so as to represent
`a. consistent family;
`however, the lines are nearly parallel. The effect of the
`slight difierences in slopes is somewhat evident in Fig. 2:
`if the lines had been made Consistent, these curves would he
`smoother.
`
`Discussion
`
`For comparison of experimental results with the-
`ory, the phase diagram Figs. 3a and 4a were calcu-
`lated using the values in Table I. The properties
`of the three components were taken as constants
`independent of temperature. The sp1veut—so1vent
`diagram was calculated by the method of Scatch-
`ar .
`The ideal solubility
`of phenanthrene was calcu-
`lated using its heat of fusion” AB? =
`43D0 calf
`(9) G. Sent-chard. Tnis Jounsnn. Gil. 24% (1940).
`H0} Inirrrirofioruni Critical Tables. 5, I34 U\‘.lcGraw-'il'i1l.
`
`l$I‘.“.IJ
`
`Aug. 20, 1952
`
`EN£I:t.\'{.‘i~2D SOLUBII.I1‘Y IN PHEN.\NrHRe.\:e—CrcLoiiEx.t:~:I-:~1\»'IEriirLt~:Ne looms
`
`4139
`
`temperatures are shown in Fig. 2; the enhanced solubility in
`certain mixtures is obvious.
`
`31.0
`
`OH
`
`020
`
`
`
`
`
`HOLEfllthflkPIEIIINIHREIIE 0.0!
`
`and
`
`X.
`CHgI3
`CaH1:
`Fig. 2.~—Solubilily of phenanthrene at selected tempera-
`tures as u function of composition of
`the eyclohexane—
`methylene iodide mixture.
`(Circles are not
`the original
`experimental measurements,
`but
`are
`taken from the
`smoothed curves of Fig. 1.)
`
`The two component system methylene iodidc-cyclohex-
`one was irwestiguterl by cooling various iuixtures and ob-
`serving the temperature at which heterogeneity (milkiness)
`appeared. The resulting phase diagram is giver: in Fig. :3.
`The freezing points of various two—cornponent mixtures
`were deterruinetl by inserting a. thermocouple in the mixture
`and recording the maximum temperature obtained after
`supercooling. The experirnental three-component diagram
`is given in Fig. 4.
`in addition to the experimental points
`shown. The freezing point: in the three-component region
`were obtained in part from interpolation between two two-
`component curves.
`Approximately forty solubility determinations were made
`for eight different solvent ratios including the two pure sol-
`vents and another forty experimental points were deter—
`mined in the solvent separation and freezing regions of the
`ph::=.e rliagmm.
`
`MYLAN PHARMS. INC. EXHIBIT 1048 PAGE 2
`
`MYLAN PHARMS. INC. EXHIBIT 1048 PAGE 2
`
`
`
`—l I-ll)
`
`LEONARD }. Gonnon AND ROBERT L. Scott
`
`Vol. 74
`
`CI-\|'l|o
`
`Fig. -l.—Experin1cnt:1l phase diagranl for the tertiary system
`phenanthrenu.--niethylene ioL|i(lc~cyclohexane.
`
`
`
`
`Hm
`
`cc”:
`CH: '2
`ternary system
`Fig. 4A.—CalcI.1late-rl phase diagram for
`pheiianthrene-methylene iodide—eyclohexane.
`
`mole at 96.3." No correction for change of An"
`with temperature was made.
`’l‘mn.e I
`
`Llonipuneut
`
`V. cc. “mule
`
`.5
`
`8!}
`Cflglg
`108
`Cyclohexane
`1-13
`Phenanthrene (solid at 111.13.)
`13.8”
`150“
`(supercooled liquid at 25 °l
`" A somewhat arbitrary value estimated from expansion
`of solid upon melting.
`9 Estimated from solubility data:
`_T. H. I-Iildebraud. E. T. Elleison anrl C.
`ll‘. Beebe. THIS
`_lf)T.?R.".'AL, 39, 2301 (1917).
`
`l. l . 8
`8. 2
`
`There is complete qualitative agreement between
`the calculated and observed phase diagrams; how-
`ever, there are several quantitative differences.
`Table II gives a comparison of the maximum ob-
`served solubility at each temperature, taken from
`Fig. 2, with the predicted ideal.
`The observed solubility of phenanthrene in CH=Ig
`(11) G. 3. Parks, Ind. Eng. C.‘-’mn.. 23, 1133 (|‘.'43lJ. There is some
`variation in the value of the melting point reporter] in the iircratnre.
`Use of a melting point of 101}: would make the idmil
`:~UllliJilil'._\' at 25''
`0.32? instead of 0.244
`
`TABLE II
`SOLUBILITY OF PRENASTHRENE
`
`-All concentrations expressed as mole fractions)
`Calculated
`,
`Observed
`ldenl
`‘Cam. In
`Z\-Iaximun:
`"ideal
`solubility
`3c.3,,in
`solubility
`C111-I::a
`Solvent"
`C:|H:u
`best solvent
`0.153
`0.49
`0.103
`0.42
`.200
`.-19
`.141
`.38
`.255
`.49
`.196
`.38
`.324
`.49
`.255
`.35
`. -l-03
`-lil
`. Iiiifi
`. 39
`
`I‘.
`°K.
`280
`290
`300
`3.10
`320
`
`Average
`
`0 . 38 :1: U. 02
`
`agrees with the theoretical prediction. The solu-
`bility in cyclohexane is far below that predicted.
`The solubility curve found here, however, is still
`"regular"' and if 6 = 9.8 is accepted as correct for
`phenanthrene, the apparent 6 of cyclohexane can be
`calculated from the theory (see Table III). Using
`the value 710, the predicted “idea1" solvent is found
`to have the molar composition 0.37 cyclohexane,
`0.63 CHQIQ. This agrees with the observed value
`within experimental error.
`TAELE Ill
`
`.‘5oLt's1L:rv OF PHE:\'.»\.\'THR.E.\'F. I.\' Cvcconaxana
`Mole fraction
`phenamhrenzv,
`xC14'iI:n
`0.0217
`.0328
`.0333‘
`.0415
`_tll5l'li:i
`_t}£J:5$)
`.144
`.173
`.404
`
`'1". ‘K.
`2S6.U
`293. ?
`298.1
`3Il2.8
`311.7
`317.5
`326.7
`328.2
`341.!)
`
`5: — fin
`2.92
`2.825
`2.89
`2.92
`2.8‘.-4
`2.79
`2,8~l
`£2.‘Fl
`2.74
`
`8.;
`(eyclohexanel
`6.9
`6.9
`6.9
`6.9
`5.9
`7.0
`7.0
`7.1
`?.1
`
`7.0 i 0. I
`
`It should be noted however that the 6 value 8.2
`for cyclohexane gives almost exact agreement be-
`tween theory and experiment for the so1vent—sol—
`vent critical temperature.
`The theory predicts that the ideal solubility will
`occur in a liquid having a composition such that
`5n = 51. Examination of the complete three-phase
`diagram reveals that no such "ideal" solvent exists.
`The mixture of solvents having the desired stoichio-
`metric composition is near enough to the solvent-
`solvent critical composition and temperature that
`extensive clustering must occur; consequently the
`solute is actually dissolved not in a homogeneous
`“ideal" solvent, but rather in a solvent which con-
`sists of microscopic regions richer either in cyclo-
`hexane or methylene iodide than the over-all aver-
`age composition would indicate. At higher temper
`atures where clustering is less significant we may
`expect the mixed solvent to be more nearly "ideal."
`Vife may try to allow for this clustering effect by
`using the more refined "quasi-chen1ical" equations
`for the free energies. Preliminary calculations (to
`be published later) indicate that this will account
`for part but not all of the discrepancy between the
`"ideal" solubility and the observed maximum solu-
`bility in the best mixed solvents.
`Los A.\-i,i«:J,1-:.-9. C.J\l.IFr m\'1.\
`
`
`
`MYLAN PHARMS. INC. EXHIBIT 1048 PAGE 3
`
`MYLAN PHARMS. INC. EXHIBIT 1048 PAGE 3