`8 Elsevier Scientific Publishing Compzny, Amsterdam - Printed in Belgium
`
`DETERMINATION
`PURITY
`CALORIMETRY
`
`BY DIFFERENTIAL
`
`SCANNING
`
`ERWIN E. MAR-i-I
`Central Research Serrices Department, C&z-Geigy Lrd., Bosel (Srcitzerland)
`
`(Received April 3rd. 1972; revised June 23rd, 1972)
`
`is
`for purity determination
`literature on the DSC method
`A review of the
`aspects,
`i-c- theory,
`sampIe
`presented, with a discussion
`of the most
`important
`handling,
`calibration
`of the instrument,
`evaluation of melting curves, and the con-
`ditions and accuracy of the measurement
`of eutectic
`impzlrities.
`for
`equilibrium
`A number of mathematical
`descriptions
`of the solid-Iiquid
`eutectic binary systems is applied to the calculation of theoretical phase diagrams and
`specific heat functions, which are then compared with experimental phase diagrams
`and melting curves. The applicability of the DSC method
`to systems of soiid solutions
`is discussed.
`by computer methods
`the evaIuation
`and
`procedure
`the experimental
`Both
`required
`to obtain accurate
`impurity determinations
`by DSC are presented. A number
`of practical examples
`is included_
`
`The measurement of the melting point of a substance as a method of identifica-
`tion dates back
`to the early days of chemistry. Many different observations
`on
`organic and inorganic
`substances were made during the thermal
`treatment necessary
`for a melting point determination.
`like
`in terms of phenomena
`The observations were summarized and interpreted
`polymorphism,
`sublimation,
`thermal decomposition,
`solid solutions, eutectic systems,
`congruentIy-melting
`compounds,
`glass
`transitions
`and others. Koffer’
`turned
`the
`meiting point deter=rnination by microscopic
`observation
`into an extremeIy useful
`method
`in the field of anaiyticaI chemistry. Kotier’s
`treatise on purity determinations
`is excellent, but of course,
`today,
`it is not easy to agree with the statement
`in T/rermo-
`method
`of purity determination
`with
`the microscopical
`observation of the melting point, however, will finaliy replace all the others”. Some-
`how, the development of the analytical methods
`for purity determination
`since 1950
`has appeared
`to prove
`the opposite, namely
`that ali the other analytical methods
`would replace
`the melting point determinations.
`Koffer’s meIting point method
`is
`nowadays performed with many different
`types of apparatus. The method
`is used
`
`mikromethoden: “The
`
` P. 1
`
`UT Ex. 2031
`SteadyMed v. United Therapeutics
`IPR2016-00006
`
`173
`
`
`because it is the simpkst analytical method for getting information about the purity
`and about
`the crystaI form of the sample under investigation. The melting point
`method
`is based on the determination
`of the absohrte
`temperature of the substance
`assuming an infiniteIy sma.II amount of solid substance
`in the solid-liquid equilibrium.
`A reference standard of a high purity is required to make the temperature measure-
`ment only a reIative one. This high purity standard
`is also used for the relation
`between the purity and the melting point difference given in Eqn. (1)
`
`AT=T,-2-s=xo-kr ('1
`
`where
`is the melting point difference in ‘K, 2-r is the meIting point of the high
`purity standard
`in “K, T, is the meiting point of the sampIe in “K, x0 is the moIe
`fraction of the impurity, and
`is the cryoscopic
`constant
`in “K.
`The cryoscopic
`constant
`is defined as
`
`.
`
`k, =RT:
`A%
`
`(3
`
`where R is the gas constant and AHfSI is the molar heat of fusion of the high purity
`standard, and is experimentally determined by means of Eqn. (1) or with a measure-
`and the melting point of the reference standard.
`ment of the heat of fusion
`
`MTRIC OXIDE
`
`Fig. 1. Heat capacity of nitric oxide measured by Johnston and Giauquef. (Mdting point TX =
`109_49%.)
`
` P. 2
`
`UT Ex. 2031
`SteadyMed v. United Therapeutics
`IPR2016-00006
`
`AT
`k,
`AH,,,
`EEAT CAPACITY C!F
`/
`
`
`175
`
`Today, a second method seems to repIace at Ieast partiahy the microscopic observation
`of the melting point. This second method
`is known as differential scanning calori-
`metry @SC). The DSC method measures the endothermic amount of ener_q which is
`afforded by the premehing process of substances.
`-The method of premelting as a
`purity determination dates back to the 1920’s in a form used by Eucken and Karwat’
`and Johnston and Giauque3
`for the measurement of the heat capacity of nitric oxide
`in the melting point region.
`In 1929, Johnston
`and Giauque3
`reported
`from
`the
`Chemical Laboratory
`of the University of Cahfornia
`in BerkeIey on the heat capacity
`of nitric oxide from 14°K
`to the boiling point.
`The paper of Johnston and Giauque is interesting enough for a brief discussion_
`In Fig. I, the heat capacity of nitric oxide is shown as a function of temperature,
`according
`to the measurements
`of Johnston
`and Giauque. The extremely
`sharp
`melting region of the nitric oxide sample at about 110°K should he noted. The nitric
`oxide used by Johnston
`and Giauque was produced by the reaction of potassium
`nitrite and potassium
`iodide in distilled water. The generated nitric oxide was purified
`over sever-a1 distillation steps.
`As an exampIe, the same purified sampie, containing no = 3.769 moles of nitric
`oxide, was used for the premeIting measurements and aIso the measurements of the
`heat capacity,
`the heat of fusion, and the melting point.
`Johnston
`and Giauque
`measured
`the following va.Iues for this sampIe of nitric oxide: moIar heat of fusion,
`AH,_, = 549.5 + 1 .O c&mole-
`’
`melting point, T, r T, = 109.49+0.05”K.
`The purity of the nitric oxide was caIcuiated by applying Eqn. (3), which holds
`for low concentration
`of impurities
`
`==$T&r
`1
`
`where xo,2 is the eutectic impurity of the sample as mole fraction, AH,,, is the molar
`heat of fusion of the pure nitric oxide, TI is the melting point of the pure nitric oxide,
`T is the temperature of the solid-Iiquid equilibrium,
`r is the molten fraction of the
`system at temperature T, and R is the gas constant.
`rise of the soiid-Iiquid
`The heat of premelting A%, necessary for a temperature
`equilibrium
`from T’ to T”, is related
`to the corresponding molten fractions of the
`sample r’ and r”. We can write the equation
`
`= AHfs,no(r”--t’)
`
`(41
`
`and Giauque enabIes the measurement of the totaI
`The method of Johnston
`amount of heat Aq for a temperature
`rise of the substance from T’ to T”. This to&I
`amount of heat is the sum of the heat of premelting Aq, and an amount Aqc given by
`the spe&c heat of the substance and the known temperature
`interval AT = T”- T’
`
`The calculation of the heat of premelting
`
`(Aq,) is possible from Eqn. (S), with the
`
` P. 3
`
`UT Ex. 2031
`SteadyMed v. United Therapeutics
`IPR2016-00006
`
`;
`x0,2
`(3)
`A%
`
`
`176
`
`of the
`and with an extrapolation
`measurement of the heat capacity of nitric oxide
`specific heat from a region \i-ith practicaIIy no premelting
`into the seIected region of
`premeiting. The eutectic
`impurity of the nitric oxide is calcuIated
`for a corresponding
`set of temperatures
`and molten fractions
`(7-‘, r’; T”, I”) and with the aid of Eqns. (3)
`and (4).
`
`x0.2 = 4, _U’-r-‘XT,--‘)
`no RT:=
`
`T”- T’
`
`to
`it is possibIe
`on nitric oxide,
`With Eqn. (6) and the values of the measurements
`calculate
`exactly
`the same values of eutectic
`impurities as found by Johnston
`and
`Giauque. The values and results are presented
`in Table
`I.
`
`I
`TABLE
`PREMELTING
`
`Temperatures
`
`(‘K)
`
`MEASUREMEhTS
`
`ON NITRIC OXIDE
`
`Heat of premelting
`&ettreen T’ arrd T’,
`
`Eutectic imptkties
`x-o_2 (mole fraction)
`
`T’
`
`4% (caz)
`
`104.71
`107.63
`
`108.59
`109.15
`
`0.171
`0.365
`
`7.9 x 10-e
`6.4 x 1O-6
`
`that the nitric oxide used in their
`and Giauque came to the conclusion
`Johnston
`contained
`less than
`!Os3 mole percent of eutectic
`impurities, or, the
`measurements
`so-caIIed purity
`is of the order of 99.999%.
`The authors excIuded
`the possibility of
`noneutectic
`impurities because of the method of preparation of the nitric oxide used
`for these
`investigations.
`Johnston
`and Giauque
`explained
`that no analyses of the
`purified gas were made since accurate meiting point and heat capacity data provide a
`more sensitive
`test of impurity
`than that given by chemicaI analysis.
`Johnston
`and
`Giauque made an equivaIent
`statement
`to Kofier’s
`about
`the measurement
`of
`impurities by the meIting point method.
`It seems
`to be clear
`that such excelient
`investigators
`as Giauque and Koffer did not emphasize
`the melting point and pre-
`mefting method
`in such a way without being deepIy impressed by the possibiiities of
`these two methods.
`laboratory
`the exceIIent work from the low temperature
`If we want to compare
`in Berkeley
`(the Iaboratory was named Giauque Hal1
`at the University of CaIifomia
`in 1967) with the premelting measurements, mainly DSC and DT.4, performed
`in the
`1970’s, we have to consider
`several points. The difference between
`the calorimetric
`methbd of Johnston
`and Giauque and the DSC or DTA method
`is not in thermo-
`dynamics but rather
`in the instrumentation
`and in the properties of the methods of
`measurement.
`II we compare
`In Table
`DSC-IB
`of the Perkin-Elmer
`
`some of the aspects of the two methods,
`Corporation
`for the second group.
`
`seiecting
`
`the
`
` P. 4
`
`UT Ex. 2031
`SteadyMed v. United Therapeutics
`IPR2016-00006
`
`(dq)
`(6)
`;r ’
`
`
`177
`
`TABLE 11
`COMPARISON OF THE PREMELTING METHOD OF JOHSSTON AND GIAUQUE
`AND THE PURITY DETERMINATI3N
`WITH THE DSC-IB
`
`Condition or properly measured
`
`Culorimetric method
`of Johnston and Giauque3
`
`DSC-IB
`(Perkin-Eimer
`
`Corp.)
`
`Weight of the sampIe
`Accuracy of the absolute
`temperatures
`Accuracy of the relative
`temperatures
`Accuracy of the measured
`heat of fusion
`Accuracy in the purity
`salue for high-purity substances
`Time for a premelting
`measurement
`
`1mg
`
`*IO-2
`
`“K
`
`*2x 1o-3 “K
`
`*2x 10-l %
`
`r 10-a %
`
`24 days
`
`3mg
`
`i3XlO_‘“K
`
`IIO-z
`
`‘K
`
`F5%
`
`&5X so-2 %
`
`20 min
`
`The great disadvantage of the calorimetric method developed by Johnston and
`Giauque, especially in industriaI use, is the extremely iong running time required for
`one measurement which is of course due to the enormous sampIe weights and the
`necessity for an equilibrium between the liquid and solid phases of the sampIe at ali
`temperature points”. It is aIso clear, however, that somehow one has to pay for such a
`high accuracy in purity measurements. Between the measurements on purity with
`thermoanaIytica1 methods of the 1920’s and the 1970’s, a seat number of papers were
`published on purity measurements by the freezing point method. We mention only
`one paper, which we regard as representative of a11 the papers on thermoanalytical
`purity measurements produced during this period: Determination
`of Purity by
`Measurement of Freezing Points, by Glasgow, Krouskop, Beadle, Axilrod and
`Rossini ‘.
`these preliminary and historical remarks, we will concentrate on
`FoIlowing
`purity work performed with the DSC-IB, an instrument of the Perkin-Elmer Corpora-
`tion. The devefopment of new DSC- and DTA-systems will certainly change the issue
`of the purity determination, e-g. enhance the accuracy of the measurement of eutectic
`impurities and solid solutions without increasing the running time for one measure-
`ment.
`
`DISCUSSION ON THE DSC Ll-rERAl-UxE ON PuRlTY
`
`In this discussion we shall not attempt a complete report of the DSC Iiterature.
`We will arrange our discussion according to theoretical and experimental points of the
`DSC-purity method.
`
`(a) Theory of the purity measurements
`As far as we know, aU DSC results on purity in the literature are calculated
`
` P. 5
`
`UT Ex. 2031
`SteadyMed v. United Therapeutics
`IPR2016-00006
`
`
`
`178
`
`according
`
`to the foIlowing equation
`
`[for symbois see Eqn. (315
`(i) The
`and conditions:
`Eqn_ (7) is derived under the folIowing approximations
`components
`form a eutectic phase diagram;
`(ii) the system is at constant pressure;
`(iii) the impurity or impurities form ideal solutions with the mohen part of the main
`component;
`(ic) the impurity
`is restricted
`to low concentrations;
`and (G) the heat of
`fusion is independent of temperature_
`A second equation discussed by Driscoll and coworkers6 describes systems
`containing eutectic
`impurities and impurities
`forming solid solutions with the main
`component. The systems of solid solutions are characterized
`according
`to Driscoil by
`a partition
`coefficient,
`this bein g the ratio of the concentrations
`of the impurity
`between the solid and Iiquid phases.
`
`leading with Eqn. (7) to the reIationship
`
`T=T,-
`
`x,,yRT:
`
`1
`
`Ah-f.1
`
`+ ~
`
`K
`1-K
`
`forming
`
`soIid
`
`impurities and
`
`impurities
`
`The discussion of systems with eutectic
`sohstions is rather inconsistent.
`With regard to this relationship, DriscoII et al. state: U Systems which form true
`soIid solutions, however, cannot be handled by this method of anaIysis”.
`Joy and
`coworkers’
`declare
`in their abstract:
`“Because
`the DSC
`technique
`is “blind”
`to
`equilibrium
`solid soIution
`formation, DSC vahxes should not be used as
`soIe
`criterion of purity”_ Mastrangelo and Domte*
`reported on a mixture of 2&dimethyI-
`butane and 2,34methyIbutane_
`These
`two substances
`are known
`to form soIid
`soIutions. Mastrangeio and Domie
`find a reasonabIe agreement between the theoreti-
`caI temperature
`reIation of Eqn. (9) and the experimenta
`findings.
`We have found no compIete experimentaI proof of Eqn. (9) in the literature.
`Such a proof would require
`the independent determination
`of the parameters and
`thermodynamic
`constants
`such as temperature T, mofe fractions of the main com-
`ponert and the impurities, moIten fraction r, the partition coefficient K, the meIting
`point Tl, and the heat of fusion of the main component AH,,,
`. Investigations of this
`kind should result in an assessment of the equilibrium with respect to temperature and
`concentrations.
`
` P. 6
`
`UT Ex. 2031
`SteadyMed v. United Therapeutics
`IPR2016-00006
`
`(9)
`2
`
`
`179
`
`Reconsidering Eqn. (7), we find the limitation of this equation discussed by
`several authors with respect to the allowable concentration of impurities. The limit is
`not properly defined because the definition would require the introduction of an
`absoIute deviation between the theoretical amount of eutectic impurities and the sum
`of eutectic impurities as determined by DSC. With the Iack of such a definition it is not
`surprising rhat the Iimitation of Eqn. (7) is estimated with considerable differences:
`Davis and Porter lo assumed a limitation of Eqn. (7), with respect to the concentration
`of eutectic impurities, of 5%
`De Angelis and PaparieIIo’
`assumed a limitation of
`1%
`and Joy et aL7, one of 2%.
`These limitations on rhe amount of eutectic impurities for the premehing
`method can be overcome by a method suggested by De Angelis and PaparieIIo’
`Samples of high impurity concentration
`(> 1%) are diluted with the pure main
`component to extend the limit of the appiicability of the DSC method. Such a dilution
`method was applied by De Angelis and PaparieIIo to 4 different organic systems \vith
`actual purities of 95597.0 mole-%. The DSC purity values of these sampies without
`dilution gave results in a nar;‘ow range from 97.4 to 97.8 mole-%. The absoIute
`differences between the true and the experimenta purity values were, therefore, of the
`order of 1-2 moIe-%_ DSC results with such high inaccuracies are not suf&ient for
`a.naIyticaI purposes. The experiments of De AngeIis and Papariello performed with the
`same compounds, but with a dilution of the main impurities with the corresponding
`main component to a purity IeveI above 99 mole-%, resulted in exceiient agreement
`between DSC values and the actual purity. Schumacher and Felder” present similar
`results in DSC purity values determined directly and after diIution with a substituted
`benztriazoie as the main component.
`The differences between the actual purity and the cxperimentai values de-
`termined without dilution are expIained by the authors of both papers’ ‘.I ’ in terms
`of an inconsistency between Eqn. (7) and the actuai melting behaviour of organic
`substances in a purity region beIow 99 mole-%. We found that such an explanation of
`the differences of theoretical and experimental purities appears to be, however, only
`one of several possibilities. Another possible explanation for the differences is that the
`DSC method without diiution, used by PaparieIIo and Schumacher, is oniy applicable
`to substances with a purity of at Ieast 99 mole-%.
`In contrary to the findings of
`PaparieIIo and Schumacher, we observed
`for many substances that the method
`without dilution gave correct vafues for impurities in the case of samples with
`substantially higher concentrations of impurities_ We found that highIy accurate purity
`values can oniy be achieved by selecting a scan speed appropriate to both the impurity
`concentration and the tvaIuation procedure. Thus, using the simplest Perkin-Elmer
`type of evaluation’ 3, a scan speed of 0.625 ‘C min- ’ wiii yield valid results only in the
`purity range above 98 moie-%_ The accuracy of a meIting curve evaluation is improved
`by a data co&&on
`and evaIuation at more than the 5-7 points within the important
`melting region, as suggested by Perkin-Elmer 1 3. The more sophisticated the data
`coIlection and evaluation, the less important are the experimentai cqnditions-scan
`speed, weight of sample and sensitivity-for
`getting a purity vaIuz of a high accuracy-
`
` P. 7
`
`UT Ex. 2031
`SteadyMed v. United Therapeutics
`IPR2016-00006
`
`;
`*
`;
`‘.
`
`
`180
`
`If one observes differences between ie experimentaI and the actual purity values one
`bas to check the experimental Conditions,
`in&din, = the type of sampIe pan used, the
`data coIlection,
`the evaiuation procedure
`of the melting curve, and
`the melting
`behaviour of the substance_
`If after aI1 these investigations
`the differences
`in the
`experimentaI and the actuaI purity persist, an inconsistency between Eqn. (7) and the
`melting behaviour of this specific system is highIy probable.
`for the solution of
`The diIution method introduced by PaparieIIo’
`’ is excelent
`special probIems_ Its praticzl use in an anaIyt.icaI Iaboratory
`is, however, limited by the
`amount of work
`invoIved_ Therefore,
`the question of the limitation of the DSC
`method to a region of high purity substances
`(e.g. to a purity better than 98 moIe-%)
`has to be reexamined because such a strong Iimitation wouId diminish the vaIue of the
`whoIe method_ Such an investigation of the purity region, in which the DSC method is
`a usefu1 anaIytica1 tool, shouid be performed with binary systems. It would be very
`heIpfu1 if the phase diagrams of the seIected binary systems were known
`from
`literature_ With such a binary system, ail kinds of possible sarameters
`and conditions
`have to be varied; the ratio of the two compounds,
`the sample weight, the sample pan,
`the scan speed, the sensitivity,
`the first, second and folIowing melting curves of the
`same sampIe if possibIe, and the data coIIection and evaluation. The results thus
`obtained may be discussed with respect
`to discrepancies between
`theoreticaI
`and
`experimentai values of the purity, the heat of fusion, and the melting points. They can,
`moreover,
`reveaI properties of the two components
`such as thermal stability, high
`vapor pressure in the melting region for one or both of the compounds, polymorphism,
`and anomaIous behaviours demonstrated by the phase diagram and by the melting
`curve. Having
`compIeted
`these investigations
`on some binary systems one couId
`perform a &niIar pro-gram on multi-component
`systems. Al1 these results shouId give
`us information on the limitation of Eqn. (7).
`
`(b) Handhkg of the samples
`Gray’ 3 suggested the use of the volatile sampIe pan with an inside cover. This
`inside cover is made fro,m aluminum
`to fit into the bottom part of the voIatiIe pan
`DriscoIl et aL6, BarraI and DiIIer’“, Reubke and MoIIica”,
`and others regard the
`volatile sample pan with an inside cover as the best sofution to avoid volatilization.
`The sample handiing and
`the variation
`of the
`temperature
`treatment
`are most
`important
`for substances with polymorphism,
`in the presence of impurities with a high
`vapor pressure
`in the melting region of the main component,
`and with substances
`which are unstabIe in the melting region.
`Difficulties also arise with the sample hoiders of the DSC-IB. The sampIe pans,
`the aluminum dome Iids and the outside cover of the sample holders have to be
`carefully placed in the correct positions’“*
`Barr311 and DiIIer * 4 make a good point on the preparation of samples whereby
`great care has to be taken in selecting test samples or in mixing of low concentration
`standards, because the sample size in DSC measurements has to be in the region of a
`few miIIigrams. For qua;rtitative work w-ith the DSC-IB,
`the sampIe size should be
`
` P. 8
`
`UT Ex. 2031
`SteadyMed v. United Therapeutics
`IPR2016-00006
`
`’ 6.
`
`
`181
`
`between 1 and 5 mg. Results with a high reproducibility are onIy possible with special
`care in the handling procedure.
`
`(c) Calibration of the DSC apparatus
`The cahbration of the temperature axis of the DSC with high purity standards
`shouId be performed
`in the way indicated by Barrall and DiiIer14. The calibration of
`the sensitivity of the DSC in calories per unit area presents no problems.
`Important
`for high purity measurements
`is the careful cahbration
`of the thermal
`resistance
`between the sample pan holder and the sample pan with standards
`like indium, tin and
`lead;
`this is also shown in the interesting
`investigations performed by Barrall and
`a_
`The question arises whether or not one is allowed to use inor,oaic materiais as
`standards for the measurement of the thermal resistance, which can then be used in the
`purity determination of organic substances. However, the DSC-IB
`is nearly independent
`of the thermal resistance of the sample, as long as the sample consists of crystals of a
`rather small size”_
`
`(d) Instrumental conditions for a purity determinction
`are sensitivity or the
`The instrumental
`conditions
`for a purity determination
`calorimetric
`range, the scan speed, and the sample pan. There are mutual relationships
`between these experimental conditions and some of the properties of the instrumenta-
`tion and the sample. As an exempie,
`the appropriate
`calorimetric
`range used in a
`purity determination depends on several conditions,
`i.e. heat of fusion of the main
`component,
`sample size, scan speed, concentration
`of impurities,
`and recording
`system or data collection.
`is in general kept at the lower
`The scan speed, as indicated in the literature6*7*‘4,
`end of the range, i.e. 0.625 or 1_25”C/min_ Such low values of the scan speed are
`
`III
`TABLE
`EFFECT OF SAMPLE SIZE AND HEATIXG RATE ON CALCULATED
`(RARRALL AND DILLER14)
`
`PUIUN
`
`_ixlure
`
`Sample size
`(mg)
`
`Hearing rate
`(“Clmin)
`
`Lead in tin
`Lead in tin
`Lead in tin
`Lead in tin
`Lead in tin
`Leadintin
`kadintin
`
`3.084
`3.054
`3.084
`4.300
`6.284
`6.284
`6.284
`
`1.25
`5.0
`20.0
`I.35
`5.0
`I.25
`0.625
`
`PuriI_v (mole-%)
`
`Found’
`
`KTlOrcnb
`
`0.425
`0.185
`0.0828
`0.321
`0.857
`OS71
`OX3
`
`0.419
`0.419
`0.419
`0.419
`1.16
`1.16
`1.16
`
`Xakulated with partial areas considered to the vertex of the endotherm. *Determined by atomic
`absorption of lead with a Perkin-Elmer Model 303 spectrophotometer, using titrous oxide as
`oxidizing agent to dissociate the tin compounds.
`
` P. 9
`
`UT Ex. 2031
`SteadyMed v. United Therapeutics
`IPR2016-00006
`
`DiI!er ’
`
`
`182
`
`for high purity measurements_ Low values of the scan speed are necessary as
`required
`the sampie
`is probably not at thermal equilibrium during rapid rates of heating,
`according
`to Barrall and DiIIer’ J_ The effects of sampIe size and heating rate on the
`measured purity in mixtures of Iead in tin ‘* are presented
`in Table IIL.
`Three parameters
`are varied
`in TabIe
`III;
`sampIe size, scan speed, and purity
`IeveI. The mixture with the Iower concentration
`of lead seems to be strongly sensitive
`to changes of the heating rate with respect
`to the concentrations
`of lead calculated
`from melting curves. Conclusions
`from TabIe
`III are only typicai
`for the applied
`conditions,
`such as the data collection and evaluation procedure. Generahzations
`are
`onIy possible after performing
`the investigations mentioned
`in part (a).
`
`(e) Ecaiualion of the meltfig curres
`the
`and
`the samples,
`of
`the handling
`of
`the
`instrument,
`The caiibration
`conditions
`for obtaining
`a melting curve
`determination
`of the correct
`instrumental
`which may be easiIy handled by an evaluation procedure, are all possible with some
`care
`in the experimental work. However, understanding
`and performing
`the purity
`cakzulations
`from melting curves is rather complicated- Therefore,
`the literature about
`this subject
`is quite extensive_ No review of evaiuation methods
`is available
`in the
`Ii terature.
`in
`of a mehing curve can be checked
`for the evaiuation
`A given procedure
`severaI different ways;
`there are a great many
`internal and external checks possibIe.
`We will discuss here the external checks which are performed with the values resuhing
`from a normaI evaIuation of a melting curve;
`i.e. (i) the mehing point of the sampIe,
`(ii) the mehing point of the pure main component,
`(iii) the heat of fusion of the pure
`main component,
`and (ic) the purity value of the sample. The meIting point and the
`heat of fusion of the sampIe caIcuIated by the evahration procedure may be compared
`v&h the vaIues measured directly on the melting curve by appIying
`the calibration
`factors. The melting point and the heat of fusion of the pure main component
`can
`probably be found
`in the literature- Such
`literature vahtes permit a comparison with
`the resuhs from the evahration of melting curves.
`For test substances,
`the measured DSC purity value may be compared with the
`actual purity vaiue known
`from mixing. A second method
`is to compare
`the DSC
`purity value with the purity
`information
`obtained
`from a separate analytica
`proce-
`dure. In the case of disagreement between the DSC purity value and the actual purity
`value, severaI points must be considered with regard
`to the DSC method;
`i.e. the
`instrumental
`conditions used in getting
`the mehing curve;
`the physical and chemical
`behaviour of the main component
`and the impurities;
`and the evahtation procedure,
`and the use of the thermodynamic
`relationship
`for the description of the solid-liquid
`equiIibrium_
`in case of
`in this section, which are necessary
`given
`AI1 the considerations
`discrepancies
`between
`the values evaiuated
`from melting curves
`(i-e_ purity, meIting
`points, heat of fusion) and va!ues found in theliterature,
`receive practically no mention
`in the pubIished. work on DSC-purity determination-
`
` P. 10
`
`UT Ex. 2031
`SteadyMed v. United Therapeutics
`IPR2016-00006
`
`
`
`183
`
`is
`The evaluation of mehing curves by hand, suggested by Perkin-EImer13,
`practicable but too cumbersome
`for routine work. Computer programs used in the
`evaluation give higher accuracies
`in purity and thermodynamic
`values, and are much
`faster. Programs were developed by Driscoll et
`Scott and Gray”,
`Barrall and
`DiIIer’*, Davis and Porter”, Heuvel and Lind”, Gent2’, and others.
`The basic probiems of the evaluation of mebing curves by computer or by hand
`are the same. Referring
`to Eqn. (3, one has to fit the experimental DSC-cgrve
`to a
`straight line in the (l/r, n-diagram,
`as it was first shown by Pitzer and Scott2’.
`The evaluation procedures cited above consist of: (I) the fit of the experimental
`points from a mehing curve to a given thermodynamic
`function,
`together with the
`determination of the true heat of fusion of the main compone&“;
`(2) the Iineariza-
`tion with an appropriate mathematical method6*r8;
`and (3) the caIculation of the
`purity value and the thermodynamic
`constants of the sample and of the corresponding
`main component.
`It is not always possible to separate a given evaluation procedure
`into these three parts. However,
`the literature of the evaiuation procedures
`is more
`easily discussed by such a partition.
`
`t * 436
`
`i
`
`~1.2 finaL , * r
`$39 OK
`L38
`437
`temperature
`
`readings
`
`____t
`
`Fig.
`
`Melting
`
`of bcuzauilide with a base&e u 1.1 fiMI calculated by Heuvel and Liudxg.
`
`is
`the DSC
`the true heat of fusion exists because
`The problem of evaluating
`measuring
`the difference
`in the heat necessary
`to maintain a given and constant
`temperature
`rise in the reference and the sample cell. The baseline of the instrument
`during an exothermic
`ci endothermic
`change of rhe sample can only be determined
`
` P. 11
`
`UT Ex. 2031
`SteadyMed v. United Therapeutics
`IPR2016-00006
`
`aL6,
`2.
`trace
`
`
`approximateIy by a connection of the recorder Iines before and after such an ener_q
`change. HeuveI and Lind I9 stated, “Under certain conditions of instrument operation,
`e.g. fast scannin, 0 rates, the course of the base line deviates to a large extent from
`simpIe interporation behveen pre-transition and post-transition baseIines”_ Fig. 2
`shows the mefting trace of benzanilide
`from the paper of Heuvel and Lind. The
`indicated baselines are given by (i) a seaight Iine from point B to E, and (ii) U, ,:!
`final; a function of the heating rate, the heat capacity of the sampIe, and the therm4
`resistance frcm the sampIe holder to the sample’ ’ according to the caIcuIations of
`Heuvef and Lind 19_
`For a sharp transition, as shown in Fig. 2, both baselines give the same
`cakuIated value for the heat of fusion, which is a concIusion of the paper of Heuvel
`and Lind”_
`The discussion of the heat of fusion is presented in two parts: (I) with high
`purity substances, and (2) with substances having Iower purity vahxes.
`TabIe IV shows the heats of fusion for several high-purity substances_ The
`values are directiy calculated from melting curves in applying a straight baseline, as
`&own
`in Fig. 2.
`
`Iv
`TABLE
`HEAT OF FUSION FROM DSC MELTIXG
`A HIGH PURITY VALUE
`
`Subsranccs
`
`DSC
`
`CURVES FOR SUBSTANCES
`
`OF
`
`Hear of f&ion Prrriry (mole
`
`DSC mcorrecred
`basetine
`values ~Jbsc w I.LiL
`(colmoie-
`(caLmole-
`
`‘) Benzene Benzene
`
`99.8 235P 2349 (Ref. 22)
`
`99.05
`9925
`
`Benzamitic
`Bcnzoic acid
`Anthrzchincn
`Potassium nitrate
`Distilkd WY&X
`Butazoiidine
`
`99-94
`
`99.97
`99.56
`zk o.zsc
`
`2349 (Ref. 22)
`4999 (Ref. 23)
`4300 (Ref. 24)
`7830 (Ref. 25)
`2295 (Ref. 26)
`1434 (Ref. 27)
`
`+0.1
`-4-s
`-6.3
`-8-3
`-1.3
`f 3.3
`-2.4
`
`223-i=
`45w
`3945*
`7725b
`237@
`1w
`57rob
`f 68ob.C
`
`WSC values by Driscoli er ol. se ‘DSC V&KS by Marti and Heiber (unpubkhed). CEt-ror in a single
`measurement on 95% confidence knits.
`
`The agreement of the DSC with the Iiterature values for the heat of fusion is
`reasonabIe in the case of high purity substances. The reproducibiIity of the heat of
`fusion, according
`to measurements on butazoIidine,
`is indicative of a normal-
`precision, and certainly not of a high-precision instrument_ A better precision in the
`determination of energies are expected from new insmments, e.g. Metier DTA
`2m’S
`and Perkin-Elmer DSC-229.
`
` P. 12
`
`UT Ex. 2031
`SteadyMed v. United Therapeutics
`IPR2016-00006
`
`%)
`Lit.
`‘)
`
`
`The determination of the heat of fusion from melting curves of sampies with
`Iow purity values reveals a compIeteIy different picture compared to that presented in
`TabIe IV. The results are presented in TabIe V. The measurements were performed by
`DriscoII et aZ.’ for an impurity content of < 2.80 mole-%, and for highest value of
`impurities by a measurement in our Iaboratory. The determination of the heat of fusion
`was performed with an uncorrected baseline, as described in Fig. 2.
`
`185
`
`TABLE V
`CURVES FOR BEXZENE WITH
`HEAT OF FUSION FROM DSC MELTING
`VARIOUS
`AMOLJ%iS OF EUTECTIC
`IMPURITIES
`
`Subslance
`
`DSC
`prrriXy
`(mole-%)
`
`Heat of fusion
`DSC, uncorrected
`baseline
`
`AHI_,
`(caI.moie-
`
`‘)
`
`A&.0x-A ffr.u..
`
`AH,..,,_
`
`(%)
`
`x 100
`
`Benzene
`
`99.8
`99.05
`99.