IkmmhimSca Aaa Ekvkr Pablirbing Company, ~mseahm
DIFFERENTIAL PURITY* E F.
JOY,J_ D. BONN, AND A. J. BARMRD, JR.
Research Anaiyrical Services, 3. T. Baker Chemical Co., Phillipsburg,
New Jersey 0886S (U. S. A.)
(Received May 14th.1970)
The valueandlimitations ofdifferential scanning calorimetry in the assessment of high-purity substances has been examined_ In favorable cases, good agreement has been secured for polycychc hydrocarbons between DSC purity values and GC assay values. For some halogenated benzoic acids, used as microanalyticai reference standards, good agreement has been obtained between DSC purity values and acid-base titration results. DSC studies on cholesterol and urea, which have limited thermal stability, are presented. With the available instrument and technique, the pracrical upper limit of absolute DSC purity vaIues may be 99.95 mole%, ahhough higher numerical values can be obtained_ Because the DSC technique is Gblind” to equilibrium solid solution formation, DSC values should not be used as a soIe criterion of purity; this recommendation is of special importance for compounds purified by fractional solidification processes_ IN-l-RODtJCXION
The practical anaIysis of high-purity chemicals has been studied in depth in our Iaboratories during the past few years. By high-purity chemicals is intended those having about 500 parts per million or less of total impurity; such chemicals are sometimes termed “ultrapure”. Our studies have been pursued in the framework of an extensive program for the development of a viable line of high-purity chemicals, both inorganic and organic, distinguished by broad analytical definition and by advanced protective packaging_ The analytical portion of this program differs from earlier studies of high-purity materials by the need to keep analytical costs at a sufficiently low level that the products are not priced out of salient markets in analytical chemistry, research and deveIopment, and materials science. An overview of this analytical effort has been given’ l’ and the methods developed and perfected for the analytical characterization of high-purity EDTA have been published ‘. For the broad characterization of a high-purity chemical, the major constituent should be determined or some overall expression of purity be sought_ Additionally *Paper III in the series “Practical analysis of high-purity chemicals”.
Acta. 2 (1971) 57-68
E. F. JOY, J. D. BOai,
A. J. BARNARD,
key impurities should be determined. Also various general tests cau be performed and physical properties can be measured. For the assessment of purity, thermal methods based on melting of freezing phenomena are valuable and the calorimetric measurements of Withers and coworkers3 for certain standards have showed the power of the approach in the high-purity range. However, the method requires large samples, extended equilibration times, and high-precision thermometry. With the auiomation of thermal methods, differential thermal analysis PTA) was used by various workers to assess purity, but largely in a quaiitative way. Attempts were made to quantify zhe approach by calculations based on the shape of the melting curve as compared to that of one or more standard samples of known purity.
Differential scanning ca.iorimet$ (DSC) became avaiIabIe in terms of a practical instrument sometime after 1964. it was early recognized that the DSC technique was applicable to the assessment of purity. Comparisons with standards of known purity have been usedL6. Gray recognized that a single DSC curve could be analyzed in terms of the Van’t Hoff equation without the need for standards’. The manual cakxfations are somewhat tedious and computer processing of the data was introduced by DriscoU and coworkersa. Additional computer programs have been developed by Scott and Gray’ and by Barrall II and Dilieri”. The calculations, however performed, provide the molar impz&~. content, which is usually expressed in moIe%. This value can be substracted from 100 to yield a value for the mole% pwizy. This practice has been followed in this work; however. in the discussion of the precision or accuracy of the DSC technique reference to the mole% impurity found is somet!mes preferable. The DSC approach to the assessment of purity during the past few years has been applied to a variety of organic compounds, including aliphatic hydrocarbons*, amides, amines and carbamates”-r2, benzene derivatives’*“-i3, haiogenated compoundsS. I ‘, heterocychc compounds’ i-i ‘, malic acidi3, organophosphates’“, pesticidal chemicals”, pharmaceuticais’ ’ -’ 3, polycychc hydrocarbons and quinones*-’ I, and steroids7.i3 . The conclusions to be drawn from reported studies is that this DSC technique is most useful above 98.0 mole%. One paper, however, has
reported extension to the 95% purity Ievel by means of the addition of a high-purity form of the major component, thereby bringing the composition within the favorable 99.0-99.5 mole% region’ 3. Some attempts have been made”-’ ‘.I3 to relate the purity values found by the DSC technique with those established by other methods. Reubke and Mohica” compared the DSC value found for anthraquinone with that by GC assay; the values found were 99.69 mole% and >99.5%, respectively. These workers also compared DSC and phase solubility measurements for dialiyIbarbituric acid. The values for the two techniques were 99.80 mole% (average of duplicate) and 100.0% by weight. For methyl reserpate, they also related the DSC vahre with the purity estimated by the temperature sohrbiiity technique and indicated that the DSC technique might serve as a rapid control method. DeAngeiis and Paparieho’ 3 for various samples of a steroid with a purity above 99.0% found good agreement for DSC measurements,
ASSESSMENT OF HIGH PURITY
phase solubility analysis, and thin-layer chromatography (t.1.c.) with densitometric evaluation_ For an investigational carbamate, at a purity less than 98.0% poor agreement was found for DSC assessment, n.m.r. and t.1.c; at a purity greater than 98.0% better agreement was secured. Additional attempts have been made**’ ‘- “-* J to confirm the accuracy of DSC purity measurements by the study of known mixtures, ranging in purity frdm about 97 to 99.9%. The analytical recovery of the minor component in most studies has been within +20% relative. Driscoli and coworkers* suggested that where the recovery was poorer, as with a thiophene-benzene mixture, equilibrium solid solution formation is involved. BarraH II and DilIer’ ’ have reported that for binary mixtures of metals the DSC purity values were in excelient agreement, k-3% relative, with the impurity content established by atomic absorption photometry_ For mixtures of organic compounds, comparison of the U.V. absorbance, fluorescence photometry and tl.c. was reported to give good agreement. They emphasized that favorable agreement was secured when the DSC instrumental parameters were properly controlIed and the calculations were based on use of the ener_e-temperature curve up to the maximum. The accuracy achieved for DSC purity measurements in published studies is shown graphically in Fig. 1. In this log-log plot the impurity content found by DSC measurements is correlated with the impurity content either established by another method or caiculated from the composition of a known mixture. Some workers have emphasized that the agreement is within + 20% relative. This view is supported by the many points that fall within the +20 and -20% lines also drawn on the graph. It shouId be added, however, that this plot does not include runs for which the original workers suspected solid solution formation or other disturbance, including incomplete solubility in the melt_
Impurity taken,mde per cent Fig. 1. For published studies, correlation of content of impurity found by DSC measurements and impurity taken. Heavy line corresponds to exact correlation. Upper and Iou-er light lines correspond to +20% and -20% relative departure; 0, Ret 13; m, Ref. 8; V, Ref. 12; x, Ref. 14; 0, Ref. IO.
2 (1971) 5743
J. D. BONN,
Our studies of the DSC assessment of purity were ititiated in December 1967 and have been concurrent with most of the reports mentioned above. Our objective has been the application of this DSC technique to the practicaI characterization of high-purity substances where the mole impurity is less than 0.10% and often less than 0.05%. Much of the work was necessarily directed to establishing the validity of the technique. Our experiences reported in this paper throw light on both the value and limitations of DSC purity assessments for high-purity substances. DSC technique
All of the DSC measurements reported in this paper were obtained with the Perkin-EImer DSC-1 I3 instrument. Unless otherwise noted, the temperature scanning rate was 0_625’C/min and with maximum sensitivity (range 1). All high-purity sampies ranged from 0.5 to 2 mg in size and were weighed on a Cahn gram electrobalance. All samples were run on an “as is- basis (that is without drying) and in a nitrogen atmosphere, All materials in the program were routinely protected after their f&I purification by storage under an atmosphere of either argon or nitrogen. With a variety of high-purity volatile and non-voiatile compounds, our experience has been that better a_g-eement is secured in the purity vaIues for replicates when so-called “volatile” ceIIs are used. Following the manufacturer’s recommendations, a close fitting aluminum disc is inserted over the sample before sealing of the cell_ AU DSC purity values in this paper, unless otherwise stated, were obtained using volatile c&Is. In all cases a similar celI with the aluminum disc insert was used on the reference side of the instrument_
In this study although a few DSC purity values were obtained by planimeter measurements of the curve and desk calculations following the recommendations of Perkin-Elmer’*’ ‘, most were secured by application of the computer program developed by DriscolI, Duling and Magnotta8. The original Fortran program was re-written into a basic program operable on a time-shared computer terminal. The czdculation of impurity (or purity) content by the DSC technique is based on the Van? Hoff equation, which relates the melting-point depression to the impurity content On the basis of this equation a plot of T,, the instantaneous temperature of the sampIe, cs_ i!E where F is the fraction melted at T,, should give a straight line with an intercept of lo, the melting point of the infinitely pure compound, and a slope - RT~X&U?,, where R is the gas constant, X, the moie fraction of impurity, and &I, the heat of fusion of the compound. However, the plot of T’ fis. IlFimmediately derived from the DSC data does not afford a straight line. The cause of this departure is the failure to measure some area under the DSC energy zx time (temperature) output curve due to the inherent sensitivity Limitations of the instrument. A straight line reiation is achieved by the trial and error addition of small increments to both the partiai areas and total area. Details of this area correction and linearization process have been given by various workers 7m8*12*15.
DSC ASXSSMENTOF HIGH PURITY
Driscoll and coworkers* with computer-based calculations recommended that the largest l/Fvalue used should be 50. This value has been used in computations of ah DSC values recorded in this paper. In our experience, use of a higher I/F Limit results in a smaller area correction and an unreasonably low impurity content. These findings are to be associated with the fact that too much weight is placed on the initial; poorly defined premeiting region. Additionally the piot of T ZLL l/F is actually S shaped and with higher l/F vaiues the straight line fitted has a smaller slope, which corresponds to lower impurity content. The lower limit for I;% in the computer program of Driscoll and coworkers’ is taken as the point corresponding to one-ha of the peak height in the energy-temperature curve. A similar limit has been taken by earlier workers following the recommendations of Perkin-EImer’. Driscoll and coworkers* introduced an empirical correction for nurzequiiibrizm solid solution formation based on a previously developed formula employed in static calorimetry Id_ By this correction of the impurity content better agreement was achieved by those workers in the assessment of the moIar composition of mixtures. The limited study of mixtures in our laboratories confirms this finding (see below). Ail DSC values reported in this paper include this correction unless otherwise stated. This correction results in an increase in the impurity content assigned, that is, a decrease in the purity content, and, if nothing more, affords a more conservative assessment of the purity of a high-purity chemical. The size of the correction increases with the extent of the area correction involved in the linearization of the T, ZLS.l/F plot_ At 99-90 mole% purity, a 5% area correction Ieads to an absolute decrease in *&e mole purity of only 0.016% when the nonequihbrium solid solution correction is applied. Scott and Gray9 with their recent computer program have recommended, as an alternative, use of a “slope criterion”, that is, the Iower lir’ limit is taken as the point at which the slope of the sampie curve exceeds one-sixth of the slope of the tangent line to the calibration (indium) curve. These workers also recommended a “partial area start criterion”, that is, the upper I/F limit is i&en at the point, proceeding from the starting point, as the next point after “5 ordinate va.Iues have remained]. _. above the initial base line by one encoder unit”_ In other words, the upper l/Flimit is taken as the first point at which the signal is significantly above the noise, However, the program of Scott and Gray9 permits the user options to use either of these criteria or to seIect his o\tn I/F limits. Barrah II and DilIerrO have recently recommended a lower l/F value corresponding to the vertex of the energy curve and provide data to support this selection by their study of mixtures. It was their view that selection of this Emit was less arbitrary than others that have been suggested. DSC srudy of a zone-refined bar Differential scanning calorimetry is an attractive approach in foliowing the purification of an orgtic compound by zone melting. Where the operation has a favorable separation factor, pronounced change in impurity content can be measured Themwchim_ Act,
2 (1971) 57-68
62 aIong the iength of the bar. Additionally,
JOY, J. D- BOSS,
A. J. BAR?URD,
smah differences in impurity content
often be detected radiaIIy. For example, in our Iaboratories, the impurity distribution of zone refined bars of benziI and benzoic acid has been explored. The study of one benzoic acid bar is noteM-orthy. At the less pure end a center hollow was present and the radial impurity content varied quite [email protected]
_ In contrast at the purer end, with no center hollow present, the radial change in composition was no greater than the reproducibility of the DSC measurement of a singIe sample. By multipass zone refmin g, benziI, which was previously purified by crystallization to a DSC purity of 99.913 mole% , yielded a bar ranging from 99.996 mole% at the purer end to 99.6 mole% toward the other end. The distribution of impurity along the bar could be measured and showed reguIar progression. Srudy of binary nrimues In the introduction to this paper, some investigations were reviewed that had the goa! of conGrming the accuracy of DSC purity measurements by the study of known binary mixtures_ In this earlier work the range of the minor component was from 0. I to 3 mole?&_ With our interest in the use of DSC techniques with high-purity substances, the study of binary mixtures of compositions having 0-I mole% or less of the minor component was undertaken. The fact that various substances were being purified in our pro_errlms by multipass zone refining made this stud3 attractive. Three binary systems xere selected for examination, namely, naphthalene-biphenyl, benziIbenzoic acid, and bena&biphenyiin aI1 cases the first-named component was the major one. A cursor_v study of literature failed to reveal any references to possible equiI.ibrium solid solution formation in these systems. Using zone refined compounds, attempts were made to prepare the required mixtures. The difficulties in preparing homogeneous mixtures of the expected composition were recognized. It w-as also appreciated that the problem was magnified because a sampte of about 1 mg is appropriate in the DSC run of a high-purit>- compound. Mixtures for the naphthalenebiphenyl system were prepared by evaporation of soiutions (using bulk crystallization and flash evaporation) and also b:; drv_ grinding using a mortar and pestle. For the two benzil systems, fusion alcne and fusion followed by crushing and mising u-ere employed- Only with some samples of the naphthalene-biphenyl system, prepared by grinding, couId recoveries of the minor component as great as 75 to 80% relative be obtained in the DSC purity measurements, and only with inclusion of the correction for nonequilibrium sohd solution formation’. Since all results were in the direction of low recoveries and sometimes with the vaIues for duplicates scattered, further attempts were abandoned to prepare and to study binary mixtures with less &an 0.1 moIe% of the minor component.
of Dsc purily cahes and GC amay raiues For a number of polycyclic hydrocarbons purified by multipass zone refining followed by simpIe sublimation, the purity was established by DSC measurements of Comparison
DSC ASSESSMENT OF
duplicate samples. Additionally these coinpounds \vere assayed by gas chromatography. The resu!ts obtained by the two methods are compared in Table I, with a statement of the instrumenta conditions. Gas chromatograph)f of such high-melting hydrocarbons TABLE
has been placed on a firm footing by the xvork of Saw%% I ’ and others.
FOR SOXE ZOXE-REFIhTD
( area 9’0)
Acenaphthene Anthracene Bibenryl BiphenyI
Durene Kaphrhalene Phenanthrene Pyrene rrans-StiIbene p-Terphenyl
UHC UHC 1
UHC 325 UHC 326 UHC 327 g h UHC 315 UHC 339 I UHC 330
99.99 99.99 93.24 99.99
(I 70) (23oq (i’ _-i) (Ii’o)
99.99 99.97 59.99 99.77 99.99
(210) (250) (250) (200) (24w)
(I 40) 99.99 (!40) 99.9? t.2IO)
99.95 ; 99.96 99.97; 99.36 99.97 ; 99.96 99 43’. 9a qjc 99:96;‘99:& 99.96; 99.93 99.99; 99.97 99.75; 99.77 99.73; 99.92 99.94; 99.94 99.97; 99.95 99.97; 99.94 99.97’; 99.96’
‘Ail GC assr?ysusing F & M model 500 with flame ionization dctcction I silicone gum rubber SE-30 Chromosorb W co1umr1(6-ft length, 0.X-in ot?tsiJiedizrnct~r); sample dissolved in benzene for
injection. Column temp. used (in ‘C) bctwccn p2rcnrheses. “Ali DSC x-aiues corrected for nonequilibrium solid solution formation kxc text) and obtained rs-ith use of vofstik ce:I, nitrogen flow, and a scanning rate of 0.625 ‘Cimin. ‘So!id sampls injection. “Sampic S29-55-2; lot rejected, set twit fcr explanation. *va:ucs for nonvoiatiic cell. ‘2-ft Column, rather than 6-ft. %SampIs 529-54-5; lot rejected on basis oi DSC values below 99.95% “Sampie S?9-65-S; lot rcj cited on basis of DSC vaiu.zs beiow 99.95%. *Sample 829-54-4; fot rejected on basis of low GC assal; value.
As can be seen, in most cases, the agreement
bet--een the two methods is good
considering the quite different underlying phenomena. It should be noted that a value of 99.99 area % was assigned for the GC assay whenever no impurity peak was detected. The study of the sample of bibenql delineated some of the limitations of differential scanning calorimetry. The values for the purity content obtained by IXC measurements were 99.97 and 99.96 moIe%. These values suggested that a high-
purity product had indeed been secured. Assay by gas chromatography was then attempted. The first GC assay vaIue of 97.63 area % was obtained on a 6-fr I/bin diameter silicone SE-30 column at 190°C with the inlet port at 3OO’C and with flame ionization detection. Since this high port temperature might have caused thermal decomposition 0,F the sample and thereby a Iow GC assay vaiue, the work was repeated under milder thermal conditions, namely, with a similar 2-ft column, a Ihermochim. Ada. 2 (1971)
E. F. JOY, J_ D. BOIW,
temperature of 125’C and a port temperature of 190°C. In this way a somewhat higher GC assay value of 98.24 area% was secured. The chromatogram in each case showed only a single significant impurity peak and under the milder conditions this peak corresponded to I-76 area%_ To study this pronounced discrepancy hetwecn the GC and DSC purity va.Iues, the impurity was isolated in milligram amounts by gas chromatography employing an 8-ft 3/8-in diameter silicone SE-30 column. Then l-89 parts by weight of this isolated impurity was added to 100 parts by weight of the original bihenzyl. Assuming that the GC area% found for the impurity approximates a weight percent, t.ke total content of the impurity in the resulting mixture was about 3-58% by weight This mixture was then subjected to DSC measurements and the purity value found was 99.95 mole%, w-hichis only slightly less than the DSC value found for the original bibenzyi! These findings clearIy show that the differential scanning calorimetric technique was virtually blind to the impurity (solid solution formation). With Zranr-stilbenethe disagreement between the DSC and GC assay values was less marked than for the bihenzyi situation (see Table I), and solid solution formation was suspected, Thin-Iayer chromatography using flexible sheets is widely used in our laboratories in the characterization of high-purity compounds. The hydrocarbons listed in Table I were examined by this technique using both aluminum oxide IB-F (with carbon tetrachloride for deveIopment) and silica gel IB-F (with carbon tetrachloride and hexane separately for development). Spots were located by examination under both Iong and short wase U.V. illumination. Even at the heavy sampIe loadings employed, no impurity spot was seen for most of the hydrocarbons. In the few cases that an extraneous spot was detected, the relative intensities indicated that the impurity \kz present only in trace amounts. Ccmparim~ of DSC purity raluzs and iirrirnerric amay values Organic eIementaI analysis requires a variety of reference standards that serve to assure that the combustion or other technique affords satisfactory results. In our programs a number of halogenated benzoic acids have heen purified for consideration as microana+tical reference standards. Since such acids appeared to melt without decomposition and were being assayed by acid-base titrimetry, it was also of interest to establish their purity by DSC measurements. These compounds were also analyzed for carbon and hydrogen in duplicate and for the halogen substituent five or more times. The results of the DSC and titrimetric evaluation of these compounds are summarized in Table II. The agreement of averages in all cases was within 0.1% absoIute, It is noteworthy that the titrimetric assay as well as the determination of the halogen content would be blind to isomeric impurities; in contrast, DSC measurements might be expected to respond to such impurities. The combination of the DSC _ auessment, the acidimetric assay, and the determination of halogen content taken together affords good evidence that the purification processes employed do indeed yield high-quality microanalytical reference standards.
DSC ASSESSMENT OF HlGH PURITY TABLE
FOR SOME MICROANALYTICAL
J_ T. Baker
m-Bromobenzoic acid m-Chlorobenzoic acid p_FIuorobcnzoic acid o-Iodobenzoic acid
’ UHC 318 UHC 314 UHC 319
99.78, 99.73 99.81, 99.81 99.93, 99.94 99.97, 100.00
99.71, 99.54 999.83, 99.59 99.56, 99.95 99.83
Via volume-based acid-base titrations, except mthlorobenzoic acid by precision weight titrimetry. bSampIe No. 816-95b; lot rejcctcd for Iow assay and DSC purity; bromine content by oxygen-filled fIask combustion: Theory 39.75%; found 39.83iO.013 % (mean of 5 values and standard deviation of the mean).
DSC study Of Ureci Urea presents an interesting challenge in its specification as a reference material for the clinicai laboratory- The staff of the National Bureau of Standards explored the DSC technique with high-purity urea and for the NBS reference materiai” “an apparent purity of 99-82 +0.003 mole% n was listed. The calculation of the purity did not incIude any correction for nonequilibrium solid solution formation. An “apparent” purity value by phase-soIubiiity analysis of 99.82% by weight was also given. The modifier “apparent” was used “because neither method accounted for the moisture content of the urea”. A satisfactory method for drying was stated not to have been found. In our laboratories, DSC assessment ofhigh-purity urea was also undertaken concurrently with the work of the National Bureau of Standards, but somewhat lower DSC values were secured with a variety of purified samples and at scanning rates varying from 0.625 to 2_5”C:/min. At that time these Iow results were attriiuted to insufficient thermal stability for the compound at and near its melting point. Phase soiubility analysis was undertaken for one lot of high-purity urea when the success of the National Bureau of Standards was communicated and a value of 99.87% by weight was asssigned for the “as is” material on the basis of eight points’g. Subsequent study of the same lot by the DSC technique was then undertaken. Both the “as is” material and a portion dried for 3 h in a pistol at 108°C over phosphorus pentoxide were run at scanning rates of 0.625 and 1_25”C/min. The DSC mole purity values found can be summarized: Urea
99.66 (av. of 3) 99-76. 99.65 (av. of 3) 99.74*
99.66 99.76* 99.71 99.78.
Acta, 2 (1971) 57-68
E_ F_ JOY. J. D. BON?G, A. J. BARNARD,
The starred values are for the same DSC runs with the correction for nonequihirium solid solution omitted from the cafculations. Even these vaIues are numerically Iower than the phase solubility value by about O-I_To/a ,’ absolute. However, for an unknown impurity the comparison of mole- and Lveight-?G values has limited [email protected]
_ The DSC technique for the absolute assessment of purity faiIs for a compound of lo<- thermal stabilit): in its melting range. Our experience with urea suggests that it shows “borderIine” thermal stability with this technique and the analyticaI information secured is of limited usefulness. DSC studyof cholesterol In studies of the purification of cholesterol for use as a clinical laboratory standard, DSC purity measurements were early applied in our laboratory_ At a scznning rate of 0.625’C/min reasonabIe purity values could be obtained and the agreement of replicates was within 20.1 mole?& The XationaI Bureau of Standards subsequently undertook DSC evaluation of high-purity cholesterol and secured good agrezement between DSC vaIues and those obtained by gas chromatography using glass coIumns and by phase solubility anaIysiszo. Using simiIar gas chromatography, Goldstein” has anaI>zed bo*& the J. T. Baker high-purity cholesterol and that certified by the National Bureau of Standards_ The impurities found are consistent with the DSC pm+ values assigned in our IaboratoryOur experience with choIesteroI offers one intriguing application for the DSC technique. Flame-sealing of ampoules of high-purity cholestero1 and the use of an inert atmosphere was ccnsidered as a means of reducing moisture pickup and of maintaining the quality of the product. Cholesterol was experimentally ampouled. the ampoules onened? and the DSC purity of the contained product was compared with that of the original material. Significantly Iower DSC v&lues were found for the ampouled material, presumabIy due to the presence of smaII amounts of decomposed product. It should be added that the properties of cholesterol make it difficult to fiI1 this compound into an ampoule under drybox conditions without leaving a few small particles clinging high on the waIIs of the ampoule. As a result of this stud>-,the ampouling of choIestero1 was abandoned and containment of the product in a screw cap vial was adopted_ The DSC study of ampouled products has been extended to other high-purity compounds, [email protected]
_ the technique can be useful in storage and container studies for pharmaceuticals and Iaboratory chemicals.
In view of its rapidity, use of miWgram samples, and its application to the purity region from 98-O to 99.95 mole%, differentiai scanning calorimetry is a most valuable tool in the characterization of organic compounds_ The DSC technique can effectively compliment such established approaches as elemental anaIysis, titrimetric and _mvimetric assay, and _m chromato_eraphy_ For a thermally stable compound, a Zo2cDSC purity value, based on a satisfactory run of the instrument, is clear evidence that a compound is not of high-purity!
In contrast, a high DSC purity value cannot be taken as conclusive evidence that the compound is indeed of high purity. In the latter case, confirmatory evidence shouid be secured by techniques not dependent on melting or freezing phenomena, since the DSC technique is bIind to equilibrium solid solution formation_ This confirmation is especiaIIy important where the compound has been purified by zone refining or other solidification processes where solid solutions can persist. In the application of the differential scanning calorimetric assessment of purity to the high-purity region, the operatin, m conditions selected should approach equilibrium melting as closely as a kinetic process allows. In brief. small samples and the lowest scanning rate practical should be used. These factors require that the commercial instrument be operated at maximum sensitivity. Additionail>- thermal contact should be optimized. Above 99.90 mole% purity, the premelting behavior on which the DSC calculation is based, becomes progressively smaIIer and the purit>- value assigned becomes strongly dependent on the assumptions made in the calculation (I/F limits). The practical upper limit for absolzrte DSC measurements may therefore be about 99.95 mole% with the presently available instrument and technique. This should not be confused with the abiIity to detect differences in impurity content as IittIe as 0.005 moIe %. Indeed the rdariz-c purity of two Iots of a single compound can be assessed up to 99.99 mole%, especialIy if [email protected]
are run. ACKiUOWLEDGMEhTS
I. N. Duling and Sun OiI Co. are thanked for making their computer program’ available and R. BrilI of the Service Bureau Corporation for assistance in conversion of the program to the Basic language. -4. Foulds is thanked for the phase-solubility work with urea and for the acidimetric assay of the microanalytical reference standards, M. Zief for assistance in providing the zone-refined compounds used in the study of binary mistures, and F. M. Rabel for the thin-layer chromatographic work. REFERENCES I A- J_ 2 A. J.
BARSARD. JR_,E. F. JOY, K. LITTLE AND J. D. BROOKS. TaIuntu, 17 (1970) 785. BARFARD, JR. ASD E. F. JOY, 7?x Chemist, 47 (1970) 243. 3 F_ W_ SCI~~AE AND E. urICHERS.Temperature. IIS Measuremenf and Confrol in Science and Industry, 4 5 6 7
E. S. WATQX, hf. D. M. MAZMXOX, Thermal Analysis Thermal Andysis
8 G.L.DRIxIoLL,I.N. AnatyIicaf
J. O'NETLL.J. JUSTIS A>D N. BREEDER, Anal. Chem., f. Anrer. O_mcial AriaI_ Chemisrs, 53 (1970) 244Xelrsletters. PEREIS-ELMER, No. 4 (undated.) ,Vercsfcrrers, PERKIV-ELMER, No. 5 and 6 (1966)-
Durrsc AXD F. MAGSOTTA, in R. S. PORTER AMI J. F. Jotixsolu (JZds.).
9 L. R. SCOTT AsV A. P. GRAY. DSC Purity and the DSC4 Computer Perkin-E!ner Corp., Norwalk, Corm. (1969); A. P. GRAY ASD PERKS%-ELMER vol. 19. No. 3 (1969).
Program for Purity Anaiysis, Xerrs,
R. Scan, Instrument
IO E. M. BARRALL, II AXE R. D. DILLER, Absrr. Amer. Chew. Sot. 159th .%‘ationaf Meeting, Texas* 19io_ 1I R. REWBKE ASD J. A. MOLLICA, JR_, f_ Pharm. Sri., 56 (1967) 522. Thermochim.
Acta, 2 (1951)
E. F. JOY, J. D. BO?+ZN,A. J. BARNARD, JR.
I2 C PLATD fiD A. R. G LUGOW, JR, Anal. Gem., 41 (1969) 330. I3 N_ J. DEAXGEU~ A5D G_ J_ PAPARm. J_ Pkzrm. Sci_, 57 (1968) 1868. 14 A. P. GRAY. Microchemical Workshop, Pennsylvania State University, (1968). in M. Zm (Ed.). Pbfication on Inorganic and Organic 15 E. hi. BAKEULL, II ASD J. F. Jo-x, Mat&s, Marcel Dckker, New York, 1969. Chapter 7_ 16 J. P. MCCVLIAXIGH AMI G. WADDIXTOS. Anal. Chim. Acta, 17 (1957) 80. I7 E SAWICK~, Chem. _&al_, 53 (1964) 56. 18 Urea, NBS Certificate of Analysis, Standard Reference Material 912 (1968). 19 k J. BARXARD. JR. FL E JOY ASD F. W_ MKHELOIT, Amer. Assoc, Clin Cfiem. 22nd Nor. Meet_, Bujido, IV_ Y., JuZy 1970, Paper 51; abstracted, C’lin. Chem., 16 (1970) 527. 20 W_ ~ZKE A%D R SCHAFFER(U. S. National Bureau of Standards), personal communications. 21 G. GOLD~IEIS (Oak Ridge National Laboratory), personal communication.