June 5, 1960
`
`IONIZATION OR CITRIC ,\CIU
`ORGANIC AND BIOLOGICAL CHEMISTRY
`
`2705
`
`[CONTRIBUTION NO. 2493 FROM THE GATES AND CRELLIN LABOR.4TORIES OF CHEMISTRY, CALIFORNIA
`TECHNOLOGY]
`The Ionization of Citric Acid Studied by the Nuclear Magnetic Resonance
`
`BY A. LOEWENSTEIN~ AND JOHN D. ROBERTS
`
`RECEIVED SEPTEMBER 14, 1959
`
`INSTITUTE O F
`
`Technique
`
`The chemical shift of the methylene hydrogens in citric acid was measured as a function of the PH and the concentration
`of the solution.
`From these measurements it was possible to evaluate the chemical shifts associated with each ionization step
`in citric acid.
`Measurements were also made of methylene resonance line positions for ionized and non-ionized methyl
`esters of citric acid.
`Analysis of the results indicates that the first and second ionizations of citric acid take place predomi-
`nantly a t the terminal carboxyl groups.
`
`Introduction
`Citric acid has three ionizable carboxyl groups
`and hence three dissociation constants, here de-
`noted as PKA,, P K A ~ and PKA~. The monoionized
`citrate ion may exist in two forms, which are in
`equilibrium: in one form, the hydrogen of the car-
`boxyl group a to the methylene group is ionized;
`whereas, in the second form, the hydrogen is ion-
`ized froni the carboxyl group bonded to the tertiary
`carbon atom. In a similar way, two forms of the
`djionized citrate ion may exist. The purpose of
`this investigation was to evaluate the relative con-
`centrations of the different species in solution using
`the n.m.r. technique.
`The n.m.r. spectrum (Fig. 1) of an aqueous solu-
`tion of citric acid shows a strong doublet and two
`weak lines, each one symmetrically spaced on both
`sides of the doublet. These lines are due to the two
`methylene groups of citric acid and can be inter-
`preted as being due to the non-equivalence of the two
`hydrogens in each methylene group3 The car-
`boxylic, hydroxylic and water hydrogens exchange
`rapidly with one another and therefore appear as a
`single line.4 -4s most of the experiments described
`in this paper involved measurements of chemical
`shifts, it was found convenient to measure all shifts
`relative to the resonance of a reference material
`added to the solution. The material chosen for
`this purpose was tetramethylammonium bromide
`(TMA),5 which was added to all solutions in a con-
`centration of about 0.2 14.4. A typical spectrum of
`an aqueous solution of citric acid, containing TMA
`but not showing the exchangeable hydrogens, can be
`seen in Fig. 1.
`It was found that the chemical shift of the center
`of the strong methylene doublet (herein designated
`as 6) changes with the PH of the solution. This
`phenomenon is analogous to similar results obtained
`with other acids and bases.5 From the variation of
`6 with the PH we have estimated the chemical shifts
`associated with each ionization step in citric acid.
`These measurements alone do not, however, pro-
`vide sufficient information to evaluate the relative
`(1) Supported by the Office of Naval Research.
`(2) On leave of absence from the Weizmann Institute of Science,
`Rehovot, Israel.
`(3) Cf. P. M Kair and J. D. Roberts, TITIS JOURNAL, 79, 4565
`(1957).
`(4) E L. Hahn and D. E. Maxwell, Phys. Rev., 88, 1070 (1952);
`H. S. Gutowsky, D. W. McCall and C. P. Slichter, J . Chem. Phys , ‘21,
`279 (1953).
`( 5 ) E. Grunmald, A. Loewenstein and S. hleiboom, ibid., 27, 641
`(1957).
`
`concentrations of the different ionic forms in the
`solution. To achieve this purpose, one or more of
`the exchangeable positions in citric acid has to be
`blocked with the minimum change in the rest of the
`molecule. The materials which appear to best
`serve this purpose are the methyl esters of citric
`acid. A series of measurements was therefore
`taken on the citric acid methyl esters. Analysis of
`the results obtained from citric acid and its methyl
`esters has enabled us to estimate the relative con-
`centrations of the different ionized forms in the
`solution.
`
`Notation
`The following notations will be used to designate
`different methylene and carboxyl groups in citric
`acid and its methyl esters
`y-methylene + CH&OOCH3
`central carboxyl --+ HOOC-(!!--OH
`I
`a-methylene + CHnCOOH t- terminal
`carboxyl
`Using this notation, both methylene groups in citric
`acid are a-methylenes, whereas both methylene
`groups in the symmetrical dimethyl ester are y-
`methylenes. The designations a and y will be used
`irrespective of whether the carboxyl groups are
`ionized or not.
`
`Experimental
`The n. m. r. spectrometer used was the Varian model 4300
`B operated at a proton frequency of 60 Mc. The frequency
`calibration was accomplished by the side-band technique.
`The precision of the measurements is expressed by the stand-
`ard deviation, 6 = [Z dlZ/(n-l)]’/z, where di represents the
`residual from the average value and n is the number of
`measurements (usually about 15).
`The citric acid (A.R. monohydrate) solutions used in the
`n.m.r. measurements were prepared by titrating solutions
`of citric acid in hydrochloric acid (0.1-0.01 N) or in water,
`with a solution of tripotassium citrate in sodium hydroxide
`(0.01-1.0 N). Both acidic and basic solutions were of the
`same molarity in citrate ions and contained a constant con-
`centration of TMA (about 0.2 M). The pH of the mix-
`tures was measured with a Beckman model G pH meter us-
`ing glass and calomel electrodes.
`The trimethyl ester of citric acid was prepared by the
`procedure given by Donaldsona; m.p. 73-73.5”.
`Anal. Calcd.: C, 46.15; H, 5.98. Found: C, 45.83;
`H, 6.24.
`The n.m.r. spectrum of an aqueous solution of trimethyl
`citrate is shown in Fig. 2. The resonance lines from left to
`right (with increasing field) are attributed to the following
`groups: (I) the central 0-CHI, (11) the terminal O-CH3,
`(6) a’. E. Donaldson, K . F. McCleary and E. F. Degering THIS
`JOURNAL, 56, 459 (193-L).
`
`Page 1
`
`LUPIN EX. 1030
`Lupin v. iCeutica
`US Patent No. 8,999,387
`
`

`
`2706
`
`A. LOEWENSTEIN AND JOHN D. ROBERTS
`
`Vol. 82
`
`P
`
`PI
`
`0
`
`G/ S.
`
`40
`
`Pig. 1.-N.m.r. spectrum of aqueous solution of citric acid
`(-1 M ) a t p H -2.
`Tetramethylammonium bromide
`(TMA) was added as a reference compound to all solutions.
`The quadruplet (a strong doublet and two weak satellites)
`due to the non-equivalent methylene hydrogens is clearly
`I n this and subsequent figures the field intensity
`resolved.
`increases from left to right, and the resonance of the ex-
`changeable (acid, hydroxyl and water) protons is not re-
`corded.
`1 1
`
`E
`
`T
`
`
`
`30
`GIs..
`40
`0
`spectrum of a solution of 1 mole of sym-
`Fig. 3.--S.m.r.
`metrical dimethyl citrate in two moles of base. The hy-
`drolysis product
`is unsymmetrical monomethyl citrate.
`( I )
`The resonance lines are attributed t o these groups:
`central-OCHs, (11) terminal-OCHa, (111) methanol, (IV)
`TMA, (V) r-CH2, (VI) a-CH2.
`metrical monomethyl ester), (11) the terminal OCHa group
`in the unsymmetrical monomethyl ester, (111) the methyl
`group in methanol which is a by-product of the hydrolysis
`(IV) TMA, (V) the ?-methylene of
`the unsymmetrical
`monomethyl ester and (\'I) the a-methylene of the unsym-
`the
`metrical monomethyl ester. The above notation of
`lines will be used throughout for +he undissociated esters and
`their corresponding anions.
`The evidence presented supports the above interpretation
`of the spectrum: (1) lines corresponding to I and I1 are
`observed in ratio of 1 :2 in the spectrum of trimethyl ester of
`citric acid a t approximately the same positions relative to
`ThfX (see Fig. 2); ( 2 ) when methanol is added t o the solu-
`tion line 111 grows in intensity, but on reduced pressure
`evaporation of the solution and re-solution in water, line 111
`disappears; (3) line V appears at approximately the same
`position relative to TMA as the y-methylene resonances in
`the di- or trimethyl esters; (4) when sodium citrate is added
`to the solution doublet VI increases in intensity, this being
`due to the addition of a-methylene groups of citrate ion;
`and (5) addition of more base to the solution results in de-
`creasing the intensity of lines I , I1 and V, while lines VI and
`III grow in intensity. This can be explained as being due to
`further hydrolysis of the monomethyl ester to citric acid and
`methanol.
`Citric acid symmetrical monomethyl ester was obtained
`in solution by the basic hydrolysis of the trimethyl ester.
`One mole of trimethyl ester was dissolved in two moles of
`aqueous sodium hydroxide. The n.m.r. spectrum of this
`solution is shown in Fig. 4. The resonance of the exchange-
`able hydrogens is not included in the figure, and the same
`notation for the lines is used as for Figs. 2 and 3. The fact
`that the rapid hydrolysis gives mainly the symmetrical
`monomethyl ester is proved by the appearance of only one
`doublet (VI), which means that both methylene groups are
`located symmetrically with respect to the remaining esteri-
`fied carboxyl group.
`It can be concluded from these experiments that thc hy-
`drolysis of the di- and trimethyl esters to the monoesters is
`fast and complete within the time taken between preparation
`of the solution and recording its n.m.r. spectrum (about 10
`minutes). This agrees with previous observations.*
`Results
`As mentioned in the Introduction, the chemical
`shift (6) between the strong doublet in citric acid
`and T M A changes with the PH of the solution. A
`plot of 6 as function of the PH is shown in Fig. 5 .
`Figure 5 has the characteristics of an acid-base
`titration curve. Since, however, citric acid solu-
`(8) A. Skrabal, Z . Elcklrochcm., 53, 332 (1927).
`
`I
`
`j
`
`
`
`1
`
`1
`
`I
`
`'
`
`
`
`I
`
`I
`
`1
`1
`I
`0
`3 0
`40
`C I S
`r. spectrum of an aqueous solution of tri-
`Fig. 2.-K.n1
`methyl citrate. The resonance lines are attributed to the
`groups: ( I ) central-OCHB, (11) terminal-OCHB, (IV) TMX,
`0') ?-CHt.
`(IV) the TMA and ( V ) the y-methylene. The resonance
`line of the exchangeable hydrogens is not shown. It should
`be noted that the satellite lines expected on both sides of the
`doublet V are barely observable because of their low inten-
`sity. The same will apply to the resonance of all the meth-
`ylene groups shown in succeeding figures.
`Citric acid symmetrical dimethyl ester was prepared ac-
`cording to the procedure given by Schroeter'; m.p. 115-
`117'.
`Anal. Calcd.: C,40.30; H,5.88. Found: C,40.00; H,
`5.75.
`This material is barely soluble in water; therefore, its n.
`m.r. spectrum is recorded with difficulty. Nevertheless the
`weak spectrum obtained was in complete agreement with the
`expectations for the assigned structure. Citric acid sym-
`metrical dimethyl ester dissolves easily in one mole of base
`to give the symmetrical dimethyl citrate. The spectrum
`of this sclution shows one CH, doublet and one 0-CHa line.
`Citric acid unsymmetrical monomethyl ester was obtained
`in solution by basic hydrolysis of one mole of symmetrical
`dimethyl ester with two moles of aqueous sodium hydroxide.
`The n.m.r. spectrum of the solution, taken immediately
`after preparation, is presented in Fig. 3. The resonance of
`the exchangeable hydrogens is not shown.
`The resonance lines shown in Fig. 3, from left to right
`(with increasing field), are attributed to the following groups:
`( I ) a central OCHs group (due to a small amount of sym-
`(7) G . Schroeter, Bcr., 98, 3190 (1905).
`
`Page 2
`
`

`
`June 5, 1960
`
`IONIZATION OF CITRIC ACID
`
`2707
`
`TABLE I
`
`METHYLENE CHEMICAL SHIFTS MEASURED IN C.P.S. RELATIVE TO TMA
`Citric
`Unsymmetrical
`Symmetrical
`Symmetrical
`acid
`monoester
`monoester
`diester
`a-CHI
`YCHZ
`a-CHr
`a-CHI
`Y-CHI
`13.3 f 0 . 2
`13.3 f 0.7
`13.3 f 0 . 6
`13.3 =I= 0 . 7
`Un-ionized
`35.9 i 0.4
`35.9 f 0 . 2
`20.7 f 0.2
`32.6 f 0.6
`Fully ionized
`Not measured due to insolubility. No figure shown for spectrum.
`relative to TMA. The use of equation 1 is justified
`
`tions are strongly buffered and the ~ K A values do
`since the various citric acid species in the solution
`not differ much from each other, the inflection
`are rapidly equilibrated by the very fast exchange
`points in the titration curve of the tribasic acid are
`of the carboxyl hydrogens. Equivalent expressions
`not clearly resolved. Hence, the chemical shifts as-
`have been used previously in n.m.r. studies of sys-
`sociated with each ionization step cannot be directly
`tems which consist of several species in equilib-
`read from Fig. 5 , and a different procedure must be
`r i ~ m . ~
`used. Another feature of Fig. 5 is the independ-
`ence of S of the concentration of the acid. The sig-
`nificance of this observation will be discussed later.
`m m
`I
`PI
`
`. .. .. . . . a
`20.4 =!= 0.7a
`
`Triester
`y-CHa
`13.3 f 0 . 6
`. . . . . . . .
`
`I
`
`I
`
`I
`
`o
`
`l
`0
`
` Q
`
`35t
`c
`
`30
`
`-
`
`-
`
`t
`
`I
`
`20
`
`+
`
`X
`
`t
`
`X
`0
`
`s
`
`*o
`0
`
`Q
`
`0
`8
`0 P
`
`CITRIC ACID
`1.33M 0
`0 . 3 3 M +
`0.17 M X
`TMA-0.2M
`
`I
`1
`
`1
`1
`
`l 5
`
`0
`
`0
`
`I
`2
`
`1
`4
`
`I
`6
`
`I
`8
`
`PH.
`chemical shift 6 between the center of the
`Fig. B.-The
`strong methylene doublet in citric acid and TMA resonance,
`as function of the pH.
`
`I I I
`
`m
`
`P
`
`spectrum of an acidified solution of un-
`Fig. 6.-N.m.r.
`symmetrical monomethyl citrate (alkaline spectrum shown
`in Fig. 3). Line notations are same as in Fig. 3. The CY- and
`7-methylenes resonances here coincide.
`
`spectrum of a solution of 1 mole of tri-
`Fig. 4.--S.m.r.
`methyl citrate in two moles of base. The hydrolysis product
`is symmetrical monomethyl citrate. The resonance lines
`(I) central-OCHa,
`are attributed
`to
`these groups:
`(111) methanol, (IV) TMA, (VI) CY-CH2.
`The spectrum of the diionized, unsymmetrical
`monomethyl ester of citric acid was shown in Fig. 3.
`When two equivalents of acid are added to the solu-
`tion of the diionized, unsymmetrical monomethyl
`ester, remarkable changes are observed in the spec-
`trum as is seen in Fig. 6. This spectrum corre-
`sponds to the formation of un-ionized monomethyl
`ester in which the CY- and y-methylene resonances
`coincide.
`The spectrum of an acidified solution of the sym-
`metricai monomethyl ester shows only a shift i n the
`position of the doublet VI.
`The chemical shifts as measured from Figs. 2-6
`are given in Table I.
`Interpretation of the Data
`The results were interpreted in two steps: (1)
`evaluation of the chemical shifts associated with
`each ionization step in citric acid and (2) evaluation
`of the relative concentrations of the different forms
`of the mono- and diionized citrate ions.
`(1) The observed chemical shifts, 6, in citric acid
`can be described by the equation
`6 = XI61 f x262 + x862 + x4b4
`(1)
`where XI, x?, x3 and x4 represent the mole fractions
`of the non-, mono-, di- and triionized citrates, re-
`spectively, and &, &, 63 and 84 their chemical shifts
`
`Page 3
`
`

`
`2708
`
`-1. LOEWENSTEIN AND J O H N I). KOBEKT~
`
`VOl. 82
`
`H O O C t O H
`
`HOOC-fOH
`
`COOCH3
`
`coocn3
`
`HOOC[ON
`
`- 0 O C I O H coo-
`
`COOH H3COOfoo- coo-
`i H 3 C O O C F o H
`
`C O O H
`
`- 0 O C t
`
` OH
`
`-0OC +OH
`
`R E L A T I V E C H E M I C A L S H I F T S :
`
`F
`
`R E L A T I V E C H E M I C A L SHIFTS :
`
`representation of monornetliyl esters
`Fig. S.-Schematic
`of citric acid and notations used in calculations. The meas-
`B
`A
`Zero
`C
`ured chemical shifts, relative to non-ionized citric acid, arc
`indicated as D, E and F.
`representation of the different ionic
`Fig. 7.-Schematic
`forms of citric acid and the notations used in calculating
`different for 62 as compared to the previously de-
`their relative concentrations. The calculated chemical
`scribed procedure. The value of & = 22.5 c.~'s.,
`shifts of the mono-, di- and triionized citric acid, relative
`however, is preferred since by using it the calculated
`to the non-ionized acid, are indicated as -1, B and C.
`&values are more equally scattered (positive and
`negative deviations) along the experimental curve.
`The \dues of & and 64 can be taken directly
`The chemical shifts associated with each ioniza-
`from Fig. 5 or Table I since they are values of 6 at
`tion in citric acid therefore are
`very low and very high PH, respectively. Thus 61
`= 13.3 i. 0.2 c.!s. and 64 = 35.9 =k 0.2 c./s. The
`singleionization: 6? - 61 = 22.5 - 13.3 = 9.2 & 0.3 c.,s.
`doubleionization: 8,? - h1 = 33.5 - 13.3 = 20.2 + 0.3 c./
`values of 62 and 62 have been calculated so as to
`give the best fit with the titration curve presented
`S .
`in Fig. 5 . For the computation of xlr xz, x3 and x4
`triple ionization: aq - 6, = 35.9 - 13.3 = 22.6 f 0.3 c.,Is.
`for each pH, these PKA values were usedlO: $KA,
`It should be noted that both the use of equation 1
`= 3.13, PA-A? = 4.76, ~ K A , = 6.40.
`and the PKA values cited above are supported by
`the fact that the chemical shifts, within the experi-
`The procedure used in the calculation of 6 2 and
`mental accuracy, are independent of the concentra-
`6a from the data given in Fig. 5 is given : equation 1
`tion.
`If more than four terms contribute to 6 in
`was set up numerically for each point on the titra-
`equation 1 (ie., presence of associated species in
`and
`tion curve, using the experimental values of 6,
`the solution) or if the K A values were not applicable
`64 and the calculated values of the mole fractions
`at the concentrations measured, one would expect 6
`( x i ) . This gave one equation for each point on the
`to be concentration dependent.
`titration curve that included 6 2 and 63 as unknowns.
`(2) Figures 7 and 8 illustrate the chemical shifts,
`A11 possible pairs of such equations then were used
`species and notations involved in the calculation oi
`to solve for and 6:). In practice, only the measure-
`the relative concentrations of the ionized forms.
`ments in the pH range of 3-5.5 were chosen for the
`The symbols x, y , u and u are the mole fractions
`computation since the change of 6 with the hydro-
`such that x + y equals unity for the monoionized
`gen ion concentration at a pH lower than 3 or higher
`of the ionized species of citric acid shown in Fig. 7,
`acid and u + v equals unity for the diionized acid.
`than 5.5 was small and was measured with insuf-
`ficient accuracy for this type of calculation. The
`average results thus obtained from the measure-
`ments taken with the 0.33 and 0.17 M solutions are:
`and a3 = 33.5 + 0.2 c./s.
`= 22.5 f 0.2 c. k A A
`The procedure described above gives a minimum to
`X(6i - Ai) calculated for all the points on the
`curve, where Ai are the calculated values of 6 using
`given above.
`the values of 62 and
`Another procedure to calculate 62 and 6;+ was car-
`ried out: Different values of & and & were chosen
`by a "trial and error" procedure with the purpose of
`minimizing Li(6, - Ai)2. The best values for 8 2
`and & thus obtained were A2 = 23.5 c./s. and 63 =
`33.5 c./s. The result was identical for 63 and slightly
`!9) See for example: M. Saunders and J. B. Hyne, J. Chem. Phys.,
`29, 1314 (1958).
`i10) R . G. Rates and G. 1). Pinching, THIS JOURNAL, 71, 1274
`i I(r491.
`( I I ) T h e precision given here is not the standard deviation b u t
`the average deviation, ZiSi - A i / / % , where n is the number of points
`I t should be noted t h a t the method of least squares can
`on the curve.
`not he appliecl successfully t o this problem because of the particular
`rirriii < > f the function 6(xt, 8,).
`
`In addition, we define the following shifts measured
`relative to the methylene line of the non-ionized
`acid taken as zero
`a = chemical shift of a methylene group located between
`a non-ionized, terminal carboxyl group and an ionized, cen-
`tral carboxyl group
`b = chemical shift of a methylene group located betweeii
`an ionized, terminal carboxyl group and a non-ionized, cen-
`tral carboxyl group
`c = chemical shift of a methylene group located between
`two (central and terminal) ionized carboxyl groups
`The following assumptions were made and are in-
`cluded in Figs. 7 and 8: (a) The methylated car-
`boxyl group behaves with regard to the chemical
`shift as if it were a non-ionized carboxyl group.
`This assumption is supported by the observation
`that in all the non-ionized methyl esters only one
`doublet is observed and its chemical shift is the
`same as for non-ionized citric acid.12 (b) The
`(12) This observation could also be explained by assuming fast ex
`change of the alkyl radical of the ester group5 between possible posi-
`tions; this, however, is not corisiderrd to lx prrhalile.
`
`Page 4
`
`

`
`Julie 5, 1960
`
`2709
`
`OF CITRIC ACID
`IONIZATION
`tain 1 9 . 3 ~ + 1 4 . 9 ~ = 20.2. This result obviously is
`
`From numerical substitution in equation 3 we ob-
`
`inconsistent with equation 4-b and the demand that
`v and u should be positive. However the magni-
`to 0.9
`tude of
`the inconsistency
`is reduced
`C.P.S. (and not far beyond our experimental error) if
`v is taken to be unity, which means that the diion-
`ized acid is solely in the symmetrical form.13 It is
`very possible that besides experimental error, the
`observed inconsistency of the results arises from
`slight differences in the values of a and b between
`the doubly ionized acid and the methyl esters.
`Such differences will be critical since it will be re-
`membered that a and b were assumed to have the
`same values in all comparable species. Differences
`in a and b might arise from molecular association
`and hydrogen bonding which should play a more
`important role in the highly ionized species than in
`the esters.
`The method of treatment of the data for citric
`acid was checked successfully through an analogous
`set of measurements and computations using suc-
`cinic acid and its monomethyl ester. The results
`are presented in Appendix A.
`Several factors may be effective in determining
`the location of
`the non-ionized protons:
`(1) a
`statistical factor of two will favor the unsymmetri-
`cal forms of the mono- and diionized species over the
`symmetrical forms ; ( 2 ) maximum separation of
`the charges will lower the electrostatic interaction
`energy (this is important for the diionized ion and
`will tend to localize the undissociated proton on the
`central carboxyl group) ; (3) the alcoholic hydroxyl
`and the two methylene groups attached to the ter-
`tiary carbon will be operating as electron-attracting
`groups (negative inductive effect') and, thereby,
`should promote ionization of the central relative to
`the terminal carboxyl groups; and (4) if the hy-
`droxyl group is intramolecul.irly hydrogen bonded
`to the terminal carboxyl groups, then ionization of
`these groups will strengthen the hydrogen bond,
`thus lowering the total energy.
`The experimental results indicate that the bal-
`ance of the above (or other) effects is to favor both
`initial and secondary ionization at the terminal car-
`boxyl groups. Which effects are the more impor-
`tant is not now known.
`L. is indebted to the
`Acknowledgment.-A.
`Conference Board of Associated Research Councils
`for a Fulbright Travel Grant. We thank Drs. G.
`Fraenkel and P. R. Shafer for helpful discussions.
`Some preliminary, n.m.r. work on the citric acid
`problem was done by Drs. Donald L. and Jenny P.
`Glusker at the suggestion of Dr. A. L. Patterson
`Appendix A
`Succinic Acid and Monomethyl Succinate.
`Measurements on succinic acid and its methyl ester
`were made to provide a check on the calculations
`and on some of the results obtained for citric acid.
`From a titration curve similar to that shown in
`Fig. 5 for citric acid the chemical shifts associated
`with the first and second ionizations in succinic acid
`were obtained. The procedure used to analyze the
`curve was the same as for citric acid. Equation
`z' = 1 2 =t 0 2, and u = - 0 2 =
`(13) Calculation of u and v gives
`0 2 .
`
`chemical shift of a methylene group is not affected
`much by the terminal carboxyl located a t the other
`end of the molecule, whether ionized or not. The
`validity of this assumption has been verified by the
`measurements of a and c in different species.
`We proceed now to evaluate a, b and c using the
`data given in Table I. From the data given for the
`diionized, unsymmetrical monomethyl ester, we ob-
`tained the values for a and c since
`a = E = 20.7 - 13.3 = 7.4 i 0.7 c./s.
`
`and
`
`c = D = 35.9 - 13.3 = 22.6 zk 0.8 c./s.
`The value of c also can be obtained from the meas-
`ured chemical shift in triionized citrate since
`c = C = 35.9 - 13.3 = 22.6 It 0.3 c./s.
`which agrees with the value obtained above.
`The value of a can also be obtained from the chemi-
`cal shift of the y-methylene group in the ionized,
`symmetrical dimethyl ester, which gives a = 20.4 -
`13.3 = 7.1 f 0.7 c./s. The small difference be-
`tween this value and the value obtained above from
`the unsymmetrical monoester (a = 7.4 c.,/s.) may
`reflect the shift due to ionization in the far-end car-
`boxyl group (in the unsymmetrical monoester) or be
`due to experimental inaccuracy. An average value,
`a = 7.25 zk 0.5 c./s., will be used in further compu-
`tations.
`The value of b was obtained from the shift of the
`methylene group in the fully ionized, symmetrical
`monoester since
`b = F = 32.6 - 13.3 = 19.3 i 0.8 c . / s .
`One might also obtain b from the methylene shift
`in the ionized, unsymmetrical dimethyl ester ; this
`compound, however, was not prepared. Hydroly-
`sis of the triester with one mole of base gave a com-
`plex mixture of several hydrolysis products, to-
`gether with non-hydrolyzed material.
`The values obtained for a, b and c show that the
`strongest effect on the chemical shift of the methyl-
`ene group is caused by ionization in the terminal
`carboxyl group, whereas ionization in the central
`carboxyl group is less effective. This is to be ex-
`pected since one more bond separates the methyl-
`ene group from the central carboxyl as compared
`to the terminal carboxyl.
`Two equations can now be set up which will de-
`scribe the observed chemical shifts in the mono-
`and diionized citrate ions
`ax + by/2 = A
`bv + (a + c ) u / 2 = B
`and by definition
`
`(2)
`(3)
`
`(44)
`x + y = l
`(4-b)
`u + u = l
`Equations 2 and 3, like equation 1, arc based on
`the chemical equilibrium existing between the ionic
`species as a result of fast exchange of the hydrogens.
`2, we obtain 7.25, + 9 . 6 5 ~ = 9.2. With the aid of
`Substituting the values of a, b and A in equation
`equation 4-a we can solve for x and y, the result
`being
`x = 0.2 rt 0.2 and J = 0.8 + 0.2
`This means that the monoionized citrate ion is
`about SO% in the unsyirimetrical form.
`
`Page 5
`
`

`
`2710
`
`h O S J. LEFFLER AND EUGENE G. T E A C H
`1 simplifies in this case to
`6’ = X1’6,’ + x2’62’ + x3’63’
`(1’)
`where 6‘ is the measured chemical shift, &’, &’ and 83’
`the chemical shifts of the non-, mono- and diion-
`ized species, andxl’, xn’, x3’ their corresponding mole
`
`fractions. The ~ K A values which were used to cal-
`culate xl, x2 and x s are14: PKAI = 4.19 and PKA’I
`= 5.48. The results are: &’ = 30.9 =t 0.3 c./s.,
`= 42.3 f 0.2 c./s.ll and 6,’ = 47.8 * 0.3 c./s.
`Hence, the chemical shifts associated with the ioni-
`zation are
`single ionization: 42.3 - 30.9 = 11.4 i 0.4 c./s.
`double ionization: 47.8 - 30.9 = 16.9 3~ 0.4 c./s.
`In the monomethyl succinate ion the chemical
`shift of the 0-methylene group was measured and
`(14) H. C. Brown, D. H. McDaniel and 0. Hafliger in E. A. Braude
`and F. C. Nachod, “Determination of Organic Structure b y Physical
`Methods,” Academic Press, Inc., New York, N. Y., lQ5.5, p. 624.
`
`Vol. 82
`
`found to be 11.3 A 1.7 c./s. This value is in good
`agreement with the value of 11.4 c./s. obtained from
`the analysis of the titration curve. That these two
`values should be the same is obvious since in the
`monoionized succinic acid both carboxyl groups
`have equal ionization probabilities. It is also rather
`close to the value of b 12 as given above. The p-
`methylene group in the methyl succinate ion is
`shifted only 4.4 c. Is., which indicates the small ef-
`fect of the negative charge at the far end of the ion.
`In the non-ionized methyl succinate, there is no
`chemical shift between the CY- and &methylene
`groups (same as in the non-ionized methyl citrates).
`This indicates again that the methyl ester group
`behaves like a non-ionized carboxyl group with re-
`spect to the chemical shift of the adjacent methyl-
`ene group.
`PASADENA, CALIF.
`
`PROM THE RICHMOND LABORATORY OF THE STAUFFER CHEMICAL Co., RICHMOND 4, CALIF. ]
`[CONTRIBUTIOX
`Preparation and Some Reactions of Tris-trimethylsilylphosphinel
`BY Axos J. LEFFLER AND EUGENE G. TEACH
`RECEIVED NOVEMBER 20, 1959
`
`Tris-trimethylsilylphosphine was prepared by the reaction of sodium dihydrogen phosphide and trirnethylchlorosilane in
`a non-ammoniacal medium.
`A number of reactions were investigated to determine the characteristics of this compound.
`I t was found slowly to decompose thermally above 100” in a sealed vessel.
`
`The preparation and some of the reactions of
`tris-trimethylsilylphosphine
`recently have been
`described in the literature.2 This work was car-
`ried out independently and differences in the
`method of synthesis and unreported reactions
`have prompted us to publish the new information.
`In this work sodium dihydrogen phosphide was
`prepared directly from the alkali metal and phos-
`phine without any intermediate metal alkyl step.
`This was accomplished by dispersing sodium in an
`aromatic hydrocarbon, and adding on cooling an
`equal amount of a glycol ether such as Ansul 121
`or 141, and then passing in PH3 until the reaction
`was complete.
`In this system there was no iso-
`lation of the intermediate mono- and bis-trimethyl-
`silylphosphines. It is assumed that the reaction
`is a stepwise one in which the intermediates react
`with NaPHz or disproportionate to form the ter-
`tiary compound as
`(CH3)aSiPH2 + NaPHz + (CH3)zSiPHNa + PHI
`+ (CH3)sSiCI ----f [(CHs)&],PH + NaC1, etc.
`or
`2(CHI)SSiPHz --+ [(CH&Si]*PH -+ PH3, etc.
`(2)
`It seems unlikely that reaction occurs between the
`intermediate and the silyl halide
`to produce
`hydrogen chloride, since NaPH2 was always in ex-
`(CH3)sSiPHs + (CH&SiCl+
`[(CH&Si]tPH + HC1
`cess and no hydrogen chloride was ever noted in
`this reaction.
`(1) Presented in Dart a t the 136th A.C.S. hfeetina. Atlantic Citv
`
`(1)
`
`The phosphine reacted with diborane a t room
`temperature to give the adduct, (SiMe3)3P.BH3,
`which split out trimethylsilane on heating giving
`a complex mixture. Treatment with methanol
`left a white crystalline powder that analyzed for
`the formula [(SiMe3)zPBHzlx that is assumed to
`be the analog of Burg’s phosphinoborine~.~ The
`material had an extended melting range and is
`presumed to be a mixture of different size rings.
`The material that reacted with methanol was
`found to be a highly cross linked polymer contain-
`ing boron, phosphorus and trimethylsilyl groups.
`Heating split out more and more trimethylsilane
`until the product was a hard, brittle, infusible
`solid.
`A similar product was formed with pentaborane-
`9. There was no reaction at room temperature,
`but on heating to 90” trimethylsilane, hydrogen
`and methane were evolved. It was noted in two
`experiments that the reaction ratio of (SiMe3)3P/
`BsH9 was 1.40 but the significance of this is not
`certain. No ring compounds were isolated; only
`the brown cross linked polymer was found.
`It was expected that the phosphine would forni
`complexes with transition metal halides, but with
`cobaltous chloride in
`tetrahydrofuran a black
`powder was formed that became stable to air only
`after first growing warm. Analysis showed this to
`be very finely divided c03P2 which was formed by
`the reaction
`2(SiMe~)aP + 3Coc12 + CO~PZ + 6SiMeaC1
`
`search for partial support of this work.
`(2) G. W. Parshall and R. V. Lindsey, Jr., THIS JOURNAL. 81, 6973
`(1959). We wish t c thank the author for a preprint of his paper.
`
`and Pattock4 with phosphine and nickel Ghloride.
`(3) A. B. Burg and R. I. Wagner, ibid., 1 5 , 3873 (1953).
`
`Page 6