Gas chromatographic and nuclear magnetic resonance determination

Adam W. Franz , Helmut Kronemayer , Daniel Pfeiffer , Roman D. Pilz , Gänther Reuss , Walter Disteldorf , Armin Otto Gamer , Albrecht Hilt. 2016,1-34...
0 downloads 0 Views 393KB Size
tems considered could be a result of anomalous behavior of the systems or the inherent crudeness of a first-generation model. I t appears, for example, that the model should fit the d a t a for systems where KGLC= K = KNMR,such as is likely for aliphatic alcoholln-electron donor hydrogenbonded complexes (16). T h a t , in itself, would be noteworthy. Nevertheless, in addition to the inconsistencies described above, it presently lacks the support of independ e n t experimental evidence (even inferential) for the existence of local aggregates of A and S in the mixed solvent, or separate clustering of A and S about D when solute is added. A similar two-phase thermodynamic model has been used to interpret solute activity coefficients in nematic liquid-crystalline solvents (32). In t h a t case, however, there was some supporting evidence for solvent clustering from x-ray and light scattering experiments. In short, more experimental evidence and additional refinement are needed before a compelling case can be made for the model.

ACKNOWLEDGMENT Stimulating discussions and correspondence with J. H. Purnell are gratefully acknowledged.

LITERATURE CITED (1) J. H. Purnell, "Gas Chromatography: 1966". A. E. Littlewood, Ed., Elsevier, Amsterdam, 1967, p 3. (2) D. E. Martire and P. Riedl, J. Phys. Chem., 72, 3478 (1968). (3) D. F. Cadogan and J. H. Purnell, J. Chem. SOC.A, 2133 (1968). (4) D. F. Cadogan and J. H. Purnell, J. fhys. Chem., 73, 3489 (1969). (5) C. Eon, C. Pommier, and G. Guiochon, C. R. Acad. Sci., 168, 1553 (1969). (6) C. Eon, C. Pomrnier. and G. Guiochon, J. Phys. Chem., 75, 2632 (1971).

(7) D. L. Meen. F. Morris, and J. H. Purnell, J. Chromatogr. Scb, 9, 281 (1971). (8) R. Vivilecchia and E. L. Karger. J. Am. Chem. SOC., 93, 6598 (1971). (9) C. Eon and B. L. Karger. J. Chromatogr. Sci.. 10, 140 (1972). (10) J. P. Sheridan. D. E. Martire, and Y . B. Tewari, J. Am. Chem. SOC.,94, 3294 (1972). (11) J. P. Sheridan, M. A. Capeless, and D. E. Martire, J. Am. Chem. SOC., 94, 3298 (1972). (12) J. P. Sheridan, D. E. Martire. and F. P. Banda, J. Am. Chem. SOC., 95, 4788 (1973). (13) H. L. Liao, D. E. Martire, and J. P. Sheridan. Anal. Chem., 45, 2087 (1973). (14) J. H. Purnell and 0. P. Srivastava, Anal. Chem., 45, 1111 (1973). (15) C. A. Wellington. Adv. Anal. Chem. Instrum., 11, 237 (1973). (16) H. L. Liao and D. E. Martire, J. Am. Chem. SOC., 96, 2058 (1974). (17) G. M. Janini, J. W. King, and D. E. Martire, J, Am. Chem. SOC., 96, 5368 (1974). (18) D. E. Martire, Anal. Chem., 46, 1712 (1974). (19) C. Eon and G. Guiochon, Anal. Chem., 46, 1393 (1974). (20) R . J. Laub and R. L. Pecsok, Anal. Chem., 46, 1214 (1974). (21) R . J. Laub, V. Ramamurthy, and R. L. Pecsok. Anal. Chem., 46, 1659 (1974). (22) R. J. Laub and R. L. Pecsok, J. Chromatogr., 113, 47 (1975). (23) J. H. Purnell and J. M. Vargas de Andrade, J. Am. Chem. Soc.. 97, 3585 (1975). (24) J. H. Purnell and J. M. Vargas de Andrade. J. Am. Chem. SOC.,97, 3590 (1975). (25) D. E. Martire. J. P. Sheridan, J. W. King, and S. E. O'Donnell, J. Am. Chem. SOC.,in press. (26) S. D. Christian, J. D. Childs. and E. H. Lane, J. Am. Chem. Soc., 94, 6861 (1972). (27) E. H. Lane, S. D. Christian, and J. D. Childs, J. Am. Chem. Soc., 96, 38 (1974). (28) E. A. Guggenheim. Trans. faraday SOC.,56, 1159 (1960). (29) J. E. Prue, J. Chem. SOC.,7534 (1965). (30) R. L. Scott, J. Phys. Chem., 75, 3843 (1971). (31) G. M. Janini and D. E. Martire, J. Phys. Chem., 78, 1644 (1974). (32) L. C. Chow and D. E. Martire. Mol. Cryst. Liq. Cryst., 14, 293 (1971).

RECEIVEDfor review August 25, 1975. Accepted October 23, 1975. This research was supported by a grant from the National Science Foundation.

Gas Chromatographic and Nuclear Magnetic Resonance Determination of Linear Formaldehyde Oligomers in Formalin Wim Dankelman" and Jacq.

M. H. Daemen

Akzo R e s e a r c h Laboratories, Corporate Research Department, Arnhem, The Netherlands

Up to now, no method has been published for determining the oligomer distribution of polyoxymethylene glycols, present in formalin solutions. We found that this distribution can be determined up to the heptamer (HO(CH20),H, n = 7 ) by direct silylation with BSTFA (N,O-bis(trimethylsily1)trifluoroacetamide), followed by GLC analysis on a column filled with 10% OV-1 on Chromosorb W. The results were corroborated with a 220-MHr NMR analysis. Only at 220 MHz is the water signal sufficiently separated from the methylene hydrogen absorptions. The exact amounts of the oligomers with n = 1 and n = 2 and the sum of n 3 3 can be determined by NMR. The results are in accordance with the GLC analysis. Methanol, added as a stabilizer to avoid precipitation of paraformaldehyde breaks down high molecular oligomers of polymethylene glycols, thereby forming more soluble compounds.

According to Walker ( I ), formalin consists of free formaldehyde (- the sulfite or the peroxide method ( I 1.

RESULTS AND DISCUSSION Figure 1. Gas chromatogram of silylated polyoxymethylene glycols HO(CH20)nHin 50% formalin

-0CH20CH20H. Therefore it is not possible t o determine t h e average value of n. Moreover, all formalin samples were prepared from paraformaldehyde or trioxane in deuterated hydrochloric acid, water and/or acetic acid. In formalin, water is a major part of the sample hiding the signals of the different methylene groups a t 60 MHz (or even a t 100 MHz). In this paper, a GLC method is described by which the distribution of formaldehyde oligomers can be determined. T h e results are compared with those of a 220-MHz NMR analysis. T h e mechanism by which methanol stabilizes formalin could be inferred by adding methanol to formalin and comparing GLC and NMR results before and after addition. EXPERIMENTAL Apparatus. Gas chromatograms were obtained on a Varian 1840-3 gas chromatograph. Use was made of a glass column (1 m.

i.d. 2.5 mm) filled with 10%OV-1 on Chromosorb LV,AN', DMCS, 80-100 mesh. Temperature programming was started after 4 min isotherm at 50 "C at a rate of 8 "Cimin and terminated at 230 " C . The carrier gas was nitrogen (15 ml/min). The flame ionization detector was fed by H2 (30 mlimin) and air (300 mlimin). The temperature of the injector and detector was 250 "C. The peak areas were measured with a Hewlett-Packard 3370 B integrator. The GLCiMS spectra were recorded on a L'arian 1700 gas chromatograph coupled to a Varian Match 5 mass spectrometer. A glass column ( 2 m, i.d. 2.5 mm), filled with 10% OV-1 on Chromosorb IV. AIV, DMCS was used. The temperature was programmed at a rate of 10 OC/min from 50-250 O C . The injector temperature was 230 "C, that of the detector (second ion source in the mass spectrometer) 250 "C. Helium was used as a carrier gas (30 ml/min) and was removed before the sample entered the mass spectrometer. A Varian 220 MHz (HR 220) nuclear magnetic resonance spectrometer was used to record the 'H SMR spectra of formalin samples as such or after addition of an equal volume of DMSO-dt;. Reagents. For the silylation reaction .V,O- bis(trimethylsily1)trifluoracetamide (BSTFA) and DMF (silylation grade) from Pierce were used. Procedure. Prior to the GLC analysis. the silylation procedure was carried out as follows. Into a reaction vessel 100 p l DMF and 200 pl BSTFA were introduced: 10 pl of sample was added, the solution was homogenized and left t o stand for 5 min at room temperature, after which 300 pl ethyl acetate was added. Of this solution, 1.5 p1 was injected into the gas chromatograph. From the peak areas, the relative amounts of formaldehyde oligomers can be obtained. These response factors are only determined by the carbon atoms of the trimethyl silyl group ( 2 ) ,and therefore 01

Gas chromatograms of silylated concentrated formalin (50-586 m/m. containing 10.0 methanol) mostly showed seven peaks (Figure 1). apart from those of the solvent and excess reagent. Not only did the retention times occur a t regular intervals but their intensity decreased monotonically indicating the homogeneous nature of the series of peaks. T h e mass spectra from these peaks. obtained with GLCMS all had a similar fragmentation pattern, but with slightly different peak intensities. The fragmentation patterns can be explained from the spectrum of silylated methylene glycol ( M = 192, fragmentation peaks a t m / e = 191, 177, 147, 103. and 7 3 ) . From these results, it was concluded that the peaks in the gas chromatograms were members of a homologous series of silylated polyoxymethylene glycols:

where n varies between 1 and 7. This means that in formalin. polyoxymethylene glycols are present with 1 < n < 7 . unless equilibrium shifts have taken place during the silylation procedure. T h e equilibrium reactions in this system are slow compared with those in the reaction between urea and formaldehyde leading to urea-formaldehyde resins ( 9 ) . Silylation of such a resin takes place without interreaction of the compounds present in the resin ( I O ) . Since this latter silylation reaction is performed under more severe conditions (1 hr, 40 " C ) , it is safe to conclude that during the silylation of formalin. n o shift in the equilibria takes place. T h e quantitative results of the analysis are depicted in Table I. These results are averages of the analyses of six formalin samples from the plant of Methanol Chemie Nederland v.o.f. ( a joint venture of Akzo and DSM) a t Europoort. IH-NMR spectra of formalin were recorded at 220 MHz. because a t 60 or 100 MHz. no separation could be obtained between the signals of the different methylene groups and water present in formalin. T h e spectra were recorded at fi6 "C because cooling of the (hardly stabilized) samples initiates precipitation. This can be avoided by adding an equal volume of DMSO-d6 to the formalin, but this dilution causes a slight change in the distribution of polyoxymethylene glycol homologues (see also Table 111). In t h e ~~

(mass of monomer in sample) = 1 X peak area of peak No. 1 (Figure 1)

402

p~~~~

_

~

~

~~~

_

.

~

~

ANALYTICAL CHEMISTRY, VOL. 48, NO. 2, FEBRUARY 1976

H O ( C" .(I),, I i If

=

1

2

3

4

5

6

7

~

_

Table I. Distribution of Polyoxymethylene Glycols in 50% Formalin at 65 'C, Determined by GLC 8

Relative % ( m / m ) 1 4 . 1 30.1 26.0 15.3 9.0 4.1 1 . 2 0.1

.

~

p

~

I i

166

475

185

340

530

310

PPU-

Flgure 3. 220 MHz NMR spectrum of formalin, to which methanol had been added (see Table II for description of the signals) Figure 2. 220 MHz NMR spectrum of 50% formalin (see Table II for description of the signals)

spectrum of 50% formalin (Figure 2) four signals are found around 6 = 4.7 (A-D), originating from t h e -0CH20groups. I n some samples signals at 6 = 9 and 6 = 5 point t o t h e presence of formic acid and trioxane, respectively. T h e signals of t h e methyl groups of t h e hemiacetals (CH:jO(CH*O),H) around 6 = 3.2 (K-N) are hardly visible (methanol content about 1%)and are therefore also shown a t a lower attenuation. If t h e sample contains more methanol, these signals are clearly visible and also four new signals around 6 = 4.7 (E-H) arise (Figure 3). T h e distinguishable compounds in formalin together with their chemical shifts are shown in Table 11. I t is t o be noted t h a t methylal and free formaldehyde are present in a maximum concentration of 0.5%. From t h e integrals of t h e spectra t h e relative amounts of the oligomers can be calculated. This is shown in Table 111. From Tables I and 111, it can be concluded t h a t t h e results of t h e N M R and the GLC analysis correspond very well. Such a reciprocal check of ‘H N M R and GLC results cannot be obtained for 37% formalin containing 8 or 14% methanol, because in t h e N M R spectrum a t room temperature, t h e hydroxyl signal overlaps t h e CH2 signals. These samples can nevertheless be completely analyzed. From the integrals of the methyl groups a n d t h e total amount of methanol (II 1, t h e exact amounts of the different hemiacetals can be calculated. T h e difference between t h e total amount of formaldehyde and t h a t present in t h e hemiacetals gives t h e amount of polyoxymethylene glycols of which a n oligomer distribution can be determined via GLC. T h e results of two types of methanol, containing formalin, are given in Table IV. Walker ( I , p 79) states t h a t addition of methanol causes the formation of hemiacetals, which are more soluble than the corresponding glycols, thus preventing precipitate formation HOCH20H HOCHzOCH20H

+ CH:3OH s HOCH20CH3 + H20

+ CH30H e HOCHzOCHzOCH3

+ HzO

We have found t h a t another mechanism probably plays a

more important role in preventing paraformaldehyde precipitation. T o 50% formalin 20% (v/v) methanol was added and t h e mixture was kept at 65 “ C for 16 hr. After silylation, t h e relative amounts of formaldehyde oligomers were determined as described for t h e original formalin. T h e results, corrected for t h e dilution caused by t h e addition of methanol, are shown in Table V. Analogous data on formalin without methanol are added. By comparing these results i t can be concluded t h a t 35% of the total amount of formalin has reacted with methanol, which exactly corresponds to the amount of methanol added. I t must be stressed t h a t t h e peaks of silylated hemiacetals with n = 1 and n = 2 coincide with those of the silylating reagent, so t h a t these peaks are not visible in the gas chromatogram. T h e sample was also analyzed by 220 MHz N M R and t h e Table 11. Chemical Shifts of Compounds Present in Formalin to Which Methanol Has Been Added Compound

Chem. shift ( 6 )

HOCH,OH A HOaCH,OCH,OH B B HOCH,O(CH,O)nCH,OH

C

A : 4.65

B: 4.72 C : 4.71 D : 4.78 E : 4.55 L: 3.23 F: 4.63 G : 4.69 M : 3.25 H : 4.7.5 N : 3.26 K : 3.21

D

CH,OCH,OH La E’ CH,OCH, OCH, OH M F G CH,OCH,OCH,(OCH,), OCH,OH N F H D G CH,OH

Table 111. Distribution of Polyoxymethylene Glycols in 50% Formalin at 65 “C, Determined by NMR Relative % ( m / m ) of HO(CH,O),H

Pure solution Formalin, diluted with an equal volume of DMSO-d,

n = l

n = 2

n > 3

15.6 18.2

28.0 31.9

56.4 49.9

ANALYTICAL CHEMISTRY, VOL. 48, NO. 2, FEBRUARY 1976

403

Table IV. Distribution of Polyoxymethylene Glycols and Hemiacetals Present in Stabilized Formalin Samples at Room Temperature, as Determined by GLC and NMR Relative c/c ( m / m ) CH ,O(CH,O), H

HO(CH,O),H

Compound I1

37% formalin + 8% methanol 37% formalin + 14% methanol

1

2

3

4

5

6

I

1

2

>3

Free MeOH

5.1

8.9

6.7

2.7

1.4

0.6

0.3

7.8

5.2

5.7

0.8

5.0

7.1

4.2

1.3

0.8

...

...

13.4

8.6

8.8

1.7

ones and their concentration does not decrease so fast as that of the higher homologues); 2 ) formation of more soluble products (hemiacetals).

Table V. Distribution of Polyoxymethylene Glycols in 50% Formalin with and without Methanol Added, Determined by GLC Relative

4 ( m / m ) of

ACKNOWLEDGMENT

HO(CH,O),H ~~

n

1

3

2

4

5

6

7

S

u

m

Pure formalin 17.0 29.4 26.6 14.3 8.2 3.6 0.9 100.0 20% ( v / v ) 17.0 22.0 16.1 6.1 2.4 0.5 . . . 64.1 methanol added

following conclusions could be drawn: 1) T h e average chain length of the formaldehyde oligomers decreases. 2) Hemiacetals CHsO(CH20),H, mainly with n = 1 and n = 2 are formed. These GLC and NMR results suggest that the following mechanism is operative: HO(CH20),H

+ CH3OH e HO(CH20),-1H

HO(CH20),-1H

+ CHsOCH20H

+ CH30CH20H

2

HO(CH20),-2H

+ CH30(CH20)2H

Methanol therefore stabilizes formalin solutions by: 1) inducing depolymerization; as a consequence, t h e concentration of the higher and less soluble homologues decreases (the lower homologues are also formed from t h e higher

404

T h e authors thank A. J. J. de Breet, F. A. Buytenhuys, W. G. B. Huysmans, H. Kenemans, P. T. van Rens and W. T h . van Wijnen for their valuable contributions t o this work.

LITERATURE CITED (1) J. F. Walker, "Formaldehyde", 3rd ed., Reinhold Publishing Company, New York, 1964. (2) H . L. Gruber and H. Plainer, Chromatographia, 3, 490 (1970). (3) K. J. Bombaugh and 4. F. Bull, Anal. Chem., 34, 1237 (1962). (4) F. OnuHka. J. Janak, S. Dura;, and M . KrEmarova, J. Chromatogr., 40, 209 (1969). (5) E. L. Styskin, Zh. Anal. Khim., 29, 398 (1974); Chem. Abstr., 81, 20618h (1974). (6) K . Moedritzer and J. R. van Wazer, J. fhys. Chem., 70, 2025 (1966). (7) M . Hellin. J . Delman. and F. Coussemant, Bull. SOC.Chim. Fr., 3355 (1967). (8) T. M. Gorrie. S. Kalyana narnan, H . K . Ronette, and H. Zollinger. Helv. Chim. Acta, 56, 175 (1973). (9) Ref. 1, p 72. (10) W . Dankelman, J. M. H . Daemen, A . J. J. d e Breet, J. L. Mulder, W. G. E. Huysmans. and J. d e Wit, Makromol. Chem., in press. ( 1 1 ) E . Smolkova, V. KolouBkova, and L. Feltl, 2. Anal. Chem., 202, 262 (1964).

RECEIVEDfor review August 4, 1975. Accepted October 3, 1975. T h e authors acknowledge the support of this work by Methanol Chemie Nederland v.0.f. ( a joint venture of Akzo and DSM).

ANALYTICAL CHEMISTRY, VOL. 48, NO. 2, FEBRUARY 1976