sample-injection solenoid. The period required for higher-boiling compounds to emerge or for cooling the oven after a temperature-programmed run can easily be provided by leaving a suitable length of blank tape after the last punched hole. To control the time cycle of a temperature-programmed oven, a third slit, phototransistor, and set of punched holes could be added. A further extension of the basic idea of this programmer is the provision of a
Thermometer
number of channels, each of which is connected to its own solenoid valve and associated trap. This would allow separated components to be directed to any chosen trap, an arrangement which would be valuable where only a few components are to be collected from a large volume of a mixture. Provided chromatograms are reproducible on a time basis, this programmer also has potential application t o liquid column chromatography especially in
conjunction with the highly sensitive liquid chromatography detectors recently developed. ACKNOWLEDGMENT
The author is indebted to E. Bourn for assistance in constructing the programmer. THIS programmer is the subject of Australian Patent Application No. 6389/ 66.
for Proton Magnetic Resonance Studies of Aqueous Solutions
D. N. Glew, H. D. Mak, J. S. Mclntyre, and N. S. Rath, Exploratory Research Laboratory, Dow Chemicalof Canada, Ltd., Sarnia, Ontario, Canada investigation of the proton Iof magnetic resonance chemical shifts aqueous nonelectrolytes between 0' N A RECENT
and 30' C. using a Varian Associates A-60 spectrometer equipped with a V6057 thermostat attachment, the full instrumental accuracy of k 0 . 3 Ha. could not be fully utilized due to sample temperature cycling, which led to systematic water proton chemical shift variations of k0.7 Hz. Further, the reconimended method ( I ) of sample temperature determination, by substitution of a methanol standard tube for that of the sample, was of uncertain accuracy due to the considerable time lapse of about 20 minutes between the measurements of the sample and that of its thermally equilibrated methanol thermometer substitute. To eliminate these uncertainties without resorting to instrumental modifications, it was required to develop a thermometer to be contained within the normal 5-mm. sample tube, which would provide a rapid and precise definition of the sample temperature in the coil region. To achieve this with convenience, it was decided that the thermometer temperature should be d 4 fined by proton magnetic resonance chemical shift differences of two signals
Table 1.
EXPERIMENTAL
The study of aqueous solutions required that the region 0 < 6 < 6.0 p.p.m. should be available for sample investigation, so that the thermometer signals had to be limited t o the region 6.0 < 6 < 8.3 pap.m. Study of numerous potential liquid mixtures showed that the ternary mixture 3 mole yo tetramethylsilane, 61 mole % m-chlorophenol with 36 mole % trifluoroacetic acid provided a suitable spectrum. In this mixture the m-chlorophenol aromatic protons furnish a complex multiplet signal which changes with temperature but which always contains a strongest sharp signal a t 6 = 6.86 p.p.ni. which moves little with temperature, while the trifluoroacetic acid exchanges rapidly with the phenolic protons t o give a single, sharp, temperature-sensitive peak which varies in the region 7.5 < 6 < 8.3 p.p.m. The temperature-sensitive phenolic peak is overlapped by the aromatic ring protons multiplet when no trifluoroacetic acid
Standard Errors of Methylene and Water Proton Magnetic Resonance Signals using Thermometer No. of
Mole % solute
observations
0 0.10 ethylene oxide 4 . 5 0 ethylene oxide 10.0 ethylene oxide 0.05 dioxane 10.0 dioxane 0 . 2 0 tetrahydrofuran 4 . 5 0 tetrahydrofuran
18 10 9
10.0 tetrahydrofuran Mean of data
1964
from an external standard, measured in the same scan as the sample signals. With such a thermometer it should then be possible to correct all chemical shifts to a standard sample temperature, thereby removing the parasitic effects of temperature fluctuations.
ANALYTICAL CHEMISTRY
Methylene signal error (Ha.)
Water signal error ( H a . )
0.288
0.300 0.274 0.441 0.399 0.342 0.266 0.292 0.344 0.304 0.326
7
14 8 17 13 8
...
is present. the composition given is particularly suitable for the temperature range - 10' t o $30" C. For use in higher temperature ranges, additional trifluoroacetic acid can be added to prevent overlapping of the aromatic multiplet by the phenolic acid peak a t the highest temperature. The thermometer sheath was axially symmetric and thin-walled as colddrawn from 6-mm. diameter borosilicate-glass tubing. It was filled with the liquid mixture via a fine capillary, then freeze-outgassed, and sealed a t its upper end. The final thermometer was a hemispherical bulb of 4.0-mm. diameter a t the bottom gently tapering to a 3.3-mm. diameter at a 6-mm. height, 2.7 mm. at 12 mm., 2.3 mm. at 18 nim., 2.0 mm. at 24 mm., becoming almost parallel at 1 mm. diameter at 100 mm. to the 170 mm. height of the seal. With this tapered thermometer, relative peak height changes of the sample and thermometer signals were readily made by vertical adjustment of the sample tube position. The thermometer locates itself accurately along the sample tube spinning axis by the close fit of the bulb at the bottom and by a centrally pierced Teflon disk surrounding it a t the top. RESULTS A N D DISCUSSION
The thermometer was calibrated using the instrument thermostat a t 14 temperatures between - 10' and +30' C. by the substitution method using a methanol standard ( I ) , which gave a standard error of k0.84' 6.on a single defined temperature. The thermometer accurately represented the calibration temperature t o C. by the linear expression.
t
= 87.961
- 1.0736
(VOH
- VA~H)
in which uOH and V A ~ Hare respectively the phenolic and the aromatic ring proton chemical shifts in Ha. The temperature sensitivity of the thermometer is 1.8 times greater than that of the methanol standard, so that with a
standard error of zk0.3 Hz. on the measurement of each peak frequency, a precision of &0.4' C. should be obtained on a single temperature measurement. Table I shows the standard errors in Herzogs for single determinations of proton magnetic resonance chemical shifts determined on a series of dilute aqueous solutions using the thermometer between 0' and 30' C. The first column gives the solute mole percentage; the second, the number of nieasurements of each chemical shift; the third, the standard error on a single determination of the methylene chemical shifts (which have small temperature coefficients d v ~ d t= zk0.05 Hx./O C.); and the final column, the standard error on a single determination of the water proton chemical shift (which has a large temperature coefficient d v p / d t = -0.55 Hz./"C.). All peaks were sharp
and could be readily measured. The standard error on the methylene proton determinations provides an essentially temperature-independent blank for the overall instrument-measurement accuracy, while the standard error on the water signal includes the temperature error component (dvJv/dt).u ( t ) . It is immediately apparent that the use of the thermometer permits the water signal to be measured with an error of the same magnitude as that for the temperature insensitive methylene groups. The standard error for a single water determination is 10.326 Hz. for the set of 104 determinations, while that for a single methylene group determination is 10.288 Hz. for the set of 124 determinations: the square root of the variance difference of the water and methylene signals is the temperature error component, from which it is found that the thermometer follows the
sample temperature with a standard ) hQ.28' C. error ~ ( t = Over a 9-month period, the thermometer liquid changed color from a light yellow to a dark red-brown; however, no change of the proton magnetic resonance spectrum was observed and regular temperature checks by the methanol standard showed that the thermometer calibration was accurately retained. Convenient and successful use of the thermometer in our particular study of dilute aqueous nonelectrolytes for 1000 chemical shift determinations, lead us t o believe that this proton magnetic resonance thermometric technique may have more general application to other aaueous and alcoholic solution work. LITERATURE CITED
(1) Vmian Assoc., Technical Information Publication 87-100-1 10.
Nuclear Magnetic Resonance Analysis of Carbon-1 3 Enriched Methane Ralph R. Eckstein and Albert Attalla, Monsanto Research Corp., Mound Laboratory, Miamisburg, Ohio
carbon-13 by NhIR A (nuclearofmagnetic resonance) deNALYSIS
scribed in this paper is obtained by a ratio method involving hydrogen atoms bonded to both carbon-12 and carbon-13. The most abundant isotope of carbon (98.792% carbon-12) does not exhibit nuclear magnetism ( 1 ) . Although carbon-13 (1.108% natural abundance) possesses this characteristic, its nucleus is not as sensitive to NMR absorption as the hydrogen atom and therefore gives a less intense absorption signal for equal numbers of nuclei. I n addition t o the ease of detecting the proton signal, the ratio method of determining carbon-13 enrichment obviates the necessity of preparing standard samples and calibrated working curves.
Figure 1.
NMR Dewar sample tube
Shaded portion is internally silvered
EXPERIMENTAL
The carbon-13 enriched methane was separated in thermal diffusion columns by W. >I. Rutherford of this laboratory from Matheson ultra-pure (99.96 mole %) methane gas. The enriched gas was stored in a 500-ml. round-bottom flask which was fitted with a vacuum stopcock and ball joint for transferring the sample to the Dewar system. A Varian Associates DA 60 EL (60 m.c.p.s. dual purpose instrument with external proton lock field stabilization) nuclear magnetic resonance spectrometer was used to obtain the radiofrequency absorption spectra of the liquefied gases. The spectra were produced on a flat-bed Varian recorder a t 500-C.P.S.sweep width and 100-second sweep time I
To improve the sensitivity of detection of the XhfR absorption signal and thereby increase the accuracy of the analysis, a special apparatus was designed and constructed which permits liquefaction of the methane gas (in the NMR probe), its measurement by NMR as a liquid, and, finally, recovery of practically all of the gas. The equipment as designed and built consists of a sample tube of Dewar glassware necessary for the transportation of both cooling gas and sample gas to the NMR probe. The outer and inner walls of the Dewar sample tube (Figure 1) were constructed from 15-mm. to 10-mm. borosilicate glass tubing, respectively. Both walls were continued into the 19/38
taper male joint that was evacuated from thermal insulation. The Dewar glassware used for transporting the sample and cooling gases and for supporting the sample tube in the NLIR probe is illustrated in Figure 2. The cooling gas (dry nitrogen passed through two copper heat exchangers submerged in liquid K2)enters the Dewar system a t point A and flows through the cold finger to exit a t B. After the sample chamber is evacuated, the carbon-13 enriched gas enters a t C. The cold finger, acting as a condenser, pulls the gas into the chamber where it liquefies on the cold finger and fills the sample tube. The liquid level is maintained by constant flow of cooling gas through the cold finger. Thus, the gas can be kept as a liquid in the NhIR probe for an indefinite period of time. To maintain the sample in a liquid state when not under visual observation, the proper chamber pressure was previously determined and the flow rate adjusted to maintain this pressure. No provision was made for spinning the sample tube. RESULTS A N D DISCUSSION
The proton K;MR spectrum of naturally occurring methane gas consists of a large central absorption peak surrounded by two very small, approximately equally spaced, peaks on either side of the central peak. The large central peak is due to the absorption of r j energy by all protons bonded to carbon-12 atoms whereas the two small side peaks are due t o those protons bonded to carbon-13. Because the natVOL 38, NO. 13, DECEMBER 1966
1965
