A primary density standardization of each glass standard is necessary. The calibration consists of a modified pycnometric sink-float procedure (2). With proper precautions the absolute density of an individual bead may be measured to 1 part in I0,ooO. The gradient tube used is similar to that described by Oster and Yamamoto (I) except that the dimensions are reduced considerably. The gradient tube is IO mm i d . borosilicate glass tubing and is 40 cm in length. The reservoir volumes are 50 ml each. Each column covered a relatively wide density range of 2 gramslcc. Organic and inorganic columns worked equally well for the carbon-coated microsphere. High densities approaching 5 gram/cc are obtained by saturating H20 with a thallous formate-malonate double salt and maintaining the columnat about 85°C. A cathetometer is generally used to obtain position measurements on the standards and samples. The cathetometer used cm. No in these experiments measured to within f 5 X movement could be detected with the cathetometer on the 100-pg carbon-coated microspheres after a 5-minute period in the column. Neither could the effects of slight temperature variations along the column length be detected with the cathetometer. The accuracy of this method of measurement was tested by measuring the density of small high-purity KCI crystals in an organic gradient column containing benzene and tetrabro(2) M. Gordon and I. A. Macnab, Tmns. Faraday Soc., 49, 31-9
(IYS3).
mwthane. The density agreed to within 2 parts per thousand of the value given in the International Critical Tables. Because the plot of position us. density is not linear for any one gradient column, curve fitting is required. This method is subject to some error because of graphical misinterpretations as well as being time-consuming when many samples are being determined. Therefore, a computer program was written to analyze, tabulate, and plot the data. A typical plot returned from the C.D.C. 1604A computer is shown in Figure 1. The open circles represent standard particles while the smaller black dots represent sample particles. The computer fits the best third-order polynomial to these standard points by a least squares method and then plots the sample particles on the curve. This plot refers to a 50-particle batch of microspheres representing the density range of a single coating run. Other useful output information which accompanies this computer plot is the calculated value of the standard data points from the “best fit” third-order polynomial equation, the variance of each standard point as calculated from all of the standard data points, sample density, mean sample density, sample standard deviation, and the standard error of fit. The standarderror of fit tells quantitatively how well the standard data fits the calculated equation. Mathematically, it is analogous t o the standard deviation of the height. RECEIVED for review December 14, 1966. Accepted March 3, 1%7. Research sponsored by the U S . Atomic Energy Commission under contract with Union Carbide COrD.
Micropipet Encapsulation for Gas Chromatography Samples Daniel S. Berry Nuclear Chicago Corp., 333 Eost Howard Aue., Des Plaines, Ill THEI ~ O D U C T I O N of sub-microliter quantities of volatile liquids into a gas chromatograph with a syringe is complicated by the variable amount of material which may be boiled out of the needle (/). In work with radioactive compounds this source of error is troublesome, since normalizing curves are not generally used in the quantitative analysis of radioactivity. Although it is possible to produce consistent peak heights for given sample volume, an analysis of contained radioactivity indicated that with variable volumes, inconsistent results are obtained. Methods of introducing sealed samples into gas chromatography columns have been previously described (I, 2). The relatively elaborate equipment and technique required to measure the quantity of encapsulated material make these methods time consuming and expensive. The use of precise bore capillary tubing in the form of disposable micropipets (Figure I ) simplifies the measurement and encapsulation of sub-microliter samples. These pipets used in conjunction with a heated crushing device makes precision of a few per cent readily obtainable over a range of volumes less than one microliter. A machinist’s rule is used to make the measurement of the contained volume. (1) A. G . N e r h e i m , A ~ n ~ C. H E M . , 1686(1Y64). ~~, (2) R. L.Bowman and A. Karmen, Nature, IR2,1233 (1958).
692
-
ANALYTICAL CHEMISTRY
60018
. I
L---.:i
Figure 1. men(
z
.
-.~ ._
One-microiirer micropipers in position lor measurcEXPERIMENTAL
Drummond Scientific Co., Microcaps, disposable micropipets were used to encapsulate and measure the samples. Two separate lots of one-microliter tubes were used for this work, having overall lengths of 33.8 mm and 31.8 mm, respectively. Accurately standardized “C-tagged toluene was used as the test material. An ampoule crusher (Figure 2) similar to that described by Bowman and Karmen ( 2 ) was fitted in the place of the standard injection block of a Wilkens Model 1520 gas chromatograph. Radioactivity measurements were made with a Nuclear-Chicago Model 4998 gas chromatography proportional detector system and with a Nuclear-Chicago Model 6725 liquid scintillation spectrometer.
Table I. Results of Liquid Scintillation Counting of Micropipet Encapsulated 14CToluene" Volume Net count Disintegrationc of rate rate liquid plb cpm + 10' dpm/pl + lo4 4.70) 0.698 2.82 4.67 0.719 2.88 2.98 4.631 0.752 4.76~ 0.683 2.79 0.594 2.37 4.63 4.63 0.390 1.55 2.10 4.62 0.528 4.65' Mean disintegration 0.519 2.07 4 . 6 2 i' r a t e = 4.66 X 104dpm/pl 0.799 3.14 0.582 2.33 4.63'Rel std dev = 0.9z 0.701 2.80 4.60; 0.594 2.38 4.62 0.629 2.54 4.65' 0.487 1.99 4.71I 0.563 2.27 4.65i 4.71 0.437 1.78 2.14 4.7l 0.522 3.15 4.63) 0.786 4.66; 0.629 2.54 a Specific activity, 4.64 X I O 4 dpm/pI b Liquid volume = length of liquid column x volume of' micropipet length of micropipet c Calculated using the liquid scintillation channels ratio method. ~
i
~
~
- i +! Figure 2. Pipet crushing device with heat tape and insulation not shown
A number of disposable onemicroliter pipets were partially filled in the following manner. A glass melting point tube was dipped into the test material and used to transfer a quantity of solution to the micropipets. A touch of the tip of the micropipet to the end of the transfer tube caused it to fill immediately by capillary attraction. Care was taken t o ieave sutficient length (at least 3 mm) at each end t o permit sealing. A range of sample!,from about one quarter to three quarters full was used for these measurements. Each tube was sealed immediately after filling. The contained liquid was brought toward the middle of the microtube by tilting, and one end slowly inserted into the side of a microburner flame. The giass fused in about two seconds. The other end of the tube was sealed in a similu manner with a minimum of heating to prevent pressure from developing in the tube. After a few triais, the entire encapsulation procedure was found to take n o more than a minute per sample. A machinist's rule having one-half millimeter divisions was ilsed to measure the, lengths and hence the volumes of the sample (Figure 1 ) . .First a number oi'empty micropipets was measured to the nearest 0.1 mm, and then the lengths of the liquid columns in the sealed tubes were similarly measured. T!he ratio between the liquid column length and the total tube length was multiplied by one microliter to obtain the volume. T w o lots of tubes were found to havc different lengths, but as determined by liquid scintillation counting they were found to containthe same volunie. To determine the a8:curacy of the procedure described above, >I number of measured and sealed micropipet sampies of high specific activity ''C-(01uene were counted by liquid scintillation. The sealed m ciotubes were piaced IP. counting vials, covered with a toiuene-based counting solution, and carefully broken with a g ass rod. After rinsing the glass rod, the vials were sealed, mixed by shaking, and counted. Efficiency
Table 11. Encapsulated Carbon-14 Toluene Counted Subsequent to GLC Using a Heated Proportional Counter Net total Total counts Volume integrated per ~ r of l Ot' sample + I O 4 toluene, p1 counts + I O 4 0.680 1.45 2.13 mean = 2.15 X I O 4 0.689 1.48 2.14counts/pi 0.695 1.52 2.19 0.633 1.35 2.13 Re! std dev = 1.3% 0.562 1.22 2.17 0.428 0.638 0.421 0.755 0.535
0.85 1.19 0.83 1.48 1.08
0.711
1.31
0.770 0.292 0.277 0.808 0.497 0,437
1.55 0.60 0.53 1 55 I .011 0 83
1 .9? 1 86 mean = 1.94 >c 10' 1 .97 counts/pi
1.96 2.02 Re1 std dev 1.84
=
3.87;
2.01, 2 . 0 5 mean
= 1.97 X IO4 i .91 counts/pl 1.92 2 01 Rei std dev = 3 . 3 x 1.90
was determined by the channels ratio method (3). The results are given in Table I. Another series of encapsulated and measured samples was counted subsequent to gas chromatography analysis by being passed through a heated 85-ml proportional counter attached t o the chromatograph ( 4 , 5). The counter, maintained a t 280" C, had a 400-V plateau using a I-mV input sensitivity. A stream of heated propane was mixed with the helium carrier prior to counting in order to form a suitable counting gas. Total flow rates of approximately 90 mljmin were used. (3) E. T. Bush, ANAL.CHEM., 35, 1024 (1963). (4) L. Bruzzi, A. Castelli, and A . Cervellati, Nurl. Ittstr. and Mefiz0d.r.26. 305 (1964). (5) K. Wolfgang and F. S . Rowland, ANAL. CHEhi.: 30,903 (1958). VOL. 39, NO. 6, M A Y 1967
693
RESULTS AND DISCUSSION
Since the results of this gas counting is directly dependent on the average length of time the sample spends in the counter as the mixed gas stream mrpes through it, a total integrated count per sample was recorded rather than a count rate. The residence time in the counter is difficult to determine accurately except by comparisons using standardized samples; therefore no attempt was made to express the results in the form of the specific activity of the sample. Also, because of the difficulty in precisely duplicating flow rates from d a y to
day, only those results obtained on a given day were used t o calculate the precision of the measurements. The results are listed in Table 11. The difference in precision of the results obtained by counting radioactive samples in a liquid scintillation counter and those obtained with a flow-through proportional counter probably results from small changes in gas flow rates, and from short counting times in the gas counter with consequently greater statistical variation. RECEIVED for review January 19, 1967. Accepted March 14, 1967.
A Pyrolysis Oven Which Utilizes a Preheated Helium Stream R . A. Prosser, J. T. Stapler, and W. E. C. Yelland Materials Research Brunch, U. S . Army Nutick Laborutories, Nutick, Muss.
As PART OF A STUDY of the thermal degradation of polymers, attempts were made to determine the effects of additives such as metals and metallic oxides on pyrolysis reactions and products. This requires that polymer samples be pyrolyzed under known, reproducible, and reasonably uniform conditions so that the results can be subsequently compared with those obtained from mixtures of polymer and additive. The pyrolysis should be done in such a way that the additive will not seriously affect the temperature distribution throughout the sample. Although these conditions can be achieved by placing a sealed vessel containing the charge in a hot oven a t temperature equilibrium, the approach suffers from a serious disadvantage. The first products of degradation may undergo further reactions. .4s a result, the establishment of a mechanism of decoinposition becomes more difficult, and the effects of the additives may be obscured. Secondary degradation can be reduced by sweeping the sample with an inert gas as is the case when the pyrolysis is carried out in the carrier stream of a gas-liquid chromatograph. The decomposition products are thereby quickly removed !tom the oven and can be either swept directly onto the column ofthe gas-liquid chromatograph or cooled quickly and stored .it a low temperature. Sweeping as normally carried out can create a problem, because i f the temperature of the inert gas is different from that of the oven, a considerable variation in the temperature throughout the sample may exist. The initial experiments showed that the temperature difference between the helium and the wall at the sample position was as high as 100" C , depending on the flow rate. Serious temperature differences have been reported by others ( I ) . -!'he presence of an additive can also affect the temperature distribution in the sample. For a powdered polymer sample containing a finely dispersed metal, conduction and radiation from the oven wall are important sources of heat. Also, the presence of the metal in the polymer mixture tends to equalize the temperature throughout the sampie. I n a sample containing a metallic oxide, the time required for the entire charge to reach pyrolysis temperature will generally be increased. The pure polymer, when molten, is in many cases transparent and will absorb little radiant energy. Therefore, the surface temperature of the sample will be close to that of the inert gas; the bottom layer, which is in contact with the container, will be a t (1) I3. A. Vassalln, ANAL. CHEM., 33, 1823 (1961).
694
ANALYTICAL CHEMISTRY
In
Figure 1.
Pyrolysis oven
( A ) Hot Watt heater ( B ) External heater ( C ) Stainless-steel test tube (D) Sample holder ( E ) 34/45 ground joints ( F ) Concave region ( G ) Controlling thermocouple (H) Power leads (He) Helium (Q) Quartz wool insulation
the same temperature as the container wall (especially when bulk charges of 3 grams or more of the polymer are pyrolyzed). When the temperature of the inert gas is well below that of the wall of the oven, the time to reach temperature equilibrium and the temperature distribution in the various samples could be sufficiently different to change the relative amounts of pyrolysis products. This could obscure or even mask the chemical and/or catalytic effects of the additive on the pyrolysis, especially when attempting to ascertain the initial degradation products by carrying out the pyrolysis just above the threshold of decomposition. Madorsky ( 2 ) found wide differences in the temperatures obtained by several workers in similar uegradation studies of polystyrene. It is difficult to assess the degree of nonuniformity of temperature in a charge. One way to minimize this problem is t o place the charge in an oven so hot that pyrolysis takes place immediately. However, Radell and Strutz ( 3 ) found that the pyrciyzate of a pure polymer is not reproducible above a n upper temperature limit. Another way is to heat the sample
(2) S. L. Madorsky, J. Res., Natl. Bur. Std.. 62, 219 (i95Y). (3) E. A. RadeIl and H. C. Strutz, ANAL. CHEM., 31, 1890 (1959).
