A Belt-Detector for Quantitative Application in the Fractionation of Polymers by Liquid Chromatography H. W. Johnson, Jr., E. E. Seibert, and F. H. Stross SheN DeoeIopment Co., EmeryoiIIe, CaIf. The belt-detector described here is intended for the analysis of polymers and other nonvolatile materials of broad as well as of narrow molecular weight distribution, as they are eluted from a chromatographic or other separation column. The instrument is designed so that the response can be interpreted in a quantitative manner by simple calibration. This is achieved by using a belt that retains the column effluent completely, and by preventing free diffusion between the carrier gas that moves the components of the effluent, and the atmosphere outside the instrument.
INTHE PAST few years a new type of detector for liquid chromatography (LC) has been described, which is based on flash evaporation of the eluent (which thus has to be volatile relative to the sample) and subsequent pyrolysis of the sample remaining behind. In the chain-, cord-, or belt-detectors operating on this principle (1-7),the LC effluent is deposited on a conveyor, from which the eluent is removed by flash evaporation; the residue is then carried to a pyrolysis zone, and the products of pyrolysis are determined by a sensitive gas chromatographic detector. As liquid chromatography usually involves samples having too low a volatility to be handled by gas chromatography, the volatility requirement mentioned above is usually satisfied. The concept of the conveyor-type detector was independently conceived at Emeryville and a variety of designs have been tested. The objective has been to construct a detector suitable for the very broad peaks commonly encountered in gel permeation chromatography (GPC) and gradient elution fractionation (GEF). Three distinctly different problems are encountered in monitoring chromatographic effluents: base lines, sharp peaks, and broad peaks. These are successively more difficult to resolve, and a design that gives good performance for the first two situations does not ensure satisfactory broad-peak performance. Thus, if a flame ionization detector is used, all conveyor detectors will have good base lines provided the eluent is completely removed before pyrolysis. Sharp peak representation will usually be considered satisfactory in LC if the peak does not significantly overlap the next peak. Electronic filtering can be used to smooth the curve because the area, and not the shape, is important. In fact, spikes and unevenness in these peaks will probably not have much effect on peak area and can usually be tolerated without filtering. Some of the published conveyor detector curves show these defects. (1) E. Haahti and T. Nikkari, Acfa Chem. Scand., 17, 2565 (1963). (2) A. Karmen, T. Walker and R. L. Bowman, J. Lipid Res., 4, 103 (1 963). (3) A. T. James, J. R. Ravenhill and R. P. W. Scott, Chem. Znd. (London), 1964,746. (4) S. Lieberman, U. S. Patent 3,128,619 (1964). ( 5 ) J. E. Stouffer, T. E. Kersten, and P. 1LI. Krueger, Biochim. Biophys. Acfa, 93, 191 (1964). (6) A. Karmen, ANAL.CHEM., 38,286 (1966). (7) R. P. W. Scott, British Patent 998,107 (1965); U. S. Patent 3,292,420 (1966).
By contrast, a single peak lasting up to several hours is usually obtained in G E F and GPC, and the area under a curve must be measured for small increments of retention volume in order to obtain quantitative results. Heavy electronic filtering would continuously shift some of the area to later intervals and cause errors in calculated molecular weight averages and peak shapes, which are very important in these applications. Control of the following experimental problems would appear to be necessary in order to achieve satisfactory broadpeak response. 1. The column effluent must be applied to the conveyor at the same uniform rate as that at which it leaves the column; bends, kinks, and irregularities in the mass and surface texture of the conveyor have been found to cause difficulty. 2. Pyrolysis in the presence of air-contaminated helium gave low and variable response with a flame ionization detector; this was confirmed with gaseous hydrocarbon samples injected directly into the pyrolysis zone. 3. In detectors employing a carrier gas to sweep pyrolysis products to a flame detector, the flow of carrier gas must be very uniform. Variations cause surges in the feeding of material to the flame and also cause transient changes in the pyrolysis temperature profile, which must be maintained constant to afford repeatable flame detector response. A conveyor detector has been designed which successfully overcomes these problems and gives the desired performance with G E F and GPC. The first problem is overcome by using a smooth, flat belt for the conveyor; the second by isolating the pyrolysis zone from the atmosphere; and the third by maintaining the carrier gas in the pyrolysis zone at a relatively high pressure and feeding the gas through a constriction to the flame detector. The resulting design is more complex than the reported conveyor detectors but is not difficult to build or maintain. EXPERIMENTAL
Apparatus. The main part of the detector is housed in a gas-tight enclosure made of aluminum alloy, and consists of three basic parts: an endless belt, a pulley system for suspending and driving the belt, and a series of compartments, each containing a chamber through which the belt passes. The nomenclature and arrangement of components in the beltdetector are shown in Figure 1. In the stripper chamber, a site on the belt is heated to a temperature that will evaporate eluent without removing sample components of interest. In the pyrolyzer chamber, the site is again heated to pyrolyze any sample components. The pyrolysis products are swept into a flame ionization detector in a stream of helium. An additional compartment, the prepyrolyzer, is provided to reduce noise by heating the belt site to slightly higher than pyrolysis temperatures before it re-enters the stripper-pyrolyzer section. During routine operations it is usually not necessary to use the prepyrolyzer. The three chambers are located in compartments, as shown in Figure 1. The stripper chamber is made of a machinable or moldable material of low heat conductivity, because it is necessary to maintain a sizable temperature gradient inside it for effective evaporation of the eluent. A synthetic ceramic VOL. 40, NO. 2, FEBRUARY 1968
0
403
House Vacuum
Line w i t h Heater
Figure 1. Schematic of belt detector (M-600 Lava, D. M. Steward Mfg. Co., Chattanooga, Tenn.) and a ceramic compound that could be molded and fired (Sauereisen Insolute No. 1) both were used successfully for this purpose, As shown in Figure 2, the front side of the stripper chamber is a thin quartz cover plate which is removable. The ends of the chamber have narrow slits extending to the front for insertion of the belt. In operation, this window is seated on a silicone rubber gasket imbedded in the edges of the chamber so as to make the chamber quite tight along its sides and to prevent diffusion of the evaporating solvent into the compartment. The pyrolysis and prepyrolysis chambers, in which it is not necessary to have pronounced temperature gradients, are machined from stainless steel. The removable bottom part of each chamber (to permit convenient exchange of belts) is held in place with simple clamps, as shown in Figure 3. Effluent is applied to the belt by stainless steel tubing inch OD X 0,010 ID) that passes through a small hole in
Figure 2. Detail of stripper chamber 404
ANALYTICAL CHEMISTRY
the top of the stripper chamber. To prevent precipitation of sample components from the column effluent in the line through which it is being moved, the line leading from the column to the stripper chamber is passed coaxially through larger tubing, in which oil held at the proper temperature is circulated. Alternatively, the line can be heated by means of resistance wire insulated by fiberglass sleeving. The line carrying the evaporated solvent from the stripper chamber is heated to prevent condensation and fluctuation of pressure within the detector system. The stripper chamber heaters needed to provide heat for the stripping operation are made of 32 gauge Chrome1 wires wound on quartz rods and fit loosely (to permit easy replacement) into the chambers. They are held in place by small spots of ceramic cement, which can be broken easily when disassembly is required. The pyrolysis and prepyrolysis chambers are each heated by two 85-watt tubular heaters held in place as indicated in Figure 3. Very fine-gauge thermocouples are located in each chamber to monitor the heaters. The heater and thermocouple leads
Figure 3. Detail of pyrolyzer chamber
Filed Square After Positioning and Locking in Jew
', :.
.
. Welder Jaws Closed
Welder J a r s O w n Ribbon Surfacer xythin Jaws Sanded and Cleaned Before Folding
Figure 5. Details of welding procedure
Figure 4.
Belt welcler
are passed through the compartment walls via insulated glands (Conax Corp., Buffalo, N. Y.). The belt is formed by welding the ends of 'Iginch by 0.036 in ribbon of nickel-chromium alloy (Tophet A, Wilbur B. Driver Co., Newark, N. J.) approximately 60 inches long. Fabrication of the belt is described below. The belt is supported by two grooved pulleys, one of which is driven by a synchronous motor so as to produce a belt speed of 1 inch/ second. The bearing for the idler is mounted on a pivoted support weighted so as to provide a belt tension of approximately 0.25 pound. The finished belt is slipped into operating position through the side openings and slots in the chambers and compartments. A shop-built flame ionization detector (FID) and a compatible electrometer amplifier are used without modification. The pyrolysis exit line is connected to the helium inlet of the detector. The other two detector inlets are supplied with air and hydrogen through micrometer needle valves, which are calibrated with a soap film flowmeter. Helium (Air Reduction Co.) is used as carrier gas without additional purification. The helium flow rate at the F I D is checked with a calibrated rotameter before igniting the flame of the detector. While the flame is lit, the helium flow rate is estimated by measuring the pressure of helium gas in the pyrolysis compartment with a water manometer. The belt welder is a modified thermocouple butt welder for 8- to 20-gauge wire (Therm0 Products Co., Milwaukee, Wis., Type BW-1). It is shown in Figure 4. An autotransformer connected to the primary is used to reduce the power output to a more suitable value, and the original spring is replaced by a softer one to reduce upset pressure. Bakelite supports are added to facilitate alignment. A grinder is used to remove the weld head produced during the welding operation. A 2-inch wheel, I/, inch thick, made of 80-mesh, resin-bonded alundum, is driven at 7000 rpm during the grinding operation. The belt is clamped to the periphery of a semicircular support (3-inch radius), which is pivoted at the center. A crosswise notch is filed in the support to accept tbe weld bead so the belt is not distorted at the weld. The grinding wheel is mounted above the support; clearance between belt and wheel is controlled with an adjustment screw.
Procedure. BELTPREPARATION. A 64-inch length of belt ribbon, free from kinks and twists, is used. This allows for a few weld failures, and will still enable one to produce the minimum 60-inch length of belt. The ends of the belt are sanded, tightly folded, and placed in the welder as shown in Figure 5. The weld zone must he purified carefully with helium hy means of a plastic film bubble suitably taped in norifinn I..nnd the i s mid- irrnrdino to P i a n d a d r.l-_ morer_" ...-weld ..-.-."...----_-I.-... ._I dure except that optimum voltage and upset pressure must be established by trial and error on ribbon test strips. Some results of improper welding procedure are shown in Figure 6. If a weld is defective, the operation is repeated. If the weld appears acceptable, the 7Neld is around flat in the grinder, so that the site of the weld is virtually indistinguishable from the rest of the belt. If on., (In%.#. or- d;c?l,.cd during grinding, or if the weld area turns out to be less than 0.005 inch thick, the belt is cut at the weld isnd the operation is repeated. First, the cold 1traps are cleaned DETECTOR OPERATION. and supplied with the appropriate coolant. The column IS disconnected from the effluent applicator, and pure eluent is passed through the column to establish the steady flow needed when the detector is ready for use. The heater for the flame detector, the water supply for the coolers, the helium supply, and the vacuum necessary to produce a slightly reduced pressure at the stripper outlet are turned on. The flame detector must be kept hot enough to prevent detectable deposition of pyrolysis products; a block temperature of 230" C has been found adequate in our work so far, but this feature is being studied further. After 30 minutes, the belt motor and the remaining heaters are turned on. The pyrolyzer and prepyrolyzer heaters are turned on gradually to avoid upsets in the compartment pressures. When the desired temperatures have been attained, the flow of helium to the flame detector is measured by means of a rotameter temporarily connected to the jet by means of rubber tubing. The optimum temperatures for the chambers and lines depend on the nature of the solvents and of the polymer samples. In the stripper chamber, the temperature in the vicinity of the effluent applicator should be somewhat below the boiling point of the solvent used, but the far end of the stripper should be nearly 100" C above it to ensure complete vaporization of the eluent. This condition is achieved quite easily by suitable winding of the stripper chamber heater. In the pyrolysis chamber, temperatures between 450'and 550' C have been found suitable. The optimum temperature for any given polymer was found by determining the maximum response of the flame detector. The maximum prepyrolysis temperature is kept at a level about 30' C higher than the pyrolysis temperatu
"..,
bu
..ll.lo
~~
.
405
A helium flow through the detector jet of about 50 mllmin, a stripper outlet flow of around 300 ml/min, and a prepyrolyzer flow of 50 ml/min have been used mostly in our work. After removing from the flame detector the tubing leading to the flowmeter, the detector is put into operation in the usual manner, and the recorder is turned on to establish a base line under conditions of gas flow only. When a drift- and noisefree base line is established at high sensitivity settings, the column is connected to the applicator, and the base line is allowed to re-establish itself. This normally does not take more than a few minutes, and a small amount of background and noise may remain at the highest sensitivities, particularly if the eluent contains components of relatively low volatillity, such as oxidation inhibitors or the like. CALIBRATION.In order to obtain the uniform flow rates re!iulting from the resistance of the usual type of column, buit yet to avoid having to wait for a calibration sample to pass thirough the column, a bypass-type sample injector valve is in!rerted between column and detector whenever calibration sa1mples are to be tested. The bypass injector valves (Biotron C
