One of the main difficulties in trace analysis is masking of the trace zones by the main component. Chemical reaction methods enabling the retention time of the main component (or its derivatives) t o be varied permit, in some instances, this difficulty to be obviated.
Let us first consider methods based on the use of reagents that form with the main component compounds exhibiting low volatility. The method is based on the fact that the main component forms with the reagent a new compound that is virtually involatile under the experimental conditions (temperature, sorbent, etc.), so that the trace com- ponents can be distinctly separated from the main component. The chemical reaction may be carried out at any stage of the chromatographic analysis, from sample injection to detection.
This method was used for the first time by Ray [6] to determine non-olefinic impurities in ethylene. The sample (10-25 ml) was first fed into a reactor (19 x 1.1 cm) filled with activated charcoal saturated with bromine (40%). The resulting liquid bromina- tion products of ethylene were securely retained on charcoal at room temperature. The zone of non-olefmic impurities (permanent and saturated hydrocarbon gases) moved in a flow of carbon dioxide (carrier gas) from the reactor into a chromatographic column (40 x 0.2 cm I.D.) packed with activated charcoal. A nitrometer was used as the detector [39, 401. The method permitted the determination of trace concentrations of l o - ' - lo-*% in ethylene. The use of a more sensitive detector should substantially lower the detection limit.
In a different modification [ 9 ] the method was further elaborated for the deter- mination of hydrocarbon impurities in toluene. Non-polar stationary phases ensure the
r -- -i
25 1
Time (min)
Fig. 8.1. Chromatogram of impurities in toluene on a column containing Apiezon with use of a reactor containing concentrated sulphuric acid (the broken lines show the peaks given by benzene and toluene in ordinary analysis). Peaks: 1 = methane; 2 = n-pentane; 3 = benzene; 4 = n-heptane; 5 = methyl- cyclohexane; 6 = ethylcyclopentane; 7 = methylheptane; 8 = 1,2,4-trimethylcyclopentane; 9 = 2-methylheptane + n-octane; 10 = 4-methylheptane; 11 = 1,l-dimethylcyclohexne; 12 = trans-1,2- dimethylcyclohexane; 13 = trans-l,3-dimethylcyclohexane; 14 = cis-l,2-dimethylcyclohexane; 15 = n-propylcyclohexane; 16 = ethylcyclohexane; 17 = n-nonane; 18 = 1,1,3-trimethylcyclohexane;
19 = cyclooctane. From ref. 9.
satisfactory separation of trace substances but the zone of toluene (main component) masks some of them. For the selective removal of toluene use was made of a reactor (20 cm long) filled with firebrick wetted with concentrated sulphuric acid. Preliminary experiments established that the introduction of a reactor into the chromatographic system does not affect trace separations.
The chemically active sorbent for the reactor was prepared by applying sulphuric acid (involatile liquid) on the support in a fluidized bed [41]. This permits the process to be carried out in a dry atmosphere, which is particularly important in view of the high hydroscopicity of concentrated sulphuric acid.
The use of a reagent that forms a poorly volatile compound with the main component has made it possible to analyse impurities qualitatively and quantitatively in toluene of various grades. Eighteen impurities were identified. Fig. 8.1 is a chromatogram of impurities in toluene obtained using a sulphuric acid reactor.
When the reaction rate is slow, the main component should be absorbed not in a flow but rather under quasi-static conditions in the reactor upstream of the chromato- graphic column. For example, Janik and Novik [42] , in the analysis of trace components in butadiene-l,3, provided for absorption of the main component by maleic anhydride
rKAC1 ANALYSIS BY ANALYTICAL REACTION GAS CHROMATOGRAPHIC METHODS Reacting
compound compounds
Chanpes in the chromatographic characteristics of the sample Changes in the detection characteristics of the sample compounds
Retention timc increases Retention time decreases Detection sensitivity increases Detection sensitivity decreases Main (1) Separation of trace
coin po nen t components and the main component, forming a
poorly volatile compound with the reagent [ 6 , 9 ] concentration [ 10-121 components and the main components yielding a compound which is retained to a greater extent than the trace components [ 1 3 ] Trace (4) Concentration of trace component components by using
(2) Frontal chemical (3) Separation of trace
chemical absorbers forming a poorly volatile compound with the trace component [ 14,151 (5) Separation of the main
component and a trace component yielding a compound which is retained to a greater extent than the main component [ 161
(6) Separation of trace components and the main component, forming a highly volatile compound
1171
(7) Separation of the trace and main components by converting the trace components to a volatile compound, using a liquid bubbler [ 18, 191 and a tubular reactor [ 201 (8) Separation of the main and
trace components with conversion formation of the latter into a compound which is retained to a greater extent than the main component [21]
(16) Detection of trace components against the background of the main component yielding a non-detectable compound [ 371
(9) Transformation of non- detectable trace components into compounds detectable by highly sensitive detectors, through one-step 1221 and two-step [23] conversion of carbon-free compounds amplification method [24]
(10) Using the chemical
Carrier
gas of the trace and main
component zones as a result of chemical binding of part of the carrier gas [ 2 5 , 2 6 ] carrier gas with a reactive component reacting with the sample compounds which are detected as the decrease of the concentration of the reactive component in the carrier gas [27]
(1 3) Introduction into the carrier gas of a complexing reagent to give a volatile component
[ 2 8 ] , to increase its stability and to minimize adsorption (12) Using a two-component
[29-311
(14) Introduction into the carrier gas of a more reactive compound for reactive protection of the sample compounds against moisture [321
Liquid stationary phase
(15) Introduction into the stationary phase of compounds suppressing the reactivity of the sample compounds [ 331 and reactively protecting them [ 341. Using, as the stationary phase, compounds reacting with the sample compounds to yield volatile
N
products [ 35,361 W VI
in the reactor. The sample (50-1OOml) was fed at 100-1 10°C for 30-60 sec into the reactor filled with Kieselguhr on the surface of which was applied maleic anhydride (3%) with small amounts of benzidine (2.5%). The non-reacting trace components were then separated on a column of Alusil (sodium aluminosilicate) modified with 20% of dimethylformamide. This technique made it possible to determine the contents of the following trace components in butadiene-1,3 : ethylene, propane, propylene, isobutane, n-butane, n-butene, isobutene and trans- and cis-butene-2. Although only trace amounts down t o 2 . could be determined, the sensitivity of the method can be considerably improved b y using a more sensitive detector.
In order to determine stable trace components, the main reactive component should preferably be retained in a selective manner. For example, to determine hydrocarbons in arsine the latter is combined with copper in a reactor at 130"C, and the trace com- ponents are separated on a chromatographic column with subsequent flame-ionization detection [43].
To absorb water, Jacobs [44] used a copper column (40 x 0.6 cm I.D.) packed with phosphoric anhydride on firebrick. A single charge (3.7g of phosphoric anhydride) is sufficient t o absorb water contained in fifty 10-1.11 samples of aqueous solutions.
To prevent the trace zone from being overlapped by that of the solvent, the latter should be chemically absorbed in the reactor column. For example, acetic acid (the solvent used as the sorbent in the n i e t h d of concentration of trace components from air) is removed by introducing into the column up to 20% of sodium hydroxide and
10-15% of polyethylene glycol [45].
In the analysis of reactive compounds, common sorbents that selectively absorb the main component can be used as reagents. Such an approach was developed by Turkeltaub et al. [46] for the analysis of trace components 1 ~ 1 boron trichloride, which was adequately chemisorbed on a column of dinonyl phthalate.
To determine trace components in arsenic trichloride the inlet portion of the column was filed with 15% of tricresyl phosphate on Chromosorb W [47]. In this portion arsenic trichloride was completely retained as a result of the formation of a stable complex with tricrysyl phosphate. The tract components were separated in the remainder (main portion) of the column packed with 1% of polyethylene glycol 400 (PEG-400).
Trace amounts of chloro derivatives of hydrocarbons in boron trichloride were determined using a thermionic detector after boron trichloride had been combined with diphenylamine applied in an amount of 20% on INZdOO [48].
Hydrogen bonding between the main component and the stationary phase may also be regarded as a particular case of chemical reaction. Such methods are advantageous with the possibility of rapid regeneration of the absorber by heating [49] or back- flushing [50]. In the former instance [49] hydrogen bonding was used for the deter- mination of trace components. Higher alcohols were selectively retained by the column, whereas the nonalcoholic compounds were rapidly eluted from the column in the carrier gas flow.
The reactor can be substantially reduced in size and an appropriate stationary phase can be used for trace separations (including a stationary phase on which the retention of the main component is not selective) if the reactor (filled with a sorbent selective for the main component) is arranged in series with the chromatographic column only
255 for a short period of time necessary for separation of the trace components from the main component and the flow in the reactor is quickly reversed without stopping the separation of trace components in the column. The use of back-flushing and semi-back- flushing methods, which have been described in detail elsewhere [ 5 0 ] , is also often recommended for trace analysis.
Swoboda [51] applied the semi-back-flushing method t o the analysis of low- molecular-weight alcohols in aqueous solutions. After sample injection the gas flow passes successively through a reactor containing diglycerol(20%) then a chromatographic column containing PEG400 (10%). After the water zone is in the preliminary column (reactor) and the trace components of interest have reached the main column, the gas flows are switched over so that the flow through the reactor is reversed and that through the chromatographic column continues in the same direction.
A limitation of the above simple and effective method of separating trace components from the main component is the impossibility of using it for the analysis of trace com- ponents with the same chemical nature and reactivity as the main substance.
to lo-*% suffers from certain difficulties even if highly sensitive detectors are used. For the analysis of some highly pure compounds one must often pre-concentrate the impurities.
An effective means of concentrating light trace components is the frontal method [52, 531. A modification of this method is frontal chemical concentration, in which a chemical reagent is used instead of the sorbent. The reagent forms with the main component a complex that is completely retained in the reactor. Compared with the method in which common sorbents are used, frontal chemical concentration offers the following potential advantages:
(1) the possibility of analysing all non-reacting trace components because, owing to the involatility of the derivative from the main component, virtually all trace com- ponents become ‘light’ and are concentrated at the front of the reagent interacting with the main component;
(2) usually a high capacity of the sorbent reagent, and hence a high degree of con- centration at the same size of the concentration column;
(3) a pronounced temperature dependence of the chemical equilibrium constant, which in some instances permits the main component to be desorbed at relatively low temperatures under conditions of thermal stability for most compounds.
Frontal chemical concentration was used in previous work [lo] during the analysis of trace hydrocarbons in carbon dioxide. The reagent was diethanolamine, which reversibly absorbs carbon dioxide (main substance) without retaining the trace components.
The concentration column (stainless steel, 400 x 0.4cm ID.) was packed with fire- brick plus 30% of diethanolamine. The column was introduced into the chromatographic system, being incorporated into the six-way gas sample valve [54] instead of a calibrated tube of a particular volume. For the separation of trace hydrocarbons the chromato- graphic column (120 x 0.2 cm I.D.) was packed with silica gel ASK modified with 1.5%
of vaseline oil. A carbon dioxide sample (25-250 ml) was injected into the concentration column through one of the ports of the six-way valve. In the column the carbon dioxide was absorbed as a result of a reaction with diethanolamine [55]. The non-reacting trace components formed a concentrated zone at the carbon dioxide front.
The direct GC determination of trace amounts from
REACTION METHODS OF TRACE ANALYSIS
With the aid of the carrier gas flow the trace zone was transferred into the analytical column. Under such conditions carbon dioxide is completely retained by diethanolamine.
The concentration and analysis were performed at room temperature. The separated trace components were sensed by a flame-ionization detector. The carrier gas (nitrogen) flow-rate was 60ml/min. For regeneration the concentration column was heated in the carrier gas flow at 100-105°C for 5-7min. For heating and cooling a special device forming part of a KhT-2M chromatograph manufactured in the U.S.S.R. was used. The duration of the total analytical cycle was 25 min. This method permits the determination of hydrocarbon gases in carbon dioxide at concentrations down t o
Note that direct elution analysis of carbon dioxide even in large samples (25ml) fails to determine trace hydrocarbons without concentration. The method may be recommended for the trace analysis of other acid gases (e.g., hydrogen sulphide, hydro- cyanic acid). A chemical concentration method based on removal of unsaturated gaseous hydrocarbons (main component) for determination of trace amounts of hydrogen and carbon dioxide was developed by Alekseeva and co-workers 156, 571. For absorption of olefms use was made of a column containing a solution of silver nitrate and mercury salts.
The extension of the range of application of analytical reaction gas chromatography is due, in particular, to the wider range of chemical reactions investigated. In previous work [I 1, 121 involving frontal chemical concentration in gas chromatography, chemical absorption was used [ 5 8 ] with a view to determining volatile carbonaceous impurities in hydrogen. For preliminary concentration of the trace components in hydrogen use was made of frontal chemisorptive concentration on palladium black, which strongly chemisorbs hydrogen at room temperature (one volume of palladium black absorbs up to 200 volumes of hydrogen). Therefore, when this sorbent is employed, all trace com- ponents (from carbon dioxide t o butane) may be regarded as ‘light’ with respect to hydrogen. It should be noted that in the adsorptive concentration of trace components from hydrogen the most difficult task is to select an appropriate sorbent that acts as an adsorbent at room temperature and, at higher temperatures, as a desorbent of com- pounds with widely differing boiling points (e.g., carbon dioxide and butene).
A hydrogen sample was injected into a concentration column packed with palladium black on asbestos (column size, 80 x 1 cm I.D.; hydrogen capacity, 1.5 1). The con- centration column was flushed in advance with an inert gas at elevated temperatures in order to desorb the hydrogen. In the concentration column, at the front of hydrogen absorption by palladium black a concentrated trace zone was formed which was fed into a chromatographic column (250 x 0.4cm I.D.) after all of the sorbent in the concentration column had been used up. Depending on the purpose of analysis, various sorbents were used in the chromatographic column (molecular sieves for separation of permanent gases, alumina modified with 3% of sodium hydroxide for separation of trace hydrocarbons, etc.). During analysis of trace hydrocarbons the separated com- ponents were detected with a flame-ionization detector, and during analysis of permanent gases by a thermal conductivity cell. The carrier gas was the hydrogen t o be andysed, whose flow-rate prior t o passage of the trace zone through the concentration column was reduced to 50ml/min. Fig. 8.2 shows a chromatogram obtained by the analysis of a mixture without concentration and that of the same mixture after concentration of trace
251
A :E +, u 0 a, 0 a In a, b U a, a, c L
I I
10 5
Time (rnin)
Fig. 8.2. Chromatogram of impurities of hydrocarbons (4 . lo-’%) in hydrogen, (A) without enrich- ment and (B) with enrichment on palladium black. Peaks: (A) 1 = methane; 2 = ethane; 3 = ethylene;
4 = propane; (B) 1 = methane; 2 = ethane; 3 = propane; 4 = butane. From ref. 12.
components. Note that concentration is accompanied by quantitative hydrogenation of unsaturated compounds to the corresponding hydrocarbons, whereas carbon oxides are hydrogenated to methane. The concentration of the component in the initial mixture was 4 1 lo-’%, and the carbon oxide detection limit according to this method is
The concentration column was regenerated by flushing it first with air then with an inert gas with simultaneous heating to 100°C. When the concentration column was flushed with air, it was also water-cooled. The total analytical cycle time was 30 min.
To concentrate trace components one can also use non-regeneratable sorbents. In t h s instance a small reactor is placed in the gas flow upstream of the chromatographic column. In determining trace amounts of permanent gases in carbon dioxide, Timms et al. [59] used a reactor filled with wet sodium hydroxide.
The frontal chemical concentration method can also be used for the intermediate concentration trace amounts of heavy components [60] in the analysis of gaseous monomers (ethylene, propylene). The concentration was conducted on a short inter- mediate column containing diethanolamine and pure carbon dioxide was used as the carrier gas. The method permits three steps t o be integrated into a single run: preliminary separation, concentration of heavy trace components and analytical determination of the composition of the concentrate. The concentration of the heavy trace components to be determined was Trace analytical methods based on selective retention of the main component are becoming more common in chromatography.
The separation of trace and main components may be achieved by making use of the formation of a volatile derivative of the main component, which is retained to a greater extent than the trace component. The separation of argon and oxygen is more difficult because of their similar adsorption properties. Most of the separation techniques that have been described are complicated [61], for example, separation of argon and oxygen on a 10-m column of molecular sieve 5A at room temperature [62] or separation on a 1.8-m column at - 72°C [ 6 3 ] . Swinnerton et al. [ 131 proposed an elegant method for determining trace amounts of argon in permanent gases. A sample was passed in a
REACTION METHODS OF TRACE ANALYSlS
flow of hydrogen through a reactor [25] filled with a palladium catalyst. The oxygen in the sample was reduced to water, which was separated at 55OC from the non-reacting gases (argon and nitrogen) on a column (40cm) containing 5% of Triton X on a PTFE support. Downstream of the column the first cell of the thermal conductivity detector detected two peaks: a common nitrogen and argon peak and a water peak whose area was proportional to the oxygen concentration in the sample. Then the separated components passed in the carrier gas flow via a drying tube in which water was retained and into a column of molecular sieves 5A, on which the common peak was separated into argon and oxygen. The products of separation were detected by the second thermal conductivity detector cell. The detection limit was lo-' mol. The method is applicable to the deter- mination of dissolved gases in liquids.
Let us now consider methods based on the longer retention of trace components or their derivatives.
Preliminary concentration is often a necessary step in the GC analysis of trace com- ponents, particularly inorganic compounds, for which highly sensitive and reliable detectors have not yet found wide application.
The widely used adsorptive trace concentration method (see, for example, Pomazanov and Nesterov [33]) is usually limited t o heavy trace components; in addition, in desorption one should expect thermal degradation of trace components (desorption is normally conducted at high temperatures) and their incomplete elution from the sorbent.
Therefore, a promising means of concentrating trace components is the chemical concentration method [14, 151. Such methods are based on the selective absorption of trace components from the sample as a result of the formation of involatile or poorly volatile compounds with subsequent isolation of the starting compounds (or their con- version products) by heating or exposing them to different reagents.
Their prototype is believed t o be the method of determining water in butane [64], based on the selective absorption of water in a reactor of polyethylene glycol as a result of hydrogen bonding. A gas sample was passed at 10°C through a trap (30.5 x 0.63 cm) filled with firebrick and PEG-200 (30%). Butane passed through the trap without being retained on PEG. After the water had been concentrated, the trap was heated to 90"C, connected to a gas chromatographic system and the water was desorbed in a helium flow into the chromatographic column (30% of PEG-200 on firebrick; 61 cm long).
In this column the water was separated from other possible trace components (methyl mercaptan, benzene, etc.). The detection limit was 2 . lo-'% for a sample size of 10 1.
A similar technique for the determination of water in isopentane was developed by Baranova et al. [65]. Water was absorbed in a trap containing triethylene glycol at 25"C, then desorbed at 100°C. The minimum detectable water concentration was I .
When large samples are used in isothermal analysis using PEG, water forms a chromato- graphic zone in the form of a step, which simplifies calibration and improves the accuracy
[66j. This method permits the determination of water at concentrations of 0.2-1.2%
(w/w) using a thermal conductivity cell. The application of reaction methods for the conversion of water in organic compounds using a flame-ionization detector or selective electrochemical detectors that are highly sensitive to water will undoubtedly enable the detection limit to be lowered.