The carrier gas has a substantial effect on the electron therrnalization reactions and thus directly determines the probability of other reactions occuring. A gas with a high ionization cross-section is most suitable for optimizing the measuring conditions (see Table 4.2). Nitrogen or argon with the addition of polyatomic gases (CH,, CO,, H,O, C,,HZnt2, etc.) are thus preferable.
When rare gases are used as carrier gases, metastable atomic states are formed in the detector and ionize the eluted substance. When helium is employed, then the gases added to it cause de-excitation of metastable atomic states and are ionized instead. Consequently, metastable states cannot affect the E C D detection mechanism (equation (4.6)) and the linear dynamic range of the measurement is broadened. If argon is used as the carrier gas, the energy of its metastable states (EArm = 11.7 eV [28]) is insufficient for ionization of the added gases and direct ionization of eluted substance by these metastable atoms may occur i n the effective volume of the detector.
It can be assumed that large additions (up to lo;/,) of CH, to argon [54] play a role in direct thermalization reactions of primary electrons, owing to their high ionization cross-section.
Hence it is preferable to add higher hydrocarbons (e.g., C,H,, ZP = 11.1 eV, or C,H,,, I P = 10.6 eV) or generally gases with ionization potentials below 11.7 eV.
These gases have high ionization cross-sections and thus are very effective during electron thermalization; low ionization poentials exclude the presence of metastable atomic states. Simultaneously, the very low electron affinity of hydrocarbons ensures that the free electron concentration in the effective volume of the detector is not lowered and therefore the detection mechanism is favourably affected.
Trace amounts of oxygen and water vapour are also present in the carrier gas containers. It has been found that as little as 10 ppm of H,O in nitrogen, i.e., an amount which cannot be removed by using a molecular sieve, causes about 90%
of the thermal electrons to be bound in ion-molecule species, i.e., (H,O),O;, where n = 0 - 3 depending on the water content of the gas. When an electronegative substance is eluted, competitive reactions of the eluted substance and impurities present in the carrier gas with liberated electrons may occur [25]. It is evident that oxygen and water in the carrier gas decrease the concentration of free electrons and thus also the probability of the formation of a negative ion of the eluted substance.
Therefore, it is necessary to employ carrier gases that are free from oxygen and water vapour and, in general, of all electronegative substances, if the linear dynamic range of the electron capture detector is to be broadened.
When hydrogen is employed as the carrier gas, it strongly affects the emission of tritium from the source [21]. It has been found that inactive hydrogen molecules are exchanged for active tritium molecules adsorbed on the support, the source activity
80
thus being decreased. Therefore, hydrogen cannot be recommended as a carrier gas;
it also has a low ionization cross-section, which leads to ineffective thermalization.
It follows from the discussion of equation (4.9) that the E C D signal depends on the type of eluted substance. As the concentration of eluted substances increases during the elution, the concentration of carrier gas simultaneously decreases and the sum of the molar fractions remains equal to unity. As the ionization cross-sections are pro- portional to the molecular size, the inequality Qs > Qc is almost always valid;
thus, with increasing amounts of eluted substance, direct ionization and recombina- tion will oppose one another, so that the signal may be negative, zero or positive, depending on the conditions. A considerable increase in the linear dynamic range would be achieved for Qs < Qc. Then SECD would decrease and the device would be more sensitive for substances with small molecules [17].
Maggs et al. [41] obtained various shapes for the dependence of the response on the carrier gas flow-rate. Under certain conditions a maximum was reached, but the response generally increased exponentially with the flow-rate. A similar dependence was found by Eisentraut et al. [12]. It can be assumed that this dependence is corre- lated with the magnitude of the effective volume of the detector, as follows from the discussion of the conditions for gas-phase coulometry by thermal electron attachment [38]. For the best results, the carrier gas flow-rale must be optimized. Data on the effect of the carrier gas flow-rate on the ECD background current and signal can be found in the literature [9, 12, 411.
Eflect of temperature on the ?nagnitude of the ECD s i g n a l
The detector temperature affects the number of electrons emitted from the radio- active source, their energy and the electron capture mechanism. It follows from the experimental results that SECD = T3" and that the detector temperature must be maintained within AO.1 "C for accurate quantitative analysis [41].
The detector temperature is limited by the p-source. With tritium, temperatures exceeding 200 "C are not recommended, although a maximum temperature of 325 "C was reported using a scandium support [16, 211 (see Chapter 3). For higher temperatures, 63Ni [65] or "?Pm on a gold foil [39] is suitable; 239Pu [19], 241Am [38] and 137Cs [56] have also been used. These sources can be employed up to 400 "C without any loss in activity.
4.3.2 Construction of the ECD
The electron capture detector is basically an ionization chamber with a p-source providing primary electrons. This source is placed in a suitable manner on the internal cylindrical walls of the chamber, through which the collecting electrodes protrude (Fig. 4.4). p-Source activities vary from 10 to 1000 mCi; in general, an increase in the p-source activity leads to an increase in the sensitivity and to broad- ening of the linear dynamic range, as the electron capture reaction is enhanced by
a higher concentration of thermal electrons. Therefore, the detector sensitivity can be compared only for identical source activities [21,40] (see p. 66).
Study of the effect of the electric field on the ion collection shows that a homogene- ous field is most suitable (p. 62). Bearing this in mind, an ECD has been constructed with the electrodes located in parallel [9, 27, 671, for both d.c. and pulsating voltages.
t
FIG. 4.4. Scheme of the ECD; 1 - anode, 2 - insulation, 3 - 63Ni layer, 4 - heated detector block, cathode, 5 - column holder.
However, these constructions require a relatively large detector volume so that detectors with coaxial electrodes are more frequently employed. Here the detector mantle is usually the cathode and the cylindrical anode occupies a certain part of the chamber volume; the effective volume of the detector is less than 1 cm3. It has been found that the best results with coaxial electrodes are obtained by using a homoge- neous electric field [ 3 5 ] .
The magnitude of the polarization voltage is a critical ECD parameter. The electrons are strongly accelerated in the electric field and hence their energy changes.
A number of detector constructions have been reported in which the ECD detector was converted into a cross-section detector by a change in the position of the elec- trodes or in the voltage on the collecting electrodes. In order to achieve as high an electron capture yield as possible, it is necessary to make measurements at low electric field intensities and, if practical, in a pulse circuit.
In a pulse circuit, the ionization current is measured at a constant frequency with pulse intervals of 5 to 300 psec [l, 151. The pulse width varies from 0.5 to 2.0 p e c with an amplitude of 10 to 60 V. During the pulse, the electrons migrate to the electrode surface and their concentration in the effective volume of the detector decreases. Between pulses, the electron concentration increases; the lower the pulse frequency, the higher is the mean electron concentration and the more probable the electron capture reaction becomes. Fig. 4.5 depicts the variation in the electron
82
concentration, the ECD background current and the ECD signal with varying pulse frequency. It is evident that the higher the pulse frequency, the closer the measurement approaches measurement with d.c. voltage.
A
- bc I
pulse per!od xc]
I I
FIG.4.5.Thedependenceof the concen- tration of free electrons in the effective volume of the detector, of the back- ground current and of the detector response on the pulse interval [ I ] .
FIG.4.6. A - ThemagnitudeoftheECDsignal as a function of varying pulse interval [41];
B - The magnitude of the E C D signal for oxygen with constant-current and constant- frequency circuits.
Recently, increased attention has been paid to measurements with a constant ionization current, where the measured signal corresponds to variations in the pulse frequency [41,60]. This approach is marked by widening of the linear dynamic range of the detector, as the measured signal is directly proportional to the eluted substance concentration. The signals of ECDs using pulse circuits are compared in Fig. 4.6.