Chapter 3.36. Ion-chamber detectors

Ion chambers, also known as ionization chambers, have been used as detectors in X-ray absorption spectroscopy for decades. In ion chambers, incident X-rays are absorbed by an atom in a fill gas, leading to a cascade of ionizations which are then collected by placing a transverse potential difference, typically of the order of 10⁴ V m⁻¹, across the chamber. Neglecting energy losses (which are primarily to heat and to fluorescent photons escaping the chamber), the current I produced is given bywhere e is the charge on an electron, N/t is the time rate of absorption of incident X-rays, E_X-ray is the energy of the incident X-ray photon and E_ionization is the average energy of ionization for the gas. Values of E_ionization for common fill gases may be found in Thompson (2009). E_ionization decreases monotonically with increasing atomic number, from 41 eV for helium to 22 eV for xenon. While dry air can in principle be used as a fill gas, ambient air is unsuitable, in part because of its conductive properties (Carlon, 1988).

Ion chambers have the advantage that they can handle high count rates (up to approximately 10¹² counts s⁻¹) over a broad energy range while maintaining good linearity. Traditional ion chambers, however, have relatively slow response times, typically of the order of 100 ms, due to the time necessary for charged ions resulting from the cascades to drift to the collector plate. This combination of high count rates but slow response time make ion chambers a good choice for measuring the intensity of X-rays upstream of the sample (I₀), even when other kinds of detectors are used downstream. For traditional X-ray absorption experiments, in which the dwell time at each measured energy is a second or more, ion chambers are also appropriate for measuring X-ray intensities downstream of the sample.

Ion chambers have also been designed to measure fluorescence (Stern & Heald, 1979). These `fluorescence ionization chambers' generally have the electrodes positioned longitudinally along the photon path, so that fluorescent photons must pass through at least one electrode in order to be measured. This geometry facilitates the use of a large solid angle, increasing the fraction of fluorescent photons captured.

The absorption coefficient of detector fill gases increases rapidly with atomic number and decreases rapidly with the energy of the incident X-ray photons. Total absorption in the ion chamber also depends exponentially on the length of the chamber along the direction of the beam, in accordance with the usual Beer–Lambert law. If the absorption in the chamber is too low, Poisson noise will become an issue.

There are three issues that may arise if the absorption in the chamber is too high.

(i) If there are other aspects of the experiment downstream of the chamber in question, then a high absorption will reduce the number of photons available further down the line. For example, I₀ detectors are upstream of the sample, the transmission detector (if present) and perhaps reference foils and detectors. For this reason, absorption by I₀ detectors is usually chosen to be of the order of 0.1 of the absorption length, leaving approximately 90% of the photons to interact with the sample.

(ii) While ion chambers are capable of handling high count rates, this ability is not unlimited. Above roughly 10¹¹ counts s⁻¹ cm⁻³ (Thompson, 2009), recombination and the effect of space charge (i.e. shielding of the electric field by the charges separated by the cascades) introduce significant nonlinearities. These rates can be achieved by many insertion-device beamlines, sometimes necessitating a choice of fill gas that yields less than maximum absorption.

(iii) Ion chambers typically include guard electrodes at each end to reduce the effect of current leaking through the insulating material that spaces the electrodes. Absorption near the ends of the chamber may therefore fail to be collected. If the absorption in the chamber is too high, then much of the absorption may take place near the front end of the detector. This may create nonlinearities and will also act to reduce the number of counts measured by the detector. For this reason, absorption by detectors downstream of the sample is usually chosen to be approximately two absorption lengths.

To obtain the desired absorption in each chamber, mixtures of gases may be used.

Variations in the gas mixture or pressure (and thus in the temperature) will change the detector response. For EXAFS experiments, in which a smooth background function is subtracted during analysis, this is acceptable as long as the changes are on timescales longer than the time it takes to collect data on any particular feature of the spectrum. For XANES, these changes need to take place on a timescale that is long compared with the data-collection time for an entire spectrum to avoid introducing distortions into the measured spectrum.

In order to assure linearity, care must also be taken with the potential difference applied across the detector plates. If the resulting electric field is too low, ions and atoms may recombine before reaching the collector plates. If the electric field is too high, the additional energy imparted to the charged particles of the cascade could be sufficient to trigger further ionizations, resulting in an energy-dependent, nonlinear amplification of the signal. Between those two extremes, the current generated by a given flux of X-rays should be independent of the potential difference applied, a condition which can be easily confirmed experimentally (Bunker, 2010).

An ion chamber collects the charges produced during the ionization of a gas by an applied electric field, as explained in Section 1. When the electric field is weak, not all of the ions produced can be collected. One reason for this is that the ions diffuse toward the electrodes with the same polarity. The second reason is the loss of ions through initial (columnar) recombination, by which the positive ions that are produced by gas ionization recombine with electrons or negative ions along the same charged particle track. These two effects, however, are negligible for high-intensity X-rays such as synchrotron radiation. In such cases, the dominant effect is general (volume) recombination. This type of ion recombination occurs between different particle tracks, and therefore the probability of recombination is higher for higher intensity X-rays. An important point to be noted is that when there is significant recombination the ionizing current is no longer proportional to the X-ray intensity, i.e. the linearity is lost. To avoid this situation, rapid ion collection is necessary, i.e. a strong electric field must be applied. Initially, when the electric field strength is increased, there is a corresponding increase in the current. Once a specific field strength has been reached, however, there will be no further change in the current even for greater applied voltages, as depicted in Fig. 1. This current saturation is necessary for linearity.

Figure 1 Current saturation curve, in which Is denotes the saturation current.

The synchrotron radiation from a storage ring takes the form of pulsed X-rays with a pulse width of the order of several tens of picoseconds. The pulse interval, however, is also short, of the order of several nanoseconds, when the storage ring is operated in the electron multi-bunch mode. Therefore, these X-rays can be considered as continuous X-rays for ion chambers. The recombination rate near the current saturation point is expressed as (Nariyama, 2006)where I_s is the saturation current (A), I is the measured current (A), e is the electron charge (1.602 × 10⁻¹⁹ C), k₁ and k₂ are the positive and negative ion mobilities (m² s⁻¹ V⁻¹), respectively, α is the ion-recombination coefficient (m³ s⁻¹), D is the length of the collecting plate (m) and E is the collecting electric field strength (V m⁻¹). Thus, the recombination rate is proportional to the saturation current, i.e. the X-ray intensity, and is inversely proportional to the square of the field strength. Fig. 2 shows the relation between E² and I_s/D, with ion losses of 1% and 2%, obtained experimentally using two sizes of free-air ionization chambers (Nariyama, 2006). A linear relation between E² and I_s/D was observed in the low-intensity region. Whenever the probability of recombination is greater than 5%, there will be a significant distortion of the electric field by the space charge (Boag, 1987).

Figure 2 Relation between E2 and Is/D with ion losses of 1% and 2% (Nariyama, 2006). Polynomial fits were used to model the data.

Here, the constant term in equation (2) can be replaced withThe β value of air was obtained experimentally as 3.5–6.6 × 10¹³ V² A⁻¹ m⁻¹. As expected from equation (2), two methods can be used to reduce ion recombination. The first method is to narrow the gap between the electrodes and/or to increase the applied voltage, as previously mentioned. A `micro' ion chamber, in which the gap is narrowed to several millimetres, is the typical solution. By shortening the collection time, more charges can be collected in free air as they are still in the form of free electrons without being attached to the oxygen molecules. This has been experimentally shown to yield a smaller β value, as indicated in Fig. 3.

Figure 3 β values of air as a function of the electric field (Nariyama, 2006).

The second method for reducing ion recombination is to choose a gas that has a small value of α, i.e. a gas with a lower electron affinity, such as nitrogen and the rare gases. Air contains oxygen, which has a large electron affinity. Free electrons have a greater chance of becoming attached to neutral gas atoms, resulting in the creation of negative ions. Therefore, significant recombination occurs between the positive and negative ions. Conversely, for nitrogen and the rare gases, there is a smaller chance of electron attachment, leaving a sufficient number of free electrons that can be collected. The value of α between positive ions and free electrons is several orders of magnitude smaller than that between positive and negative ions (Knoll, 2010). For this reason, it is much easier to saturate an ionization chamber that contains nitrogen or rare gases at a low voltage, and therefore a nitrogen or rare-gas flow is often used in beamlines.

To estimate the recombination rate for X-rays, the saturation current I_s of the ionization chamber has to be predicted. When air is assumed to be the gas in a chamber with collecting electrode length L, the absorbed energy per second in the collection volume is given by (International Commission on Radiation Units and Measurements, 1970)where n is the photon intensity (photons s⁻¹), E₀ is the photon energy (eV), (μ_en/ρ)_air is the mass energy-absorption coefficient of air and ρ_air is the density of air. For a short electrode length or high-energy photons, this reduces to (International Commission on Radiation Units and Measurements, 1970)

Therefore, the saturation current is expressed as (Greening, 1985; Attix, 1986)where W_air is the W value of air (in eV), i.e. the energy required to produce an ion pair. The W value of air is 33.97 eV. Here, ρ_air is obtained fromwhere ρ₀ is 0.001293 g cm⁻³, T is the temperature on the Celsius scale and P is the atmospheric pressure (in Pa) (Attix, 1986). ρ_air is proportional to the atmospheric pressure and depends weakly on temperature.

For example, when 15 keV X-rays of 10¹⁰ photons s⁻¹ are incident on the free-air ionization chamber with a collection electrode length of 3.3 cm at 25°C and 1.01325 × 10⁵ Pa, the saturation current will be 3.7 nA, using (μ_en/ρ)_air = 1.334 cm² g⁻¹ (Hubbell & Seltzer, 1995). Under these conditions, when a voltage of −200 V is applied to the ionization chamber with a gap of 14 mm, the recombination rate will be 1.9%, as determined from equation (2) with β = 3.5 × 10¹³ V² A⁻¹ m⁻¹.

When the synchrotron radiation is discontinued using a chopper, the pulse interval can be greater than milliseconds, which is often longer than the ion-collection time, such as that produced from an X-ray free-electron laser. Under these conditions, the X-rays behave as pulsed X-rays at the ion chambers, i.e. the recombination rate is inversely proportional to the applied voltage, and it becomes more difficult to reduce the rate of recombination even by applying a higher voltage (Nariyama, 2013).

At synchrotron-radiation beamlines, the ionizing current is converted and digitized for management using a personal computer. To cope with the time variability of the current source, for example, the current was integrated using an IVC102 precision integrating amplifier and the output voltage was digitized using an analog-to-digital converter (ADC; Ahmed et al., 2000). Generally, the current is amplified and converted into a voltage using a current amplifier. One method for digitization is to convert the output into a frequency using a voltage-to-frequency converter (VFC; Frahm, 1989; Kocsis & Somogyi, 2003). A counter is used to count the logic pulses of 5 V from the VFC during a constant time interval. This is a type of integrating ADC. For example, when the peak values of the current amplifier and VFC are 10 V and 1 MHz, respectively, an output voltage of 1 V is converted into 100 kHz, i.e. 10⁵ pulses are counted per second. More recently, 100 MHz VFCs have become available (Hino et al., 2013). For faster data acquisition, the output from the current amplifier is directly digitized using an ADC (including a digital multimeter). The use of this method has increased, especially in time-resolved quick X-ray absorption fine-structure (XAFS) systems (Uruga et al., 2007; Müller et al., 2016). A block diagram for these processes is shown in Fig. 4. To improve the signal-to-noise ratio, low-pass filters are used.

Figure 4 Block diagram of a typical measurement system.

An inverting amplifier that uses an operational amplifier is utilized for current amplification. This process is illustrated in Fig. 5. The input current signal is fed to the feedback resistor of an operational amplifier, and the current is converted into a voltage by the feedback resistor and the operational amplifier. The feedback lowers the input impedance, thus making it favourable for current measurement. The rise time is set to a shorter time for the quick XAFS measurement, which increases the current noise.

Figure 5 Current-to-voltage conversion circuit.

The advantages of the ion chamber are its wide dynamic range and excellent linearity. A wide dynamic range of current measurement is necessary for XAFS measurements. This is the reason why ion chambers are extensively used as monitors. Fig. 2 shows the dynamic range of a saturation current covering seven decades in X-ray intensity. The linearity of the overall measuring system is determined by the linearity of the current amplifier, which is approximately 0.1% under normal conditions. The linearity of the VFC is as low as 0.01%.

In order to mitigate the influence of unwanted noise, the signal cable should be as short as possible. To prevent triboelectricity from producing floating charges, unnecessary load to the cable should be avoided. The use of low-noise coaxial cables that have an electrically conductive carbon membrane inserted between the insulation layer and the shield wire is recommended.

The influence of a high voltage applied to the ionization chamber, in relation to a small ionizing current, should also be as small as possible. For this purpose, the electrodes are supported using insulating materials, while the leakage current flows through both humidity and dirt. Guard plates that are grounded are used to prevent the inflow of leakage current to the measuring system.

Placing the incident (I₀) detector as close to the sample as possible minimizes absorption by the intervening medium (typically air, but sometimes other gases, particularly at low energies), but exacerbates another issue: X-rays can fluoresce or scatter off the sample into the detector. Fluorescence is generally of greater concern than elastic scattering, as for concentrated samples the energy dependence of the fluorescence intensity will be similar to the energy dependence of the absorption being measured. Since the measured absorption is computed from ln(I₀/I), this can exaggerate features in the absorption spectrum. The effect is not linear, however, as can be seen by considering the mathematics of self-absorption (see Best & Chantler, 2024). If the sample is optimized for transmission, it will be in neither the thick limit nor the thin limit, and the distortions introduced into the spectrum are a complicated function of composition and geometry.

Contamination of the I₀ measurement by the sample can be reduced by allowing a greater distance between the detector and the sample. Because fluorescence is isotropic, the intensity will fall off as the square of this distance (or faster if absorption by the intervening medium is also included). In addition, the downstream window on the I₀ detector should be large enough to pass the entire collimated X-ray beam, accounting for any tendency of the beam position to drift (for example as the monochromator is scanned), but not much larger.

As an example, a detector with a 2 × 1 cm exit window placed 1 mm from the sample will intercept nearly 50% of the photons fluorescing off the sample. At a distance of 5 cm away, it will intercept less than 1% of the fluorescing photons.

Absorption of the incident beam en route to the sample, on the other hand, follows the Beer–Lambert law, decreasing exponentially with distance. Air at 10 keV, for example, has an absorption length of well over a metre, and thus the incident beam loses less than 5% of its intensity over 5 cm. With soft or tender X-rays, however, the loss due to absorption by air over 5 cm could be quite significant, so the necessity of making sure that the detector is far enough away to subtend a small solid angle as viewed from the sample may prompt the insertion of a tube containing helium into the beam path.

As mentioned previously, X-rays may also scatter off the sample. While scattering processes are not strictly isotropic they are less of a concern, as the energy dependence is generally monotonic and gradual. These scattered photons can still degrade the signal to noise of the measurement, however. In any case, the intensity of scattered photons in the I₀ detector will also be considerably reduced by placing the detector far enough away from the sample to ensure that it subtends a small solid angle.

Fluorescence and scattering can also impact other detectors in the chain. For this reason, the transmission detector should likewise be positioned so that its front window subtends a small solid angle as seen from the sample. If a reference foil is placed downstream of the transmission detector, it is particularly important that it is far enough away so as to subtend a small solid angle as seen from the downstream window of the transmission detector. Otherwise, fluorescence from the reference foil could distort the measured spectrum of the sample.

In cases where geometric considerations do not sufficiently reduce the impact of fluorescence and scattering, techniques used for fluorescence measurements, such as the use of filters or energy-discriminating detectors, can also be contemplated.