Electrometer offset current due to scattered radiation

Abstract Relative dose measurements with small ionization chambers in combination with an electrometer placed in the treatment room (“internal electrometer”) show a large dependence on the polarity used. While this was observed previously for percent depth dose curves (PDDs), the effect has not been understood or preventable. To investigate the polarity dependence of internal electrometers used in conjunction with a small‐volume ionization chamber, we placed an internal electrometer at a distance of 1 m from the isocenter and exposed it to different amounts of scattered radiation by varying the field size. We identified irradiation of the electrometer to cause a current of approximately −1 pA, regardless of the sign of the biasing voltage. For low‐sensitivity detectors, such a current noticeably distorts relative dose measurements. To demonstrate how the current systematically changes PDDs, we collected measurements with nine ionization chambers of different volumes. As the chamber volume decreased, signal ratios at 20 and 10 cm depth (M20/M10) became smaller for positive bias voltage and larger for negative bias voltage. At the size of the iba CC04 (40 mm³) the difference of M20/M10 was around 1% and for the smallest studied chamber, the iba CC003 chamber (3 mm³), around 7% for a 10 × 10 cm² field. When the electrometer was moved further from the source or shielded, the additional current decreased. Consequently, PDDs at both polarities were brought into alignment at depth even for the 3 mm³ ionization chamber. The apparent polarity effect on PDDs and lateral beam profiles was reduced considerably by shielding the electrometer. Due to normalization the effect on output values was low. When measurements with a low‐sensitivity probe are carried out in conjunction with an internal electrometer, we recommend careful monitoring of the particular setup by testing both polarities, and if deemed necessary, we suggest shielding the electrometer.

investigate the polarity dependence of internal electrometers used in conjunction with a small-volume ionization chamber, we placed an internal electrometer at a distance of 1 m from the isocenter and exposed it to different amounts of scattered radiation by varying the field size. We identified irradiation of the electrometer to cause a current of approximately −1 pA, regardless of the sign of the biasing voltage. For low-sensitivity detectors, such a current noticeably distorts relative dose measurements. To demonstrate how the current systematically changes PDDs, we collected measurements with nine ionization chambers of different volumes. As the chamber volume decreased, signal ratios at 20 and 10 cm depth (M20/M10) became smaller for positive bias voltage and larger for negative bias voltage. At the size of the iba CC04 (40 mm³) the difference of M20/M10 was around 1% and for the smallest studied chamber, the iba CC003 chamber (3 mm³), around 7% for a 10 × 10 cm² field. When the electrometer was moved further from the source or shielded, the additional current decreased. Consequently, PDDs at both polarities were brought into alignment at depth even for the 3 mm³ ionization chamber. The apparent polarity effect on PDDs and lateral beam profiles was reduced considerably by shielding the electrometer. Due to normalization the effect on output values was low. When measurements with a low-sensitivity probe are carried out in conjunction with an internal electrometer, we recommend careful monitoring of the particular setup by testing both polarities, and if deemed necessary, we suggest shielding the electrometer.

| INTRODUCTION
Many different detectors are available for small field dosimetry, for example, for output factor measurements or beam profile and percent depth dose curves (PDD) acquisitions. Ionization chambers are frequently used for such measurements in radiation therapy. Since larger active detector volumes can lead to volume averaging effects, ionization chambers have recently been produced with an active volume as small as 3 mm³. 1 The sensitivity of such a detector drops with its volume, thereby resulting in a worse signal-to-noise ratio for chambers with smaller active volume. In addition, small ionization chambers are susceptible to effects that have not been observed for larger chambers, including effects on measured PDDs. Discrepancies in measurements of PDDs with small-volume ionization chambers have been observed by Sarkar et al. 2 They noticed PDD discrepancies at different polarities when measuring with an electrometer placed in the treatment room, which were not present with an electrometer placed outside of the treatment room. Although they ruled out a number of influence parameters, it remained unclear where exactly the interference was originating.
Polarity effects manifest themselves as different readings between a positive and a negative bias voltage. For micro-ionization chambers, this effect is voltage dependent and increases as their volumes decrease due to higher relative changes in the collecting volume. 3 TRS 483 demands a polarity effect of less than 0.4% of the chamber reading as a specification for reference class ionization chambers for absolute dosimetry. 4 The same maximum polarity correction of 0.4% from unity at any energy is recommended in the addendum to the TG-51 protocol. 5 Smaller chambers often do not meet this criterion; for example, according to Hyun et al. the mean polarity correction factor averaged over several beam qualities at 5 cm depth in a 10 × 10 cm² field for a CC01 (10 mm³) micro-chamber was 1.011. 6 While certain chambers have shown depth-dependent polarity effects, 7-10 the anomalous PDD behavior observed by Sarkar et al. for small-volume chambers vanished when using an external electrometer, suggesting it was likely not caused by the chamber or the polarity effect.
We encountered similar problems to Sarkar et al. when scanning PDDs at either a positive or negative bias with the electrometer placed inside the treatment room. This type of electrometer setup will be referred to as an "internal electrometer"; while an electrometer placed outside of the treatment room will be referred to as an "external electrometer". We investigated the cause of this effect and present ways to mitigate and correct it to increase the accuracy of PDDs, profiles, and output factors obtained using small-volume chambers.

2.A | Radiation effects on an internal electrometer
To evaluate the origin of the observed deviations, two ionization chambers were connected to a Tandem internal electrometer (PTW, Germany). A Semiflex 31013 chamber (PTW, Germany, 0.3 cm³) was connected to the field channel while a Famer type 30013 (PTW, Germany, 0.6 cm³) was connected to the reference channel.
Using two 8 m cables, the chambers were positioned just outside of the treatment room with the doors closed. The electrometer was positioned on the treatment couch at a certain electrometerto-isocenter distance EID (Fig. 1). In addition, a second Semiflex 31013 was placed next to the other chambers outside the treatment room door and connected to a Unidos external electrometer (PTW, Germany). A third Semiflex 31013 was inserted into a plastic cube, put on top of the internal electrometer and connected to a Unidos E electrometer (PTW, Germany) outside the treatment room. To generate realistic scatter conditions, a MP3-XS phantom (PTW, Germany) was filled with water and placed as if measurements at a SSD of 100 cm were carried out. The MP3-XS has a volume of 34 × 34 × 42 cm³.
A 6 MV beam from a Primus linac (Siemens, Germany) was used for the measurements. The linac operates at a repetition rate of 250 MU/min (±1%), and is calibrated to deliver 1 cGy/MU at the isocenter at a depth of 5 cm in water. All four chamber signals were zeroed while the beam was off. For the following measurements, the current was integrated over 30 s for all chambers nearly simultaneously. The integrated value was read directly from the Unidos electrometers' display and obtained from the water tank software (Mephysto mc² TanSoft, PTW, Germany, version 1.4) for the internal electrometer. Measurements were repeated 3 to 5 times. A bias voltage of 400 V was used for all chambers. All measurements were repeated with a negative bias voltage applied to the internal electrometer. After the bias voltage was changed, the current was stabilized before the measurements were continued. F I G . 1. Measurement setup. Two ionization chambers were placed outside of the treatment room (Point 1 external) and connected to an internal electrometer placed at a distance EID from the isocenter (Field Channel, Reference Channel). A third detector was placed next to the first two detectors (Point 1 external) and connected to an external electrometer outside the treatment room. A fourth detector was placed on top of the internal electrometer (Point 2 internal) and connected to an external electrometer. A water-filled phantom was placed at the isocenter. In some experiments, lead bricks were placed between the electrometer and the phantom or the internal electrometer was rotated by 180°.
Measurements at a constant electrometer to isocenter distance EID = 100 cm were carried out for different nominal field sizes from 1 × 1 to 20 × 20 cm², as well as for beam off, using both polarities.
Measurements were then carried out for a 10 × 10 cm² field at the EID of 100 and 150 cm, and additionally for the latter with 10 cm of lead as a shielding between the electrometer and the water phantom. Note that the cables were not shielded explicitly. At EID = 150 cm, the phantom is positioned at the head end of the table and the electrometer at the feet end to achieve the largest distance that still fits the electrometer on the treatment couch. Finally, at an EID = 100 cm and a field size of 10 × 10 cm², the effects of exchanging the chambers between the reference and the field channel as well as rotating the electrometer by 180 degrees were tested. To rule out the malfunction of the used individual device, measurements for this configuration were also repeated with a second Tandem electrometer at both polarities. To quantitate the effect of the offset current on relative dose measurements, the depth dose, lateral profile, and relative output factor measurements were performed at different internal electrometer positions. All of the following scanning measurements were carried out using the small MP3-XS phantom and the same electrometer and software described previously in this section. For the smallest chamber, the CC003, depth dose curves were measured at EID = 100 cm and EID = 150 cm for both polarities. In addition, at EID = 150 cm the Tandem electrometer was shielded. For comparison, a conventional scanning chamber, the CC13, was used in the setup with the shielded electrometer.

2.D | Output factor measurements
Output factor measurements were taken for both polarities at an SSD 90 at 20 cm depth with the CC003 chamber for nominal square field sizes of 10, 6, 4, 2, and 1 cm, 100 MU, and three readings per field size using the TanSoft software. Directly afterwards, a Semiflex 31013 ionization chamber was placed outside the treatment room and connected to the Tandem electrometer field channel inside to quantitate the effect of scattered radiation on the electrometer in exactly the same setup for all field sizes used. Measurements were repeated three times with 100 MU. The measured charges obtained were then corrected for the electrometer offset determined individually for each field size. As the charge induced by electrometer irradiation was always negative, its absolute value was added to correct the signed electrometer readings obtained with TanSoft. Output factor measurements were also carried out at SSD 90 and 10 cm depth   larger the difference between the two polarities used. The mean between positive and negative bias was closer to the value obtained with the larger chambers than each individual polarity. For the smallest CC003 chamber, the M20/M10 value was 7% lower for the positive than for the negative bias voltage. With the added shielding, the difference decreased to 0.3%.   tbaScan only reports absolute values, the curve appears to increase rather than decrease at approximately 80 mm and further from the field center.
The results of output factor measurements are shown in Fig. 6. The measured current for 100 MU increased in all fields and for both polarities when the shielding was added [ Fig. 6(a)].
The corrected values, that is, the unshielded measurements corrected by the measured electrometer offset, were slightly above the shielded ones, indicating that the shielding had a large effect but did not entirely protect the electrometer from irradiation.
Small systematic deviations between measurements with an internal and an external electrometer were also visible for the normal-

| DISCUSSION
An electrometer inside the room was shown to produce a current that was reduced with shielding and distance from isocenter, indicating the current is due to stray radiation interacting within the electrometer (Fig. 2). Approximately, the same current was measured with a Farmer type and a thimble type ionization chamber, suggesting that the signal is independent of the detector connected to the electrometer.
With decreasing detector size, depth dose curves measured with those detectors deviated further from curves obtained with planeparallel chambers or large ionization chambers, which are assumed to produce the correct curve (Fig. 4). What seems like a polarity problem at first sight can be explained by the currents induced by scattered radiation hitting the electrometer: When this negative background current is added to the signal of the detector, the behavior is dependent on the sign of the measurement signal. For positive polarity, the measurement signal is positive, so the constant current is always subtracted. When one normalizes the measurements, one gets a steeper depth dose curve than without the extra current. For negative polarity, the measurement signal is negative, so the additional background current always increases the absolute signal. Consequently, normalization will yield a shallower depth dose curve. This is exactly what is seen in Fig. 4. It also explains the observation that the apparent polarity effect is reduced when the electrometer is shielded by lead (Figs. 5 and 6).
For output factor measurements, the situation is a little more complex. The dose to the electrometer is not fixed, but also shows a field size dependence. As a consequence, the effect is partly mitigated when normalizing to a reference field. Nevertheless, systematic deviations are introduced into the measurements, for example, in the form of different apparent polarity correction factors k Pol .
There are differences between the signal ratios with internal and external electrometers [ Fig. 6 30 MV beam. 9 They found a depth-dependent effect that is small except close to the surface for the studied Exradin A16 chamber (7 mm³) and suggested electron contamination to be the cause. For electron beams, polarity effects seem to depend on a variety of parameters such as the cable, stem length, energy and field size. 12 This suggests that the interplay between polarity, detector construction details and electron contamination near the surface needs to be further analyzed.

| CONCLUSION S
It was shown how an internal electrometer's response to scattered radiation influences relative dose measurements. A current induced by radiation reaching the electrometer results in distorted relative dose ratios becoming apparent when measuring with both polarities.
The effect increases when the detector signal decreases and becomes visible when the measurement signal is of an order comparable to the electrometer offset current. In a typical measurement setup in a 10 × 10 cm² field, M20/M10 values measured with a CC003 chamber deviated more than 7% from the value obtained with larger chambers.
To prevent the introduction of erroneous data into the treatment planning system, special care must be taken when small ionization chambers are used in combination with an internal electrometer. We recommend to always test at both polarities and to compare some results to point measurements carried out with an external electrometer or to quantitate the effect of scattered radiation on the electrometer for the specific setup used. To reduce the influence of scattered radiation on the electrometer, shielding the electrometer with lead may be a sufficient, practical solution in many cases.

ACKNOWLEDG MENTS
The Publishing.

CONFLI CTS OF INTEREST
There are no conflicts of interest.