Commissioning of a PTW 34070 large‐area plane‐parallel ionization chamber for small field megavoltage photon dosimetry

Abstract Purpose This study investigates a large‐area plane‐parallel ionization chamber (LAC) for measurements of dose‐area product in water (DAP w) in megavoltage (MV) photon fields. Methods Uniformity of electrode separation of the LAC (PTW34070 Bragg Peak Chamber, sensitive volume diameter: 8.16 cm) was measured using high‐resolution microCT. Signal dependence on angle α of beam incidence for square 6 MV fields of side length s = 20 cm and 1 cm was measured in air. Polarity and recombination effects were characterized in 6, 10, and 18 MV photons fields. To assess the lateral setup tolerance, scanned LAC profiles of a 1 × 1 cm2 field were acquired. A 6 MV calibration coefficient, ND ,w, LAC, was determined in a field collimated by a 5 cm diameter stereotactic cone with known DAP w. Additional calibrations in 10 × 10 cm2 fields at 6, 10, and 18 MV were performed. Results Electrode separation is uniform and agrees with specifications. Volume‐averaging leads to a signal increase proportional to ~1/cos(α) in small fields. Correction factors for polarity and recombination range between 0.9986 to 0.9996 and 1.0007 to 1.0024, respectively. Off‐axis displacement by up to 0.5 cm did not change the measured signal in a 1 × 1 cm2 field. ND ,w, LAC was 163.7 mGy cm−2 nC −1 and differs by +3.0% from the coefficient derived in the 10 × 10 cm2 6 MV field. Response in 10 and 18 MV fields increased by 1.0% and 2.7% compared to 6 MV. Conclusions The LAC requires only small correction factors for DAP w measurements and shows little energy dependence. Lateral setup errors of 0.5 cm are tolerated in 1 × 1 cm2 fields, but beam incidence must be kept as close to normal as possible. Calibration in 10 × 10 fields is not recommended because of the LAC's over‐response. The accuracy of relative point‐dose measurements in the field's periphery is an important limiting factor for the accuracy of DAP w measurements.


| INTRODUCTION
The dosimetry of small megavoltage (MV) photon fields is difficult.
The physical and theoretical challenges are outlined in several recent articles. [1][2][3] These challenges have been recognized by major medical physics organizations and an IAEA and AAPM joint task group (TG 155) has been formed to provide guidance for medical physicists with respect to small field dosimetry. Meanwhile, a draft of the German Industry Standard DIN 6809-8, detailing procedures for small field dosimetry, has become available for public comment. 4 Overall, there is a considerable amount of research activity investigating the measurement and modelling of small photon fields, with two recent publications discussing those issues in detail. 5,6 Radiotherapy treatment planning systems require measured beam data, such as output ratios (ORs), percentage depth dose curves (PDDs), and tissue-phantom-ratios (TPRs). It is challenging to measure these data accurately for small photon fields, because of uncertainties in detector alignment with the field's axis as well as volumeaveraging effects as the field size approaches detector size. Lateral electronic disequilibrium and changes in energy spectrum of the beam with subsequent changes in detector response need also be considered.

1.A | Dose-area-product and large area chambers
An alternative to central-axis point-dose measurements is the Dose-Area-Product (DAP). Detectors with sensitive volumes many times larger than the beam have the advantage of integrating dose with high precision independently of uncertainties in lateral alignment with the field axis. In practice, a Large-Area plane-parallel Chamber (LAC) has a response proportional to the absorbed Dose-Area-Product in water (DAP w ). The accurate measurement of DAP w allows a precise determination of on-axis dose output of small photon fields, provided that the relative two-dimensional dose distribution is accurately known from measurements with radiosensitive film or scanning detectors.
Several publications describe the use of a LAC, the PTW Bragg Peak chamber type 34070-2.5 (PTW Freiburg, Germany), for measuring integral dose in small megavoltage photon fields. 7-10 Djouguela et al first described remarkable properties of DAP w measurements of small fields, such as the similarity of depth-dose curves at constant source-to-detector distance (DAP w -TPR) and constant source-to-surface distance (DAP w -PDD) and the invariance of those curves with field size. 7 Sanchez-Doblado et al derived OR of small fields by combining measurements of DAP w and two-dimensional dose distributions measured with radiochromic film, and found good agreement with diode measurements and Monte Carlo calculations. 8 The Bragg Peak chamber has even found use as an upstream in-line beam monitor by Lechner et al; 9 and Heidorn et al have introduced the LAC as a quality assurance device in order to verify the constancy of the beam area for a Cyberknife Iris Variable Aperture Collimator. 10 Since the PTW type 34070-2.5 chamber has shown some potential useful applications for small field dosimetry measurements, 7,8 it was chosen as one of the detectors to measure DAP w at the Australian Radiation Protection and Nuclear Safety Agency's (ARPANSA) linear accelerator, which is the reference source for dosimetric calibration and audit services provided by ARPANSA.
DAP w is also a quantity of interest investigated by the MedExtRT project (http://radiotherapy-emrp.eu/), which aims to improve the dosimetry for small and composite MV photon beams. In this context, DAP w has been investigated in the setting of various other primary standard laboratories, such as the National Physics Laboratory (NPL), UK, the Laboratoire Nationale Henry Becquerel in Paris, France, and the Istituto Nazionale di Metrologia delle Radiazioni Ionizzanti in Rome, Italy. 11-18

1.B | Requirements for a suitable DAP chamber
We investigated whether the PTW type 34070-2.5 chamber has the performance characteristics desirable of a reference DAP w detector, by measuring signal reproducibility, linearity, long-term stability, effects of recombination and polarity, angular sensitivity, and extracameral effect. We used a microCT to determine whether the distance between the LAC's electrodes is uniform, and we measured the energy-dependency of response in 6, 10, and 18 MV beams.
We calibrated the chamber in terms of DAP w by cross-calibration in a 6 MV photon field with known DAP w , and compared this to published values. We performed this calibration in fields that were either smaller or much larger than the LAC.

2.A | LAC performance as a reference detector
The Bragg Peak chamber Type 34070-2.5 (PTW Freiburg, Freiburg, Germany) is a waterproof, large-area plane-parallel ionization chamber (Fig. 1). The cylindrical lacquered PMMA body has an outer diameter of 103.95 mm and a height of 12.95 mm. The internal cylindrical air cavity is nominally 2.0 mm thick with a diameter of 84.0 mm. According to specifications, the entrance window has a physical thickness of 3.47 mm, comprised of 0.1 mm outer lacquer layer, 3.35 mm PMMA and the 0.02 mm graphite electrode, which in total corresponds to a water-equivalent thickness of 0.4 g/cm 2 .
The graphite collecting electrode is distal to the entrance window and has a diameter of 81.6 mm, hence the volume of air over which ionization is collected is 10.5 cm 3 and its sensitive area is 52.25 cm 2 . The sensitive volume is guarded by a 1.1 mm wide guard ring and is vented to the atmosphere via a 1.5 m long waterproof cable. A schematic drawing of the chamber can be found in Djouguela et al. 7 We used the center of the inner surface of the entrance window as the effective point of measurement (EPOM).
The chamber was irradiated with 6, 10, and 18 MV pulsed photon radiation generated by ARPANSA's linear accelerator (Elekta Synergy, Elekta AB, Stockholm, Sweden). This linac has main jaws in the in-plane direction, and 1 cm wide Multi-Leaf Collimators (MLC) with backup jaws in the cross-plane direction.
A reference-class Unidos Webline electrometer (PTW Freiburg, Freiburg, Germany) was used to provide the polarizing voltage and collect the charge M LAC produced in the LAC. An operating voltage of +400 V was used, with the collecting electrode at negative potential (CEN). This voltage is the maximum voltage recommended by the manufacturer. It results in minimal recombination losses while the LAC is operating in the ion-chamber region. M LAC was corrected for changes in air density by applying a correction factor k TP . 19 The reference values for standard temperature and pressure are 20°C and 101.325 kPa, respectively. The so-corrected charge is denoted as M LAC,cor .

2.A.1 | MicroCT analysis
The DAP w -method requires a uniform dose-response of the LAC across its sensitive area, and this in turn calls for a constant electrode separation. In order to measure electrode separation across the chamber area and also to assess details of the LAC's internal geometry, a high-resolution computed tomography (CT) image of the LAC has been obtained (HR-pQCT, XtremeCT II, SCANCO Medical AG, Br€ uttisellen, Switzerland). The spatial reconstruction accuracy was verified with a test object of known geometry (Phantom KP70, SCANCO Medical AG). X-ray transmission signals (60 kVp) of the LAC were reconstructed to a 3D CT data set of 1274 images with voxels of 0.082 mm side length. All images were inspected visually for high-density material.
The thicknesses of the chamber body and the internal air cavity were measured by performing an edge analysis of the CT data using an in-house routine written in MATLAB â (The MathWorks Inc., Natick, MA, USA). While keeping the resolution of the voxel space at 0.082 mm in the chamber's front-to-back direction, the data were resampled by averaging over 20 9 20 voxels (1.64 mm 9 1.64 mm) along the chamber's longitudinal and transverse axes to reduce image noise and to increase edge detection accuracy. The resulting voxel columns along the chamber's front-to-back direction were analyzed for edges by finding the 50% values between air (voxel value = 0) and chamber body (average voxel value = 2200). The thickness of the chamber body was calculated from the average distance between the outer edges of all columns, and this value was compared to calliper measurements. The thickness of the internal air cavity was calculated from the average distance between the inner edges of all voxel columns within a 4.08 cm radius around the chamber center. The internal thickness of the air cavity was mapped and inspected visually for any systematic variations.

2.A.2 | Linearity and reproducibility
The reproducibility and linearity of response were determined in 6, 10, and 18 MV beams of 10 9 10 cm 2 field size at 100 cm SSD and . We did not expect a change in the LAC's linear behaviour when reducing the field size below 10 9 10 cm 2 , and we did not perform a separate assessment of linearity in fields that are smaller than the LAC's sensitive diameter.
The reproducibility of M LAC,cor in the 5 cm diameter calibration field was expressed as the relative standard deviation of M LAC,cor . geometry. The LAC was placed on a 5 cm thick slab of Plastic Water â (CIRS Inc., Norfolk, Virginia, USA) to provide uniform backscatter. A holder, constructed from poly-vinyl chloride (PVC), suspended the source at a distance of 5 cm from the entrance window. M LAC,cor was measured over 120 s and corrected for source decay.

2.A.4 | Saturation and polarity correction
The saturation correction factor k s for the LAC was measured by changing the bias voltage U between +80 V and +400 V in 10 9 10, 4 9 4 and 1 9 1 cm 2 6 MV beams, at a depth in water of 1.5 cm and a SSD of 100 cm. Variations in linac output were accounted for with a monitor chamber (Farmer-type, FC65-G, IBA Dosimetry GmbH, Schwarzenbrueck, Germany), suspended laterally and upstream from the LAC within the 10 9 10 cm 2 field. For the 1 9 1 and 4 9 4 cm 2 field size, the monitor chamber was suspended in water 4 cm below the LAC on the beam central axis. The results were analyzed using a Jaffe plot. M LAC,cor (U) À1 was normalized to M LAC,cor (+400 V) À1 and plotted against U À1 . A linear regression analysis was performed to assess whether the LAC's nominal operating voltage of +400 V is appropriate. k s was derived from the intercept of the fitted curves at U À1 = 0. k s was also measured using the twovoltage technique and quadratic fit coefficients provided in TRS-398, at voltage ratios U 1 :U 2 of 1:2, 1:4, and 1:5, where U 2 = +400 V. 19 Supplementary values of k s (voltage ratio U 1 :U 2 = 1:4) have been determined for field sizes of 1 9 1, 3 9 3, 5 9 5, and 10 9 10 cm 2 at SSD and depth combinations of 100 and 10 cm, 90 and 10 cm as well as 80 and 20 cm, without a monitor chamber. Under the same setup, the polarity correction factor k pol was determined using the method described in TRS-398. 19

2.A.5 | Response anisotropy and extra-cameral effect
The change of response with change of angle of beam incidence a was measured in air without build-up. The linac was set to gantry angle a = 0°and a large 6 MV field (20 9 20 cm 2 ) was selected.
Using a large floor retort stand, the LAC was mounted with the entrance window toward the source, so that its EPOM was on the beam central axis and at the linac's gantry rotation isocenter, at 100 cm from the source. The chamber was irradiated with 100 MU at varying gantry angles a between 0°and 15°, M LAC was recorded and normalized at a = 0°. The procedure was repeated with a

2.B | Determination of LAC calibration coefficient
The aim of determining a calibration coefficient N D,w,LAC for the LAC is the ability to perform measurements of DAP w in small 6 MV photon beams. Under knowledge of DAP w , it is then possible to derive the output factor for a small field from the relative dose distribution, as demonstrated by Sanchez-Doblado et al. 8 A further use may be that of using the LAC as a transfer dosimeter to enable the calibration of small field detectors in a small reference field against a primary standard in primary standards laboratories, as outlined by Dufreneix et al. 15 For clarity, the methods required for determining N D,w,LAC are described first. Table 1 where M LAC,cor is the charge collected by the LAC and corrected for air density; and k i are correction factors for ion collection efficiency (k s ), polarity effect (k pol ), and electrometer collection efficiency (k elec ).
The relationship between DAP w and the central-axis dose, D w, CAX , is given by where R(r) is the relative two-dimensional dose distribution, normalized to unity at central axis (r = 0), in the plane of the LAC detector.
The double integral extends over A LAC , the LAC's sensitive area. Ð Ð A R(r)dr has been determined with radiochromic film and scanned dose profiles in the 5 cm diameter calibration field at 100 cm SSD and at 10 g/cm 2 depth, as described in the following two sections.

2.B.2 | Relative dose integral -film method
Pre-and postexposure images were spatially coregistered using marks on the film. The preexposure OD was subtracted from the postexposure OD to obtain DOD, the net-optical density. DOD was converted to dose (unit: Gy) using the formula where a 0 = 0, a 1 = 5.29081, a 2 = 62.94971, a 3 = À295.56757, a 4 = 908.96498, a 5 = À981.35406. The coefficients a i are based on a 5 th order polynomial fit to the D w vs DOD curve ranging from 0 to 2.8 Gy, which has been obtained by cross-calibration against a reference ionization chamber.
To improve the accuracy of the film dosimetry in the low-dose region, we employed a method similar to that presented by Sanchez-Doblado et al. 8 Two pieces of film were exposed at a high

Uncertainty of the film dose measurement
For the film response within the geometric field boundaries, we have assumed a value of 2% (k = 1) for the local response of the EBT3 film, which is achievable using stringent film dosimetry protocols based on net-optical density. In the area outside the geometric field boundary but still inside the LAC's sensitive diameter, we have added an extra 5% uncertainty to account for a potential underresponse of the EBT3 film. In that area, the mean photon energy is approximately 0.5 MeV, based on Monte Carlo data for a comparable setup (5 9 5 cm 2 6 MV field, 100 cm SSD, 10 cm depth) depicted in Fig. 2(c) in the publication by Chofor et al. 24 Approximating from data presented by Dufreneix et al, the EBT3 response is reduced by a factor of 0.95 to 1.0 compared to 6 MV, and we have assumed 0.95 as the worst case. 16 Hence, a 5% uncertainty term was added to the out-of-field EBT3 response uncertainty, increasing it to 7.0% (k = 1).
The relative uncertainty of ∫∫ A R(r)dr was calculated as the sum of the in-field and out-of-field uncertainty, weighted by the relative contribution to the total integral.

2.B.3 | Relative dose integralprofile method
If the dose distribution across the circular integration area has a circular symmetry, then an alternative integration method using highresolution scanned dose profiles can be employed. 15,25,26 Table 1 lists the type of detectors used for each instance where we used this method, a general description of which is given here. In order to obtain the relative dose integral, a small-volume detector is scanned in water at the calibration depth in a star pattern across the beam.
The dose profiles R prof (r) must intersect with and be normalized to the beam's central axis, where r = 0 cm. With R prof (r) and annuli of area A annulus (r,dr) as defined in Fig. 2, the relative dose integral over a circle with radius r LAC is calculated from each half-profile as the annular area-weighted average dose multiplied by p r LAC 2 , thus The thickness, dr, of the annulus is determined from the sum of the halfway distances to the two adjacent measurement points in the profile. The results from several transverse scans in a star pattern may be combined to improve the accuracy of this method and account for asymmetries in the beam profile.
∫∫ ALAC R(r)dr was measured for the 5 cm circular calibration field with two small detectors: a miniature thimble ionization chamber with a detector volume of 0.13 cm 3 and operated at +300 V (CC13), and an unshielded electron diode (EFD, IBA Dosimetry GmbH).
Detector choice was based on availability, well-known properties, stability and adequate size for the purpose, as well as to provide two independent data sets. Each detector was scanned in the inplane, cross-plane and in both diagonal directions.

2.B.4 | Determination of calibration coefficients in
a field smaller than the LAC's sensitive area The calibration coefficient for the LAC, N D,w,LAC , was determined in a 6 MV calibration field, for which DAP w was previously measured.
The field has a diameter of 5.0 cm at 100 cm SSD and is collimated Because in our case the calculated spectra for our 5 cm diameter calibration field were not yet available, we have not applied a correction factor for energy dependence and instead have added a relative uncertainty term of 0.2%.
We repeated the LAC calibration procedure in a 4 9 4 cm 2 6 MV field (without a stereotactic cone) to highlight differences in the resulting total uncertainty. The setup was identical to the one described above, but, due to the noncircular field shape, ∫∫ ALAC R(r) dr was derived from EBT3 only.

3.A.2 | Linearity and reproducibility
The reproducibility of M LAC was better than 0.04% in the broad field and 0.07% in the 5 cm diameter calibration field, respectively. The linearity for absorbed doses between 0.5 and 4 Gy, measured as the residual square error of a linear regression analysis, was R 2 = 1.0000 for 6, 10, and 18 MV photon beams. Maximum and average relative deviation of M LAC from linearity were 0.17% and <0.1%, respectively.

3.A.3 | Long-term response stability
The reproducibility of setup with the check source was 0.4% (1 SD).
Rotating the source resulted in changes <0.02%. The variation in chamber response to the check source over 9.5 months was within a range of 0.6%.

3.A.4 | Saturation and polarity correction
Jaffe plots are shown in Fig. 4. The intercept with the 1/Q axis lies at 0.9991 AE 0.0001, which corresponds to a value of k s = 0.9991 À1 = 1.0009 for 6 MV ( Table 2). Further values for k s determined with the two-voltage method at a number of different field sizes and depths in water are listed in Table 3.
Values for k pol are within 0.05% of 0.9990 for an operating voltage of +400 V (CEN), as shown in Table 3.

3.A.5 | Response anisotropy and extra-cameral signal
The LAC's response in the 6 MV broad beam increases slowly with a at a rate of 0.25% per 5° (Fig. 5). For the small 1 9 1 cm 2 6 MV field, a very significant additional increase in response is evident. An explanation for this effect, which was not investigated in other MV photon beam energies, is given in the discussion and the thus predicted response anisotropy is shown in the figure.
The extra-cameral effect was not detected, as its influence was less than the reproducibility of the LAC's response (0.1%).

3.A.6 | LAC response vs lateral off-axis displacement
The flat region in the center of the lateral response of the LAC in a 1 9 1 cm 2 beam shows that there is less than 0.2% variation of M LAC with lateral displacement of up to 0.5 cm (Fig. 6). At depth of d max , the out-of-field LAC signal is approximately equal to 4% of the

3.B | Determination of LAC calibration coefficient
All quantities used to determine the dose-area calibration coefficient   4. Jaffe plot normalized to Q(+400 V) À1 for field sizes of 10 9 10, 4 9 4, and 1 9 1 cm 2 at 100 cm SSD. Least square fit of linear curves from +80 V to +400 V; the linear curve equations are shown in the figure. Error bars correspond to 1 SD. The LAC was irradiated with 6 MV at 1.5 g/cm 2 depth in a water phantom at 100 cm SSD.
T A B L E 3 k s and k pol measured in broad and small square fields of side length s at different combinations of SSD, depth and beam energy. k s was determined with the two-voltage method using bias voltages of +400 V and +100 V.

SSD, depth (cm)
Field side length s (cm) 6   Repeating the film analysis without the median filter resulted in a relative dose integral that was 0.2% less than with the 30 9 30 pixel median filter, and this difference is small compared to the overall uncertainty of the experiment.
Repeated calculations of ∫∫ ALAC R(r)dr from film data while shifting the center of the integral laterally, resulted in a less than 0.05% change per 1 mm of lateral displacement of the integral for distances of up to 3 mm in the in-plane or cross-plane direction.

3.B.2 | Determination of calibration coefficient in a
field smaller than the LAC's sensitive area

| DISCUSSION
Uniform thickness of the sensitive volume is a requirement for correct planar dose-integration, as even a change of as little as 0.01 mm will change the volume of air (and hence the signal) by 0.5% in a 2 mm wide air cavity. By analyzing microCT data, we were able to verify the vendor-specified data and to visually inspect for any systematic variations, which were not found. High-resolution microCT was also an appropriate tool to scan for high-density material inside the chamber and revealed the presence of a small air cavity near the stem, a feature that is not seen on the drawings in the LAC's manual.
Our lateral profile measurements produced a flat response with off-axis positions of a few mm in a 1 9 1 cm 2 6 MV beam. The tolerance is about 0.5 cm, which is less than quoted by Douguela et al. 7 We were somewhat surprised by the relative magnitude the out-offield signal, which is approximately 4% of the central-axis signal at d max for all three photon energies and presumably arises due to scatter and leakage. Even for a 1 9 1 cm 2 beam, the 8 cm diameter LAC is not sufficient to encompass all of the beam. Caution is recommended in any DAP w analysis that assumes the entire beam is measured, or that the value of the dose at the edge of the detector is zero.  The underline signifies that the value in this row is a product of the two preceding rows. It is an intermediate result.
The bold signifies that this is the final result. I intended to draw the readers attention to that row foremost. a = 2.5°would cause a 0.1% increase in signal for fields smaller than the LAC's sensitive area. In cases where the LAC is positioned with its entrance window facing toward a photon source at 100 cm distance, any photon beam originating from the source will intersect with the LAC at an incident angle of a < 2.5°, and the increase in signal due to beam incidence would be less than 0.1%.
Ion recombination losses are generally very small. The linear slope of Q À1 vs U À1 at operating voltages between +100 V ≤ U ≤ +400 V for 6 MV indicates that the LAC operates in the ion-chamber region and that the two-voltage method can be used to derive k s if U 2 > 100 V (Fig. 4). The associated correction factor k s increases slightly with beam energy and with field size. The latter effect can be explained by the higher dose rate in larger fields, whereby an increased charge-density within the LAC's sensitive air volume leads to an increase in general recombination. Because the values for k s are generally so close to unity, it is appropriate to use a k s value determined in a 5 9 5 cm 2 field and apply this value to any field size down to 1 9 1 cm 2 without compromising measurement accuracy.
Measurements of the relative dose integral with EBT3 film were consistent with measurements using scanned profiles, as long as the two-film method was used to obtain in-field and out-of-field dose.
Applying a 30 9 30 pixel median filter during the film analysis did not significantly alter the calculated dose integral compared to no filter, despite an apparent penumbra broadening. There was very good agreement of the calculated relative dose integral for the small ionization chamber (CC13) and the unshielded diode (EFD), despite their different sizes in sensitive volume. This is because the numerical integration process removes the effect of any blurring due to detector size, as long as the integration area is much larger than the size of the small detector.
From an analysis of the dose map obtained with film, it is evident that the setup alignment of the LAC in the calibration beam does not require a lateral displacement accuracy of better than 2 mm without loss of calibration accuracy.
The calibration values N D,w,LAC derived here in a 5 cm diameter field are consistent with those determined in a 4 9 4 cm 2 MLC field.
However, in a 10 9 10 cm 2 square MLC field, the coefficient is reduced by 3.0%, which indicates an over-response of the LAC. The increased response of the chamber may be attributed to an additional ionization caused by secondary electrons scattering laterally into the sensitive volume after crossing the 1.1 mm wide guard ring, which is not wide enough to effectively screen those electrons.
There is also a remote possibility that ionization of air within the small cavity near the stem contributes to the measured charge.
The calibration coefficient measured in the broad 6 MV field (158.9 mGy cm À2 C À1 ) differs by 5 Compared to the response in a 6 MV beam, we noted an increase in the LAC's response of 1.0% and 2.7% in 10 and 18 MV, respectively.
Djouguela et al observed a broad-field calibration value for 15 MV of (1.700 AE 0.002) 9 10 8 Gy cm À2 C À1 at 995 hPa and 23°C, and this marks an increase in response of 1.7% compared to their 6 MV beam, which is consistent with the trend seen in our values. 7 This increase can largely be explained by the changes in the restricted Spencer-Attix water/air stopping power ratio, s w,air . Following TRS-398, the known

| CONCLUSION S
A large-area plane-parallel ionization chamber (LAC) with a sensitive area of 8.16 cm diameter was investigated for MV photon beam dosimetry. A calibration coefficient N D,w,LAC based on dose-area product in water (DAP w ) was determined in a circular field of 5 cm diameter and in the standard 10 9 10 cm 2 reference field.
This particular LAC was found to be suitable for the measurement of dose-area products in 6 MV beams of 5 cm diameter or less. Even when the LAC is not exactly aligned with the beams' central axis, its response is nearly constant: With a signal drop of just 0.05% per mm of lateral misalignment, precise DAP w measurements should therefore be achievable without a scanning water tank. Compared to point dosimeters, this would make the LAC a more practical dosimeter for routine measurements in small fields. However, users should verify this for their measurement geometry where it is different from ours, because the LAC also responds to the scattered and leakage radiation outside the primary beam. In addition, care must be taken to orient the chamber at right angles to the source in order to avoid over-response due to the additional amount of exposed air within the sensitive volume.
The uncertainty of N D,w,LAC was 1.3% (k = 1) when the calibration was performed in a field collimated by a 5 cm diameter cone.
Although the uncertainty of N D,w,LAC appears much reduced when the calibration is done in the standard 10 9 10 cm 2 field, we do not recommend using those large fields for the calibration of this detector due to an observed over-response of the LAC.
The DAP w -based calibration method requires a relative dose measurement with high precision over two or more orders of magnitude.
For a circular field, DAP w results from profile scans with small pointdose detectors (EFD, CC13) are consistent with those obtained with radiochromic film measurements (EBT3). Film also is useful for the measurement of DAP w in fields of noncircular symmetry, but in any case, the results from multiple films exposed to at least two dose levels should be combined. In order to reduce the overall uncertainty further, the EBT3 response should be corrected for the changes in photon spectrum that occur in the periphery of the calibration field.

ACKNOWLEDGMENTS
The author would like to thank Chris Oliver, Ivan Williams, Jessica Lye, Peter Johnston, and Peter Harty from ARPANSA for many helpful discussions. The author would also like to thank Ali Ghasem-Zadeh, from Austin Health's Endocrinology Centre of Excellence, for his skillful assistance in obtaining the high-resolution CT scans. The support of this work through an Australian Government Research Training Program Scholarship is gratefully acknowledged.

CONF LICT OF I NTEREST
No conflict of interest.