Applied Radiation and Isotopes 102 (2015) 15–28

Contents lists available at ScienceDirect

Applied Radiation and Isotopes

journal homepage: www.elsevier.com/locate/apradiso

Precise measurements of the absolute γ-ray emission probabilities of

223
Ra and decay progeny in equilibrium
S.M. Collins a,n, A.K. Pearce a, P.H. Regan a,b, J.D. Keightley a
a
National Physical Laboratory, Hampton Road, Teddington, Middlesex TW11 0LW, United Kingdom
b
Department of Physics, University of Surrey, Guildford, Surrey GU2 7XH, United Kingdom

H I G H L I G H T S

Discrepancies found within currently published γ-ray emission probabilities.

Absolute γ-ray emission probabilities of decay series in equilibrium determined.

Significant improvement in precision of measured values.

Closer agreement between deduced and experimental α transition probabilities.

Correlation coefficients presented for γ-emissions of 223Ra, 219Rn and 211Pb.

art ic l e i nf o a b s t r a c t

Article history: Precise measurements of the absolute γ-ray emission probabilities have been made of radiochemically
Received 25 September 2014 pure solutions of 223Ra in equilibrium with its decay progeny, which had been previously standardised by
Received in revised form 4π(liquid scintillation)-γ digital coincidence counting techniques. Two high-purity germanium γ-ray
3 March 2015
spectrometers were used which had been accurately calibrated using a suite of primary and secondary
Accepted 14 April 2015
radioactive standards. Comparison of the activity concentration determined by the primary technique
Available online 15 April 2015
against γ-ray spectrometry measurements using the nuclear data evaluations of the Decay Data Eva-
Keywords: luation Project exhibited a range of  18% in the most intense γ-ray emissions ( 41% probability) of the
Nuclear data
223
223
Ra decay series. Absolute γ-ray emission probabilities and standard uncertainties have been de-
Ra
219 termined for the decay of 223Ra, 219Rn, 215Po, 211Pb, 211Bi and 207Tl in equilibrium. The standard un-
Rn
215 certainties of the measured γ-ray emission probabilities quoted in this work show a significant im-
Po
211
Pb provement over previously reported γ-ray emission probabilities. Correlation coefficients for pairs of the
207
Tl measured γ-ray emission probabilities from the decays of the radionuclides 223Ra, 219Rn and 211Pb have
Gamma-ray emission probabilities been determined and are presented. The α-transition probabilities of the 223Ra have been deduced from
Gamma-ray spectrometry P(γ þ ce) balance using the γ-ray emission probabilities determined in this work with some agreement
Correlation coefficients observed with the published experimental values of the α-emission probabilities.
Radioactivity Crown Copyright & 2015 Published by Elsevier Ltd. All rights reserved.

1. Introduction
Radium-223 is a naturally occurring radionuclide, occupying
the later stages of the primordial decay series of 235U (see Fig. 1),
that makes up approximately 0.7200(51)% of naturally occurring
uranium (Rosman and Taylor, 1998). With a half-life of 11.4354
(17) d (Collins et al., 2015), 223Ra has undergone investigations for
use as a radiopharmaceutical, with successful clinical trials for
targeted radiotherapy of bone metastases and bone palliation that
occur from late-stage castration resistant prostate cancer
n
Corresponding author.
E-mail address: smc1@npl.co.uk (S.M. Collins).
(Michalski et al., 2013; Nilsson et al., 2007; Parker et al., 2013). As a
group II element, radium shares many chemical properties with
calcium and exhibits a high level of incorporation into metaboli-
cally active osteoblastic bone and tumour lesion sites (Bruland
et al., 2008; Nilsson et al., 2007). Coupled with the relatively short
dose deposition range of α-emissions this allows a highly targeted
cytotoxic dose of ionising radiation to a specific cancer site with
reduced damage to the bone marrow and other surrounding
healthy tissue, giving this treatment obvious advantages over the
relatively long energy deposition range of β-emitting bone-tar-
geting radiopharmaceuticals e.g. 89Sr, 166Ho and 153Sm, that have
been used historically. Additionally, as a naturally occurring
radionuclide it is of interest as a potential radiotoxic hazard from

http://dx.doi.org/10.1016/j.apradiso.2015.04.008
0969-8043/Crown Copyright & 2015 Published by Elsevier Ltd. All rights reserved.
16 S.M. Collins et al. / Applied Radiation and Isotopes 102 (2015) 15–28
Fig. 1. Decay series of
naturally occurring radioactive material (NORM) and technologi-
cally enhanced NORM (TENORM) (Kathren, 1998).
The 223Ra nucleus decays by 100% α-emission to excited states
of 219Rn (Chechev, 2011c), the decay scheme of which is shown in
Fig. 2, with a small branch (4.7  10  10 relative to the α-decay
branch (Kutschera et al., 1985)) decaying via the spontaneous
emission of a 14C nuclear cluster to 209Pb (Rose and Jones, 1984).
The 219Rn further decays via a series of relatively short-lived (T1/
2 o37 min) α- and β-emitting decay progeny each decaying via
their respective excited states (Chechev, 2011a, 2011b, 2011c; Ni-
chols, 2011; Kondev, 2013a, 2013b; Luca, 2010; Luca, 2011) with
associated γ-ray emissions, with the series terminating at the
stable nucleus of 207Pb. The 14C cluster decay mode of 223Ra has
not been investigated in the scope of this work.
In view of the growing importance of 223Ra in these applica-
tions the National Physical Laboratory (NPL) undertook a course of
work to provide an absolute standardisation of 223Ra (Keightley
et al., 2015) that would provide traceability to the SI unit of the
Becquerel. Two solutions of 223Ra were independently standar-
dised, henceforth referred to as S1 and S2, using liquid scintillation
(LS) absolute standardisation techniques, the first performed in
2013 and the second in 2014. Initial measurements of the solutions
were made by high purity germanium (HPGe) γ-ray spectrometry
of the thirteen most intense γ-ray emissions (41%) of the 223Ra
decay series, listed in Table.1. The activity of each selected γ-ray of
the 223Ra decay series was determined using the nuclear data from
the evaluations of the Decay Data Evaluation Project (DDEP)
(Chechev, 2011c; Nichols, 2011; Kondev, 2013a; Luca, 2010). Whilst
235
U.
the weighted mean activity of the individual γ-ray emissions were
in good agreement with the activity determined by the LS stan-
dardisation techniques (see Fig. 3), the activity concentration cal-
culated from the various γ-ray emissions had a range of  18%, as
shown in Table 1 and Fig. 4. We considered this indicated a sta-
tistically significant discrepancy in the existing published γ-ray
emission data, and that further measurements were required. This
conclusion reflected the findings of Kellett and Nichols (Kellett and
Nichols, 2013) who highlighted a disagreement between the α-
transitions and values deduced from the P(γ þ ce) balance within the
219
Rn excited levels.
A review of the currently published normalised γ-ray emission
values (Blaton-Albicka et al., 1976; Briançon and Leang, 1968; Da-
vidson and Connor, 1970a; Hesselink, 1972; Kossert et al., 2015;
Krien et al., 1970; Sheline et al., 1998) for the decay of 223Ra was
performed. In all cases, with the exception of Krien et al. (1970), no
specific information was provided regarding the full-energy peak
(FEP) efficiency calibration of the γ-ray spectrometers used. Such
FEP efficiency calibrations are critical to the accuracy of γ-ray
emission probability measurements. Comprehensive details of the
methodology used and the resulting FEP efficiency calibration
curve are therefore presented in detail in this article.
Maintaining a chemical separation of the different radio-
nuclides present in the decay series of 223Ra is problematic due to
the short half-lives of the decay progeny and evolution of 219Rn;
therefore the measurements were made of the decay series in
equilibrium within an aqueous solution. Investigation of the γ-ray
emissions from the decay of 223Ra and its decay progeny indicated
 S.M. Collins et al. / Applied Radiation and Isotopes 102 (2015) 15–28 17

223 219
Fig. 2. Decay scheme of Ra to states in Rn (Chechev, 2011c).

Table 1.
γ-ray nuclear data from the DDEP (Chechev, 2011a; Nichols, 2011; Kondev, 2013a;
Luca, 2010) and determined activity per unit mass of the S1 solution by HPGe γ-ray
spectrometry.

Energy/keV Radionuclide Intensity /% Activity/kBq g  1

223
122.3 Ra 1.238(19) 55.4(9)
223
144.3 Ra 3.36(8) 54.0(13)
223
154.2 Ra 5.84(13) 53.9(12)
223
269.5 Ra 14.23(32) 49.0(11)
219
271.2 Rn 11.07(22) 50.9(10)
223
323.9 Ra 4.06(8) 47.1(9)
223
338.3 Ra 2.85(6) 48.0(10)
211
351.0 Bi 13.00(19) 53.1(8)
219
401.8 Rn 6.75(22) 50.8(17)
211
404.8 Pb 3.83(6) 54.7(9)
211
427.2 Pb 1.81(4) 54.1(12)
223
445.0 Ra 1.28(4) 50.0(16)
211
832.0 Pb 3.5(5) 52.1(8)
Final result 52.0(8)

Fig. 3. S1 activity per unit mass determined by various measurement techniques.

the measurements of the primary γ-ray emission of each radio-

nuclide would be essentially unaffected by interferences from the
decay of the other radionuclides present, although some potential
convolutions would be present in the energy range of the main γ-
ray emissions of 223Ra (269.5 keV) with 219Rn (271.3 keV) and
219
Rn (401.8 keV) with 211Pb (404.8 keV).
A full uncertainty budget is detailed within this paper. All un-
certainties are stated as standard uncertainties or combined
standard uncertainties as defined in the Guide to the Expression of
Uncertainty in Measurement (GUM) (BIPM, 2008).
2. Experimental method
2.1. Sample preparation
Two separate active solutions were supplied by Algeta ASA
(Norway) with a nominal activity of 50 MBq 223RaCl2 in a 10 mL
sodium citrate pH buffer solution. Following previous experience
at NPL with 223RaCl2 in this chemical format, the material was
diluted with an aqueous solution of 1 M HCl to reduce the risk of
hydrolysis and any associated loss of activity during transfer of the
active solution between vessels. Solution S1 was diluted to a
nominal activity concentration of 50 kBq g  1 in 1 M HCl, with 1 g
18 S.M. Collins et al. / Applied Radiation and Isotopes 102 (2015) 15–28

2.3. HPGe γ-ray spectrometry

Two LN2 cooled HPGe γ-ray spectrometers, identified as BART

and LORAX, were used to perform the measurements. BART is an
n-type HPGe γ-spectrometer with a resolution (FWHM) of
1.79 keV at 1.3 MeV and a relative efficiency of 28%, LORAX is an
n-type semi planar HPGe γ-spectrometer with a resolution of
0.6 keV at 0.122 MeV. Both detectors were contained in identical
1.5 m  1 m  1 m Pb shields comprised of 10 cm thick aged Pb
walls covered with a 0.5 mm Cd and 0.7 mm Cu graded liner to
reduce effects from background radiation and Pb fluorescence
X-rays in the spectra. Aluminium optical bread boards were
mounted along the horizontal axis of the coffin with a kinematic
mounting plate holding a precision engineered sample holder at-
tached to enable highly reproducible geometric source positioning
in front of the detector window.
The energy calibration of both detectors was performed using
the most intense γ-rays ( 41%) of 152Eu, covering an energy region
from 121.8 keV to 1408.0 keV. The energy calibration resulted in
peak centroids within 0.05 keV of the 152Eu γ-ray energies eval-
uated by the DDEP (Vanin et al., 2004).
The spectra were collected using a chain of CANBERRA analo-
gue electronics (AFT Research amplifier 2025, Analogue-to-Digital
Converter 8715, AIM) connected to a PC running the CANBERRA
GENIE 2000 v2.1c software. The net peak area losses due to dead
Fig. 4. Activity per unit mass of the individual γ-ray emissions of 223Ra and decay
time were corrected for by the electronics using the integrated
progeny in equilibrium compared to the absolute activity per unit mass determined
for S1 (solid line) with the standard uncertainty (dashed lines). pile-up rejection (PUR)/live-time correction (LTC) circuit. An ad-
ditional correction was required for pulse pile-up occurring from
aliquots dispensed to three 2 mL ISO ampoules (ISO, 2010) for random coincidence summing events which had not been cap-
measurement by γ-ray spectrometry and dispensed to sixteen tured by the integrated PUR/LTC circuit of the analogue electronics
34 mL Wheaton type liquid scintillation vials. Solution S2 was di- (Rajput, 2010). This was determined empirically for each detector
using the decay corrected counts of a 99mTc source.
luted to a nominal activity concentration of 330 kBq g  1 in 1 M
All spectra were analysed using CANBERRA GENIE 2000 v2.1c. The
HCl and dispensed to a further three 2 mL ISO ampoules (ISO,
fit to each peak was manually reviewed and adjusted where neces-
2010) for γ-ray spectrometry and sixteen 22 ml Wheaton type li-
sary using the CANBERRA GENIE Interactive Peak Fit application. The
quid scintillation vials. In order to minimise the risks associated photopeak areas were corrected for background and integrated decay
with emanation of 219Rn, the ampoules were flame sealed and the (Harms and Jerome, 2004) during the counting period.
vial screw threads wrapped with PTFE tape.
2.3.1. Full-energy peak (FEP) efficiency calibration
The FEP detection efficiencies were determined using a suite of
2.2. Absolute standardisation traceable primary and secondary standards of γ-ray emitting
radionuclides to cover the photon energy range from 14 keV to
The S1 solution was standardised using the CIEMAT/NIST effi- 1836 keV. The calibration geometry was 1 g aqueous solution in a
ciency tracing method (Broda et al., 2007) and the 4π(LS)-γ digital 2 mL ISO ampoule at perpendicular distances from the detector
coincidence counting (DCC) technique (Keightley and Park, 2007; window of 30 cm and 16.5 cm for detectors BART and LORAX
Keightley and Watt, 2002). The S2 solution was standardised using respectively.
the 4π(LS)-γ DCC technique only. A description of the measure- Each calibration point was measured with multiple sources,
ment methodology has been covered in-depth by Keightley et al. collecting at least 105 counts in the photopeaks of interest. Using
(2015). The results of these measurements can be found in Table 2. the evaluated nuclear data from the DDEP (BIPM, 2004) for each
The activity per unit mass for the two solutions were de- radionuclide, the FEP efficiency was estimated using the equation:
termined as AS1 ¼ 52.35(17) kBq g  1 and AS2 ¼ 329.2(11) kBq g  1. N × Ct × Cd × Cp × Cc
ε keV =
t × m × A 0 × Iγ (1)
Table 2 where,
223
Activity per unit mass of Ra solutions S1 and S2 determined by multiple
techniques.
εkeV photopeak efficiency
Measurement technique S1 activity/ S2 activity/kBq g  1 N total counts collected in photopeak
kBq g  1 Ct Integrated decay correction
Cd Self-absorption correction to H20
CIEMAT/NIST efficiency tracing 52.4(4) –
4π(LS)-γ DCC 52.34(18) 329.2(11) Cp Pulse pile-up correction
HPGe γ-ray spectrometry 52.0(8) 325(5) Cc True coincidence summing correction
4π APPC (α)-γ coincidence counting 52.4(4) t Live time of measurement
4π APPC (α þβ)-γ coincidence 52.5(10) – m Mass of active material
counting
Final result 52.35(17) 329.2(11)
A0 Activity per unit mass of radioactive material
Iγ γ-ray emission probability per decay
 S.M. Collins et al. / Applied Radiation and Isotopes 102 (2015) 15–28 19

Table 3 2.3.2. Measurements of 223Ra and detection of impurities

Full-energy peak efficiency calibration points for the HPGe γ-ray spectrometers The S1 samples were measured solely on BART over a period of
BART and LORAX.
approximately 20 d with a total of eight measurements; referred
Energy/keV Radionuclide Full-energy peak efficiency /% to as dataset S1B. Each source was measured at least twice with a
typical measurement time of 50,000 s. A dead time of 0.50% was
BART LORAX recorded at the start of the measurement campaign and 0.23% at
2 the end of the measurement campaign.
14.41 57
Co 2.96(10)  10 9.0(3)  10  2
26.24 241
Am 1.29(5)  10  1 3.44(12)  10  1 The S2 samples were measured over a period of 29 d and 18 d
35.49 125
I 1.52(4)  10  1 – on both BART and LORAX respectively, with a total of nine mea-
46.54 210
Pb 1.647(23)  10  1 4.26(6)  10  1 surements on each detector; the corresponding datasets are re-
59.54 241
Am 1.674(10)  10  1 4.33(3)  10  1
ferred to as S2B and S2L. Each sample was measured three times
88.03 109
Cd 1.67(4)  10  1 4.29(8)  10  1
122.06 57
Co 1.601(11)  10  1 3.88(3)  10  1 on each detector, with a typical counting time of 86400 seconds.
136.47 57
Co 1.537(24)  10  1 3.54(6)  10  1 Dead times of 3.22% and 7.76% were recorded at the beginning of
165.86 139
Ce 1.416(12)  10  1 2.893(25)  10  1 the measurement campaign and 0.58% and 2.43% at the end of the
320.08 51
Cr 8.54(6)  10  2 1.107(10)  10  1
measurement campaign for BART and LORAX respectively. An ex-
391.70 113
Sn 7.16(6)  10  2 8.23(7)  10  2
514.00 85
Sr 5.61(5)  10  2 5.63(5)  10  2 ample of a spectrum collected on BART can be seen in Fig. 7.
661.66 137
Cs 4.53(4)  10  2 4.06(3)  10  2 All peak areas were corrected for dead time, integrated decay,
834.84 54
Mn 3.716(14)  10  2 3.038(12)  10  2 pulse pile-up and background. Self-absorption corrections were
898.04 88
Y 3.499(20)  10  2 2.793(13)  10  2
made to convert the FEP efficiency from H2O to 1 M HCl, this
1115.54 65
Zn 2.929(22)  10  2 2.162(16)  10  2
1173.23 60
Co 2.822(8)  10  2 2.045(6)  10  2 correction was less than 0.2% at γ-ray emission energies greater
1332.49 60
Co 2.534(14)  10  2 1.770(5)  10  2 than 100 keV. The count rates of those γ-ray emissions produced
1836.05 88
Y 1.93(9)  10  2 1.218(5)  10  2 by the decay of 211Pb, 211Bi, 211Po and 207Tl have been corrected to
take into account that these radionuclides exist in a transient
equilibrium with that of 223Ra due to the half-life of 211Pb being
approximately 36.1 min (Sargent, 1939). This correction was de-
As the chemical composition of the aqueous calibration stan- termined using the Bateman equation (Bateman, 1910):
dards varied, as is necessary to obtain a stable solution of the
λp
various elements, corrections were calculated for each data point × BR
λ p − λd (4)
to compensate for the difference in expected self-absorption
compared to H20. The corrections were determined using a far- where λp and λd are the decay constants of the parent and
field model for cylindrical samples (Parker, 1984). The linear at- daughter respectively and BR is the branching ratio of the parent
tenuation coefficients were derived from the NIST XCOM database to daughter. The correction of the γ ray emission probabilities for
(Berger et al., 1998). these decay progeny was approximately 0.22%.
The samples were set at large distances from the detector S1 and S2 were investigated for the presence of γ-ray emitting
window, such that the solid angles were low (approximately impurities and the precursors 227Ac and 227Th. The presence of
227
0.026 sr and 0.075 sr for BART and LORAX respectively, assuming Ac – T1/2 ¼7952.1 d (Browne, 2001) – and 227Th – T1/2 ¼18.68 d
the sample as a point source). Hence, no cascade summing coin- (Browne, 2001) – in small quantities could affect the accuracy of
cidence corrections were applied. The measured FEP efficiencies the determination of the γ-ray emission probabilities as they will
for each calibration point are listed in Table 3. support the activity of the 223Ra, thus perturbing the expected
The FEP detection efficiency curve fitted to the calibration data radioactive decay rate. Neither solution showed any indication of
set was of the form: impurities or the presence of 223Ra precursors, though additional
consideration of the precursor 227Ac was required as this radio-
⎡ N ⎤
⎢ ⎥ nuclide exhibits relatively weak α- and subsequent γ-ray emis-
⎢ ∑ c jbj(x, λ)⎥
ε(x, C) = e⎣ j = 1 ⎦ (2) sions that render α- and γ-spectrometry impractical. The activity
of S1 was measured over a period of 72 d (3.6 times longer than
where the measurement period for the γ-ray emission measurements)
E using an ionisation chamber. There was no exhibited perturbation
x = loge
511 (3) in the radioactive decay or trends within the residuals of least
squares fit; Indicating that there was no significant presence of
with the FEP efficiency denoted ε and the basis functions bj (x) 227
Ac or 227Th within the solution. It has been assumed by com-
selected were quartic b-splines (Cox, 1972). The only adjustable
parison of the measurements of S2 over the measurement period
parameter, for each value of j, was C. A series of six knots (where and the agreement between the calculated γ-ray emissions of S1
the piecewise polynomials are joined at λ = loge(Eλ/511)) were and S2 that there was no significant presence of 227Ac in this so-
selected at intervals along the energy range; these knots were lution either.
positioned based on a visual inspection of the shape of the FEP
efficiency curve. The knot set was augmented by placing an ad-
ditional four replicated knots both at the beginning and end of the 3. Results
sequence; the total number of knots was therefore fourteen and
the number of fitted coefficients (N) was 9. The best fit was de- The absolute γ-ray emission probabilities of 223Ra and its decay
termined by a generalised least squares fit (Aitken, 1936) to the progeny were derived from the activity per unit mass determined
calibration points spanning the range of energies that were cov- by absolute standardisation techniques and measurement by the
ered for each detector, 14 keV(57Co)–1836 keV (88Y) for BART and two HPGe γ-ray spectrometers. The absolute γ-ray emission in-
46 keV (210Pb)–1836 keV (88Y) for LORAX. The respective fits and tensities for each measurement were calculated using the equa-
residuals of the fits can be seen in Fig. 5 and Fig. 6. tion:
20 S.M. Collins et al. / Applied Radiation and Isotopes 102 (2015) 15–28
Fig. 5. Full-energy peak efficiency curve and residuals of the B-spline fit for the
HPGe γ spectrometer BART.
Fig. 6. Full-energy peak efficiency curve and residuals of the B-spline fit for the
HPGe γ spectrometer LORAX.
N × Ct × Cd × Cp × Cc
Iγ =
t × m × A o × ε keV (5)
The weighted mean of the measured absolute γ-ray emission
probability from each measurement for a dataset was deduced
using the uncertainties derived from the peak fitting software,
dead time and radioactive decay to determine the weight of con-
tribution from each measurement. The final value was deduced as
the weighted mean of the three datasets with the FEP efficiency
uncertainty combined with the standard uncertainty of the
weighted mean of the γ-ray emission probability uncertainty from
each dataset. The reduction in the standard uncertainty of the
weighted mean was limited to the smallest of the standard un-
certainty of the three datasets. The final uncertainties were de-
termined by adding the additional systematic uncertainties in
quadrature to the standard uncertainty calculated from the
Fig. 7. An annotated spectrum of 223
Ra and decay progeny in equilibrium.
weighted mean. The uncertainty budget is described in detail in
Section 4.
As the photopeaks below 100 keV are mainly the convoluted
X-ray emissions of 223Ra and its decay progeny, only results for the
observed γ-ray emissions from the decay of 223Ra and the resulting
decay progeny that have energies greater than 100 keV are pre-
sented in Table 4. These are presented as the absolute γ-ray
emission intensities per 100 disintegrations. Additionally the
radionuclide source of the γ-ray emission is listed as deduced from
the evaluations of the DDEP (Chechev, 2011a, 2011b, 2011c; Ni-
chols, 2011; Kondev, 2013a, 2013b; Luca, 2010, 2011).
The relative γ-ray emission probabilities of this work and those
previously published for 223Ra (Blaton-Albicka et al., 1976;
 S.M. Collins et al. / Applied Radiation and Isotopes 102 (2015) 15–28 21

Table 4
223
Absolute γ-ray emission probabilities per 100 decays of Ra and decay progeny in equilibrium.

Energy Source Iγ (%) Energy Source Iγ (%) Energy Source Iγ (%) Energy Source Iγ (%)
(keV) (keV) (keV) (keV)

223 223 215 211

103.9(5) Ra 0.0119(6) 323.9(6) Ra 3.655(18) 438.8(6) Po 0.0533(7) 675.4(6) Pb 0.0058(6)
223 223
106.7(4) Ra 0.0213(11) 328.4(6) Ra, 211Po, 0.2021(16) 445.0(6) 223
Ra 1.218(6) 676.9(6) 219
Rn 0.0184(5)
207
Tl
223 223 219 211
110.8(5) Ra 0.0512(10) 333.9(6) Ra 0.0756(6) 462.8(6) Rn 0.0011(5) 704.6(7) Pb 0.498(3)
223 223 223 219
122.3(5) Ra 1.312(6) 338.3(6) Ra 2.605(13) 487.3(5) Ra 0.0083(3) 707.8(7) Rn 0.0034(4)
219 223
130.6(5) Rn 0.1478(10) 342.9(6) Ra, 211Pb 0.1958(21) 500.2(6) 223
Ra 0.0013(5) 711.4(7) 223
Ra 0.0037(3)
223 211 211 223
144.3(5) Ra 3.481(16) 351.1(6) Bi 13.17(7) 504.1(6) Pb 0.0022(4) 727.4(7) Ra 0.0024(7)
223 223 219 211
154.2(5) Ra 6.02(3) 355.5(6) Ra 0.0124(15) 517.6(6) Rn 0.0453(5) 766.4(7) Pb 0.685(4)
223 211 223 211
158.7(5) Ra 0.749(4) 361.7(6) Pb 0.0341(7) 522.6(6) Ra 0.0021(6) 831.9(7) Pb 3.448(16)
223 223 223 219
175.6(5) Ra 0.01578(10) 363.0(6) Ra 0.0192(9) 527.6(6) Ra 0.0659(8) 835.6(7) Rn 0.00364(19)
223 223 223 211
177.4(5) Ra 0.0426(8) 368.4(6) Ra 0.0134(4) 531.4(6) Ra 0.0028(9) 865.8(6) Pb 0.00540(21)
223 223 223 219
179.7(5) Ra 0.1613(10) 371.7(6) Ra 0.435(3) 537.5(6) Ra 0.0033(6) 891.3(7) Rn 0.00107(20)
223 223 223 211 207
221.4(5) Ra 0.0304(10) 372.9(6) Ra 0.1133(13) 542.1(6) Ra 0.0026(6) 897.8(7) Po, Tl 0.2725(15)
219 223 223 211
224.0(5) Rn 0.0056(14) 376.2(6) Ra 0.0056(4) 545.9(6) Ra 0.0028(6) 1014.7(7) Pb 0.0171(4)
223 223 219 219
249.4(5) Ra 0.0375(9) 383.3(5) Ra 0.0023(6) 555.9(5) Rn 0.0026(7) 1074.5(7) Rn 0.00044(12)
223 223
251.9(5) Ra 0.0640(11) 386.3(5) Ra, 219Rn 0.0052(7) 564.4(5) 219
Rn 0.0035(4) 1080.1(7) 211
Pb 0.01228(21)
223 223 211
255.1(5) Ra 0.0499(13) 390.1(5) Ra 0.0053(7) 569.6(7) Po, 207Tl 0.0043(5) 1103.3(8) 211
Pb 0.00380(12)
223 219 223 211
269.5(6) Ra 13.37(7) 401.8(6) Rn 6.57(3) 573.7(7) Ra 0.0029(13) 1109.5(8) Pb 0.1113(7)
219 211
271.3(6) Rn 10.75(6) 404.8(6) Pb, 215At 4.011(19) 598.6(7) 223
Ra 0.0867(12) 1196.2(8) 211
Pb 0.01052(17)
223 211 223
288.2(6) Ra 0.1498(16) 427.1(6) Pb 1.890(9) 609.3(7) Ra, 219Rn, 0.0543(7) 1234.3(8) 211
Pb 0.00092(8)
211
Pb
219 223 211 219 211
293.6(5) Rn 0.0688(7) 430.4(6) Ra, Pb 0.0206(19) 619.8(6) Rn 0.0056(12) 1270.7(8) Pb 0.00624(19)
211 223 223
313.7(6) Pb 0.0276(5) 432.4(6) Ra 0.0297(14) 623.4(5) Ra 0.0082(8)

Briançon et al., 1968; Davidson and Connor, 1970a; Hesselink,
1972; Krien et al., 1970; Sheline et al., 1998), 219Rn (Blaton-Albicka
et al., 1976; Briançon and Leang, 1968; Dalmasso and Maria, 1967;
Davidson and Connor, 1970b; Krien et al., 1970; Liang et al., 1998),
215
Po (presented as absolute values for comparison) (Briançon and
Leang, 1968; Davidson and Connor, 1970b) and 211Pb (Blaton-Al-
bicka et al., 1976; Briançon and Leang, 1968; Cothern and Connor,
1965; da Silveira et al., 1971; Dalmasso and Maria, 1967; Davidson
et al., 1967; Giannini et al., 1962a; Gorodetzky et al., 1968; Ha-
milton and Davies, 1968; Hindi et al., 1988; Mead and Draper,
1965; Vandenbosch et al., 1963) are presented in Tables 5–8. The
269.5 keV γ-ray emission of the 223Ra decay is convoluted with the
271.3 keV γ-ray emission of the 219Rn decay. Hence, the relative γ-
ray emission intensities of 223Ra were deduced as the γ-ray
emission intensity normalised to the 154.2 keV γ-ray emission
probabilities, as this transition has no known convolutions and
should therefore be a reliable normalisation point. The 271.3 keV
γ-ray emission has been used as the normalisation for the γ-ray
emissions of 219Rn. Though it is convoluted with the 269.5 keV of
223
Ra, the only other significant emission of 219Rn, the 401.8 keV γ-
ray emission, is also convoluted by the 404.8 keV γ-ray emission of
211
Pb. The 211Pb values have been deduced relative to the 351.1 keV
γ-ray emission.
4. Uncertainties
4.1. Uncertainty components of the γ-ray emission probabilities
A summary of the uncertainty components for the absolute γ-
ray emission probability of the 269.5 keV γ-ray emission is pre-
sented in Table 9. The assigned uncertainty values of the compo-
nents were combined in quadrature to determine the final un-
certainty value. The dominating uncertainty components for the
most significant γ-ray emissions of the 223Ra decay series (where
the statistical uncertainties are less than 0.1%) are due to the
standard uncertainty of the standardisation of the activity per unit
mass, the FEP efficiency calibration and efficiency stability of the
detector chain.
The FEP efficiency uncertainties were calculated from the least
squares fit of the photopeak efficiency calibration including the
input efficiency uncertainties (see Table 3) and taking into account
the correlations and co-variances due to nuclear data, standardi-
sation, etc. The calculated uncertainty of the FEP efficiency cali-
bration can be seen in Figs. 5 and 6 as the dashed line in the re-
sidual plots.
A systematic uncertainty of 5% was estimated for the dead-time
and pile-up correction, which when propagated using the median
recorded dead-time resulted in an overall uncertainty component
of 0.05%. An additional uncertainty component was incorporated
in the dead-time and pile-up uncertainty to take into account the
perturbation on the photopeak shape that occurs at increasing
count rates (Rajput, 2010). Within the range of dead times ob-
served this was estimated to be insignificant.
A radioactive decay uncertainty was determined using the
evaluated uncertainty of the 223Ra half-life, in previous work by
the authors (Collins et al., 2015), of 11.4354(17) d, using the
median decay period of the measurements to the standardisation
reference time to propagate the radioactive decay uncertainty.
An estimated uncertainty of 0.10% was incorporated to account
for the small but inevitable quantity of true coincidence summing
events that may occur. As described previously in Section 3, due to
the small solid angle subtended by the detector the number of
events that would occur would be insignificant and therefore the
assigned uncertainty value would be sufficient to account for these
events.
It cannot be assumed that the detector response will remain
constant with small fluctuations that may occur in the detector
itself or in the detector electronics chain; these effects can be small
and hence obscured within the measurement series. This effect
should be accounted for especially if the measurements are made
over an extended period; the long term stability uncertainty has
been estimated using a 152Eu source which had been measured
over the measurement campaigns.
The geometric reproducibility uncertainty for the positioning of
the source in an identical geometry is 0.10% as the use of precision
engineered source holders in tandem with the kinematic optical
mounting system allows a very high level of positional
reproducibility.
22 S.M. Collins et al. / Applied Radiation and Isotopes 102 (2015) 15–28

Table 5
Normalised γ-ray emission probabilities of 223Ra observed in this work and published values. All values normalised to the 154.2 keV γ-ray emission. Energies marked with an
n
have interferences with γ-ray emissions from decay progeny in equilibrium.

Energy/keV This work Briançon et al. Krien et al. Davidson and Hesselink Blaton-Albicka Sheline et al. Kossert et al. χ 2 /(n − 1)
(1968) (1970) Connor (1970a) (1972) et al. (1976) (1998) (2015)

103.9(5) 0.198(10) – – – – 0.35(6) 0.34(5) – 7.8

106.7(4) 0.354(18) 0.43(8) 0.34(7) 0.42(11) 0.43(14) 0.41(6) 0.42(3) – 0.8
110.8(5) 0.850(17) 1.04(17) 0.81(7) 1.08(15) 0.47(20) 1.05(15) 1.02(8) 0.93(5) 2.3
122.3(5) 21.79(11) 21(4) 21.3(6) 26(4) 19.6(15) 20(3) 21.2(6) 21.63(12) 0.9
144.3(5) 57.8(3) 57(7) 57.9(17) 61(9) 62(6) 57(8) 57.3(19) 57.53(24) 0.2
154.2(5) 100(–) 100(–) 100(–) 100(–) 100(–) 100(–) 100(–) 100(–) –
158.7(5) 12.27(6) 13.0(16) 12.2(4) 14.7(22) 11.9(11) 12.1(17) 12.2(4) 12.22(8) 0.3
175.6(5) 0.2620(17) – 0.24(7) – 0.34(9) – 0.34(7) 0.284(16) 0.9
177.4(5) 0.708(13) 0.54(19) 0.83(8) – 0.79(14) – 0.83(8) 0.632(18) 4
179.7(5) 2.678(18) 2.8(8) 2.6(7) 2.9(5) 2.6(4) 2.7(4) 2.68(5) 2.54(5) 1.2
221.4(5) 0.504(16) 0.65(19) 0.54(10) – 0.56(10) – 0.63(10) 0.463(18) 1.3
249.4(5) 0.622(14) 0.67(19) – – 0.7(3) – 0.68(17) 0.60(5) 0.1
251.9(5) 1.062(19) 1.3(3) 0.66(17) 1.1(4) 1.1(4) 1.24(23) 0.73(24) 1.05(3) 1.2
255.1(5) 0.829(21) 1.1(3) – 1.0(4) 0.54(16) 0.87(21) 0.93(12) 0.73(3) 2.1
269.5(6) 221.9(11) 260(40) 243(7) 260(40) 225(16) 260(30) 244(8) 218.2(13) 4.2
288.2(6) 2.488(27) 3.0(5) 2.82(13) 2.8(5) 2.1(4) 2.8(4) 2.80(10) 2.42(3) 3.6
323.9(6) 60.7(3) 69(9) 71.5(20) 70(10) 60(5) 71(8) 70.0(21) 60.71(25) 7.3
328.4(6)n 3.36(3) 3.7(5) 3.70(18) 3.1(7) 2.7(5) 3.7(5) 3.7(12) 3.28(3) 1.6
333.9(6) 1.255(9) 1.6(3) 1.85(15) 2.4(6) 1.6(3) 1.42(24) 1.78(11) 1.244(22) 6.6
338.3(6) 43.25(21) 50(6) 51(12) 50(8) 43(3) 49(6) 49.8(16) 43.35(17) 2.8
342.9(6)n 3.25(4) 3.7(5) 4.14(23) 3.9(11) 1.6(4) 3.9(5) 3.9(3) 2.90(3) 15
355.5(6) 0.206(25) – – – – – 0.073(24) – 15
363.0(6) 0.318(15) – – – – – 0.49(12) 0.292(10) 2.3
368.4(6) 0.223(5) – – – – – 0.15(7) 0.252(19) 1.7
371.7(6) 7.22(5) 10.1(13) 8.66(24) 10.5(19) 9.5(11) 8.3(10) 8.5(4) 8.3(3) 10
372.9(6) 1.882(22) – – – – 1.9(3) 0.878(21) 0.85(3) 440
376.2(6) 0.092(5) – – 0.23(8) – – 0.22(7) 0.093(5) 2.0
383.3(5)n 0.038(9) – – 0.29(11) 0.07(5) – – – 2.7
386.3(5) 0.086(11) – – – 0.23(9) – 0.27(10) – 2.8
390.1(5) 0.088(11) – – – 0.05(4) – 0.12(5) – 0.9
430.4(6)n 0.34(4) – – – 0.32(9) – 0.34(10) 0.393(4) 1.2
432.4(6) 0.493(23) 0.622(91) 0.63(10) – 0.54(14) 0.49(9) 0.61(5) 0.496(21) 1.3
445.0(6) 20.22(10) 22.5(20) 26.8(20) 25(4) 20.7(20) 22(3) 22.7(9) 20.27(8) 3.0
487.3(5) 0.137(5) 0.184(39) 0.24(7) – 0.18(9) – 0.195(25) 0.100(11) 3.7
500.2(6) 0.021(8) – – – – – 0.024(10) – 0.1
522.6(6) 0.034(9) – – – – – 0.024(10) – 0.6
527.6(6) 1.094(13) 1.30(16) 1.31(12) 1.3(3) 1.1(3) 1.08(13) 1.24(8) 1.056(18) 1.9
531.4(6) 0.047(16) – – – – – 0.024(10) – 1.5
537.5(7) 0.055(10) – – – – – 0.037(12) – 1.4
542.1(7) 0.043(9) – – – – – 0.024(10) – 1.9
545.9(7) 0.047(10) – – – – – 0.020(10) – 4.0
573.7(7) 0.049(21) – – – – – 0.020(10) – 1.6
598.6(6) 1.441(19) 1.48(19) 1.85(17) 1.8(4) 1.5(3) 1.65(19) 1.66(8) 1.425(14) 2.3
609.3(7)n 0.902(10) 0.93(13) 1.31(17) 1.21(22) 0.68(25) 0.98(12) 1.00(6) 0.507(12) 102
623.4(5) 0.135(14) 0.15(8) – – – – 0.15(7) – 0.3
711.4(7) 0.061(5) 0.065(19) – – – – 0.063(17) – 0.1
727.4(7) 0.040(12) – – – – – 0.0049(24) – 8.2

Table 6
219
Normalised γ-ray emission probabilities of Rn observed in this work and published values. All values normalised to the 271.3 keV γ-ray emission.

Energy/keV This work Dalmasso and Briançon et al. Davidson and Krien et al. Blaton-Albicka Liang et al. Kossert et al. χ 2 /(n − 1)
Maria (1967) (1968) Connor (1970b) (1970) et al. (1976) (1998) (2015)

130.6(5) 1.375(9) 1.40(14) 1.18(23) 1.05(25) 1.21(10) 1.10(11) 1.7(2) – 2.3

224.0(6) 0.052(13) – – – – – 0.013(2) – 8.8
271.3(6) 100(–) 100(–) 100(–) 100(–) 100(–) 100(–) 100(–) 100(–) –
293.6(5) 0.640(7) 0.64(6) 0.70(14) 0.59(15) 0.51(27) 0.72(5) 0.680(42) – 0.7
401.8(6) 61.1(3) 58(6) 61(4) 65.2(65) 69(3) 58.4(27) 59.0(23) 60.9(5) 1.4
462.8(6) 0.010(5) – – – – – 0.0015(3) – 3.3
517.6(6) 0.421(4) 0.44(10) 0.44(4) 0.22(5) – 0.41(3) 0.40(22) – 3.5
555.9(5) 0.024(6) – – – – – 0.0005(3) – 14
564.4(5) 0.033(3) – – – – – 0.014(3) – 19
619.8(6) 0.052(11) – – – – – 0.003(1) 0.0469(78) 25
676.9(6) 0.171(4) 0.21(3) 0.21(2) 0.06(3) – 0.15(1) 0.160(20) – 4.7
707.8(7) 0.031(4) – – – – 0.003(1) – 60
835.6(7) 0.0339(17) – – – – 0.015(3) – 30
891.3(7) 0.0100(19) 0.015(7) – – – – 0.007(2) – 1.0
1074.5(5) 0.0041(10) – – – – 0.003(1) – 0.6
 S.M. Collins et al. / Applied Radiation and Isotopes 102 (2015) 15–28 23

Table 7

χ 2 /(n − 1)
Absolute γ-ray emission probabilities per 100 decays of 215Po observed in this work

1.2
8.0

2.2
0.9

5.3

1.0
1.8
2.3

8.7
5.9
6.8
1.1

21
59

431

2600

180
and published values.

Energy/keV This work Briançon Davidson and Kossert χ 2 /(n − 1)

et al. (1968) Connor et al.

3.776(28)
14.44(10)

25.91(17)
5.166(38)
0.198(15)
Kossert

30.6(4)
(1970b) (2015)

(2015)
et al.
438.8(6) 0.0533(7) 0.048(5) 0.064(2) 0.054 (4) 9.1

–
–

–
–
–
–
–
–

–
0.045(6)

0.042(6)

0.095(6)

0.043(4)
0.033(4)

0.072(5)
4.94(16)

0.129(8)
29.3(9)

26.7(8)

0.90(3)
13.9(4)
(1988)

3.6(1)
Hindi
et al.
4.2. Uncertainty correlations of γ-ray emission probabilities

–
–

–
bicka et al.
The majority of the uncertainty components are significantly

Blaton-Al-

0.326(24)

0.049(5)
25.4(20)

0.090(7)

0.081(6)
0.033(4)

0.057(5)
30.2(14)

0.122(8)
0.130(8)
0.20(3)

0.82(6)
14.2(7)
correlated between the solutions S1 and S2 (as the standardisation

(1976)

3.6(3)
5.1(4)
methodology is identical) as well as between the two detectors.

–
This is due to the process of determining the efficiency curve

da Silveira
where the standard solutions used to calibrate the detectors were

0.120(12)
0.040(6)
5.55(28)

0.050(8)

0.070(7)
3.77(19)

0.010(2)
0.24(3)

30.0(9)

29.8(7)
13.5(6)
0.12(2)

0.14(1)

0.10(1)
of similar provenance i.e. ionisation chamber, and identical nuclear

1.15(8)
(1971)
et al.
data used. Failure to account for these correlations will often un-

–
dermine the quality of the measurements and calculations relying

Bi.

et al. (1968)
211
on such data. These effects are not routinely dealt with in pub-

0.053(15)

0.083(10)

0.079(15)
lications of γ-ray emission probabilities.

0.128(15)
0.173(15)
Briançon
Pb observed in this work and published values. All values are normalised to the 351.1 keV γ-ray emission of

0.005(2)
0.048(8)
3.68(23)
5.04(30)

0.023(5)
25.6(23)
30.8(15)
0.30(8)

0.79(8)
14.3(8)
0.19(4)
In this work the calibration standards were all of NPL prove-
nance and therefore the off-diagonal terms of the matrix used to

–
weight the generalised least-squares fit were estimated from the
ionisation chamber calibration factor uncertainties for the radio-

et al. (1968)
Gorodetzky
nuclides concerned. Correlations due to sample preparation decay

29.6(20)
13.7(10)

24.1(17)
0.25(5)

0.07(2)

0.08(1)

0.81(6)
0.08(1)

0.08(1)
0.10(5)

0.15(1)
and dead time corrections were not included as these were rela-

3.7(3)
4.9(3)
tively small contributors to the final uncertainty. Correlations in

–
published γ-ray emission values may be more significant; however
Hamilton and
Davies (1968)
data were not available to allow the covariance to be estimated.

0.0006(1)
0.0025(1)
The correlation matrix R as defined in GUM supplement 2 (BIPM,

28.6(11)

0.011(1)
27.4(4)
11.6(7)
2011) contains the correlation coefficients associated with pairs of

2.9(1)
4.5(1)
measured γ-ray emission probabilities such that:
–
–

–
–

–
–

–
–
⎡ r (P , P ) … r (P , P ) ⎤
Dalmasso and
Maria (1967)

⎢ 1 1 1 N
⎥
R=⎢ ⋮ ⋱ ⋮ ⎥
28.0(28)

23.0(23)

0.70(15)
14.0(14)

⎢⎣r (PN , P1) ⋯ r (PN , PN)⎥⎦

0.26(5)

0.03(1)

0.08(2)

0.08(2)

0.05(1)
0.13(2)
(6) 3.0(3)
4.0(4)
–

–
–

–
where
et al. (1967)

Cov(xi , xj )
0.0347(14)

0.076(14)
0.125(21)
Davidson

r (x i , x j ) = 0.104(14)

0.042(7)
29.9(35)

26.4(35)

0.87(10)
13.9(17)
0.21(7)

Var(xi )Var(xj ) (7)

3.3(4)
5.1(6)
–

–
–

–
Following the determination of the fitted parameters C, from Eq. (2),
and uncertainty matrix UC the covariance of the fitted parameters was
Mead and

14.5(14)
27.4(12)

27.4(12)

1.03(10)
0.13(12)

determined by the linear propagation of uncertainties for multivariate

0.04(2)

0.06(2)
Draper

0.14(2)

0.11(3)
(1965)

3.7(2)
5.2(2)

models as described in supplement 2 of the GUM (BIPM, 2011). The

correlations between pairs of the most intense measured γ-ray
–
–

–
–

emissions (41%) associated with the decay of the radionuclides

223
Ra, 219Rn and 211Pb in the series are given in Tables 10–12.
Vandenbosch
et al. (1963)

24.8(25)
12.5(25)

1.07(16)
3.8(11)
26(5)

5. Discussion
211

–
–

–
–

–
–
–
–

–
–
–
Normalised γ-ray emission probabilities of

5.1. Results
Giannini et
al. (1962a)

34.2(13)

0.38(19)

1.46(19)

A total of 83 absolute γ-ray emission probabilities have been

5.5(4)
6.1(4)
34(4)
22(3)

determined in this work. The determined probabilities show sig-

–
–

–
–

–
–

–
–
–

nificant improvement in their precision compared to some pre-

0.0932(16)

0.0799(11)
0.0410(16)

0.0474(14)
Energy/keV This work

0.0070(6)
0.0289(9)
3.782(21)
30.45(14)

26.17(13)
0.044(3)
0.209(3)
0.259(4)

0.845(5)
14.34(7)
0.017(3)

0.130(3)

viously published values. For example the uncertainty of the

5.20(3)

269.5 keV γ-ray emission probability is a factor of 4 lower than

previously quoted by Sheline et al. (1998) and a factor of 2 more
precise than quoted by Kossert et al. (2015). Previously no detailed
1234.3(8)
1270.7(8)
1080.1(7)
1014.7(7)

1103.3(8)
1109.5(8)
1196.2(8)
404.8(6)

865.8(6)
675.4(6)
704.6(7)
766.4(7)
504.1(6)
313.7(6)
361.7(6)

831.9(7)
427.1(6)

uncertainty budgets have been published therefore it is difficult to

Table 8

specify the exact reasons for the improvement. One possible rea-
son is the use of higher activity samples of 223Ra that has allowed
24 S.M. Collins et al. / Applied Radiation and Isotopes 102 (2015) 15–28

Table 9
223
Uncertainty budget for the 269.5 keV γ-ray emission of Ra.

Uncertainty component Relative uncertainty

(k ¼ 1)

Standard uncertainty of the weighted mean (includes 0.24%

the full-energy peak efficiency)
Activity per unit mass 0.33%
Dead time and pile-up correction 0.050%
Radioactive decay correction 0.010%
Geometric reproducibility 0.10%
Gravimetric 0.10%
Detector stability 0.20%
Peak fitting 0.10%
True coincidence summing 0.10%
Total uncertainty 0.50%

Table 10
223
Correlation coefficient matrix of the most intense γ-ray emissions (4 1%) of Ra.

Energy/keV

122.3 144.3 154.2 269.5 323.9 338.3 445.0

Energy/keV 122.3 100% – – – – – –

144.3 85% 100% – – – – –
154.2 75% 96% 100% – – – –
269.5 67% 75% 75% 100% – – –
323.9 72% 74% 72% 96% 100% – –
338.3 73% 75% 72% 95% 97% 100% –
445.0 77% 77% 74% 85% 89% 90% 100%

Table 11
219
Correlation coefficient matrix of the intense γ-ray emissions ( 41%) of Rn.

Energy /keV

271.3 401.8

Energy/keV 271.3 100% –

401.8 85% 100% Fig. 8. Results of the (a) 269.5 keV and (b) 271.3 keV absolute γ-ray emissions for
the three datasets. The solid line indicates the final weighted mean result reported
and the dashed lines indicate the final uncertainty of the absolute gamma emission
probability. The error bars are composed of the statistical and efficiency un-
certainties only.
Table 12
211
Correlation coefficient matrix of the intense γ-ray emissions ( 41%) of Pb.

Energy/keV

404.8 427.1 831.9

Energy/keV 404.8 100% – –

427.1 95% 100% –
831.9 83% 83% 100%

the collection of greater than 106 counts in the photopeaks,

therefore a low statistical uncertainty, of the most intense γ-ray
emissions. This has also allowed many of the relatively small γ-ray
emission probabilities to be quantified, which have only previously
been seen in γ–γ coincidence measurements (Sheline et al., 1998).
The 269.5 keV absolute γ-ray emission values determined for the
three datasets of S1B, S2B and S2L are in good agreement, shown in
Fig. 8(a), with only the statistical and FEP efficiency uncertainty in-
cluded in individual dataset uncertainty. The agreement between S1B
and the results of S2B and S2L indicate a consistency in the standar-
disations of the solutions S1 and S2. The agreement between S2B and Fig. 9. The peak fits of the convoluted photopeaks of the 269.5 keV and 271.3 keV
γ-ray emissions.
S2L indicates that there is no significant difference between the FEP
efficiency calibrations of the two detectors. These two observations
 S.M. Collins et al. / Applied Radiation and Isotopes 102 (2015) 15–28
Fig. 10. The reduced χ2 of the published literature for the most intense γ-rays
normalised to the 154.2 keV (hollow squares) and the 269.5 keV (hollow triangles)
γ-ray emissions.
Fig. 11. The relative FEP efficiency curves for BART determined using the relative γ-
ray emission probabilities of 75Se; using the nuclear data of Krien et al. (1970)
(hollow triangles) and Negret and Singh (2013) (hollow squares).
provide supporting evidence in the confidence of the reliability in the
presented absolute γ-ray emission probabilities.
A potential source of error could be from the de-convolution of
the 269.5 keV and 271.3 keV γ-ray emissions by the peak fitting
software. Although the use of high resolution γ-ray spectrometry
allowed the individual peaks to be relatively well resolved (see
Fig. 9) any inaccuracy in the peak shape fitting calibration could
lead to inaccuracies in the determined peak areas. Fig. 8(b) shows
the results for the 271.3 keV absolute γ-ray emission from the
219
Rn decay, showing a consistency in the results between the
datasets. It can therefore be inferred that no significant errors
occurred from the de-convolution of these two photopeaks by the
software.
The effect of the diffusion of 219Rn from the aqueous phase to
gas phase in the air space at the top of the ampoules was con-
sidered and then discounted, as this effect has been previously
investigated by Cessna and Zimmerman (2010) and Bayer (2012).
25
Fig. 12. Difference and z-score of the γ-ray emission probabilities (normalised to
the 154.2 keV γ-ray) of Krien et al. (1970) and this work before (hollow squares) and
after (hollow triangles) adjustment for the 75Se nuclear data.
It is the authors' hope that the provision of the correlation data
will help to inform any future decay data evaluation of 223Ra.
While the magnitude of the correlation will to some extent be
dependent on both the choice of radionuclides for the efficiency
calibration and on the selection of the fitting function, it is sug-
gested that by assuming the same level of correlation is present
across other measurements of the emission probability, a more
realistic evaluation, than the usually applied assumption of no
correlation between the emission probabilities, could be achieved.
5.2. Published literature of 223
Ra γ-ray emission probabilities
A comparison of the values reported for the 223Ra γ-ray emis-
sions in this work to those in the currently published literature,
which had been normalised to the 154.2 keV γ-ray emission, are
shown in Table 5. A review of the reduced χ2 for each of the main
γ-ray emissions (41%) shows that for γ-ray emissions of the
122.3 keV, 144.3 keV and 158.7 keV all the reported values are in
good agreement; with reduced χ2 values less than one. Alter-
natively, the reduced χ2 values for the γ-ray emissions of the
269.5 keV, 323.9 keV, 338.3 keV and 445.0 keV indicate significant
differences within the datasets. The reported values of this work
and Kossert et al. (2015) are in good agreement for these γ-ray
emissions. In the majority of cases the reported values of the re-
maining published literature show a significantly positive bias;
10% and 15% for the 269.5 keV and 323.9 keV γ-ray emissions
respectively.
Normalisation of the γ-ray emission to the 269.5 keV γ-ray in-
dicates that the discrepancies in the datasets reverse; whereby the
datasets of the 122.3 keV, 144.3 keV and 158.7 keV γ-rays become
discrepant and vice-versa. This is shown in Fig. 10.
The switch in the discrepancies would suggest that the relative
ratio of the efficiency in the region of the 269.5 keV to the region
of 154.2 keV differs between that of the FEP efficiency calibration
used in this work and Kossert et al. (2015) to the earlier publica-
tions. In the majority of the previous literature it is impossible to
test the veracity of this supposition as only Krien et al. (1970) has
reported any details about the methodology of the FEP efficiency
calibration. The details provided by Krien et al. (1970) have
26 S.M. Collins et al. / Applied Radiation and Isotopes 102 (2015) 15–28
Table 13
Comparison of published experimental probabilities of the α transition to those deduced from the P(γ þ ce) balance determined from the evaluations of the DDEP (Chechev,
2011a) and the measured γ-ray emission probabilities determined in this work.
α-particle energy/MeV Feeding level/keV This work Chechev (2011c)
Iα Iα
5.747 126.77 10.94(24) 10.0(3)
5.716 158.64 52.0(8) 49.6(9)
5.607 269.48 24.3(3) 25.8(6)
5.540 338.27 9.70(11) 10.60(17)
5.502 376.26 0.67(2) 0.74(3)
5.434 445.03 þ 446.82 2.19(6) 2.10(9)
allowed investigation of the difference in the FEP efficiency cali-
brations to be made.
5.3. Krien et al. (1970)
Krien et al. (1970) describe that the FEP efficiency calibration
was performed using a 75Se source to determine a relative effi-
ciency over the photon energy range 66–400 keV. While 75Se is
commonly used for performing efficiency calibrations of HPGe γ-
ray spectrometers, there are significant problems with the use of
75
Se in this case, where true coincidence summing corrections can
play an important role; this can be a source of significant error.
The relative γ-ray emission probabilities of 75Se used by Krien
et al. (Krien et al., 1970) differ significantly from the evaluated
relative γ-ray emission probabilities determined by Negret and
Singh (2013). The differences in the relative γ-ray emission prob-
abilities for the most intense γ-rays range from 0.4% to 14%.
To estimate the effect of this difference in the nuclear data, a
75
Se source was measured on BART. The relative FEP efficiencies
were determined using both sets of emission probabilities (as-
suming that the 125.99 keV γ-ray reported by Krien et al. (1970)
was a typographical error and should represent the 136.0 keV γ-
ray). The relative FEP efficiency calibration points were fitted with
a 3rd order and 4th order polynomial for the Krien et al. (1970) and
Negret and Singh (2013) data respectively. The respective fits of
the data points are shown in Fig. 11. It was observed that the FEP
efficiency calibration, relative to the 269.5 keV, using the Krien
et al. (1970) 75Se nuclear data was significantly different to that
using the nuclear data evaluated by Negret and Singh (2013).
Using the two FEP efficiency curves, an estimate of the effect on
the γ-ray emission probabilities reported by Krien et al. (1970) was
made, and the results are shown in Fig.12. While the estimated
corrections do not completely resolve the differences in the re-
ported values, the z-scores (Devore, 2011) show an improvement
in the agreement between the γ-ray emission probabilities of Krien
et al. (1970) and this work.
The analysis shows that the inaccuracy of the nuclear data used
and hence the accuracy of the FEP efficiency calibration can lead to
a significant error on the accuracy of the final values. No additional
information is available at this time to investigate further the
differences in the values e.g. true-coincidence summing effects.
Hence, these corrections should only be considered as qualitative
and not as a ‘true’ correction.
5.4. α-transitions
As previously reported by Kellett and Nichols (2013) there is
disagreement between the P(γ þ ce) deduced α transition prob-
abilities and those determined experimentally (Davidson and Con-
nor, 1970a; Giannini et al., 1962b; Pilger, 1957; Walen et al., 1962). A
Pilger (1957) Giannini et al. Walen et al. (1962) Davidson and Connor (1972)
(1962b)
Iα Iα Iα Iα
10.5(–) 10.2(–) 8.85(18) 9.50(58)
50.4(–) 48.0(–) 52.2(11) 52.8(8)
23.6(–) 25.7(–) 25.3(5) 24.2(4)
10.3(–) 10.2(–) 8.85(18) 9.16(30)
0.86(–) 1.3(–) 0.78(–) 1.00(15)
2.4(–) 2.5(–) 2.24(–) 2.27(20)
provisional rebalancing of the decay scheme was attempted to see if
the observed differences in absolute γ-ray emission probabilities
could explain this disagreement. The energy, positioning and mul-
tipolarity of the γ-ray transitions were taken from the Evaluated
Nuclear Structure Data File (ENSDF) adopted levels and γ-rays
(Browne, 2001). The total conversion coefficients were recalculated
using the BrIcc code v2.3S (Kibédi et al., 2008) based on the adopted
energies and multipolarities. For γ-rays not observed in this work,
relative intensities were also adopted from ENSDF. The compiled
data were assembled into an ENSDF file and processed with the
GABS code (Browne and Baglin, 2004). The results are presented in
Table 13. The differences between the α-emission probabilities thus
calculated and the experimental data are generally lower than
previously reported (Chechev, 2011c). It should be noted these va-
lues are purely indicative and a full decay-scheme evaluation in-
corporating data from this work as well as that from PTB (Kossert
et al., 2015) is merited to determine to what extent the dis-
crepancies have been truly resolved.
6. Conclusion
A total of 83 absolute γ-ray emission probability values have
been determined experimentally for the 223Ra decay series, de-
rived from absolute standardisations of two 223Ra solutions. The
use of two high accuracy calibrated HPGe γ-spectrometers has
allowed the precise measurement of the γ-ray emission prob-
abilities presented, with reduced quoted uncertainties of the most
significant γ-ray emissions of the decay series. In this work sig-
nificant discrepancies have been shown in the current absolute γ-
ray emission values of 223Ra decay series.
Correlation coefficient matrices have been presented for the
first time showing the correlations within the uncertainties be-
tween pairs of the most intense γ-ray emission probabilities of
223
Ra and decay progeny, to enable improved decay data evalua-
tions in the future.
The differences in the experimental α-transition probabilities
and those deduced by the P(γ þ ce) balance using the γ-ray emission
probabilities are reported here and show some improvement in
the agreement to the published experimental values of Davidson
and Connor (1970a). There remains a paucity of precise mea-
surements of the α-transition probabilities and further investiga-
tion of the 223Ra nuclear data is recommended to finalise the 223Ra
decay scheme.
Disclaimer
Identification of commercial services or products does not
imply recommendation or endorsement by the National Physical
 S.M. Collins et al. / Applied Radiation and Isotopes 102 (2015) 15–28
Laboratory, nor does it imply that the products or services iden-
tified are necessarily the best available for the purpose.
Acknowledgements
The authors would like to thank Andrew Fenwick, Kelley Fer-
reira and Lynsey Keightley for the preparation of the samples,
Dr. Stefaan Pommé for providing the figure of the decay scheme,
Mily Brewer for translation of texts and Algeta ASA (Norway) for
provision of the 223Ra solutions. This work was supported by the
National Measurement System Programmes Unit of the UK's De-
partment for Business, Innovation, and Skills and the European
Metrology Research Programme (EMRP). The EMRP is jointly
funded by the EMRP participating countries within EURAMET and
the European Union.
References
Aitken, A.C., 1936. On least squares and linear combination of observations. Proc. R.
Soc. Edinb. 55, 42–48.
Bateman, H., 1910. The solution of a system of differential equations occurrig in the
theory of radioactive transformations. Proc. Camb. Philos. Soc. 15, 423–427.
Bayer, 2012. Radium-223 Dichloride: Bayer responses to NRC questions, Bayer.
Berger, M.J., Hubbell, J.H., Seltzer, S.M., Chang, J., Coursey, J.S., Sukumar, R., Zuck-
er, D.S. , Olsen, K., 1998. XCOM: Photon Cross Section Database, NIST Physical
Measurement Laboratory.
BIPM, 2004. Monographie BIPM-5-Table of Radionuclides, Seven Volumes, CEA/
LNE-LNHB, 91191 Gif-sur-Yvette, France and BIPM, Pavillon de Breteuil, 92312
Sèvres, France. 〈http://www.nucleide.org/DDEP.htm〉.
BIPM, 2008. Evaluation of measurement data – Guide to the expression of un-
certainty in measurement, JCGM 100:2008, Bureau International des Poids et
Mesures.
BIPM, 2011. Evaluation of measurement data – Supplement 2 to the "Guide to the
expression of uncertainty in measurement" – Extension to any number of
output quantities, JCGM 102:2011, Bureau International des Poids et Mesures.
Blaton-Albicka, K., Kotlinska-Filipek, B., Matul, M., Stryczniewicz, K., Nowicki, M.,
Ruchowska-Lukasiak, E., 1976. Precision gamma-ray spectroscopy of the decay
of 223Ra and its daughter products. Nukleonika 21, 935–947.
Briançon, C., Leang, C.F., 1968. Schema des niveaux de 219Rn. J. Phys. Colloq. 29,
184–185.
Briançon, C., Leang, C.F., Walen, R., 1968. Etude du Spectre gamma Emis par le
Radium-223 et Ses Derives. C.R. Acad. Sci. (Paris) 266B, 1533–1536.
Broda, R., Cassette, P., Kossert, K., 2007. Radionuclide metrology using liquid scin-
tillation counting. Metrologia 44, S36–S52.
Browne, E., 2001. Nuclear Data Sheets for A ¼ 215, 219, 223, 227, 231. Nucl. Data
Sheets 93, 763–1061.
Browne, E., Baglin, C., 2004. ENSDF–Evaluations: Methodology and Worked Ex-
amples On Nuclear structure and decay data: Theory and Evaluation Manual,
Part 2, 13.
Bruland, O.S., Jonasdottir, T.J., Fisher, D.R., Larsen, R.H., 2008. Radium-223: from
radiochemical development to clinical applications in targeted cancer therapy.
Curr. Radiopharm. 1, 203–208.
Cessna, J.T., Zimmerman, B.E., 2010. Standardization of radium-223 by liquid scin-
tillation counting. Appl. Radiat. Isot. 68, 1523–1528.
Chechev, V.P., 2011a. Table of Radionuclides (Vol. 6 – A ¼ 22 to 242), Monographie
BIPM-5, Bureau International des Poids et Mesures, pp. 79–84.
Chechev, V.P., 2011b. Table of Radionuclides (Vol. 6 – A ¼ 22 to 242), Monographie
BIPM-5, Bureau International des Poids et Mesures, pp. 85-89.
Chechev, V.P., 2011c. Table of Radionuclides (Vol. 6 – A ¼ 22 to 242), Monographie
BIPM-5, Bureau International des Poids et Mesures, pp. 125-138.
Collins, S.M., Pearce, A.K., Ferreira, K.M., Fenwick, A.J., Regan, P.H., Keightley, J.D.,
2015. Direct measurement of the Half-Life of 223Ra. Appl. Radiat. Isot. 99, 46–53.
Cothern, C.R., Connor, R.D., 1965. Studies in the active deposit of actinium. Part II:
the decay of 211Pb (AcB). Can. J.Phys. 43, 383–403.
Cox, M.G., 1972. The numerical evaluation of B-Splines. IMA J. Appl. Math. 10,
134–149.
da Silveira, E.F., de Pinho, A.G., de Barros Leite, C.V., 1971. Esquema de Niveis Do
Bi211. An. Acad. Bras. Cienc. 43, 97.
Dalmasso, J., Maria, H., 1967. Gamma radiation of 219Rn, 215Po and 211Pb. C.R. Acad.
Sci (Paris) 265B, 822–825.
Davidson, W.F., Connor, R.D., 1970a. The decay of 223Ra and its daughter products.
(I): the decay of 223Ra. Nucl. Phys. A 149, 363–384.
Davidson, W.F., Connor, R.D., 1970b. The decay of 223Ra and its daughter products.
(II): the decay of 219Rn and 215Po. Nucl. Phys. A 149, 385–391.
Davidson, W.F., Cothern, C.R., Connor, R.D., 1967. Studies in the decay of the active
deposit of actinium. III: Levels in 211Bi and its daughter products. Can. J. Phys.
45, 2295–2313.
27
Devore, J.L., 2011. Probability and Statistics for Engineering and Science, eighth ed..
Brooks/Cole, California, pp. 346–348.
Giannini, M., Prosperi, D., Sciuti, S., 1962a. Decay scheme of 211Pb. Nuovo Cim. 25,
1227–1238.
Giannini, M., Prosperi, D., Sciuti, S., 1962b. Intensity measurements of alpha groups
from 211Bi, 211Po, 219Rn and 223Ra by means of solid state counter techniques.
Nuovo Cim. 25, 1314.
Gorodetzky, S., Beck, F.A., Byrski, T., Knipper, A., 1968. Détermination de moments
angulaires et de mélanges multipolaires dans le 211Bi. Nucl. Phys. A 117,
208–230.
Hamilton, W.D., Davies, K.E., 1968. The low energy levels of 211Bi. Nucl. Phys. A 114,
577–588.
Harms, A.V., Jerome, S.M., 2004. On the integrated decay and ingrowth equations
used in the measurement of radioactive decay families: the general solution.
Appl. Radiat. Isot. 61, 367–372.
Hesselink, W.H.A., 1972. Spin Assignments in the Transitional Nuclei 215Po, 219Rn,
and 223Ra from Alpha-Gamma Directional Correlation Measurements, NP-
19781.
Hindi, M.M., Adelberger, E.G., Kellogg, S.E., Murakami, T., 1988. Search for the l-
forbidden beta decay 207Tl–4 207Pbn(570 keV). Phys. Rev. C 38, 1370–1376.
ISO, 2010. Injection Equipment for Medical Use – Part 1: Ampoules for Injectables,
ISO 9187-1:2010, International Organization for Standardization.
Kathren, R.L., 1998. NORM sources and their origins. Appl. Radiat. Isot. 49, 149–168.
Keightley, J.D., Park, T.S., 2007. Digital coincidence counting for radionuclide stan-
dardization. Metrologia 44, S32–S35.
Keightley, J.D., Pearce, A.K., Fenwick, A.J., Collins, S.M., Ferreira, K.M., Johansson, L.E.,
2015. Standardisation of 223Ra by liquid scintillation counting techniques and
comparison with secondary measurements. Appl. Radiat. Isot. 95, 114–121.
Keightley, J.D., Watt, G.C., 2002. Digital coincidence counting (DCC) and its use in
the corrections for out-of-channel events in 4(π)(β)-(γ) coincidence counting.
Appl. Radiat. Isot. 56, 205–210.
Kellett, M.A., Nichols, A.L., 2013. Library of Recommended Actinide Decay Data
International Atomic Energy Agency, Vienna, Austria.
Kibédi, T., Burrows, T.W., Trzhaskovskaya, M.B., Davidson, P.M., Nestor Jr, C.W.,
2008. Evaluation of theoretical conversion coefficients using BrIcc. Nucl. In-
strum. Methods Phys. Res. Sect. A: Accel. Spectrom. Detect. Assoc. Equip. 589,
202–229.
Kondev, F.G., 2013a. Table of Radionuclides (Vol. 7 – A ¼ 14 to 245), Monographie
BIPM-5, Bureau International des Poids et Mesures, pp. 135–141.
Kondev, F.G., 2013b. Table of Radionuclides (Vol. 7 – A ¼ 14 to 245), Monographie
BIPM-5, Bureau International des Poids et Mesures, pp. 113–118.
Kossert, K., Bokeloh, K., Dersch, R., Nähle, O., 2015. Activity determination of 227Ac
and 223Ra by means of liquid scintillation counting and determination of nu-
clear decay data. Appl. Radiat. Isot. 95, 143–152.
Krien, K., Günther, C., Bowman, J.D., Klemme, B., 1970. Experimental investigation
of the decay 223Ra - 219Rn. Nucl.Phys. A 141, 75–96.
Kutschera, W., Ahmad, I., Armato, S.G., Friedman, A.M., Gindler, J.E., Henning, W.,
Ishii, T., Paul, M., Rehm, K.E., 1985. Spontaneous 14C emission from 223Ra. Phys.
Rev. C 32, 2036–2042.
Liang, C.F., Paris, P., Sheline, R.K., 1998. Level structure of Pb-211. Phys. Rev. C 58,
3223–3227.
Luca, A., 2010. Table of Radionuclides (Vol. 5 – A ¼ 22 to 244), Monographie BIPM-
5, Bureau International des Poids et Mesures, pp. 41–46.
Luca, A., 2011. Table of Radionuclides (Vol. 6 – A ¼ 22 to 242), Monographie BIPM-
5, Bureau International des Poids et Mesures, pp. 79–84.
Mead, R.O., Draper, J.E., 1965. Energy Levels of Bi-211. Phys. Rev. C 139, B9–B16.
Michalski, J., Sartor, O., Parker, C., Shan, M., Garcia-Vargas, J., Aksnes, A., Vogelzang,
N.J., 2013. Radium-223 dichloride (Ra-223) impact on skeletal-related events,
external beam radiation therapy (EBRT), and pain in patients with castration-
resistant prostate cancer (CRPC) with bone metastases: updated results from
the phase 3 ALSYMPCA trial. Int. J. Radiat. Oncol. Biol. Phys. 87, S108–S109.
Nichols, A.L., 2011. Table of Radionuclides (Vol. 6 – A ¼ 22 to 242), Monographie
BIPM-5, Bureau International des Poids et Mesures, pp. 95–103.
Negret, A., Singh, B., 2013. Nuclear Data Sheets for A ¼75. Nucl. Data Sheets 114,
841–1040.
Nilsson, S., Franzén, L., Parker, C., Tyrrell, C., Blom, R., Tennvall, J., Lennernäs, B.,
Petersson, U., Johannessen, D.C., Sokal, M., Pigott, K., Yachnin, J., Garkavij, M.,
Strang, P., Harmenberg, J., Bolstad, B., Bruland, Ø.S., 2007. Bone-targeted ra-
dium-223 in symptomatic, hormone-refractory prostate cancer: a randomised,
multicentre, placebo-controlled phase II study. Lancet Oncol. 8, 587–594.
Parker, C.C., Pascoe, S., Chodacki, A., O’Sullivan, J.M., Germá, J.R., O’Bryan-Tear, C.G.,
Haider, T., Hoskin, P., 2013. A randomized, double-blind, dose-finding, multi-
center, phase 2 study of radium chloride (Ra 223) in patients with bone me-
tastases and castration-resistant prostate cancer. Eur. Urol. 63, 189–197.
Parker, J.L., 1984. The Use of Calibration Standards and the Correction for Sample
Self-attenuation in Gamma-ray Nondestructive Assay, LA 10045, Los Alamos
National Laboratory, pp.159–194.
Pilger, R.C., 1957, Nuclear Decay Schemes in the Actinium Family (Thesis), UCRL-
3877, University of California .
Rajput, M., 2010. Random coincidence summing of gamma rays and dead time
count loss corrections in HPGe-detector based gamma ray spectrometry.
Radiochim. Acta 98, 99–107.
Rose, H.J., Jones, G.A., 1984. A new kind of natural radioactivity. Nature 307,
245–247.
Rosman, K.J.R., Taylor, P.D.P., 1998. Isotopic compositions of the elements. Pure
Appl. Chem. 70, 217–235.
28 S.M. Collins et al. / Applied Radiation and Isotopes 102 (2015) 15–28

Sargent, B.W., 1939. The β- and g-rays of actinium B and actinium C. Can. J. Res. 17a, A ¼ 151 to 242), Monographie BIPM-5, Bureau International des Poids et Me-
82–102. sures, 1–19.
Sheline, R.K., Liang, C.F., Paris, P., 1998. Configurations and level structure of Rn-219. Walen, R.J., Nedovessov, V., Bastin-Scoffier, G., 1962. Spectrographie α de ∗223Ra
Phys. Rev. C 57, 104–112. (AcX) et de ses dérivés. Nucl. Phys. 35, 232–252.
Vandenbosch, S.E., Baba, C.V.K., Christensen, P.R., Nielsen, O.B., Nordby, H., 1963.
The decay scheme of Pb211. Nucl. Phys. 41, 482–496.
Vanin, V.R., de Castro, R.M., Browne, E., 2004. Table of Radionuclides (Vol. 2 –