Lowenthal G., Airey P. Practical Applications of Radioactivity and Nuclear Radiations

Подождите немного. Документ загружается.

geometry (Section 3.4.5). However, there are many applications when g rays

have to pass through attenuating material before reaching the detector, and

in such cases Compton scatter is unavoidable.

When g rays are scattered in low Z number materials (Z < 15), signi®cant

Compton scatter begins at energies above about 50 keV (Figure 3.7(a)), when

g rays de¯ected from the beam can be scattered back into the detector so

reducing the peak-to-total ratio.

If the intensities of g rays transmitted in broad beam geometry (Figure

4.6(b)) are to be measured reproducibly via the countrate in the full energy

peak, two points have to be noted. (1) Since Compton scatter reduces the

peak-to-total ratio of the spectrum, it must be kept constant during a series of

comparisons by ensuring a constant environment. (2) Results obtained using

published attenuation coef®cients, e.g. those shown in Figure 3.7(a), could

overestimate the attenuation since they assume narrow beam geometry.

Build-up is de®ned as the ratio of the g ray countrate following attenuation

in broad beam geometry (Figure 4.6(b)) to the countrate that would have

been observed for attenuation in narrow beam geometry (Figure 4.6(a)).

4.4 Gamma ray counting 111

Figure 4.6. Gamma ray attenuation in narrow and broad beam geometry. (a)

Narrow beam geometry. A collimator prevents g rays scattered out of the beam by

Compton interactions from reaching the detector. (b) Broad beam geometry when

Compton scattered g rays could be scattered back into the detector. (c) Gamma rays

attenuated in iron in both narrow beam and broad beam geometry, illustrating

build-up of the broad beam.

Build-up factors for various materials are available in the literature (e.g.

Shleien, 1992). These factors increase with the thickness of absorbing material

as shown in Figure 4.6(c) for iron. Every gamma ray scattered prior to

detection reduces the peak-to-total ratio in the spectrum as shown in Figures

4.7(a) and (b).

Figure 3.15 demonstrates the effect of a high rate of Compton scatter on

the appearance of spectra. The Compton region of the spectrum, i.e. the

Nuclear radiations: a user's perspective112

Figure 4.7. Pulse height spectra due to technet ium-99m g rays. Peak heights

were normalised to a maximum of 20 000 counts per interval. (a) Spectrum due to

a thin source in air and in narrow beam geometry, and (b) with the

99m

dissolved in a water phantom when the peak-to-total ratio is greatly reduced by

Compton scatter (courtesy of S. Eberl, Royal Prince Alfred Hospital, Sydney,

Australia).

section between the backscatter peak and the Compton edge (Figure 3.6(f ))

is, in this case, a much more prominent feature of the spectrum than the 411.8

keV full energy peak. In a scatter-free environment the peak-to-total ratio at

412 keV is close to 0.6 (Figure 4.5) whereas it is only about 0.12 in Figure

3.15.

4.5 Corrections and precautions, part 1

4.5.1 A summary

Results of measurements of ionising radiations are nearly always incom-

plete and could be misleading until corrected for a number of effects,

notably radionuclide decay and unwanted or lost counts. Here it is only

practical to deal with the most commonly occurring corrections which

include:

1. dead time corrections

2. corrections for pulse pile-up, random and coincident summing

3. decay corrections

4. corrections to account for unwanted radiations.

Corrections 1 to 3 will be dealt with in this section and the fourth group in

is covered in Section 4.6. The order does not imply an order of importance.

4.5.2 Dead time corrections

If nuclear radiations trigger a pulse counter, they initiate a train of events

which has to go to completion before the counter can respond to another

ionising event (Section 5.3.3). The time during which it cannot process

incoming radiations is known as its dead time. Ionisation chambers operating

in the saturation region (Section 5.2.1) are not subject to dead time as de®ned

here. Semiconductor detectors are not subject to dead time either, but they

are subject to pile-up and summing effects.

Ionisation chambers are not subject to dead time because the electric ®eld

across the chamber continues to collect ions formed in the gas regardless of

what went on before. Errors can occur if the ®eld is not strong enough to

collect all the ions before they recombine, so causing the extent of the

ionisation to be underestimated.

The dead times of GM detectors range from about 50 to over 300 ms

(Section 5.3.2). The dead times of proportional counters similar to those

shown in Figures 5.2 or 6.1 are likely to be in the range 2 to 4 ms (Section

4.5 Corrections and precautions, part 1 113

5.3.3). The dead time of NaI(Tl) crystals is only about 1 ms while that of

organic scintillators (Section 4.2.4) can be shorter than 0.1 ms.

If a radionuclide is available which has a well known and suf®ciently short

half life, one can measure the effective dead time for a detector as follows.

Beginning with a high countrate which is subject to a large dead time

correction, one takes periodic counts down to countrates where the dead time

correction is small enough to be neglected. Using the known decay rate of the

radionuclide, one can calculate the `true' countrate for each measured result

and obtain an average value for the dead time t using Eq. (4.2).

An accurately known dead time can be important when countrates

(>500 cps) have to be accurate within about 2%. If so, equipment should be

available to make the dead time adjustable and closely reproducible. This can

be done using a commercially available dead time gate (marked DT in Figure

4.2). For other methods of measuring dead times readers are referred to

NCRP (1985, Section 2.7).

As a rule, dead times can be assumed to be non-extending, being

unaffected by ionising radiations entering the detector while it is dead.

Corrections for nonextending dead time losses can be calculated from the

simple expression

N=N'/(17N't), (4.1)

where N cps is the corrected countrate, N' cps the observed countrate and t is

the dead time in seconds. If both N and N' are known t can be calculated by

rearranging Eq. (4.1) to obtain:

t =(N7N')/NN' (4.2)

Dead time corrections should always be applied before other corrections

are applied.

4.5.3 Pulse pile-up, random and coincidence summing

Randomly occurring effects

Pile-up occurs in the ampli®er. Unlike pulse counters, ampli®ers remain open

to incoming signals regardless of what went before, but their response is

subject to a resolving time. Two pulses arriving within the resolving time are

processed together leading to the pile-up and/or summing effects respectively

shown in Figures 4.8(a) and (b). A resolving time close to 1 ms is short enough

to avoid dif®culties due to pile-up until countrates exceed 2000 to 3000 cps,

depending on circumstances.

Nuclear radiations: a user's perspective114

Pulse distortions due to pile-up distort the spectrum while random pulse

summing can cause counts to be lost as two pulses combine into a single

pulse. If possible, these effects should be avoided by reducing countrates,

since they are dif®cult to correct for. An example of pulse summing is shown

in Figure 4.8(b) and Figure 4.8(c) shows a sum peak obtained with a

source and also the continuation of the summing region to lower energies

(this section of the spectrum has to be calculated).

The spectrum in Figure 4.8(c) was taken in conditions where there is both

random and coincidence summing. Random summing could have been

minimised by reducing the countrate, e.g. by increasing the source±detector

distance. Although coincidence summing can in this case be reduced in the

same way, it is normally subject to other rules (see below).

Effects due to random summing are also reduced or avoided by reducing

the ampli®er resolving time. A relatively long resolving time, say 4 ms,

improves the energy resolution of the detector, but it leads to a higher

probability for pile-up. If high countrates are more important for an

application than high-energy resolution, the resolving time should be as short

4.5 Corrections and precautions, part 1 115

Figure 4.8. (a) Random pulse pile-up before and after pulse shaping by the

ampli®er. (b) Pulse superposition of two pulses to pro duce a sum pulse. (c) Pulse

summing in a NaI(Tl) detector for g rays from yttrium-88. The sum spectrum below

about 2 MeV was obtained by calculation, allowing for the source±d etector

geometry.

Channel number

as practical. High-class spectroscopy ampli®ers incorporate pile-up rejector

and base line restorer circuits to effect the type of improvement shown in

Figure 4.8(a) (Mann et al., 1980, Ch. 7).

Countrate losses due to random summing in the ampli®er are indistinguish-

able from losses due to dead time in pulse counters. Corrections can be made

using a 50 cps precision pulser. The pulses are injected into the preampli®er

(Figure 4.2) and the output rate is measured using an SCA and scaler or an

MCA. For most routine applications, the percentage loss in the countrate

obtained from the spectrum can be taken as equal to that observed for the

pulser for which the `normal' rate is known.

Coincidence summing

The daughters of many radionuclides emit two or more g rays within

picoseconds or less following the decay (Figure 3.4(b)). Although the

directions in which these g rays are emitted are often correlated, the

assumption of randomness will rarely cause signi®cant errors during indus-

trial applications.

If two coincident g rays are both detected, the resultant pulses will sum

since they arrive at the detector well within its resolving time. While random

summing is a function of the rate at which photons reach the detector,

coincidence summing is a consequence of the decay scheme of the radio-

nuclide and of the detection ef®ciency.

The ef®ciency of radiation detectors is the higher the larger their surface

and the smaller the source±detector distance. If

Co radiations are detected

in a large well-type NaI(Tl) crystal, the countrate, which in this case would be

strongly affected by coincidence summing, could be lower than 60% of the g

ray emission rate. By contrast, if the source±detector distance exceeds about

30 cm, the detection of more than one of the coincident g rays is highly

improbable. For a 50650 mm detector the probability to detect two coin-

cident g rays is suf®ciently small to be ignored for many purposes, even if the

source±detector distance is only 15 cm. To avoid the need for what are likely

to be uncertain corrections, counting should be carried out in conditions

when coincidence summing is negligible.

Yttrium-88 decays are followed by two coincident g rays, as are

decays. The conditions shown in Figure 4.8(c) are clearly such that the g

rays are subject to coincidence summing as well as the random summing

mentioned earlier. A comprehensive discussion of summing effects is given

by Debertin and Helmer (1988, Sections 4.5 and 4.6); they deal with

semiconductor detectors, but much is applicable to work with NaI(Tl)

detectors.

Nuclear radiations: a user's perspective116

4.5.4 Decay corrections

Results of radioactivity measurements must refer to a reference time T

which has to be stated the more accurately the shorter the half life of the

radionuclide. If T

and T

1/2

are known, a result can always be corrected to tell

operators what it would have been at the time of interest, say T

. Such

corrections are most easily made if T

, T

and T

1/2

are expressed as fractions

of the year (365.25 d) rather than in the conventional way.

A measured countrate N

is corrected to what it was or will be at some

other time T

, using equations (1.6) and (1.7) (Section 1.6.2) when it can be

shown that N

, the countrate at time T

is related to N

, the countrate at the

reference time T

by:

= N

exp7(0.693 Dt /T

1/2

) (4.3)

where Dt = T

7 T

.IfT

precedes T

, Dt becomes negative when the exponent

becomes positive. The uncertainty in this correction is the greater the greater

the ratio (T

7 T

)/(T

1/2

). T

could be chosen at any convenient time but

preferably at T

, the time when counting starts, or at the mean of the

counting time, especially so if the half life of the radionuclide is not well

known.

4.6 Corrections and precautions, part 2

4.6.1 Unwanted radiations, a summary

Ionising radiations emitted by the radionuclide of interest never have the ®eld

to themselves. They are invariably accompanied by radiations from other

sources, here called unwanted radiations which fall into one or more of three

groups:

radiations from radioactive daughters or parents (including isomers)

radionuclidic impurities

background radiations due to a variety of sources.

4.6.2 Radioactive parents and daughters

When using a radioactive parent, its radioactive daughter could be respon-

sible for unwanted radiations. To what extent this is so depends on the ratio

between their half lives and this should always be veri®ed.

For cases (b) and (c) in Section 1.6.1 and Figure 1.7, the parent cannot be

4.6 Corrections and precautions, part 2 117

used free from radiations due to the radioactive daughter. The daughter

could be eluted and used on its own, but regrowth into the parent is too fast

to permit the latter to be used free of daughter activity. If both parent and

daughter are present, one should make measurements only when the two

activities are in equilibrium (if possible), to help with avoiding errors when

making decay corrections.

Parent activities are readily used if the half life of a recently eluted daughter

is long enough to keep its activity negligibly small. This is so for two

frequently used radionuclides, technetium-99m (T

1/2

= 6.0 h) and americium-

241 (T

1/2

= 432 y). The daughters are the b

particle emitting technetium-99

1/2

= 2.14610

years), and the a particle emitting neptunium-237

1/2

= 2.14610

years). Their percentage contribution to the decay rate of

the parents is suf®ciently small to be disregarded.

4.6.3 Radionuclidic impurities

The magnitudes of the impurities dealt with in this section are assumed to

exceed the uncertainty estimate applying to the measurement (see Section

6.5.2), but not be high enough to be easily identi®able.

Measurements employing radioactive sources could be affected by che-

mical (non-radioactive) impurities, or by radionuclidic (radioactive) impuri-

ties when the radionuclide of interest is contaminated by an unwanted

radionuclide. Chemical impurities could affect radioactivity measurements,

e.g. if they increase source self-absorption or affect the chemical stability of

source material. Radionuclidic impurities are likely to be the more dif®cult to

detect, the more similar their half lives are to that of the principal radio-

nuclide.

Gamma or X ray emitting impurities introduced accidentally when

working with g ray emitters can be identi®ed from their energies when there is

access to a g ray spectrometer, preferably with a semiconducting detector.

For many radionuclidic impurities detection is only possible by monitoring

the half life of the principal radionuclide.

If a radionuclidic impurity is due to a pure b particle emitter with a high

enough activity to cause errors during measurements, it is likely to affect

measurements of the half life of the principal radionuclide. This could be

detected by monitoring its half life as long as practical using the method

which was employed for the application. To what extent such tests could be

successful depends on the respective lengths of the two half lives and the

difference between the b particle energies. If results of half life measurements

give no indications of an impurity, radionuclidic impurities are unlikely to be

Nuclear radiations: a user's perspective118

signi®cant. However, tests to verify radionuclidic purity should be continued

for as long as a radionuclide is employed.

Reputable suppliers provide documentation stating all radionuclidic impu-

rities that were detected but could not be eliminated. Users then must look up

the decay scheme data of any residual impurities (Section 4.2.1), consider the

characteristics of their detection equipment and estimate what has to be done

to avoid errors.

4.6.4 The gamma ray background

Table 2.5 lists natural and man-made background radiations (background

for short), but knowledge of how the background affects the results of activity

or countrate measurements is also important. In addition there could be

unwanted counts due to nearby radioactive sources. In practice, it is g rays

that are responsible for nearly all background counts. Charged particles from

external sources have short ranges in matter and will only penetrate into the

detector if emitted `close by', but their possible presence should not be

overlooked.

The background due to g rays recorded in a built environment is commonly

relatively high. An unshielded 50650 mm NaI(Tl) detector operated for

integral counting in an industrial or hospital laboratory rarely records less

than 200 cps, with rates of over 400 cps quite possible (see below).

A large fraction of the photons that account for the background are

emitted from naturally radioactive materials at energies up to 2.6 MeV

(Tables 2.5 and 4.2). The concentration of naturally occurring radionuclides

(Fig. 1.5), is, on average around 10 parts per million in the soil and in stones

and minerals used for buildings. Radiations escaping from these materials are

nearly always multiple Compton scattered by walls, ¯oors and ceilings prior

to reaching a detector. As a result the initially energetic g rays are reduced in

energy, on average to around 100 keV as is evident in the typical background

spectrum shown in Figure 4.9, obtained with a 50650 mm NaI(Tl) detector

near the centre of a laboratory. However, the loss of g rays energy is

compensated by build-up due mainly to Compton scatter (Section 4.4.4). The

countrates would have been higher near the walls.

Detectors used for g ray applications and especially for g ray spectrometry

should be shielded by 50 mm thick lead bricks shaped to prevent `shine'

through gaps between the bricks. Similar structures serve to minimise

radiation escaping from stores of radionuclides. Adequate shielding should

also be used around liquid scintillation detectors since liquids are also

ef®cient detectors of background g rays.

4.6 Corrections and precautions, part 2 119

Nuclear radiations: a user's perspective120

Table 4.2. Gamma ray energies and intensities emitted by uranium-238,

thorium-232 and their radioactive daughters.

(a)

238

(b) 232

(b)

Daughter Energy

(c)

Intensity Daughter Energy

(c)

Intensity

nuclide (keV) (%) nuclide (keV) (%)

234

Th 63 3.8

228

Ac 338 11

226

Ra 186 3.5

224

Ra 241 4.0

214

Po 295 18

212

Pb 87 7.9

352 35 239 44

214

Bi 609 45

212

Bi 727 7.0

1120 15

208

Tl 510 23

1238 5.8 583 85

1378 3.9 2614 100

1765 15

2204 5.0

(a)

The data in this table are based on NCRP, 1985, Appendix A3.

(b)

Only those daugh ter nuclides are listed which emit g rays at a rate exceeding 3.0%

of the decay rate. The

235

U chain is disregarded, being of too low intensity

(Section 1.4.1).

(c)

Energies and intensities exceeding 10% have been rounded to the nearest

integer.

Figure 4.9. Background spectra obtained with a 50650 mm NaI( Tl) detector.

Curve (a) in a laboratory, with the detector unshielded. The peak near 100 keV is due

to multiple Compton scattered g rays. Curve (b) was obtained after covering the

detector with 5 mm steel; the background rate is reduced by nearly a factor of two.