Nuclearadiationshielding

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 26

ANNUAL

REVIEWS Further
Quick links to online content

NUCLEAR RADIATION SHIELDINGl


By E. P. BLIZARD
Oak Ridge National Laboratory, Oak Ridge, Tennessee

PREFACE

The building of nuclear radiation shields is a necessary part of preparing


for most nuclear research. With the present emphasis on reactors and high­
Access provided by New York University - Bobst Library on 06/06/15. For personal use only.

yield accelerators the shields are both thicker and more efficient by far than
was the case fifteen years ago. Also, we are now much more conscious of the
Annu. Rev. Nucl. Sci. 1955.5:73-98. Downloaded from www.annualreviews.org

the dangers of radiation exposure and consequently take much greater pains
to avoid them.
In spite of improved design, these larger shields are inevitably more
expensive; they require greater space in buildings, which are themselves
expensive; and they are inconvenient, for apparatus must be made to extend
further between radiated and shielded locations. In addition, there are
military applications for which the reactor and its shield must be mobile,
light, and small. For these there is a strong incentive to develop the very
best shield designs.
The changes are strongly noticeable throughout the laboratories of the
world. The old shield walls of five-gallon water cans are nearly all gone.
No longer is it possible to enter radiation areas without express permission
of an operator. Shields are carefully checked for leaks or other inadequacies.
There are federal regulations on permissible radiation levels outside radio­
active materials as they are mailed or shipped. The philosophy that the shield
is a simple adjunct which can be added nearly as an afterthought, and that
it is cheaper to add another foot or two of concrete than to eliminate the need
by careful design, is extant no longer. Instead, at least four handbooks and
two textbooks are being written on the subject, and the material which is to
fill these publications is much sought-after and largely heeded in modern
design.
The present paper is in no sense designed to be a handbook but it is a
report on the fundamental work which has been of most use in preparing the
handbooks. The large body of work which has been done on the subject of
radiation attenuation cannot possibly be described in a single review article,
so only that which has been of use in the presently accepted shield design
methodology will be treated and emphasis will be given to reactors, which
present the greatest problem. Since no new physical phenomena are involved,
interest in shielding lies in the end result. However, some of the work which
must be done in the course of shielding research, such as the development of
sensitive and special detectors and the invention of new mathematical
techniques, will surely be more generally applicable. For example, methods of
1 The survey of literature pertaining to this review was completed in March, 1955.
73
74 BLIZARD

calculating attenuation of -y-rays and neutrons may apply to cosmic ray


shower theory and vice versa.
The scope of the present treatment will include descriptions of the im­
portant radiation sources and of the individual attenuation processes. Fol­
lowing these are the calculations of attenuation in simple homogeneous
shields as well as other problems in shield design, such as penetration after
slant incidence, scattering, and the effects of voids in shields. Following this
come some special methods for translating shield experiments into designs.
Lastly, experiments and actual shield designs will be described briefly.
Access provided by New York University - Bobst Library on 06/06/15. For personal use only.

INTRODUCTION
Annu. Rev. Nucl. Sci. 1955.5:73-98. Downloaded from www.annualreviews.org

Many modern shields, especially reactor shields, are required to attenu­


ate by from 8 to 12 factors of ten. This fact alone sets the problem apart from
most, for it means that the neutron or photon which penetrates the shield
is extremely unusual. The average behavior of radiation may well be of
little or no interest, for processes which are relatively uncommon may be
just those for which the attenuation is least and the process may therefore
dominate in penetration. An example of this lies in the prompt fission neu­
trons in a reactor. Most of these neutrons are born at about 1 Mev energy,
and of course the behavior of the majority determines reactor criticality.
On the other hand, most neutrons that escape from reactor shields are pro­
duced at energies in excess of about 7 Mev in spite of the fact that these
constitute no more than about 2 per cent of all prompt neutrons. This fact
comes about, of course, because of the general trend to lower cross sections
with increased neutron energy.
The exact solution of the shielding problem is at present beyond the
capabilities of even our best machines, except for the simplest cases. Approxi­
mate solutions based on semi-quantitative analysis, however, have· been
found for most problems. Analytic methods are required for calculation of
most of the real situations, and they must be foreshortened for expedience.
This is done by carefully examining the sources of radiation and the attenua­
tion processes to determine which are most likely to be involved in the radia­
tion which penetrates the shield. In addition to the more obvious mechanisms
for shield penetration, there are special situations which arise from unusual
designs. For example, a circulating fuel reactor will have delayed neutrons
released in a heat exchanger and perhaps thereby activate a secondary cool­
ant. If this is not taken account of in the design, the primary shield may well
be nearly superfluous. For these reasons the first part of any design process
consists of a careful examination of the radiation sources.

THE SOURCES OF RADIATION

Since they are by far the most penetrating forms of radiations, only
neutrons and -y-rays or x-rays are considered. The energy distribution is,
of course, very important.
Prompt fission neutrons.-In the fission of U235 approximately 2.5 fast
neutrons are released. For other fissionable isotopes the number varies
NUCLEAR RADIATION SHIELDING 75

somewhat. The U235 thermal fission neutrons are distributed in energy ac­
cording to an empirical formula suggested by Watt (1, 2):

N(E) =
12-'Ire sinh v2E
- e-E 1.

where N(E) = number of neutrons per energy interval, Mev-i, at energy E,


Mev, per neutron produced, e=Naperian logarithm base, E=neutron
energy, Mev.
The expression is not exact but appears to be accurate to about ± 10
per cent throughout the measured region from 0.075 to 17 Mev (1 to 5).
Access provided by New York University - Bobst Library on 06/06/15. For personal use only.

Delayed neutron s.- Del ayed neutrons are not of importance in compari­
son with prompt neutrons except for those cases in which the reactor fuel is
Annu. Rev. Nucl. Sci. 1955.5:73-98. Downloaded from www.annualreviews.org

circulated out of the primary shield. They are released from certain fission
products, each component being monoenergetic with a given radioactive
half life. Hughes' collection of data (6, 7 ) is given in Table I.

TABLE I
DELAYED NEUTRONS

Absolute Yields per 104 Neutrons Emitted


Half Life, Energy, (prompt and delayed)
sec. key
U23li U233 PU239

55.6 250 2.5 1.8 1.4


22. 0 5 70 16.6 5.8 to.5
4.5 1 4 12 2 1. 3 8.6 12.6
1.52 6 70 24. 1 6.2
0.43 400 8.5 1.8 1 1.9

Total 73. 0 24. 2 36. 4

Prompt fission ')'-rays.-In the fission of U235 prompt ,),-rays are given
off, some of which are quite energetic. These have been measured by Gamble
(8,9) to be distributed according to the following empirical formula:
N(E) = 8.2e-E 2.

where N(E) = number of photons per energy interval, Mev-i, at energy E,


Mev, per fission; E=energy of photon, Mev.
The expression above varies by as much as 30 per cent in places from the
actual data. Gamble also reports the following interesting integrals which are
obtained from his experimental data:
fBHev
N(E) dE= 7.59 photons/fission
fBHev
0.2

EN(E)dE = 7.49 Mev/fission


0.1
76 BLIZARD

Fission product "I-rays.-The fission products are in many shielding prob­


lems the dominant source of "I-rays, although they are overshadowed as a
source in a reactor core during operation because of their low energy. The
most energetic common fission product "I-ray is from Rb88, which emits a
2.8 Mev photon. The most successful fit to the fission product "I-emission
data is that formulated by Way & Wigner ( 10):
ret) = 1.26 r1.2, 10 < t < 107 sec. 3.

where r(t) ="I-ray energy emission rate, Mev/sec., after one fission; t=time
after fission, sec.
Access provided by New York University - Bobst Library on 06/06/15. For personal use only.

Two treatises are available which give the emission spectra as functions
of reactor operation history for the total agglomeration of fission products.
Annu. Rev. Nucl. Sci. 1955.5:73-98. Downloaded from www.annualreviews.org

Moteff ( 1 1) has classed 24 nuclides according to energy and calculated the


spectra, and Clark (12) has treated some 40 nuclides similarly except that
he permitted each to retain its identity in the results of his calculation.
Secondary radiation sources.-On interaction with matter, neutrons cause
secondary "I-rays to be emitted. These are either capture "I-rays or inelastic
scattering "I-rays. Of the two, the former are generally the more important.
Capture "I-rays : A discussion of the neutrons capture process (96) and
the concomitant release of capture "I-rays is out of place in this paper, except
for a discussion of their importance in shielding. For this purpose, nuclides
are classed according to the energy distribution and yield of their capture
"I-rays as follows:
Class I: Following capture, the compound nucleus emits all or most of
its excitation energy in a single transition, giving a predomi­
nant single hard "I-ray of energy equal or nearly equal to the
neutron binding energy.
Class I I : Several photons are given off per capture with the energy dis­
tributed between them so that the spectrum appears to cover
the whole energy range up to the binding energy.
Class I I I: A charged particle is given off following capture, so that the
capture "I-rays are weak or nonexistent.

In the first class there are many light elements, except that three light
nuclides fall in the third class. In addition, the magic nuclei are in the first
class, and, effectively, aluminum, iron, nickel, manganese, and copper fall
into this category, although they do have multiple-photon emissions.
In the second class fall the more complicated heavy nuclei, although such
intermediate elements as sodium, magnesium, silicon, phosphorus, sulphur,
chlorine, potassium, and calcium are members.
The third class has two bona fide members, liS and BIO. The former ex­
hibits no capture "I-ray, and the latter only gives a single photon of 0.478
Mev in 93 per cent of the disintegrations ( 13). It is of interest to note, from
a shielding point of view, that both these nuclides have large capture cross
sections, which makes them useful for capture "I-ray suppressors, and they
NUCLEAR RADIATION SHIELDING 77

are often added to shields just for this purpose. The third member of this class
is NU, which gives a proton and no capture -y-ray in 94 per cent of the cap­
tures. In other cases, it gives a total of 10.8 Mev in one, two, or more photons.
I ts capture cross section is low (1. 7 barns) by comparison with lithium and
boron. These properties mitigate the utility of nitrogen as a capture -y-ray
suppressor, although in some special cases it may be useful in this respect.
Inelastic scattering -y-rays: Another important secondary source of -y­
rays is the inelastic scattering of neutrons. This process is most important
with heavy nuclei, for which the cross sections are high ("-'3 barns) and the
Access provided by New York University - Bobst Library on 06/06/15. For personal use only.

thresholds for neutron energy are low ("-'200 kev for tungsten). Not enough
experimental evidence is yet available to say with certainty just how these
Annu. Rev. Nucl. Sci. 1955.5:73-98. Downloaded from www.annualreviews.org

-y-rays are distributed in energy, but in general they would be expected to


follow the same trend as the capture -y-rays. That is, the heavier nuclei will
be expected to give more but softer photons than the lighter ones.

INTERACTION PROCESSES

-y-rays.-It has been pointed out by Fano (14 to 16) and by Goldstein &
Wilkins (17) that there are no less than 12 types of interaction of -y-rays
with matter. Fortunately, only three of these need be studied in connection
with shielding problems. The photoelectric effect dominates for low energy
photons and high atomic number materials (E,<O. 15 Mev for Z =29;
E,<O.5 Mev for Z =82) . Pair production is the dominant effect for high
energy photons and high atomic number materials (E >80, 15, or 5 Mev for
Z = 1, 13, or 82, respectively). In the intermediate energy regions the Comp­
ton process is most important. Two excellent tabulations have been made of
-y-ray cross sections by White (18) and by Davisson & Evans (19).
While a detailed discussion of the -y-ray interaction processes is left to the
references, it is perhaps pertinent to mention the relative advantages from a
shielding point of view of different materials. The cross sections per atom
vary with atomic number for photoelectric, Compton, and pair-production
processes, approximately as Z·, Z, and Z(Z+I), respectively. This means
that on a weight basis the highest elements in the periodic table are best,
but in the energy region in which the Compton process dominates, there is
essentially no difference. An exception is hydrogen, for which ZjA = 1,
whereas for all other elements ZjA �!. For low-energy -y-rays (E:(,0.1 Mev)
the weight advantage is very strong for high Z materials; for high energy
(E >5 Mev) the advantage is definite but not so strong.
Neutrons.-The detailed cross section information which is needed for
the calculation of neutron attenuation is not available. Nevertheless this pat
excuse for lack of progress in neutron calculations is losing much of its original
force. On one hand, our totality of cross section information obviously is
increasing, and new techniques are being used for measurement in the energy
region from 6 to 10 Mev (20), a most pertinent region which was barren for
a long time. Also measurements of differential cross sections are beginning
to appear (21 to 23), On the other hand, progress is being made in developing
78 BLIZARD

the nuclear rationale on the basis of which predictions of differential cross


sections are made. In addition to the many texts in the general field of nu­
clear physics, there has been a project at Nuclear Development Associates,
Inc., under United States Atomic Energy Commission contract, for the
collection, interpretation, and evaluation of fast-neutron experimental data
in order to enable the prediction of cross sections in the region of interest for
shielding (24). Unfortunately much of this work has been rendered obsolete
by the newer "clouded crystal ball model" (25), but the latter will soon be
applied to cross sections of interest in shielding.
Access provided by New York University - Bobst Library on 06/06/15. For personal use only.

An example of the filling in of experimental gaps by application of the


theory is evident in the work of Goldstein (26) in deriving a complete picture
Annu. Rev. Nucl. Sci. 1955.5:73-98. Downloaded from www.annualreviews.org

of the oxygen cross section which he made preparatory to calculations of


neutron penetrations in water (27).
As is implied above, it has been demonstrated that with fission sourC(;S
the neutrons which penetrate thick shields have initial energies of the order
of 8 Mev. This means that the cross section at this energy has particular
significance. On this hypothesis, Welton & Albert (28) have shown that it
is possible, using a single parameter, to describe the additional attenuation,
introduced by lead and iron when slabs of these materials are added to a
water shield. This parameter they called the "effective removal cross sec­
tion." Subsequently many such cross sections have been measured at Oak
Ridge National Laboratory, and a more definite definition has been given
(29). A recent collection of removal cross section data by Chapman & Storrs
(30) shows the close correspondence between the removal cross section and
the total cross section for 8-Mev neutrons (Fig. 1). Roughly, it is found
that the removal cross section, O'rem, is about three-quarters of the total cross
section:
O'rom � 3/40' total. 8 Mev 4.

While this removal cross section applies well to slabs of material placed close
to the source and followed by a thick ("'120 cm.) water (or other hydrog­
enous) shield, other configurations are more common, e.g., those cases in
which the material is distributed uniformly throughout the shield. While
it is true that in general the cross section which applies in this case is about
nine-tenths of that previously described, the situation is more complicated,
and discussion is reserved for a later section.

THE ATTENUATION OF -y-RAYS

Fundamental to all shielding problems is the calculation of the attenua­


tion of radiation (neutrons or -y-rays) from a point isotropic or other simple
source in an infinite homogeneous medium. This is of course only a start on
the problem, for shields and their reactors are seldom homogeneous and never
infinite. The effects of laminations, of boundaries, and of holes in the shield
will remain after the simpler problem is solved.
The basic attenuation probtem.-Given the attenuation coefficients it is a
NUCLEAR RADIATION SHIELDING 79

simple matter to calculate the unscattered photon flux. Thus, for a point
isotropic source in an infinite homogeneous medium, the uncoIlided photon
number flux is:
Se-I'R
N°= -- 5.
41rR2

where S=source strength, photons per unit time; R=distance from source;
IJ. = total linear absorption coefficient for 'Y-rays of the source energy in the
medium.
I t is because of those collisions in which photons are scattered and not ab­
Access provided by New York University - Bobst Library on 06/06/15. For personal use only.

sorbed that the usual calculation of 'Y-ray penetration is complicated. The


Annu. Rev. Nucl. Sci. 1955.5:73-98. Downloaded from www.annualreviews.org

UNCLASSIFIED
2-04.,..057-0-93 Rt

T OTAL CROSS-SECTION VALUE AT a Mev


0
5 (TAKEN FROM LA -1655 )
REMOVAL CROSS-SECTION VALUE DERI V ED
A
FROM MEA SUREMENTS BEHIND COMPOUND �

2 OF THE ELEMENT
'.
REMOVAL CROSS-SECTION VALUE FROM

MEASUREMENTS BEHIND ELEMENT
�= REMOVAL CROSS SECTION OR T OTAL
� CROSS SECTION A S DESIGNATED
0.5
"

t\I
'\. p = DEN SITY OF MATERIAL
-
E
2
0.2
I'\.

H 1\
OJ
1\
""
..... I'.. v
0.05
I--- 0.0852 � ."- �

'!' IT' �" Ie'


I---- AV3 •

j-...L. n
0.02
-."�
0.04
� � "'r;.;
0.005
� 2 5 W 20 50 WO 200 500 4000
At ATOMIC WEIGHT

FIG. 1. Empirical mass attenuation coefficients for fast neutrons and comparable
total coefficients for 8 Mev neutrons. (Total cross sections from Los Alamos Scientific
Report, LA-loSS, "Survey of Average Neutron Total Cross Sections from 3 to 13
Mev," by N. G. Nereson, S. Darden, E. Fryer, E. T. Jurney, O. Merrill, and M. M.
Thorpe.)
80 BLIZARD

totality of the complication is encompassed by the so-called buildup factor,


which is the ratio of the actual intensity to that attributable solely to un­
collided photons. Thus the photon number flux buildup factor is defined by
the following equation:
N(R) B(R)N°(R)
= 6.
where N i s the actual photon number flux for collided as well as uncollided
photons.
It is common to define buildup factors for any of a number of measurable
intensities, such as the energy flux, which is the product of the photon flux
Access provided by New York University - Bobst Library on 06/06/15. For personal use only.

and photon energy integrated over all energies :


Annu. Rev. Nucl. Sci. 1955.5:73-98. Downloaded from www.annualreviews.org

feR) = BE(R)fD(R) 7.

IO(R) EoND(R),
where
= 8.

f° EN(R, E)dE EoN°(R),


B<Eo
feR) = + 9.

E o = source energy per photon, assuming monoenergetic source; E=ener­


gy per photon of scattered radiation.
The dose, for which it is most common to calculate buildup factors, is
the integral over all energies of the product of photon flux and energy and
the energy absorption coefficient for air (or for tissue, which is about the
same). This latter factor is the "poor geometry" attenuation coefficient and is
variously defined, depending on the assumptions concerning secondary
photons. The most common assumptions are that photoelectric and pair­
production processes are genuine absorptions, i.e., that the fluorescent and
annihilation radiations are absorbed almost at once and that the Compton­
scattered photons are not. In this case the energy-absorption coefficient for
source energy Eo is:
EO - E
!La (Eo) = !Lphoto (Eo) + !Lpair(Eo)
+ f (dd{2!L) Compion E df!
4". 0
10.

where (dp./drt)compton is the differential Compton cross�section per unit solid


angle for scattering a photon of energy Eo into a direction for which the
scattered energy is E. The current practice is to define dose in terms of the
rad, which is the amount of radiation which will deposit 100 ergs per gm. of
material. There is no simple relationship between the dose as usually calcu­
lated for energy deposition in air and that for other materials. No great error
i s to be expected, however, in using air dose for tissue dose, since in both
materials the Compton process dominates so that the energy depositions per
gram will be about the same.
The calculation of buildup factors has been accomplished by approxima­
tions which yield extremely good results. The problem was first assayed by a
number of different persons working independently, each group more or less
unaware of the other's work. Perhaps the most notable feature of these early
efforts is the "straight-ahead" approximation. In this, Compton scatterings
NUCLEAR RADIATION SHIELDING 81
'
are assumed to degrade photons approxi
deviation is ignored. This of course leads to an upper limit to penetration.
The first such calculation was made by Grueling (31; see also 32). In addition
to the "straight-ahead" assumption the Klein-Nishina formula for scattering
cross section is replaced by a simpler function which allows separation of the
pre- and postscattering energy variables. Grueling's choice for this latter
approximation fits the slope and magnitude of the actual expression at the
source energy. Others (33 to 37) have suggested different approximations
which have met with varying degrees of success. These analytic approaches
Access provided by New York University - Bobst Library on 06/06/15. For personal use only.

have by now been superseded in importance, particularly by the Spencer­


Fano moments method (38, 39), their main contribution being the introduc­
Annu. Rev. Nucl. Sci. 1955.5:73-98. Downloaded from www.annualreviews.org

tion of the linear approximation to the buildup factor, B(R)"'l + AJ.l.R where
A is a factor of the order of unity. The approximation has since been shown
to be crudely correct only for media which are primarily Compton scatterers,
which is to say that it applies best to photon energies below the minimum in
the total attenuation coefficient. This is, of course, not surprising since it is
for this situation that the derivation was tailored.
The Spencer-Fano "moments" method (38, 39) of calculation of ')'-ray
attenuation, exploited so thoroughly by the groups at Nuclear Development
Associates (17, 40) and the Bureau of Standards, has been by far the most
prolific producer of attenuation data. This method therefore deserves a full
description here. Unfortunately its complexity precludes this, but Goldstein
& Wilkins (17) give a lucid and complete delineation of the method as
well as most of the results.
The calculations are aimed at a calculation of the energy flux as a func­
tion of distance in an infinite homogeneous medium for certain simple sources
or source distributions. Emphasis is put on accuracy at appreciable distances
from the source, that is, from several to 20 mean free paths. The function
which is of calculational interest is the angular energy flux,
-7-7 -7-7
J(r, fl, E) '" EN(r, {l, E) 11.

-t -t
where N( r, 0, E)dQdEdtdS is the number of photons of energy between E
-t
and E+dE, in solid angle dQ about the direction Q, passing in time dt through
-7 -7
an element of surface dS located at r and perpendicular to Q.
It is more convenient to use wavelength, X, than energy and to describe
the direction in terms of the cosine, w, of the angle between photon direction
and radius vector from the source. For the plane isotropic source, the trans­
port equation is
dl(x, A, w)
w + P.(A)J(X, A, w)
dx
0(1 + A' -}. cos"')
fA J J(x,A',w')k(A',A)
-
=
dfl'dA' + S(A,w)o(x) 12.
o 4" 2".

where x =distance from source plane; k(X', X) = Klein-Nishina formula for


Compton scattering; If =angle between 0' and Q (angle of scattering);
82 BLIZARD

S ("11., w) ==source strength, photons per unit time per unit area of source
plane. Of the two delta functions the first insures that the Compton relation
between scattering angle and energy degradation is satisfied and the second
insures that all the sources are in the plane x = O.
The solution of this transport equation in a useful approximation has been
adapted for use with the faster computing machines, and the SEAC at the
Bureau of Standards was chosen for the work. The problem had to be re­
duced to manageable proportions and this was accomplished logically enough
by concentrating the information to be obtained in the interesting spatial,
Access provided by New York University - Bobst Library on 06/06/15. For personal use only.

angular, and energy regions. The angular energy flux is expanded first as a
Annu. Rev. Nucl. Sci. 1955.5:73-98. Downloaded from www.annualreviews.org

Legendre polynomial series:


21 + 1
L -- hex, X)PI(W)
00

lex, X, w) = 13 .
1_0 411"

where
hex, X) == 27r Ii lex, x, fl)PI(w)dw
-1
14.

The zeroth moment, 10, is the energy flux, and II is the energy current den­
sity. If the transport equation (equation 12), is multiplied by Pl(W) and an
integration is carried out over all solid angles, a series of integro-differential
equations is obtained which are satisfied by the coefficients h Next it is
noted that from a recursion formula for the Legendre polynomials it is possi­
ble to express wII in terms of 11+1 and II-I, thus yielding a sequence of integro­
differential equations in which the original direction variable w does not
appear.
The derivative in the transport equation is now eliminated by the intro­
duction of spatial moments,

15.

Thus multiplication of the equations which II satisfies by x71 and integration


over all space yields, provided I, satisfies certain convergence requirements,
what Goldstein & Wilkins (17) describe as a "doubly indexed sequence of
linear integral equations of Volterra type." Certain other simplifications are
adopted, such as assuming monoenergetic sources and removing the delta
function by transforming to spatial moments for the scattered flux alone. The
integral equations are then solved for a chosen set of the moments. The desid­
eratum is Io(x, }.), and this is obtained to good approximation from the 1 == 0
spatial moments up to about n = 10. The total necessary number of moments
varies with the source geometry. It is necessary, in order to obtain a given
moment b7l/, to first obtain bn--l. 1+1 and bn--l. 1--1. (The requirement is void
when a negative index is indicated.) It is fortunate that in the case of the
plane isotropic source about half the moments are identically zero owing to
symmetry.
The reconstruction of the energy flux Io(x, }.) from the calculated mo-
NUCLEAR RADIATION SHIELDING 83

ments must be carefully carried out, since for less than the complete set of
moments the flux solution is not unique. Accordingly, the flux is expanded
about a known function which is a reasonable approximation to the desired
solution. For this function it is simplest to use the unscattered flux. Thus the
total energy flux is equated to the product of the unscattered flux and a
linear combination of appropriately chosen polynomials. Some polynomials
which are well-suited for the expansion have been given by Spencer & Fano
(38), and the appropriate coefficients are linear combinations of the spatial
moments. The results of the work include dose buildup factors and energy
Access provided by New York University - Bobst Library on 06/06/15. For personal use only.

spectra for the gamut of elements and photon source energies from 0.5 to
10 Mev. They have been compared with several experiments (41 to 45)
Annu. Rev. Nucl. Sci. 1955.5:73-98. Downloaded from www.annualreviews.org

and are found to agree extremely well.


A brief description of the results of Goldstein & Wilkins' calculations
(17) is probably in order. For the lowest energy calculated, 0.5 Mev., the
linear buildup factor is a fair approximation for an atomic number Z of be­
tween 30 and 40. For lower Z, the buildups are much higher, and for higher
Z, much lower, as is to be expected on account of the strong Z-dependence
of the photoelectric effect.
At 1 Mev, the linear buildup applies for elements with Z about 45 with
the same, though not so drastic, changes with Z. At 2 Mev the linear buildup
approximation is good only for large attenuations and Z ",45, and the
photoelectric depression of buildup with increasing Z is much less evident,
being strong only above about 40. At 4 Mev, in the low Z-region the trend is
reversed (at large attenuations the buildup is maximum for Z ",60) because
of the overriding importance of the variation of the Compton cross section
slope in energy dependence. At 6 and 8 Mev there is a stronger maximum in
the buildup factor at Z ",70 for large attenuations. For Z's greater than this
the pair production cross section accounts for the lower buildup factor.
Taylor (46) has shown that it is possible to reproduce the buildup factor
calculations of Goldstein & Wilkins (17) using a function which is the sum
of two simple exponentials with three adjustable parameters which he has
calculated to fit the original work. This is not merely interesting for the
brevity of presentation it affords. The exponential form also makes integra­
tions over sources which are distributed in space amenable to analytic attack.
Although the calculating machines were used to obtain the foregoing
results, it has been possible to obtain some analytic expressions (47, 48,49,
50) for the asymptotic trend in attenuation at very great penetrations. In
general, the laws derived are the same as those of the "straight-ahead"
approximation. It is also reassuring that for the case of lead the moment
method and Spencer's semi-asymptotic calculation (48) for large attenuations
"meet" with the same result.
An independent method for calculating '}'-ray attenuation has been used
by Peebles (50). In this work photons are classified according to the number
of scatterings they have experienced, and the transport equation is solved
successively for each component. The work would of course be prohibitive
84 BLIZARD

if many scatterings were permitted and unreliable if too few were encom­
passed. The artifice by which this dilemma is circumvented is the division
of the medium into many slabs, thin enough so that no more than three
scatterings are likely, using the transmitted photons from one slab as the
source for the next. The results of this method agree well with those from
the moment method. It has not, however, enjoyed the wide acceptance of
the latter, probably because the demonstration of its adequacy and reliability
is somewhat more difficult. It does have the advantage, however, that angu­
lar distributions of photons are available from the results, and many useful
Access provided by New York University - Bobst Library on 06/06/15. For personal use only.

calculations are presented for angles of incidence on the slab shield other
than the normal.
Annu. Rev. Nucl. Sci. 1955.5:73-98. Downloaded from www.annualreviews.org

Another technique for attenuation calculations is the stochastic, or Monte


Carlo method (51, 52, 53). In this method, the histories of many particles
(photons) are calculate-d, using the known cross sections and energy degrada­
tion laws. The probabilities for various events in the history (e.g., scattering
through any angle, absorption, etc.) are known from the physical laws;
particular events are chosen from equivalent possibilities in a random
fashion so that if sufficient samples are taken the aggregate probability of
shield penetration wiII be indicated by the fraction of histories in which
penetration is the result. Since in most shields penetration is the rare excep­
tion, it is necessary to adjust the method so that the effort is concentrated on
successful histories. If this is not done, either the work is prohibitive or the
small penetration is not a reliable index of shield performance. To this end.
photons are weighted according to their chance of penetration, and the calcu­
lations are concentrated on the most important histories. Unfortunately it is
difficult to insure that the result is characteristic of the problem and not the
chosen weighting function. The contribution of the method to the solution
of the large attenuation problem has been essentially nil. On the other hand,
it is extremely promising for special problems in which the geometry is com­
plicated and the attenuation not great, e.g., for attenuations :s 104• In this
field it stands alone and promises to become more useful with current de­
velopment. A significant success for the method is demonstrated in the recent
work of Zerby (52) on the problem of attenuation of slant-incident "I-rays in
two-region slab shields. The calculations agree very well with experiments
by Kim, Kennedy & Wyckoff (54), as well as Clifford et at. (55). The superi­
ority of the Monte Carlo method for special problems is attested by the fact
that Zerby's results include energy and directional distributions of the pene­
trating radiation.
An analytic calculation of the angular distribution of photons in a shield
has not been carried out in many cases, but Spencer & Stinson (56) have
calculated this quantity for Co6°-radiation in water. The moments method
discussed above was applied, but more general polynomials were used which
are solutions of third-order differential equations. The weighting function,
instead of being the uncollided flux, is (/.lor)! exp (-/.lor), where the power l
is the previously mentioned Legendre index. Some experiments have been
NUCLE AR R ADI ATION SHIELDING 85

performed at Oak Ridge National Laboratory (57) which seem to confirm


the calculations. In addition, Certaine (58) has calculated the angular dis­
tribution of photons arising from plane isotropic and plane monodirectional
sources. He has avoided the difficulty of slow convergence which the moments
method exhibits for this calculation (attributable, for example, to a a-function
discontinuity in the angle w in the first-scattered beam) by calculating first­
and second-scattered photons separately. The analytic forms from the mo­
ment method for these components are then subtracted off, being replaced
by the exact calculations. The calculations so performed are reported for
Access provided by New York University - Bobst Library on 06/06/15. For personal use only.

several source energies and media and for penetrations up to 10 mean free
paths.
Annu. Rev. Nucl. Sci. 1955.5:73-98. Downloaded from www.annualreviews.org

Special "(-ray attenuation problems.-In practice the ideal conditions as­


sumed in the foregoing discussion are seldom realized. As a consequence,
various schemes are devised to obtain from the basic solutions reasonable
estimates for the more difficult problems.
One complication is that the shield is always finite, and the solution is
usually required in the region of the outer boundary. The effect of this termi­
nation is to remove the back-scattered radiation. Since this is seen intuitively
to be low in number and energy, in view of the nature of the Compton scatter­
ing process, no correction is made with ,,(-rays for this effect. Measurements
of 'Y-ray dose (59) in the region of the boundary of a water shield confirm
this tenet. For materials of higher Z the effect would be even less.
Shields are not always made of a single element. The calculational meth­
ods such as the moments method can of course be applied to a homogeneous
mixture of elements with cross sections from appropriately weighted contri­
butions of the constituent elements. However, Goldstein & Wilkins (17)
offer a convenient substitute method. They propose that the cross section of
the mixture be approximated by that of some intermediate element for which
the buildup factors are already calculated. It is demonstrated that for 2-
Mev photons a 50 per cent by weight mixture of lead and water is well ap­
proximated by the element gadolinium for which Z 64. A moments method
=

calculation for the mixture agrees to within a few per cent with the buildup for
gadolinium up to 15 mean free paths.
If the shield is made up of successive layers of different materials the
calculations by the moments method do not apply. Peebles' methcd (51)
or a Monte Carlo method (52, 53) would, however, not be hampered by this
feature, although no compilation of buildup factors for these cases is at
hand or contemplated. (It would fill many handbooks.) The angular distri­
butions of photons as calculated by Certaine (58) could be used to solve the
problem of multilayered shields, since each region could be separately calcu­
lated, the leakage from one region serving as the source for the next. However,
the complexity of the calculation is so great that such an effort is not
warranted. Instead, rather simple rules of thumb are suggested.
For the case in which the last region is a fairly thick one (at least several
mean free paths) it is reasonable to assume that the buildup is characteristic
86 BLIZARD

of the material in the last layer and of the thickness in mean free paths of
the whole shield. The defense of this assumption lies in the fact that, except
for the case of quite low source energies, the harder scattered radiation is not
greatly different in intensity relative to the unscattered component regardless
of the atomic number of the medium. For this reason the harder built-up flux
would be expected to be characteristic of the total shield attenuation. On
the other hand, the softer built-up radiation is generated in rather a short
distance, so that this, in which different Z materials differ the most, would be
characteristic of the last layer.
Access provided by New York University - Bobst Library on 06/06/15. For personal use only.

THE ATTENUATION OF NEUTRONS


Annu. Rev. Nucl. Sci. 1955.5:73-98. Downloaded from www.annualreviews.org

Until quite recently there was no counterpart for neutron attenuation


calculations to the neat treatment of ')I-rays by the Spencer-Fano moments
method or any of its competitors. The reason for this lies largely in the un­
certainties regarding cross sections which are only now beginning to clear.
With the courage that stems from frustration, estimates in lieu of measure­
ments are now being made (26) of the most likely differential cross sections
for some elements found commonly in shields, and with these the moments
method is now being applied to neutron attenuation calculations.
The basic attenuation problem.-As in the case of 'Y-rays, the fundamental
calculation is for a point isotropic or other simple monoenergetic source in
an infinite homogeneous medium. Again the physical dose is the quantity
to be calculated, this being the energy deposited per unit weight of tissue.
In the case of ')'-rays the dose was rather closely related to the energy flux,
that is, the convolution in energy and direction of photon flux and photon
energy. The reason for this is that while the attenuation coefficient decreases
with increasing energy, the average energy deposited per interaction increases
strongly with photon energy so that the rate of energy deposition, or physical
dose, closely corresponds to the energy flux. In the case of neutrons in tissue,
the prime mechanism for absorption of energy is by elastic collision with
protons. In the pertinent energy range of 2 to 10 Mev the cross section de­
creases nearly inversely with energy, whereas the average energy loss in­
creases linearly with neutron energy. The result is that, unlike the ')'-ray
case, the neutron dose is roughly proportional to the number flux.
In the case of ')I-rays the first attempts at analytic treatment of attenua­
tion made use of the straight-ahead approximation, in which the energy
degradation on interaction is treated correctly but angular deviation is ig­
nored. For the case of attenuation of neutrons in hydrogen the same treat­
ment has been applied, the justification being much the same as in the ')I-ray
case, namely that the significantly deviated neutrons are appreciably lower
in energy and hence contribute little to the total transmission. Wigner (60,
61) has used this approximation to make an overestimate of the transmitted
neutrons in a hydrogen shield. Young (62) has collected and compared much
of this early work, demonstrating the feasibility of obtaining analytic solu­
tions for specific assumptions regarding energy dependence of the hydrogen
NUCLEAR RADIATION SHIELDING 87

cross section. Welton & Goertzel (63) have made some additional observa­
tions for the extension in which an isotropic scatterer is added to straight­
ahead-scattering hydrogen, obtaining a form for the asymptotic distribution
in energy as a function of distance. The straight-ahead approximation has
been improved upon by Wick (64) who has derived expressions for the asymp­
totic form of attenuation, allowing some angular deviation of the neutrons.
Welton (65, 66, 67) has extended this work and compared his results with
experiments. This latter comparison, however, is somewhat clouded by the
fact that Welton has used for his source a collimated plane distribution,
whereas the experiments have been made with isotropic plane sources.
Access provided by New York University - Bobst Library on 06/06/15. For personal use only.

Another rather more complete treatment by Holte (68, 69) has also shown
the power of the Laplace and Fourier transform methods, comparing favor­
Annu. Rev. Nucl. Sci. 1955.5:73-98. Downloaded from www.annualreviews.org

ably with earlier experimental work. Tait & Biram (70) have calculated the
neutron spectrum in a hydrogenous medium for the case of 1/v cross sections.
They used the first two spherical harmonics and show this to give a reason­
ably accurate description of the spectrum.
Some of the early work on analysis of neutron attenuation by means of
expansion in terms of the more familiar functions was carried out by Murray
(71, 72, 73). He used the continuum theory for calculation of differential­
scattering cross sections (74) and on the basis of this work was able to check
by calculation a number of effective removal cross sections which had been
measured at Oak Ridge National Laboratory (75).
The program of neutron attenuation computation at the Nuclear De­
velopment Associates, Inc. (NDA) (76, 77, 78), makes use of the Spencer­
Fano moments method, with appropriate tailoring for the different cross
sections. The first calculations were carried out on the SEAC, the computing
machine at the National Bureau of Standards; later work was and will be
calculated on the UNIVAC at New York University.
The first calculation (79) was for hydrogen, for which, of course, the cross
section was well known. Aronson, in reporting the work, makes several
interesting observations and comparisons. He verifies that the "-'8-Mev
neutrons in the fission spectrum are responsible for the penetrations to about
120 cm. (water equivalent hydrogen density), as is evident from the similar­
ity in slopes in that region for a fission source and for an 8-Mev monoenergetic
source. Comparison with a straight-ahead analytical calculation shows the
latter to be only a slight overestimate. The comparison with Welton's (65)
calculation was rather difficult because of the differences in hypotheses, but
the two methods did not appear incontrovertibly to disagree, although the
Welton buildup factors do seem to be lower at large penetrations.
The same method was used by Certaine & Aronson (80) of NDA to calcu­
late the distribution of epithermal neutrons in water from a point isotropic
fission source. The Goldstein oxygen data (26) and the Watt fission spectrum
(1) were used. For this calculation direct comparison with experiment was
possible since the work of Hill, Roberts, & Fitch (81, 82) was directly com­
parable. In so far as shielding theory is considered, the agreement is excellent.
88 BLIZARD

It is interesting to note, however, that the neutron age, or one-sixth of the


second spatial moment, was calculated to be 26 cm.2, whereas the experi­
ment gave 31 cm.2• This discrepancy has been found before in other calcula­
tions (83) and implies that another experiment is in order. A calculation
for 2.25-Mev neutrons by Volkin (84) gives 31.6 cm.2, but that energy, while
a mean for the fission spectrum, is not necessarily properly representative
thereof for a calculation of the second spatial moment.
Of more direct interest to shielding are the subsequent calculations of
penetration in water of monoenergetic (27) and fission (85) neutron sources.
The calculations were carried out to 120 cm., which thickness is of considera­
Access provided by New York University - Bobst Library on 06/06/15. For personal use only.

ble importance for many applications. The internal consistency and the agree­
ment with measurements of Caswell (86) are impressive. For the low-energy
Annu. Rev. Nucl. Sci. 1955.5:73-98. Downloaded from www.annualreviews.org

end of the spectra an analytic treatment is used, and the spatial variation for
these neutrons is taken to be that of the fast-neutron dose. For this analytical
treatment, the calculation of energy variation amounts to that for the slow­
ing down with a uniform distribution of sources, and this has a convenient
analytic form because of the simplicity of the hydrogen slowing down, which
of course predominates. The calculations again confirm the fact that the
neutrons of about 8-Mev initial energy account for penetrations to 120 cm.
The removal cross section.-It has been observed (28) that the effect of
insertion of thin slabs of materials into a water shield can be described in
terms of a simple exponential attenuation factor. The material must be in­
serted near the source, and the thickness of water between source and de­
tector must be greater than about 100 em., 140 cm. being the usual distance
chosen. Attenuation which is exactly exponential is characteristic of absorp­
tion processes alone. Of course in the case of fast neutrons other effects pre­
dominate, yet with a large thickness of water to filter out degraded or de­
flected neutrons a large fraction of the interactions are tantamount to absorp­
tion. These include all the inelastic scatterings and about half of the elastic
scatterings. The equivalent absorption cross section which describes the ef­
fect of the inserted slab of material is called the effective removal cross sec­
tion. It of course depends to some extent on the nature of the detector, the
source (although a fission source is usually assumed, the concept has been
developed independently and used by Moyer (87) for cyclotron shields), the
thicknesses of water and sample, and the location of the latter within the
shield. Fortunately it has been shown to be relatively independent of these
factors within the limits described above, but an exact definition (29) has
been used for purposes of uniformity and measurements are usually tailored
as closely as possible to give the exactly defined quantity. Calculations by
the moments method have recently been made by Goldstein & Aronson
(88) to determine the equivalent removal cross section for elements added
to a hydrogenous shield. These confirmed the validity of the concept and the
approximations used heretofore.
Suppose a unit strength source of neutrons is located in an infinite water
medium at some standard distance Ro from a detector. The reading on the
detector is then given by the kernel
NUCLEAR RADIATION SHIELDING 89
go(Ro)
Go(R0) = 16.
41I"Ro2

Now let there be inserted about the source a small sphere of the material
whose removal cross section is to be measured. Its thickness (radius) is
R"" and the reading at Ro+R", is now given by another kernel,
C(Ro + R-) =
go(Ro)g.(R.)
17.
• 411"(Ro + R.)2
The quantities Go(Ro), G(Ro+ R.,), Ro, and R", are measured. From these
can be found go(Ro) and g",(R",). g",(R",) is assumed to be of exponential form,
Access provided by New York University - Bobst Library on 06/06/15. For personal use only.

so that the macroscopic removal cross section is given by


Annu. Rev. Nucl. Sci. 1955.5:73-98. Downloaded from www.annualreviews.org

dg.(R.)

�% = -...,-
dR.
- - =
"
L 1m =
'
L 1m
1 - g.(R.) 18.
g.(R.) R.-tO R.g.(R.) R.-O R.
For measurement with R", finite, as is of course necessary, the following is
used:
g.(R.) = e-T..R.

�. = - R. 1
- log, g.(R.)

Of course,

g.(R ) • =
( Ro + R. ) 2 C(Ro + R.)
19 .
Ro Co(Ro)
There are no point fission sources, but by irradiating a disk of uranium
from the ORNL Graphite Reactor, the group at Oak Ridge National Labora­
tory have obtained a plane disk source which emits neutrons into an adjacent
tank of water. At this facility, the so-called Lid Tank Shielding Facility
(Fig. 2), many removal cross sections have been measured. For the slab
geometry that is used there, transformations are required to interpret the
data in terms of the exact quantities. For this, the kernels defined above are
assumed to apply, the distance R", being taken as the length in the slab
measured along a ray between the detector and an element of the source.
Integration of the kernel is then performed usually by some standard method
(29, 30). It is found that the effective removal cross section does not in fact
vary greatly with sample thickness up to thicknesses of five relaxation lengths
(reciprocal macroscopic removal cross sections) and that the standard water
thickness is not very critical, though care must be taken to avoid measure­
ment of softer scattered radiation with large R", and small Ro. In view of the
near approach to equilibrium in the neutron spectrum it is possible to meas­
ure removal cross sections from thermal-neutron measurements, and these
will be useful for predictions of fast-neutron dose attenuation. These cross
sections have been shown to be additive for the case of several elements
introduced together.
The foregoing applies to the case in which the material is added to a water
shield in the region near the source. In many cases, however, the material
Annu. Rev. Nucl. Sci. 1955.5:73-98. Downloaded from www.annualreviews.org
Access provided by New York University - Bobst Library on 06/06/15. For personal use only.

90
BLIZARO

..

"
'"

. �.
.

;z
I
a:

' U
' 0
u
:;;
If
c

w
NUCLEAR RADIATION SHIELDING 91

is added elsewhere. The most common other distribution is uniform. The


effective removal cross section has been measured for distributed carbon by
means of a saturated sugar solution, which is equivalent to water of normal
density with about i gm./cc. of carbon. For this measurement the removal
cross section was nine-tenths of that for a slab of graphite next to the source.
By comparing water and oil (CH2) measurements it is possible to derive a
similar quantity for oxygen, and this, too, appears lower for the distributed
than for the close slab case. Attempts have been made to explain this differ­
ence, but none have been successful.
Access provided by New York University - Bobst Library on 06/06/15. For personal use only.

The concept of the effective removal cross section has been an extremely
useful one, in that it affords a simple method of shield calculation which gives
Annu. Rev. Nucl. Sci. 1955.5:73-98. Downloaded from www.annualreviews.org

reasonably good answers and at the same time circumvents the expensive
treatments by the moments method using a high-speed computing machine.
In addition, of course, it is to be noted that the cross sections needed for the
latter are mostly missing, whereas the removal cross sections are easily esti­
mated if not already measured.
Up to this point it has been presupposed that the shield is largely water ,

or at any rate hydrogenous, and the other materials have been "added. "
At a sacrifice of exactness and some accuracy this concept can now be dis­
carded, the attenuation being treated as exponential with the removal cross
section defining the coefficient. There is, however, a limitation to this
freedom which is paramount. For this method to apply, the attenuation of
the fast neutrons must control the whole attenuation. The intermediate
(circa 0.5 Mev) and lower neutrons must not be more penetrating than the
fast neutrons from which they are produced ; at least this must not be the
case for the outermost regions of the shield. That this could be the case is
'
evident from the behavior of a pure iron shield. For this the relaxation
length of the fast neutrons is about 6 em. The intermediate energy neutrons,
on the other hand, are too slow to be further inefastically scattered, and
too fast to be absorbed. Hence by a process of multiple elastic scatterings
they diffuse outward, exhibiting a slowing down length in the neighborhood
of 30 em.
A more succinct way of putting the restriction is :
1
L. < --
--
20 .
�., + �u + �, + · · .
where L,2 = 1/6�, ;2 second spatial moment of the slo fing down density for
=

a point source of neutrons of the controlling energy ( ",g Mev), and the �'s
are the macroscopic removal cross sections of the different components of
the shield material.
This requirement is easily met by the addition of hydrogen to a shield,
and it is difficult to find shields without hydrogen which satisfy it. It is a
matter of practical importance, too, for the best shields are those for which
the fastest neutrons control, S(l the requirement should be met in any good
shield design. It is obvious that diffusion at thermal or near-thermal energy
92 BLIZARD

should not be appreciable either, but this requirement is met satisfactorily


by almost all materials, so that it is never necessary for this purpose to add
materials with high thermal-neutron absorption cross sections.
Because the total cross section of hydrogen varies so strongly with energy,
any removal cross section ascribed to this element wiII be a function of Ro,
which defines the amount of hydrogen in the shield. (Other hydrogenous
materials are treated by rather obvious substitutions.) As a consequence,
the hydrogen attenuation is usually treated separately, using either a meas­
ured attenuation kernel (89) or one of the NDA calculations.
Access provided by New York University - Bobst Library on 06/06/15. For personal use only.

Integration methods of neutron shield design.-Data on which to base


neutron shield design appear either in the form of integral measurements for
Annu. Rev. Nucl. Sci. 1955.5:73-98. Downloaded from www.annualreviews.org

certain simple source distributions, such as the Lid Tank disk, or as a col­
lection of removal cross sections which may be used in conjunction with
hydrogen attenuation data. There are two general methods for using this
information.
The Uniform Kernel Method : The simplest method to use is the homo­
geneous or kernel method (90, 91) for which the basic assumption is that
there exists a unique attenuation kernel which describes the dose at every
point in the shield (and on its surface) for each element of source and which
is a function only of the separating distance. The conditions attendant upon
this assumption are never actually met in practice. For example, the reactor
core is never made of the same material as the shield, although their proper­
ties as regards fast-neutron attenuation may not differ much. The shield it­
self is seldom homogeneous and never infinite, and this introduces an error,
for clearly the kernel is then a function of position as well as separation dis­
tance. Nevertheless, for fast neutrons at least, the kernel method gives rea­
sonable results.
The method is used to predict the dose in one situation on the basis of a
measurement in another. First it is necessary to determine the kernel itself
or some basic form which serves to define the kernel. Suppose an infinite
plane is uniformly endowed with sources of strength unity per unit area.
The dose at a distance z from it is given by a simple integration :

DPI(z, (0 ) = f
plano
G(R)ds = 211' J "'G(R)RdR

21 .

and

1 d
G(z) "" - - - DPI(Z, (0 ) 22 .
211'15 ds

For a finite disk source of radius a,

DPl(z, a) =f G(R)RdR 23 .

.JZ;+ a'

It is possible to relate the infinite plane result to a sum of disk source results
by means of j udicious choice of the integration limits. Thus
NUCLEAR RADIATION SHIELDING 93
'"

DPI(z, 00 ) = 2: Dp,(Vz! + va2, a) 24.


_0

and
.,

G(z) - 2: B (vz! + va!) 25 .


_0

where
_ __ 1 dDpl(Z, a)
B(z) =
211"z dz
Access provided by New York University - Bobst Library on 06/06/15. For personal use only.

These relations are easily verified by writing down the integrals for the sum­
mands as indicated by equation 23.
Annu. Rev. Nucl. Sci. 1955.5:73-98. Downloaded from www.annualreviews.org

It is also possible to show that for a spherical shell coated with unit strength
source per unit area the dose is related to doses from plane sources as follows :
a
D.ph(b, a) = b I DPI(b - a, 00 ) - DPI(b + a, 00 ) I 26 .

where a= radius of sphere, and b distance from sphere center to detector.


=

By other similar relations, but often with approximations, it is possible


to relate the dose in a given geometry to many others. This method is often
used when full-scale experiments are available and when the reactor-shield
complex is not too inhomogeneous.
The method suffers most in that it does not properly take into account
the relative thicknesses of different shield layers for other than a point or
plane source. The next method to be discussed attempts to do this.
The Method for Inhomogeneous Shields : Ascoli (92) has suggested a
method of shield design in which the attenuation kernel is taken to be the
product of an inverse square in the distance and an exponential whose ex­
ponent is the integrated number of relaxation lengths along the ray connect­
ing source and detector. Thus,
1
G(R) = - 2 exp ( - /1:" (x, y, z) ds) 27.
4lI"R ray

where �T(X, y , z) macroscopic effective removal cross section, assumed to


=

vary arbitrarily with position in reactor and shield, and ds element of =

length along the source-detector ray.


Simon (93) has demonstrated a convenient form for the dose in spherical
geometry using Ascoli's kernel, for the case in which the macroscopic removal
cross section of the shield is a function solely of the radial distance from
reactor core center. This method has in general proved more flexible than
the uniform kernel method, although the latter, when applicable, is much
easier to use. Simon has also demonstrated a calculational method for
determining dose from secondary radiation such as capture ')I-rays from
slowed-down fast neutrons. This is a logical extension of the Simon-Ascoli
method.
The weakness in the Simon-Ascoli method lies in the question of whether
94 BLIZARD
the number of relaxation lengths along the straight ray is a true index of
the attenuation. In general, a line integral of the same integrand along an­
other path is smaller than that along the ray. While there must be some
penalty (in the sense that attenuation is a penalty on the radiation) as­
cribed to curvature (or bending) of the path of integration, it is quite clear
that the Ascoli kernel must be in some sense an underestimate for other
than homogeneous shields. In practice the method appears to give quite
good results, for the apparent reason that most shields are nearly homo­
geneous for fast neutrons. When the method is applied to ')'-rays it also can
be expected to give nearly correct results, for in this case the degradation
Access provided by New York University - Bobst Library on 06/06/15. For personal use only.

accompanying deflection puts a rather severe penalty on an appreciably


Annu. Rev. Nucl. Sci. 1955.5:73-98. Downloaded from www.annualreviews.org

curved path of integration.


It is important to note that both this method and the kernel method of
shield design are equally applicable to ')'-ray attenuation problems. They
were devised, however, to apply to neutrons, for which there has been
available until quite recently no more sophisticated analytical treatment.
In the absence of microscopic data other than the removal cross sections
these have been essentially the only methods.
SHIELDS WHICH HAVE BEEN BUILT
The Brookhaven reactor shield (94) is one of the first shields to be de­
signed and built with the post-war attention to cost. It has in general
proved quite satisfactory, although its ')'-ray attenuation is somewhat
lower than was expected, a fact attributed to the large piece size of the steel
aggregate that was used.
Another shield in which careful design was invested is that for the Ma­
terials Testing Reactor in Idaho. For this case a barytes aggregate portland
concrete was used. This was the first application of an AscoIi-type kernel for
shield design, antedating the work of Ascoli, who formalized the method and
described its general applicability.
In general, it can be said that it is possible to design reactor and ac­
celerator shields with a fair degree of confidence, but such special problems
as voids in shields and unusual geometries, scattering, etc. are not sufficiently
well understood to afford a rationale for their solution. There is a consider­
able amount of experimental data on ducts through cyclotron shields (95)
but no universal theory connects the results.

LITERATURE CITED
1. Watt, B. E . , "Energy Spectrum of Neutrons from Fissions Induced by Thermal
Neutrons," U. S. Atomic Energy Commission Document, AECD-3073 ( 195 1 ,

date declassified) ; based on Los Alamos Scientific Lab. Rept., LA·718 (1948)
(Classified)
2. Watt, B. E., Phys. Rev. 87, 1037 (1952)
3. Bonner, T. W., Ferrell, R. A., and Rinehart, M. C., Phys. Rev., 87, 1032 (1952)
4. Hill, D. L., Phys. Rev., 87, 1034 (1952)
5. Nereson, N., Phys. Rev., 85, 600 (1952)
NUCLEAR RADIATION SHIELDING 95
6. Hughes, D. j., University of Chicago, Metallurgical Lab. Rept., CF-3596 (1946)
(Classified)
7. Hughes, D. j., Dabbs, r, Cahn, A., and Hall, D., Phys. Rev., 73, 111 (1948)
8. Gamble, R. L., "Fission Gamma-Ray Spectrum" (Physics Div. Semiannual
Prog. Rept. for Period Ending September 10, 1953) Oak Ridge Natl. Lab.
Rept., ORNL-1620, p. 15 (1953)
9. Francis, J. E., and Gamble, R. L., "Prompt Fission Gamma Rays" (Physics Div.
Semiannual Prog. Rept. for Period Ending March 10, 1955) Oak Ridge Natl.
Lab. Rept., ORNL-1879 (1955)
10. Way, K., and Wigner, E. P., Phys. Rev., 73, 1318 (1948)
Access provided by New York University - Bobst Library on 06/06/15. For personal use only.

1 1 . Moteff, j., "Fission Product Decay Gamma Energy Spectrum" (Aircraft Nuclear
Propulsion Project) General Electric Rept., APEX 134 (1953)
Annu. Rev. Nucl. Sci. 1955.5:73-98. Downloaded from www.annualreviews.org

12. Clark, F. H., Decay of Fission Product Gammas, NDA-27-39 (Nuclear Develop­
ment Associates, Inc. Rept., 1954)
13. Ajzenberg, F., and Lauritsen, T., Revs. Mod. Phys., 24, 321, esp. 351 (1952)
14. Fano, U., "Gamma-Ray Attenuation," (Reactor Handbook, Technical Informa­
tion Service) U. S. Atomic Energy Commission Document, RH-I , p. 753 (1953)
(Classified) ; U. S. Atomic Energy Commission Document, AECD-3645 (Au­
gust, 1955) (Unclassified version)
15. Fano, U., Nucleonics, 1 1 (8) , 8 (1953)
16. Fano, V., Nucleonics, 1 1 (9) , 55 (1953)
17. Goldstein, H., and Wilkins, J. E., Jr., "Calculations of the Penetration of Gamma
Rays" (Technical Information Service) U. S. A tomic Energy Commission
Document, N YO-3075 (1954) j previously issued by Nuclear Development
Associates, Inc., as NDA-15C-41
18. White, G. R., "X-ray Attenuation Coefficients from 10 Kev to 100 Mev, "
Nat. Bur. Standards Rept., NBS-I003 (1952)
19. Davisson, C. M . , and Evans, R. D., Revs. Mod. Phys . , 24, 79 (1952)
20. Nereson, N., and Darden, S., Phys. Rev ., 94, 1678 (1954)
21. Jurney, E. T. , Arnold, G. P., Schafer, W. D., Swickard, E. 0., and Zabel, C. W. ,
"Inelastic Collision and Transport Cross Sections for Some Light Elements, "
Los A lamos Scientific Lab. Rept., LA-1339 (1951)
22. jurney, E. T., and Zabel, C. W., Phys. Rev., 86, 594 (1952)
23. Walt, M., and Beyster, J. R., Phys . Rev., 98, 677 (1955)
24. Feld, B . T., Feshbach, H., Goldberger, M . L., Goldstein, H., and Weisskopf,
V. F., Final Report of the Fast Neutron Data Project, Rept. N YO-636 (Nuclear
Development Associates, Inc., 1951)
25. Feshbach, H., Porter, c., and Weisskopf, V. F., Neutron Reactions and the Forma­
tion oj the Compound Nucleus, Rept. NDA-15B-4, N YO-3076 (Nuclear Devel­
opment Associates, and Massachusetts Institute of Technology, 1953)
26. Goldstein, H., Fast Neutron Data for Oxygen, Rept. NDA-15C-15 (Nuclear De­
velopment Associates, Inc., 1953)
27. Aronson, R., Certaine, J., and Goldstein, H., Penetration of Neutrons from Point
Isotropic Monoenergetic Sources in Water, Rept. NDA-15C-60, N YO-6Z69
(Nuclear Development Associates, Inc., 1954)
28. Albert, R. D., and Welton, T. A., Westinghouse Electric Corp., A tomic Power
Division Rept., WAPD-15 (1950) (Classified)
29. Blizard, E. P., Oak Ridge Nat. Lab. Rept., CF-54-6-164 (1954) (Classified)
30. Chapman, G. T., and Storrs, C. L., Oak Ridge Nail. Lab. Rept., ORNL-1843 (1955)
(Classified) ; unclassified version to be issued later
96 BLIZARD

3 1 . Greuling, E., Appendix A in Bernstein, S., Preston, W. M., Wolfe, G., and
Slattery, R. E., Oak Ridge Natl. Lab. Rept., MonP-17Z, p. 23 (1946) (Classi­
fied)
32 . Nordheim, L. W., "Physics Division Report for Month Ending June 30, 1946,"
Oak Ridge Natl. Lab. Rept., MonP-130, p. 7
33. Bethe, H. A., General Electric Nucleonics Project Report (Knolls), A -4Z51 (Classi-
fied) (1947)
34. Hirschfelder, J. 0., Magee, J. L., and Hull, M. H., Phys. Rev., 73, 852 (1948)
35. Foldy, L. L., and Osborn, R. K, Phys. Rev., 81, 395, 400 (1951)
36. Bethe, H. A., Fano, U., and Karr, P. R., Phys. Rev., 76, 538 (1949)
Access provided by New York University - Bobst Library on 06/06/15. For personal use only.

37. Fano, U., Phys. Rev., 76, 739 (1949)


38. Spencer, L. V., and Fano, U., Phys. Rev., 81, 464 (1951)
Annu. Rev. Nucl. Sci. 1955.5:73-98. Downloaded from www.annualreviews.org

39. Spencer, L. V., and Fano, U., J. Research Nat!. Bur. Standards, 46, 446 (195 1)
40. Goldstein, H . , and Aronson, R., Status Report on Calculations of Gamma-Ray
Penetrations, NDA -15C-l, N YO-3079 (Nuclear Development Associates,
Inc., Rept., 1953)
41. White, G. R., Phys. Rev., 80, 154 (1950)
42. Dixon, W. R., Phys. Rev., 85, 498 (1952)
43. Elliot, J. 0., Farrar, R. T., Myers, R. D., and Ravilious, C. F., Phys. Rev., 85,
1048 (1952)
44. Hayward, E., Phys. Rev., 86, 493 (1952)
45. Roys , P. A., Shure, K, and Taylor, ]. J., Phys. Rev., 95, 911 (1954)
46. Taylor, J. J., Application of Gamma Ray Build-up Data to Shield Design, Atomic
Power Division Rept., WA PD-RM-Z1 7 (Westinghouse Electric Corporation ,
1954)
47. Wick, G., Phys. Rev., 75, 738 (1949)
48. Spencer, L. V., "Penetration and Diffusion of x-Rays: Calculation of Spatial
Distributions by Semi-Asymptotic Methods," Natl. Bur. Standards U. S.,
Rept. NBS-1 7ZI5 (1952)
49. Fano, U., J. Research Natl. Bur. Standards, 51, 95 (1953)
50. Peebles, G. H., Gamma-Ray Transmission Through Finite Slabs, R-240 (RAND
Corporation Rept., 1952) ; previously issued as Rept. RM-1553, Parts I (1951) ;
II (1952)
51. Berger, M. J. (Five papers on applications of the Monte Carlo and moments
methods were received too late for review ; To be published by National
Bureau of Standards, and subsequently by Physical Review.)
52. Zerby, C. D., "Preliminary Report on the Penetration of Composite Slabs by
Slant Incident Radiation," Oak Ridge Natl. Lab. Rept., CF-54-9-1Z0 (1954)
53. Shor, S. W. W., Nuclear Shielding Studies : V. Computation of Radiation Shield
Thickness by the Monte Carlo Method, Technical Rept. No. 32 (Laboratory for
Nuclear Science and Engineering, Massachusetts Institute of Technology,
1950)
54. Kim, F. S., Kennedy, R. J . , and Wyckoff, H. 0., "Oblique Attenuation of Gamma­
Rays from Cobalt-60 and Cesium-137 in Polyethylene, Concrete and Lead,
Natl. Bur. Standards Rept., NBS-2125 (1952)
55. Clifford, C. E., Blosser, T. V., and Watson, F. N., Oak Ridge Natl. Lab. (Private
communication)
56. Spencer, L. V., and Stinson, F., Phys. Rev., 85, 662 (1952)
57. Blizard, E. P., Maienschein, F. c., PeelJe, R. W., and Love, T. A., "Angular
N UC LEAR RAD IATION SHIELD ING 97

D istr ib ut ion of Photon s from a C 060 So ur ce in Wat er," Oak Ridge Natl. Lab.
Rept. (To b e p ublished )
58. C erta ine, J., A ngular Distribution of Photons from Plane Monoenergetic Sources,
NDA-15C-l0, N YO-3074 (N uclear D evelopment A sso ciat es, In c. , 1953)
59. H unger ford , H. E. , Oak Ridge Natl. Lab. Rept. , ORNL-1611 (1954) (Classified )
60. Wign er , E . P., "Prot ect ion Aga in st Rad iat ion " ( Unill. Chicago Rept., CH-137
(No date)
61. Wign er , E. P., and Yo ung , G., Oak Ridge Natl. Lab. Repl., MonP-283 (1947)
(Cla ssified )
62. Yo ung , G., Oak Ridge Natl. Lab. Rept., MonP-293 (1947) (Cla ssified)
Access provided by New York University - Bobst Library on 06/06/15. For personal use only.

63. Welton , T. A. , and Go ert zel , G. , "A symptot ic D ist rib ut ion of N eutron s from a
Po int So ur ce in a M ed ium Conta in ing H ydrogen and an Isotropic S catt er er ,"
Annu. Rev. Nucl. Sci. 1955.5:73-98. Downloaded from www.annualreviews.org

Oak Ridge Natl. Lab., U. S. A tomic Energy Commission Rept. , AECD-2796


( 1 949 ; d ecla ssified in 1950)
64. Wick, G. C . , Phys. Rev., 75, 738 (1949)
65. Bl izard , E. P., and Welton , T. A., Oak Ridge Natl. Lab. Repts., ORNL-1J33,
Part s I and II (1952) (Cla ssified )
66. Blizard, E . P., and Welton , T. A., Reactor Science and Technology, I , N o. 3,
p . 1 5 (1951) ; U. S. A tomic Energy Commission Rept., TID-73 (Cla ssified )
67. Welton , T. A., and Bl izard , E. P., Reactor Science and Technology, 2 , No. 2 , p . 73
(1952), U. S. A tomic Energy Commission Rept., TID-ZOOZ (C lassified )
68. Holt e, G., A rkill Fysik, 2(48), 523 (1951)
69. Holte, G., Arkiv Fysik, 3(14), 209 (1951)
70. Ta it , J . H ., and Biram , M . B., "The Calculat ion of the Spectrum of Neutrons at
Po int s in a S em i In fin it e H ydrog enous M ed ium , Who se Free S ur fa ce Is
Irradiat ed b y Mono en erg et ic N eut ron s Fall ing Normall y onto It ," Bntish
Atomic Energy Research Establishment Rept., A .E.R.E.T/M 92 (1953)
71. M ur ra y, F. H ., "App ro ximat e Calculat ion of t he N eutron Flux from a Plan e
So ur ce in an In fin it e M ed ium Wher e t he N ucl ear Cro ss S ect ion s Va ry wit h
t he En er gy and wit h t he D istan ce from t he So ur ce, for En erg ies above 5
M ev," Oak Ridge Natl. Lab. RePI., ORNL-218 (1948)
72. Murray, F. H . , Oak Ridge Natl. Lab. Repl. ORNL-424 (1949) (Classified )
73. M urray, F . H., "An Approximat e Calculat ion o f t he N eutron Fl ux in a Thick
S hield Contain ing Some H ea vy Elem ent s, wit h t he Aid of C ert ain Recurr ent
Relat ion s Bet ween t he Co efficient s in a L eg endr e S er ies Expan sion ," Oak
Ridge Natl. Lab. Rept., ORNL-748 (1950)
74. Murray, F. H., "Notes on t he Calculat ion o f t he Att en uat ion of N eutrons
Throug h Thick S hield s," Oak Ridge Natl. Lab. Rept., CF-53-11-64 (1953)
75. Murray, F. H., Oak Ridge Nall. Lab. RePI. , CF-53-12-60 (1953) (Cla ssified )
76. Certa in e, J., A Solution of the Neutron Transport Equation, Introduction and Part
I, NDA -15C-43, N YO-3081 (N uclear D evelopment Associat es, In c., Rept . ,
1954)
77. C erta in e, J. , Integral Term for Elastic Scattering of Particles, NDA-15C-IZ (N u­
clear D evelopm ent Asso ciat es, In c., Rept., 1953)
78. C erta ine, J., A Solution of the Neutron Transport Equation, Part III, NYO-6270
(N ucl ear D evelopm ent Asso ciates, In c., Rept., 1954)
79. A ro nson , R ., Neutron Penetration in Hydrogen, NDA -15C-39 (Nuclear D evelop ­
m ent Asso ciates, In c., Rept., 1954)
80. Certa in e, J . , and Aron son , R., Distribution of Fission Neutrons in Water a' the
98 BLIZARD

Indium Resonance Energy, NDA ·15C·40 (Nuclear Development Associates,


Inc., Report, 1954)
81. Hill, J. E., Roberts, L. D., and Fitch, T. E., "The Slowing Down Distribution
from a Point Source of Fission Neutrons in Light Water", Oak Ridge Natl.
Lab. Rept., ORNL-181 (1948)
82. Hill, J. E., Roberts, L. D., and Fitch, T. E . , J. Appl. Phys. , 26, 1013 (1955)
83. Zweifel, P. F., and Hurwitz, H., Jr., (Address given at Shielding Information
Meeting, Knolls Atomic Power Lab. , May 13-14, 1954) Technical Informa­
tion Service Rept., WASH-I74 (Classified)
84. Volkin, H. c., J. Appl. Phys., 25, 83 (1954)
Access provided by New York University - Bobst Library on 06/06/15. For personal use only.

85. Aronson, R., Certaine, J., Goldstein, H., and Preiser, S., Penetration of Neutrons
from a Point Isotropic Fission Source in Water, NDA -15C-42, N YO-6267
Annu. Rev. Nucl. Sci. 1955.5:73-98. Downloaded from www.annualreviews.org

(Nuclear Development Associates, Inc., Rept., 1954)


86. Caswell, R. S., Gabbard, R. F., Doering, W. P., and Padgett, D. W., Phys. Rev. ,
94, 786 (1954)
87. Moyer, B. J., University of California (Private communication)
88. Goldstein, H., and Aronson, R., Effective Removal Cross Sections Theory, TID-
2015, p. 147 (1954) (Classified)
89. Maienschein, F. c., "Attenuation by Water of Swimming Pool Reactor Radia­
tions," Oak Ridge Nat. Lab. Rept., ORNL 1891 (1955)
90. Blizard, E. P., Oak Ridge Nat. Lab. Rept., CF-51-10-70, Parts I and II, Revised
(1952) (Classified)
91. Blizard, E. P., "Reactor Leakage for Shielding Calculations," Oak Ridge Natl.
Lab. Rept., CF-53-7-I70 (Revised) (1953)
92. Ascoli, G. , Appendix XV in U. S. A tomic Energy Commission Document, A NP-
52, 196 (1950) (Classified)
93. Simon, A., Chapter I I I-B-l in Oak Ridge Natl. Lab. Rept., ORNL-1575, 67 (1954)
(Classified)
94. Pratt, W. W., Fox, M., Kouts, H. ]., and Rand, A. c., Brookhaven Natl. Lab.
Repe., B NL-145 (1951) (Classified)
95. Crawford, J. W., and Kintner, E. E., "Experiments to Determine the Effect of
Ducts on the Attenuation of Neutrons and Gamma Rays in the M.LT.
Cyclotron Shield," Neutron and Gamma Ray Shielding Group and Electronic
Nuclear Instrumentation Group Quarterly Prog. Rept., April-June, 1950 , p. 10
96. Mittelman, P. S., and Liedtke, R. A., Nucleonics 13(5), 50 (1955)

You might also like