Comprehensive nuclear materials 5 07 performance of aluminum in research reactors

5.07

Performance of Aluminum in Research Reactors

K. Farrell
Formerly of Oak Ridge National Laboratory, Oak Ridge, TN, USA

ß 2012 Elsevier Ltd. All rights reserved.

5.07.1
5.07.2
5.07.2.1
5.07.3
5.07.3.1
5.07.3.2
5.07.4
5.07.5
5.07.6
5.07.6.1
5.07.6.2
5.07.6.2.1
5.07.6.2.2
5.07.6.2.3
5.07.7
5.07.7.1
5.07.7.2
5.07.7.3
5.07.7.4
5.07.8
References

Introduction
Typical Applications
History of Aluminum Applications in Research Reactors
Properties of Aluminum
Practical Characteristics
Alloy Types, Temper Designations, and Tensile Properties
Fuel Elements
Corrosion
Radiation Effects
Basics
Microstructures
Fluence
Temperature
Transmutation products
Property Changes
Swelling
Mechanical Properties
Effects of Neutron Spectrum
Radiation Softening, Creep, and Stress Relaxation
Conclusion

Abbreviations
AIME
ANL
ANSI
ASM
ASTM
ATR
CRC
CTE

EBR-II
Emod
ETR
GR
HEU
HFIR
HPRR
IAEA
INL
IRV-M2

American Institute of Mining,
Metallurgical, and Petroleum Engineers
Argonne National Laboratory
American National Standards Institute
American Society for Metals
American Society for Testing Materials
Advanced Test Reactor
Chemical Rubber Company
Coefficient of thermal expansion
Experimental Breeder Reactor-II
Modulus of elasticity
Experimental test reactor
Graphite Reactor
Highly enriched uranium
High Flux Isotope Reactor
High performance research reactor
International Atomic Energy Authority
Idaho National Laboratory
Acronym for a recent Russian research

reactor

LANL
LEU
MTR

OPAL
ORNL
ORR
PIE
PIREX
RERTR
RR
SNF
STP
TRIGA
TEM
UTS
VPH
YS

144
144
144
145
146
147
149
153
158

158
159
160
161
161
166
166
166
169
170
173
173

Los Alamos National Laboratory
Low enriched uranium
Specifically, MTR is the Materials Testing
Reactor at Idaho National Laboratory.
Also used generically for materials test
reactors
Open Pool Australian Light water reactor
Oak Ridge National Laboratory
Oak Ridge Research Reactor
Post irradiation examination
Proton Irradiation Experiment facility
Reduced enrichment for research and
test reactors
Research reactor
Spent nuclear fuel
Special Technical Publication
Test, research, isotopes, general atomic

Transmission electron microscopy
Ultimate tensile stress
Vickers pyramid hardness
Yield stress

143


144

Performance of Aluminum in Research Reactors

5.07.1 Introduction
Aluminum alloys are generally too weak or have
temperature limitations that preclude their use in
reactors built to produce electricity, high-temperature
process heat, or marine propulsion. But in the milder
conditions in most research reactors (RRs) where
bulk water coolant temperatures are usually <100  C,
aluminum alloys are quite comfortable and are universally employed and have greatly contributed to
the success and longevity of the reactors. RRs are
those whose principal function is to generate neutrons for purposes of nuclear education and training,
production of medical and industrial isotopes, neutron activation analyses, neutron scattering studies,
and even semiconductor doping, neutron radiography, and food preservation treatments. RRs are also
employed to study basic radiation effects in materials
and as test beds for evaluating candidate structural materials and fuels/assemblies for power reactors. RRs come in many shapes, sizes, and types.
For descriptions of the various classes of RRs, see
and West.1 They are
generally low power, typically about a few kilowatts,
thermal, but range up to about 250 MW. According to

the recently updated list2 of worldwide RRs published
by the IAEA, a total of 674 RRs have been built in
57 countries, of which 234 are still operational, and
7 are planned or under construction. Two new ones
are OPAL, the 20 MW Open Pool Australian Light
water-cooled reactor, which opened at Lucas Heights,
Sydney, in April 2007, and the Russian 4 MW pooltype IRV-M2 commissioned in 2008.
This chapter is a review, more a tutorial, of the
behavior of aluminum alloys in water-cooled RRs.
It is a somewhat personal view, based on American
experience in the area. Because that experience
has been adopted in many countries and is still
influencing the current state of the art, this chapter
should be of interest outside the borders of the
United States.

5.07.2 Typical Applications
Aluminum is the material of choice for construction of many components in low-temperature
water-cooled-and-moderated RRs. Typical applications are the reactor tanks in open-pool reactors;
containment vessels in some sealed reactors; core
grids; pedestals; neutron beam tubes; cold neutron
source moderator vessels; shrouds to direct and

separate water flows; shuttles (‘rabbits’) and aluminum filler powder used to convey isotope target
materials and test materials rapidly in and out of
the reactor via aluminum hydraulic and pneumatic
tubes; sheaths and finned tubing for stationary longterm isotope target rods; cladding for control plates/
rods; cladding and liners for reflector materials; cladding and thermal conduction filler for fuel rods/
plates; and temporary plugs for closing idle irradiation
facilities in and around the core. Applications outside

the reactor per se are in-pool tool extension arms;
transfer gates between pool sections; restraint baskets
in some shipping casks; support beams for pool covers;
and hot cells manipulator arms.
5.07.2.1 History of Aluminum
Applications in Research Reactors
Aluminum was at the forefront of the development
of nuclear technology. It has the distinction of being
the first nonfissile, non-neutron absorber class metal
used in the world’s first continuously operating
nuclear reactor, the X-10 Graphite Reactor at Oak
Ridge, TN. The Graphite Reactor became critical
on 4 November 1943, <1 year after Fermi’s demonstration of a self-sustaining nuclear fission chain in
the graphite pile at the University of Chicago on
2 December 1942. In Fermi’s experiment, the only
metals in the pile were natural uranium and the
cadmium-coated control rods. The pieces of natural
uranium (238U containing about 0.7 at.% 235U) and
uranium oxide were bare, placed in shallow depressions carved into the upper faces of the graphite slabs,
and cooled by convection of ambient air. The power
level was about 2 kW. The X-10 Graphite Reactor
pile3 was much bigger than the Chicago pile and
was designed to operate at 1 MW thermal power,
later upgraded to 4 MW. It was built to produce
pilot plant quantities of plutonium isotopes. The
Chicago pile had no shielding; the Graphite Reactor
was shielded by a 2.2-m thickness of high-density
concrete. Aluminum made its debut in the Graphite Reactor as fuel cladding to protect the highly
chemically reactive uranium from contamination
by air and graphite during the higher power and

longer fissioning periods and to safeguard it from
attack by water during subsequent radioactive decay
in underwater storage. In addition, it trapped the
more copious volatile radiation products resulting
from the longer irradiation exposures. These
aluminum–clad pieces of natural uranium, called
‘slugs,’ were the forerunners of metal–clad fuel


Performance of Aluminum in Research Reactors

elements. A slug was made by placing a solid cylinder
of uranium in a thimble-shaped aluminum can
25 mm diameter  100 mm long with a 0.75 mm
wall. A flat Al end cap was added, and the assembly
was passed through a die to force the can walls tightly
around the fuel. Surplus wall material was cut off
above the cap, and the cap was welded all around its
edge. These slugs were pushed end to end into the
reactor via round horizontal holes through the concrete face, which were aligned with 44 mm square
holes cut through the full 7.3 m width of the cubic
array of graphite blocks. The square holes were oriented on edge such that the slugs occupied the lower
corner, allowing cooling spaces around the slugs.
Cooling was simple: two large fans at the rear of the
pile sucked ambient air through the holes around the
slugs and discharged it up a tall chimney. The slugs
exited the pile at the rear face and were channeled into
a deep water canal where they were held until shipped
to hot cells for processing to extract the plutonium.
Some early problems4 were encountered in the slugs,

including faulty welds and blisters and formation of
an intermetallic UAl3 phase by interdiffusion at the
U–Al interface, especially in the high-temperature
regions in the center of the reactor. The blistering
was attributed to fast-growing gas bubbles in the UAl3
phase. These problems were overcome by better
welding practice and the development of bonded
slugs as described next.
The next phase of exploitation of aluminum was
in the B reactor at the Hanford site in Washington
State, which went critical on 27 September 1944.
The B reactor was a scaled-up production model
of the Graphite Reactor designed to operate at
250 MW. At such power, forced air cooling would
have been inadequate. So the horizontal holes
were replaced with aluminum tubes in which
aluminum–clad uranium slugs were cooled with
flowing water from the Columbia River. To improve
the transfer of heat from the uranium to the cladding, the spaces between them were filled with a low
melting Al–12% Si eutectic alloy by melting the
eutectic in situ. A bonus of this treatment was that
it killed the formation of the UAl3 phase and associated blistering, presumably due to an inhibiting
effect of the silicon. The successes of these upgrades
established aluminum as a suitable material for use
in combined conditions of intensive irradiation and
a flowing aqueous environment. Aluminum became
more firmly entrenched in RRs with the development of advanced fuel elements, as described in
Section 5.07.4.

145

Before concluding the present subsection, another
lesser-known ‘first’ for aluminum deserves mention.
It has particular relevance to the nuclear power
industry. It is not widely known that aluminum was
involved in the earliest demonstration of electricity
produced from steam made by boiling water in a
nuclear reactor. Normally, the heat from nuclear
fission in RRs is discarded, not used to produce
electricity. However, in August 1948, two staff members at the X-10 Graphite Reactor, Mansel Ramsey
and Charles Cagle, placed an aluminum can containing ten aluminum–clad uranium slugs and some
water in a normally unused side channel of the reactor where it was exposed to reactor neutrons. The
trapped heat generated in the slugs boiled the water.
Steam from the process was piped to a small model
steam engine, rated at 1/1000 hp (0.75 W), which
rotated an armature mounted between the poles
of a permanent magnet. Sufficient electricity was
generated to light a flashlight bulb. The thermal
efficiency was estimated to be $2%. The Graphite
Reactor is now a National Historic Landmark and is
open to the public. A commemorative plaque and a
replica of the steam engine and coupled dynamo
from Ramsey and Cagle’s pioneering boiling water
power reactor are displayed in a small showcase
in the reactor lobby. The ‘official’ first production
of nuclear electricity is credited to the lighting of
another bulb on December 1951 at the Experimental
Breeder Reactor-I, Arco, Idaho, now the Idaho
National Laboratory.

5.07.3 Properties of Aluminum
Heat removal and reduced generation of heat
are major considerations in the popularity of aluminum in RRs. Most of the energy released from controlled nuclear fission appears as heat. Much of the
heat, >80%, arises in the fuel from nuclear fissions.
However, a significant portion, 5–20%, is produced
in the nonfissile materials in the core and its surroundings by bombardment with particles emanating
from the fission reactions and from decay of fission
products. For power reactors, the heat is essential to
generate the electrical output. In the case of RRs, the
heat is a nuisance product; and the goals are to
minimize heat generation from the nonfissile materials in the system and to get rid of it from those
materials and from the fuel as fast as possible.
Hence, structural materials that create the least heat
and/or conduct it away the fastest are the most


146

Performance of Aluminum in Research Reactors

favored for RRs. In this regard, aluminum is outstanding. Generally, heat production is greater with
increasing material density and with decreasing
specific heat. It is increased by high cross-sections
for neutron absorption and scattering, which also
reduce reactor efficiency by stealing neutrons from
participation in fission processes. Heat removal rate
is larger with higher thermal conductivity. Therefore, construction materials with low density, high
specific heat, high thermal conductivity, and low
nuclear cross-sections offer the best prospects for
minimizing heat generation and maximizing heat

removal. In Table 1, the relevant properties for
aluminum are compared with those of other cladding and structural materials used in power reactors
and for uranium. All values are for room temperature or 100  C. The scatter in values for a given
parameter and material is due in part to sensitivity
to chemical composition and heat treatment, etc.
These variations do not mask the large differences
between Al and the other materials. The density of
Al is 1/2–1/3 of those of the other cladding materials, and only 1/7 that of U. Its specific heat capacity
is twice as high as the other materials. And its
thermal conductivity is 5–10 times greater than
the values for the other materials. Additionally,
its neutron capture and scattering cross-sections are
much smaller than those of the other materials, except
for nuclear-grade Zr. In that respect, it should
be remembered that in the early days when Al was
establishing its foothold in nuclear technology commercial Zr was contaminated with up to 3% of the
strong neutron absorber Hf. It was also inordinately
expensive.

Table 1

5.07.3.1

Practical Characteristics

Having attractive physical properties for reactor use
is of no merit if those properties cannot be exploited
in a practical manner. The commercial and economic
attributes of aluminum that encourage its deployment in RRs are: It is ductile, plentiful, cheap, and
light weight. It is castable, machineable, and weldable, and it can be shaped readily by conventional

processes of rolling, forging, extrusion, drawing, and
cupping. It has good aqueous corrosion resistance due
to near-insolubility in water and formation of a passive, self-restoring surface film of hydrated aluminum
oxide. It is nonmagnetic and nonsparking. Although
aluminum is inherently weak, it can be strengthened
by cold work, solid solution hardening, and precipitation treatments. It has an fcc crystal structure and
no crystallographic phase changes. Its crystal structure is near isotropic, ensuring that it will not suffer
damaging directional thermal expansion and radiation growth like those exhibited by graphite and the
hexagonal metals Mg and Zr. It does not form stable
embrittling hydride phase(s) as Ti and Zr do. At low
temperatures, it has no ductile-to-brittle transition.
On the contrary, it is somewhat special in that at
cryogenic temperatures, where it gains strength, it
often gains ductility too. This combination of no
hydride phase, outstanding low temperature properties, and low neutron cross-sections make aluminum
the prime material for building cold neutron sources.
Another attractive feature is that pure aluminum has
no long-lived radioisotopes. The major source of
immediate radioactivity is from decay of 24Na produced via 27Al(n,a)24Na, decaying by g-emission

Relevant properties of reactor materials

Material

Aluminum
Zirconium
Austenitic steel
Ferritic steel
Uranium

Density
(kg mÀ3)

2700
6490
$8000
$7900
1900

Specific heat
(J kgÀ1 KÀ1)

887–963
254–285
377–565
440–494
111–167

Thermal
conductivity
(W mÀ1 KÀ1)

160–230
8–40
11–21
17–42
11–28

Melting
point ( C)

660
1852
$1425
$1525
1132

Emod (GPa)

70
88–98
190–201
200–210
176–208

CTE, lin.
(Â10À6 KÀ1)

23
5.7
$16
$12
13.9

Nuclear
cross-section
(barns)
sabs

ss

0.23
0.19
$3.0
2.5
7.6

1.5
6.4
$10
11
8.9

Sources: Matos, J. E.; Snelgrove J. L. Selected Thermal Properties and Uranium Density Relations for Alloy, Aluminide, Oxide, and Silicide
Fuels; IAEA- TECDOC-643, International Atomic Energy Agency, Vienna, 1992; pp 1–19, article Appendix I-1.1 in Research reactor core
conversion guidebook Volume 4; Fuels (Appendices I–K). Lide, D. R., Ed. CRC Handbook of Chemistry and Physics, 86th ed.;
Taylor & Francis: Boca Raton, FL, 2005–2006; Gale, W. F.; Totemeier, T. C., Eds. Smithells Metals Reference Book, 8th ed.; Elsevier and
ASM International, Amsterdam and Materials Park, OH, 2004; Cverna, F., Ed. ASM Ready Reference: Thermal Properties of Metals;
ASM International: Materials Park, OH, 2002.


Performance of Aluminum in Research Reactors

with a half-life of 15 h. In alloys, long-lived radioactivity arises from decay of isotopes produced from
alloying elements and residual impurity elements
present in the aluminum, primarily 65Zn, 51Cr, 59Fe,
with half-lives of 250, 28, and 45 days respectively.
So if low residual radioactivity is an objective it can
be met to a large extent by avoiding alloys containing
significant quantities of Zn, Cr, and Fe.

Aluminum is not without its shortcomings. It has a
low elastic modulus and low melting temperature.
The former means that in their annealed conditions
aluminum alloys have low strengths compared with
annealed austenitic steels, Zr, and bcc metals. However, aluminum can be hardened by various treatments
as described in Section 5.07.3.2. However, the low
melting temperature of 660  C imposes operating
temperature limits of 100–150  C, which are homologous temperatures of 0.4–0.45Tm where lattice
vacancies are mobile and can invoke susceptibility
to creep and stress relaxation. Even without imposed
stresses, the strength condition of prehardened alloys
can become compromised at temperatures above
150  C because of the possibility of thermal overaging
as described in Section 5.07.3.2 Aluminum has poor
abrasion resistance. It can be sensitive to localized
galvanic and pitting corrosion. It is prone to liquid
metal embrittlement, particularly Hg. Difficulties
may be encountered in obtaining leak-tight fusion
welded joints for hi-tech applications, mainly due
to porosities resulting from solidification shrinkage
(volumetric change) and dissolved gases, in particular,
hydrogen.5 In addition, aluminum does not undergo a
color change on heating, and during manual welding
may melt abruptly without warning, allowing overheating that can cause excessive sagging and dropthrough of the weld bead. The advent of a solid-state
joining process, namely friction-stir welding,6 has
largely overcome those welding troubles.
5.07.3.2 Alloy Types, Temper
Designations, and Tensile Properties
There is no universally embraced international standard system for defining the types and conditions
of aluminum alloys. The International Standards

Organization does have classifications for aluminum
and its alloys, but most countries adhere to their
own systems. The system followed in the United
States of America is ANSI H35.1-1990, instituted
by the American National Standards Institute and
supported by the Aluminum Association and ASM
International. The ANSI system and the US alloys

147

covered by it are described in reference,7 which is an
excellent source of aluminum data; it includes a short
list of alloys for other nations and their national
designations. The ANSI system is used herein. In its
entirety, it is a morass. Here, it is outlined just to the
extent that is necessary to provide an uninformed
reader with enough details to understand the nomenclature and the various processing treatments and the
upper service temperature limits those treatments
impose for maintaining stability of the processed
materials.
The system has two classifications, one for
wrought alloys and one for cast alloys. Only the
wrought alloy classification is described here. Briefly,
it is an eight-character code consisting of two groups
of four characters separated by a hyphen. The first
four characters are all numerals and they identify the
alloy group by chemical composition. There are eight
aluminum alloy groups. The first digit of the first alloy
group is 1, which represents alloys with a minimum of
99.00 wt% aluminum. In this group, the major foreign

elements are Fe and Si, which are really residues from
the aluminum extraction process and will be found to
various degrees in all aluminum alloys. The next three
digits in the group identify specific alloys in the same
series, and the group as a whole is denoted the 1xxx
series, often vocalized as the one-thousand series.
The other seven alloy series are 2xxx (major alloying
element, Cu), 3xxx (Mn), 4xxx (Si), 5xxx (Mg), 6xxx
(Mg þ Si), 7xxx (Zn), and 8xxx (other element).
An upper case X preceding the series identifier
numeral indicates an experimental alloy.
The second group of four characters in the eightcharacter designation represents the temper condition, that is, the heat treatment or degree of cold
work. The first character of the four-character temper condition is an upper case letter representing a
type of treatment. The other three characters are digits
indicating variations within the treatment. There are
many temper treatments. Only the three treatments
most likely to be encountered in RR materials are
described here. They are ‘O’ for the fully annealed
condition, ‘H’ for a strain-hardened condition, and
‘T’ for a precipitation-hardened condition. The
O condition is attained by annealing the material at
about 340  C then slowly cooling it. There are no
specified variations of the O condition. The H temper
is more complex. The first digit after the H is a 1, 2,
or 3. H1 signifies strain-hardened only. H2 is strainhardened and partially annealed. H3 is strain-hardened
and stabilized by a low temperature heat treatment.
The second digit, that is, the one following the H1,


148

Performance of Aluminum in Research Reactors

H2, or H3 designation is a number between 1 and
8 and is the degree of reduction in thickness or crosssectional area given to the alloy in its fully annealed
condition to bring it to the desired strength level.
Level 8 corresponds to a maximum reduction of
about 75%. Level 1 represents approximately oneeighth of 75%, 2 is two-eighths, and so on. The third
digit, if used, implies a variation of the two-digit
temper. Partial annealing for the H2 condition is
applied to products that are strained beyond the
desired final amounts and are then brought back
to the needed strength level by the partial anneal.
Stabilization heat treatment for the H3 condition
is applied to those products that, unless stabilized,
would gradually age-soften at room temperature.
Partial annealing also inhibits age softening. This
tendency for softening of some cold-worked aluminum alloys at room temperature is important
because such recovery requires the involvement of
mobile lattice vacancies and/or self-interstitial
atoms that promote climb and rearrangement of the
cold work dislocations. It indicates the occurrence of
atomic movement at room temperature, which, as we
shall see shortly, is a factor affecting the development of radiation damage in aluminum.
In addition to hardening by cold work, aluminum
can be strengthened by solid solution treatment
and by precipitation hardening. Only two alloying
elements, Mg and Li, have sufficient solubility
(several %) at room temperature to provide significant solid solution strengthening. Al–Li alloys are
not recommended for reactor use because natural

Li contains about 7.5% 6Li, which has a large crosssection for transmutation to 3H and 4He, both of
which can be highly detrimental to aluminum.
So the only solid solution-hardened alloys available
for reactor use are the 5xxx (Al–Mg) series. Other
metallic elements, principally Cu, Si, and Zn, have
little or no solubility in aluminum at room temperature but are modestly soluble at higher temperatures
near the melting point. This latitude permits considerable strengthening of such alloys by quenching-andaging, also known as precipitation hardening. The
ANSI designations for the precipitation-hardened
T conditions comprise ten subdivisions, T1–T10.
For all T treatments, the alloy is heated to a temperature of 500–540  C to dissolve segregated alloying
elements, followed by a rapid quench into cold water,
which gives an unstable supersaturated solid solution.
Precipitation is achieved by allowing the material to
sit at room temperature for periods of weeks called
‘natural aging’ (tempers T1–T4) or by ‘artificial

aging’ at temperatures of 160–190  C for times of
6–24 h (tempers T5–T10). Flattening or straightening treatments may be applied before or after the
aging treatment and are indicated by numbers in
the third and fourth character positions. The temper
conditions for aluminum alloys most often encountered in RRs are T4, T6, and T651. A T651
condition indicates a material that has been artificially aged then subjected to a light stretching
operation insufficient to change its mechanical
properties from those of the T6 condition. Of the
precipitation-hardened alloys, the 6xxx series hardened by precipitates of Mg2Si is by far the most
popular for RR service. The 6061 alloy in its ÀT6
and ÀT651 conditions has been approved for service as a class 1 nuclear components material under
the Boiler and Pressure Vessel Code of the American Society of Mechanical Engineers, Case N-519.8
Two types of precipitation-hardenable wrought
aluminum alloys, the 2xxx series (Al–Cu) and the

7xxx series (Al–Zn), both of which can be heat
treated to greater strengths than the 6xxx alloys,
are not usually found in nuclear reactors. Some
2xxx alloys are prone to aqueous pitting corrosion
or may release Cu ions to the coolant that could
be deleterious to other materials in the reactor
such as stainless steel. The 7xxx series alloys
have too low ductility and are the most difficult to
weld. Their high zinc contents will cause high
radioactivity.
Unlike the cold-worked 1xxx alloys that can
undergo recovery at room temperature, the
precipitation-hardened alloys are thermally stable
at temperatures up to about 150  C provided they
have been given appropriate natural or artificial
aging treatments. However, exposure to higher temperatures will cause overaging and associated reduction in mechanical strength. This softening is
illustrated in Figure 1 for 6061-T6 alloy after heating
to various temperatures for various times and testing
at room temperature.9 It can be seen that softening is
promoted by both time and temperature. For times
up to 1 h, softening commences at about 200  C and
is substantial but incomplete at about 370  C. For
a longer exposure of 1000 h, the softening begins
around the aging temperature, indicated by the
down-pointing arrow, and is essentially complete at
temperatures between 260 and 300  C. The data in
Figure 1 are for specimens reheated without load.
If reheating occurs under loads sufficient to induce
creep and stress relaxation, the softening temperatures are pushed downward.

Performance of Aluminum in Research Reactors

149

350
Softening effects of reheating temperature and time on room temperature
properties of 6061-T6Al (originally aged 18 h at 160 ЊC)

300

200

1000 h

30 min

6 min

150
50
100
Elong., 1000 h

25

Elongation

50

Elong., 6 and 30 min

0

0

100

200
Reheat temperature (ЊC)

% Elongation

0.2% yield stress (MPa)

YS
250

0
400

300

Figure 1 Softening effects of reheating temperature and time on room temperature properties of 6061-T6 aluminum
(originally aged 18 h at 160  C). Data from Structural Alloys Handbook, 1989 ed., Vol. 3, Battelle Memorial Institute, Columbus,
OH, 1989; p. 14.

Table 2

Example alloys and their room temperature tensile properties

Alloy

Composition (wt%)

YS (MPa)

UTS (MPa)

Elongation (%)

1100-O
4032-T6
5052-H34
6061-T651

<1 (Fe þ Si)
12.2Si, 1Ni
2.5Mg, 0.25Cr
1Mg, 0.6Si, 0.28Cu, 0.2Cr

35
320
210
280

90
380
260
310

40
9
16
17

Source: Aluminum Standards and Data, 10th ed.; The Aluminum Association: Washington, DC, 1990.

Table 2 gives typical tensile properties of various
Al alloys employed in RRs. The weak 1100-O alloy
is simply annealed commercial purity aluminum with
no deliberate alloy additions; it is hardenable to an
H condition by cold work if so desired. The 4032
alloy is a eutectic composition of Si in aluminum that
has been solution-treated and aged to create finely
divided precipitates of Si; this alloy is used principally as a filler wire to improve the weldability of
aluminum alloys. The 5052 alloy is a solid solution
alloy of 2.5% Mg with a small amount of Cr added to
control grain size and strengthen the grain boundaries. The particular 5052 alloy in the table has been
work hardened to a 4/8, or half-hard, condition
before stabilization. The 6061-T651 alloy has been
solution treated and artificially hardened by precipitates of Mg2Si phase and its precursors, then given a
mild stretching treatment.

5.07.4 Fuel Elements
The most crucial and demanding applications of
aluminum in RRs are in the fuel elements. There it
is used inside the fuel element as a thermal conduction matrix in which a dispersion of fuel particles is
embedded and as a cladding material that protects
the fuel from corrosive attack by the cooling water,

retains fission products, and transfers heat from the
fuel and matrix to the coolant. As RRs matured,
considerable improvements were made in the fuel
elements. Most fuel in RRs is no longer unalloyed,
metallic a-phase uranium whose orthorhombic crystal structure is prone to severe radiation growth and
swelling leading to distortion and cracking. It has
been ousted by more stable and more isotropic uranium compounds that can also better accommodate
fission gases with minimum swelling. The most


150

Performance of Aluminum in Research Reactors

common ones are U3O8, UAlx, where x can be 2, 3, or 4
but is usually considered10 to be a mixture of 3 and 4,
and U3Si2. There is also a hydride fuel, U–ZrH1.6,
which is used exclusively in the open-pool TRIGA
(test, research, isotopes, General Atomic) types where
the fuel is in the form of slugs comprised of particles of
U dispersed in the ZrH1.6 phase (see Chapter 3.12,
Uranium–Zirconium Hydride Fuel). Originally, the
TRIGA slugs were sheathed in aluminum, which has
now been replaced with stainless steel or nickel alloy.
However, TRIGA reactors still contain other aluminum components.
There is no outstandingly superior aluminum
cladding alloy. The most common aluminum cladding
alloys are 1100 and the stronger 6061. Other alloys
have been investigated in neutron irradiations,11
namely 5052; X800N, where N is 1, 2, or 3 and

whose compositions are Al–$1Ni–$1Fe; and two sintered aluminum powder alloys, M257 and M470,
which were fabricated by ball milling flake powder
of 1100Al in air until it contained a dispersion of 6%
and 10% Al2O3, respectively, then consolidating by
pressing, sintering, and hot rolling. The Mxxx alloys
were deemed to be no better than 1100 and 6061
types. They are more difficult and expensive to
make and harder to weld than regular melted-type
alloys. In Europe, particularly in France, two preferred
alloys are AlFeNi, a relative of X8001 with the composition 1Fe–1Ni–1Mg, and AG3-NET, a 5xxx-type
with 2.5–3.0Mg and low residuals. The greatest
concern for cladding is its corrosion behavior (see
Section 5.07.5).
A feature of RR fuels is that they are much more
highly enriched in 235U than those in power reactors:
12–93% versus about 2.5%. Drivers for raising the
235
U levels were extended fuel cycles; the growing
demands for industrial and medical isotopes, particularly 99Mo the parent of the all-important medical
diagnostics tool 99mTc; and the need for higher neutron fluxes for increased production of the heavy,
transuranic isotopes. The use of highly enriched
uranium (HEU) meant higher heat generation and
required improved means of removing the heat. The
solution was the development of dispersion fuels
in which particles of the enriched fuel were distributed
in a matrix of thermal conductor material, all compressed together in sealed aluminum cans. The
thermal conductor is aluminum powder, usually a
1xxx-type, often atomized powder of better than
99.5% purity and particle size <100 mesh (150 mm
maximum, 23–48 mm mean). Atomized powder particles are denser, pour more easily than milled flake

powders, and have less low conductivity surface oxide
per unit volume. The aluminum matrix may occupy
more than 50 vol% of the fuel/aluminum mixture.
A huge advance in fuel element morphology and
heat removal efficiency took place when Eugene
Wigner designed his thin, curved fuel plates for the
high flux Materials Testing Reactor (MTR) built at
Arco, Idaho. A thin plate has a number of advantages
over cylindrical slugs. The rolling treatment used to
produce the plates from a fuel slab, or from a dispersion of fuel particles in aluminum matrix powder,
sandwiched between two aluminum cladding sheets
gives superior mutual contact of cladding, matrix, and
fuel for improved heat transfer paths to the cladding.
The much larger surface-to-volume ratio of plates
provides more efficient heat transfer to the coolant,
thus permitting higher fuel loadings per unit volume.
The benefit of a curved fuel plate is that any buckling
and bowing in the plate due to irradiation will be
focused in the direction of the radius of curvature.
Thus, in a fuel element comprised of a stack of
curved plates restrained at their edges and separated
from each other by cooling channels of the same
width as the thickness of the plates, any such distortions will be accommodated cooperatively in the
radial direction without unacceptable narrowing of
the cooling channels. An MTR fuel element contained
18 plates each about 72 mm wide and about 727 mm
long bent to a curvature of 140 mm radius in the width
direction. The plate thickness was 1.27 mm including a
minimum cladding thickness of 0.25 mm on each face.

The plate edges were brazed into sturdy side panels
to seal the plate edges and impart rigidity to the
assembly. The water gap was 1.27 mm. The cladding
and side panels were made from 1100Al; the Al
brazing alloy contained about 13% Si.12 This assembly was then enclosed in a long, rectangular aluminum box fitted with end fixtures for remote handling.
The end fixtures were castings of Al–7% Si. The
reactor core was built from groups of such elements
assembled upright in rectangular arrays held together
by aluminum grid plates. Refueling was done from
the top, and any element could be replaced by a box
of the same size containing a reactor experiment or
materials for isotope production, or a beryllium
reflector or a control rod. These MTR-type boxed
fuel elements in open grid core arrangements performed very well and became very common for RRs.
To satisfy demands for higher power densities and
more sophisticated tailoring of local neutron fluxes,
the next advancement in aluminum–clad fuel elements
was the development of upright, annular


Performance of Aluminum in Research Reactors

elements using curved fuel plates in which the fuel
particles may be required to be graded in concentration across the thickness and width. Beryllium reflectors surrounding the annulus direct neutrons from
the fuel back to the hollow center, or ‘trap,’ of the
annulus where reactor experiments and isotope targets are placed. The Be also creates additional neutrons from (n, 2n) reactions. Vertical holes
bored through the reflector allow passage of cooling
water and house irradiation experiments. Two highperformance beryllium-reflected reactors using annular fuel elements are the High Flux Isotope Reactor
(HFIR) at Oak Ridge National Laboratory (ORNL),
rated at 100 MW thermal and currently running at

85 MW, and the Advanced Test Reactor (ATR) at
Idaho National Laboratory, rated at 250 MW but
lately operating at 100–125 MW. The cores of these
reactors are of uncommon designs and deserve comment. The ATR core13 is 1.22 m diameter and 1.22 m
high. It contains a continuous serpentine-like wall of
fuel elements looped around nine flux traps each
about 120 mm diameter arrayed in a square 3 Â 3
grid. In plan view, the wall forms the shape of a
four-leaf clover. It fully embraces the central flux
trap and the four corner ones. The other four traps
lie just outside the wall; each is tucked in between the
junctions of two leaves and is about half wrapped by
the wall. At each corner lobe, there are four shim
control cylinders just outside the wall and six shim
rods at the neck of the wall inside the cloverleaf.
These controls allow each of the four lobes to be
run at different power levels simultaneously, as
needed by the experiments in the traps. The remainder of the space in the core is occupied by blocks of
Be reflector containing numerous experiment holes.
The wall is built14,15 from 40 individual wedgeshaped fuel elements, each containing 19 curved
fuel plates. The cross-sectional area of an element is
a 45 sector of a circular annulus. Its outer arc, plate
#19, has a radius of 137 mm and an arc length of
100.9 mm. Its inner arc, plate #1, has a radius of
$77 mm and an arc length of 54.1 mm. The 19 fuel
plates are attached by roll-swaging to 6061-T6Al side
panels 64.6 mm wide  1257 mm long. Within the
elements, the curved plates are concentric with the
circumferences of the traps. The plates are 1.27 mm
thick except for #1 and #19, which are thicker. The

water gap is 1.98 mm. The ATR fuel is UAlx enriched
with 235U to 93%, dispersed in a matrix of Al powder
and clad with 0.38 mm thick 6061-OAl.
The HFIR core16 is more compact, about the size
of a small trash can, into which are packed 540 fuel

151

plates in quite a different arrangement than in the
ATR. The core diameter is 435 mm and it is 791 mm
tall. It has a single central flux trap, 129 mm diameter.
The fuel is granules of U3O8 enriched with 235U to
93% and embedded in Al powder. The cladding is
6061Al. The core consists of two concentric annular
arrays of involute-curved fuel plates, as shown in the
sketch of a radial segment in Figure 2. The black
region in the fuel plates is the fuel dispersed in its
Al matrix; the white area is Al filler. There are 369
plates in the outer annulus and 171 in the inner
annulus. The plates are 610 mm high with widths for
the inner and outer annulus plates of 94 and 81 mm,
respectively, before bending. The plate thickness and
coolant gaps are 1.27 mm, as in the MTR-type elements. The two annuli are fabricated separately and
are united when loaded into the reactor. In addition to
the unique radial-like orientation of the fuel plates, the
fuel particles are uniquely distributed in the plates. To
minimize the radial peak-to-average power density
ratio, the thickness of the compacted fuel mix is varied
along the arc of the involute curve as seen in Figure 2.
This shaped region is backed by filler Al containing

no fuel particles. For the inner annulus, the filler
powder backing the shaped fuel region contains

Al filler
Al + 41 wt% U3O8
1.27 mm
Coolant
channel

Outer
annulus
sidewalls

Al + 30 wt% U3O8

Al filler +
B4C poison

1.27 mm

Figure 2 Horizontal section through a small segment
of the HFIR core showing fuel plate curvatures and
fuel distributions in the plates. Modified from
Binford, F.T.; Cramer, E. N. The High Flux Isotope
Reactor; A Functional Description, Vol 1B, Illustrations;
ORNL-3572 (Rev.2); Oak Ridge National Laboratory:
Oak Ridge, TN, 1968.


152

Performance of Aluminum in Research Reactors

particles of B4C burnable poison. Two concentric
cylindrical control plates clad in Al are located immediately surrounding the core. Outside the control
plates are four concentric cylindrical Be reflectors.
Because beryllium generates copious quantities of
helium and tritium from neutron irradiation, it tends
to swell and crack, particularly at the faces of its high
neutron flux regions. To retain chips spalled from
these surfaces, the reflector and any penetrations in it
are clad with aluminum. Four horizontal 6061Al beam
tubes and numerous vertical holes penetrate the
reflector.
Like most dispersion-type fuel plates, the HFIR
and ATR plates are fabricated by what is called
a picture frame technique. This utilizes powder metallurgy methods to disperse the fuel particles uniformly in the Al matrix and press the mixture into
a hard rectangular compact. The rigid compact is
placed in a window of the same size cut in an Al
slab or frame, which is usually the same alloy as the
cladding. Sheets of cladding material are welded to
the top and bottom faces of the filled frame and the
assembly is hot rolled through a large reduction in
thickness to ensure that the cladding is fully bonded
to the fuel charge and the frame. After verifying the
location of the fuel charge, the rolled plate is cold
rolled to flatten it and bring it to the specification
thickness. It is then given a final anneal at 500  C
to reveal any blisters and rolling defects in the
cladding surfaces. After verifying the location of the

fuel region, the plates are blanked to finished size in
a press.
Of course, it is not as simple as that. Strict quality
assurance standards have to be met, and at every stage
in the operation, there are numerous inspections
and rigorous sizing and confirmation tests. To reproducibly obtain the graded fuel distributions in the
HIFR plates, a special procedure was developed.17,18
A custom-designed contoured auxiliary die plate
is mounted over the cavity of the powder press to
facilitate mounding of the fuel/matrix powder mix in
a semicylindrical hump. Another auxiliary die plate is
added to allow filler powder to be leveled on top
of the humped fuel charge. This duplex charge is
withdrawn into the press cavity, the auxiliary dies
are removed, the rectangular punch is inserted into
the die mouth and pressure applied, and the charge
is consolidated in a single cold pressing operation.
The HFIR fuel plates are bent to the desired involute
shape in an elastomer-faced punch and die press.
They are welded into the cylindrical inner and outer
sidewalls of the fuel elements. The sidewalls are

machined from extruded-type 6061 aluminum tubing
in the T6511 temper. Twenty-seven equally spaced
circumferential weld grooves are turned on one face
of each sidewall, and slots are milled at prescribed
depths and angles on the other face of the wall. The
weld grooves intrude a short way into the slots.
The fuel plates are slid into the slots and properly
spaced with the aid of temporary Teflon separators.

The plates are machine welded in place through the
grooves. A 4043Al weld filler wire and an argon shield
gas are used. End fixtures machined from 6061Al
tubing are welded to the ends of the elements, and
final machining and inspection are conducted.
These multiplate fuel elements are a testimonial
to designer ingenuity and superb fabrication skills,
and the versatility of aluminum. Manufacturing these
fuel elements is not only painstaking but also expensive. In year 2007, each HFIR element cost $$1 M.19
It is replaced after its regular lifetime operating
cycle of 26 days. With so much effort and cost
invested in it, a rejected element is a severe financial
loss. The specifications and acceptance standards are
so high that the chances of producing a fuel element
completely free of specification violations are very
low. The first 30 000 fuel plates suffered a rejection
rate of 12%, and of the first 45 fuel assemblies, only
4 inner elements passed the final inspection.20 However, the degrees of severity of the violations were
all minor or were correctable. With waivers, all 45
elements were accepted and gave exemplary service.
After operation of the first 60 fuel cores at the
full design power level of 100 MW, 4 of them were
autopsied.21 No significant faults were found. The
in-reactor performance of these complex ‘aluminumbased’ fuel elements has been incredible, surpassing all
expectations.
Development of RR fuels and fuel plates is
continuing. Concerns over the possibilities of nuclear
weapons proliferation and terrorism led to establishment of the Reduced Enrichment for Research and
Test Reactor (RERTR) program at Argonne National
Laboratory.22 The goal of RERTR is to eliminate

the use of highly enriched uranium (HEU) in RRs
by converting to the use of low enriched uranium
(LEU). HEU is defined as uranium that has the
fraction of the fissile isotope 235U greater than 20%,
LEU is less than 20%. Historically, RRs have used
enrichment levels of 235U up to 93%. RERTR
is intended to be achieved without impairing the
safety and performance of the reactors and/or jeopardizing the production of important isotopes,
and at minimum cost for changes in fuel elements.


Performance of Aluminum in Research Reactors

In some RRs with modest uranium enrichment and
low power levels the RERTR LEU goal was met by
diluting the fuel with natural uranium. For many of
the high performance RRs (HPRRs) that must retain
their 235U levels and cannot tolerate the burden of
added 238U without excessive operational penalty, the
RERTR dilution can be achieved by replacing the
HEU fuel with LEU compounds or alloys containing
higher fractions of U. To that end, the initial focus of
RERTR was on the development of uranium silicide
fuels, U3Si and U3Si2, dispersed in aluminum and
clad with aluminum.23,24 While this move has been
successful for many RRs it is not sufficient for the
most demanding HPRRs. For them, attention has
turned away from dispersion fuels to monolithic alloy
fuels where higher U densities are attainable. The goal
is to develop fuel plates built from foils of LEU alloy,

250–500 mm thick, clad with aluminum.19,25–27 In
order to prevent buckling and cracking of the foil
during multiple rolling and recrystallization treatments and to inhibit radiation growth and warping,
there must be just enough alloying metal in the
U to stabilize it in its isotropic g-phase. Several
alloying metals are suitable, but the field of contenders has been reduced to the U–Mo system. A 90%
LEU-10% Mo alloy currently holds the best prospects. Some serious hurdles are recognized. Interdiffusion between the cladding and the fuel foil during
annealing and in-reactor exposure encourages the
formation of reaction layers of uranium–aluminum
compound(s) with low thermal conductivity and low
resistance to growth of fission gas bubbles. Such
layers threaten the integrity of the fuel/cladding
interface. Development of these layers is retarded
by additions of Zr or Ti to the fuel, or Si in the
cladding. When Si is incorporated in the cladding, it
is found to segregate at the fuel/cladding interface,
acting like a diffusion barrier. Thin film diffusion
barriers of Si, Zr, and ZrN applied directly to the
surfaces of the fuel foil by co-rolling and thermal
spraying have done well in reactor tests. The current
hot roll bonding processes used for attaching cladding to dispersion fuel plates may not be fully
adaptable to barrier-coated foil fuels. Other bonding
methods such as hot isostatic pressing are under
investigation. For HFIR plates, where the foils must
be tapered in both width and length and have involute shapes, fitting and bonding diffusion films and
cladding to the fuel foil on a mass production scale
is a challenge. Hot roll bonding will not work
because the foil and the cladding will not deform
to the same extent and will result in nonuniformly

153

thick cladding, and shear deformation during rolling
may damage the diffusion barrier. It is recommended19 that the tapered foil, bent to its involute
shape and with an adherent diffusion barrier, should
be prepared separately then sandwiched in shaped
recesses in two full-length clamshells of cladding
of appropriate thickness and bonded over all mating
surfaces. Alternatively, if the clamshells can be made
from a two-ply Al sheet, like the commercial OneSide Alclad™, the inner layer of, say, 1100Al, could
contain the ingredients for a diffusion barrier. The
hot isostatic pressing route may then allow bonding
and barrier filming in a single operation and in batch
mode. If burnable poison cannot be incorporated in
the fuel foil, it may be possible to accommodate it
in the inner cladding layer with the diffusion barrier
components.
The corrosion behavior of the Al cladding on alloy
foil fuel elements will need to be explored thoroughly. A penetration of the cladding will probably
be more serious than one in current dispersion fuel
plates because the alloy fuel will likely be more reactive and soluble in water than the dispersant-type
intermetallic and refractory fuels.

5.07.5 Corrosion
Metallic corrosion, the removal of metal atoms from
the metal surface by the electrochemical action of
the environment, has many forms: uniform, galvanic,
pitting, grain boundary, crevice, etc. Uniform corrosion and pitting are the types of most interest to RRs.
The greatest worry is the aluminum fuel cladding
where the environmental conditions are most aggressive and where an unexpectedly high corrosion rate

might breach the cladding and allow release of highly
radioactive fission products throughout the water
system. Pitting corrosion is the major form of attack
on the cladding of spent fuel elements during longterm storage in water basins.28 Herein, the focus is on
uniform corrosion of cladding.
Aluminum is a very reactive metal. In dry air,
it combines with oxygen to form an adhesive,
self-healing Al2O3 film that retards further oxidation
at the metal surface. Such films are usually quite thin,
tens of nanometers, usually described as amorphous.
Films formed in moist air and water are much thicker,
1 mm or more. The water-formed reaction films developed on aluminum cladding are variously described
as ‘hydrated oxides’ and ‘hydroxides,’ and ‘oxidehydrates,’ and they are generically referred to as


154

Performance of Aluminum in Research Reactors

‘oxide films.’ In HPRRs, the films grown on the fuel
cladding may be 20–50 mm thick.29,30 The most common corrosion products28,30 reported on aluminum
cladding are boehmite, a crystalline monohydrated
aluminum oxide, Al2O3ÁH2O, and bayerite, a crystalline trihydrated oxide, Al2O3Á3H2O. At temperatures
below about 77  C, the boehmite phase is formed
preferentially but may transform to bayerite with
continued immersion. At temperatures above $77  C
and below $100  C, a pseudoboehmite structure
grows, which may age to other hydrated oxide forms
or retain its pseudoboehmite structure. Between $100
and $400  C, crystalline boehmite will form. A gelatinous boehmite is the chemical precursor of both of the

crystalline hydroxides.30 The mature hydroxides are
normally white color but other hues have been
reported and may stem from absorption of Fe, Cr, Ni,
or other metal ions leached from steels in the reactors
or in the corrosion test loops.
The corrosion film is both the reaction product
and the medium through which the corrosion process
occurs. Whether corrosion is governed by ingress of
O and OH ions through the film to the metal surface
or by egress of Al ions to the film/water interface,
it is expected to be diffusion controlled. Thus, all else
being equal, an increase in film thickness should
lower the corrosion rate by increasing the diffusion
length, and vice versa. Therefore, the corrosion rate
should be parabolic with time and have an Arrheniustype dependence on temperature. Moreover, ideally,
if all the corroded metal was retained in the corrosion film, if the chemical composition and physical
structure of the film were constant throughout
the thickness, and if all of the film was retained
on the metal, the film thickness would be proportional
to the amount of metal corroded. Alas, such ideality
does not prevail. The corrosion process is confounded by a number of interacting factors, including
the following: there is a one-sided heat flux on the
cladding; the corrosion film is a thermal insulator
compared with the Al cladding, so the temperature
of the film will increase with thickness; the film may
not be of uniform composition and/or structure;
the film is soluble to some extent in water, and its
solubility is strongly susceptible to the pH of the water,
which is related to water composition; the film is
subject to erosion in flowing water and to spontaneous

spallation above some uncertain thickness, about
50 mm in one case.31 And to further complicate the
situation, there is wide variation in the ways the corrosion tests are conducted and in the parameters that
are measured.

The tests may be carried out in open cups, closed
autoclaves, vented autoclaves, closed loops, bypass
loops, or on used fuel plates. Evaporation or consumption of the water may require that it will need
to be periodically replaced or its volume readjusted.
Except in in-reactor tests and loop test systems with
bypass monitoring and adjustment of the water, the
chemistry of the water may change substantially during the test. Few corrosion rates for cladding materials are measured directly. They are usually derived
from measurements of the thickness of the corrosion
film. A thickness measurement gives the thickness of
the film adhering to the substrate at the time of the
measurement. It will not include film that has been
dissolved and/or eroded away. On a spent fuel element, it may include film that has formed in a storage
pool over time periods much longer than it experienced in-reactor, and with no forced cooling. During
preparation for post irradiation examination (PIE) in
a hot cell, the spent element is no longer fully
immersed. It gets hot and has to be periodically
sprayed with water to cool it. It has been opined21
that the resultant steaming and thermal cycling may
cause more corrosion than in-reactor operation and
underwater storage. There is no guarantee that the
density and the composition of the film will be invariant through the film thickness. On the contrary, multilayer films are more common than not. Almost all
films have a thin, monolithic base in contact with the
Al surface, presumably associated with the ubiquitous
air-formed Al2O3 film. On top of this base, there may
be one to three distinct layers. Some films contain

pores or are cracked. Only the films on irradiated fuel
elements have been exposed to the effects of neutron
irradiation and radiolysis of the water. The way in
which the film thickness is measured may be questionable, too. At least six different methods are used,
viz.: (1) Scaled measurements by optical or scanning
electron microscopy of metallographically polished
and etched cross-sections of the corroded test piece;
(2) micrometer measurements of the thickness of the
test piece before corrosion and after the corrosion
product is removed by electrolytic polishing until the
shiny metal is seen; (3) weight gains of coupons with
film in place; (4) weight losses of coupons after
removal of the film; (5) acoustic and eddy current
measurements with instruments calibrated against
accepted standard films; and (6) temperature increases
measured with thermocouples attached to the noncorroding back surface of the test piece during the test,
and related to spot film thicknesses measured metallographically after the test.


Performance of Aluminum in Research Reactors

A neglected aspect of film measurements is that
almost all of the measurements have been made on
specimens that, deliberately or unavoidably, were
dried at room temperature or at 100  C32 before the
measurement was attempted, or before the measuring
instrument was calibrated. Until recently, nobody
seems to have determined whether such drying treatments will shrink, spall, crack, or otherwise alter the
bulk film. The gelatinous surface layer that precedes
the crystalline corrosion films will almost certainly

be altered during dehydration. It is not uncommon
for test coupons to be dried, weighed, and placed
back in the test for the next exposure period, and
so on until the termination of the campaign. That
was the method used in one seminal laboratory test
study.33 The first periods in the full exposure sequence were the shortest ones, 1 or 2 days, and they
always showed the largest weight gains, usually
60–90% of the total weight gained during the full
duration of the test, which was about 22 days. Weight
gains after the first period were linear with time and
were relatively minor. That is not parabolic corrosion
behavior. The abrupt change in weight gain indicates
that something happened during the first interruption of the test that set the scene for a sudden switch
from an initial rapid corrosion rate to a subsequent
constant low rate. Likely, the first drying treatment
irreversibly altered the structure and permeability of
the hydrated film. Recent autoclave tests34 on AlFeNi
alloy reinforce that suspicion. It was demonstrated
that during a 34-day test, interruptions made every
7 days to remove, dry, weigh, descale, dry, reweigh,
and replace the test piece in the autoclave with
refreshed water for the next exposure period had
serious consequences to the corrosion kinetics. Without interruptions, the inner and outer oxide layers
were twice as thick, the weight gain was 26% higher,
and the amount of metal removed from the substrate
was 23% higher.
Some efforts have been made to correlate film
thicknesses with corrosion rates.31–33 Tests made
under controlled conditions in a corrosion loop31
found that the thickness of the boehmite film on

1100, 6061, and X8001 alloys was about 1.4 times
the depth of penetration into the aluminum regardless of changes in test parameters that changed the
film thickness, as long as there was no stripping or
spallation of the film. Using a literature value for the
density of boehmite, it was estimated that about 70%
of the corroded Al remained in the adherent film and
about 30% was lost to the coolant. When spallation
did occur, which was usually above a film thickness of

155

50 mm, the 1100 and 6061 alloys always showed localized attack of the aluminum under the spalled area,
whereas the X8001 alloy showed only uniform attack
under all conditions. This correlation was for a
closed, single set of data. It should not be considered
representative of all data and situations. Other data
by some of the same authors,32 where the principal
variables were temperature and flow rate, showed
that the ratio of corrosion product retained to the
weight of metal corroded ranged from a high of 0.54
at a low temperature of 170  C and flow rate of
6.1–9.5 m sÀ1 to a low of 0.08 at 290  C and 29 at
32.6 m sÀ1. Another source29 quotes a retention level
of 50–80% of the oxide on the cladding surface, but it
may be citing Griess et al.31 In general, the relationship between film thickness and corrosion rate is not
well established.
Film thicknesses from laboratory tests31,35–38 display power law growth with exposure time, but the
time exponents, preexponential factors, and activation energies differ from one experimenter to another
and may be applicable only to the particular set of
data from which they were determined. Nevertheless,

the laboratory tests have established that the corrosion films are sensitive to a number of interacting
factors. They include the temperature and surface
condition of the cladding; the heat flux density on
the cladding; and the temperature, pH, flow rate, and
purity of the water. In RRs, water purity is controlled
by filtration and ion exchange systems; it is also
linked to pH. With regard to pH, the films will
dissolve if the water is strongly acidic (pH < $4.5)
or strongly alkaline (pH > $8.5); films are most stable in the range 5.0–6.5, the closer to 5.0 the better.
The pH of reactor water and spent fuel storage pool
water tends to converge toward the desired range
by carbonic and nitric acids formed from CO2 and
N absorbed from air. It can be maintained close
to 5.0 by controlled additions of nitric acid. The
strongest increase in film growth is from increase in
temperature, and the controlling temperature is that
at the hydroxide/water interface.31 To lesser extents,
increased heat flux density and water flow rate
will raise the film growth rate. For the alloys 1100,
6061, and X8001, which all corroded alike until spallation occurred,31 the rate of oxide formation at a
heat flux of 1.58 MW mÀ2 was about half of that at
3.13–6.31 MW mÀ2, other conditions being the same.
At coolant flow rates in the range 7.6–13.7 m sÀ1, the
rate of accumulation of the corrosion product was
the same for all three alloys. Corrosion rates measured
on the insides of 1100Al production tubes39 were


156

Performance of Aluminum in Research Reactors

found to be unchanged by water velocities in the range
0.305–5.58 m sÀ1. Reduced water temperatures will
reduce the film growth rate.
Despite differences in strengths, compositions,
and microstructures, the alloys 1100, 6061, and X8001
all seem to have similar corrosion behavior under
similar conditions.31,32 Spalling tends to introduce
local attack in 1100 and 6061 but not in X8001.31
The AlFeNi alloy shows good performance at
temperatures up to 250  C in autoclave tests35 and
in-reactor exposures40 at temperatures below 120  C,
but it has not been tested under high heat fluxes.
AG3-NET cladding on U3Si2 dispersion fuel plates
undergoing in-reactor tests failed41 at a heat flux of
5.5 MW mÀ2. The cladding was swollen and breeched
by a combination of a very thick corrosion film and
subfilm intergranular corrosion. Cross-section X-ray
spectroscopy analyses showed that oxygen had penetrated intergranularly all the way through the cladding to the meat. The corroded cladding was
interesting in other ways. The outer oxide layer was
monolithic and was exceptionally thick, $100 mm.
Directly beneath it was a region about 80 mm thick
containing many round 30 mm size pores. Below the
porous region, the grain boundaries were enriched in
Mg and oxygen. The plates were intended to reach a
temperature of about 180–200  C at the exterior surface of the cladding and 220–240  C in the fuel.
Temperatures estimated from the thick corrosion
layers were >300  C for the water/corrosion film
interface and >400  C for the fuel meat. The AG3NET alloy has a history of intergranular cracking in

beam tubes and other structures in the Reacteur Haut
Flux at the Institut Laue-Langevin in France.
Although that cracking occurred at high fluences,
the irradiation temperatures were low. Such low temperature intergranular cracking is a sign of pending
weakness in the alloy and does not bode well for
applications at higher temperatures as in fuel
cladding.
The influences of neutron flux and radiolysis of
water are unclear. These parameters are omnipresent
in RRs and we might imagine them to strongly affect
aqueous corrosion of fuel cladding by damaging
the cladding and its corrosion film and by altering
the activity of the water. One researcher42 writes that
reports of neutron flux effects on the hydroxide films
are few and there is disagreement; he claims that
the opinion of most (Russian) researchers is that
neutron irradiation decreases, rather than increases,
the corrosion rate. Effects of radiolysis are uncertain.
According to Golosov,42 one Russian authority argues

that radiolysis may either accelerate corrosion by
facilitating cathodic processes or reduce corrosion
by promoting anodic passivation. Data from laboratory
corrosion loop tests without radiation fields seem to be
fairly compatible with data from irradiated fuel elements in terms of oxide thicknesses, compositions,
and pH effects. There are no outlandish differences
that would immediately draw attention to radiation
effects. At least, none that has been strong enough to
insist that loop tests should be repeated in irradiation
fields. A similar conclusion was reached for aqueous

corrosion of aluminum process tubes in production
reactors.39 Therefore, irradiation effects must be
modest at worst. However, there are some troubling
reports that seem to indicate large effects of irradiation fields in nonreactor conditions. Sindelar et al.43
studied 6061Al coupons exposed to moist air at 150
and 200  C, with and without exposure to a 60Co g
source at 1.8 Â 106 R hÀ1. Weight gains and film
thicknesses were measured. The corrosion product
was patches of loosely aggregated, randomly oriented
1 mm size boehmite crystals sitting on a thin monolithic base layer, even at 100% relative humidity
where the product was permanently under a film of
water. g-Irradiation seemed to double the weight
gains and increase the film thicknesses by a factor of
10. There was substantial surface blistering of the
base layer, attributed to hydrogen gas. The paper
provided no details of the experimental conditions.
Enquiries to the authors produced a lengthier publication44 with the missing details. Those details cast
grave doubt on the conclusions drawn from Sindelar
et al.43 In particular, the experiments with the g-field
were made under radically different conditions
than those without the field. Specimens for the
g-irradiations were sealed in small stainless steel
cans of just 78 ml and each can represented an uninterrupted test for a given exposure period of 1, 4, 8,
and 12 weeks. The tests without the g-field were made
in stainless steel autoclaves of volume 37 850 ml for
15 unequal exposure periods totaling about 30 weeks.
At the end of each period, the specimens were
removed, dried, weighed, and replaced in the autoclave with a new charge of water. In light of the
effects of interruptions described in Wintergerst
et al.,34 the effects of g-irradiation described in

Sindelar et al.43 and Lam et al.44 are inconclusive.
In the other work,45 Al coupons of undeclared composition and condition were exposed to static brackish water of pH 8–9, at undisclosed temperature
for periods of up to 30 days, with and without low
dose irradiations with neutrons from a 252Cf source


Performance of Aluminum in Research Reactors

(1010 n mÀ2 sÀ1) and, separately, g-rays from a 60Co
source at 15 Sv hÀ1 (1.6 Â 104 R hÀ1). Corrosion was
determined from weight losses. It was not stated
whether the specimens were recycled from one period
to the next. The neutrons and the g-rays had the same
effects and to the same degree; they promoted formation of a grayish layer on the specimen surfaces; they
reduced the weight losses by 25–30%; and they almost
eliminated severe pitting corrosion displayed by the
unirradiated specimens. None of these three reports
mentioned whether radiation heating was a factor.
The laboratory loop tests have verified the expectation that the corrosion film is a thermal insulator compared with the Al cladding, and they have provided31
a thermal conductivity value of 2.25 W mÀ1 K for
boehmite, which is a factor of 70–100 less than Al.
However, it is not always ascertained whether a particular film is boehmite or bayerite or a mix of both.
No thermal conductivity value is available for bayerite. When insulating films build on the Al cladding
of heat sources like the fuel and long-term heavy
isotope targets, the temperatures of the sources and
their claddings or containers will rise. This temperature rise will increase the corrosion rate and
the growth rates and dissolution rates of the corrosion
films. In HPRRs, a side effect of an increase in cladding temperature by the adherent corrosion product
is the threat of plate buckling.31 As described earlier,
the strengths of the cladding and Al fuel matrix can

be decreased significantly by tens of degrees increases
in temperature, and creep rates will increase. If an
insulating corrosion film increases the temperature
gradients between the center thickness of the fuel
plate and the surface of the film, and between the
fuel-loaded portions of the fuel plates and their cooler
frames, the plates may distort. If the distortion is not in
phase from one plate to the next, it might perturb the
coolant flow and accelerate the temperature changes.
Griess et al.31 envisaged that the insulation provided by
the corrosion-product film might be more of a limitation on the use of aluminum–clad fuel elements in
high flux reactors than is corrosion damage per se and,
in the worst case, may lead to burnout of parts of the
fuel plates. Fortunately, that prophecy has not been
fulfilled. Serious plate distortion has not been a widespread issue. One case of plate distortion is described
in Shaber and Hofman.30 Plate buckling found in some
MTR elements12 was blamed on new design changes.
It is recommended30 that new fuel elements
should be prefilmed with a hydroxide film to reduce
the rate of in-reactor buildup of the corrosion layer.
Tests32 with 1100, 5154, 6061, and X8001 alloys at flow

157

rates in the range 6.1–20.4 m sÀ1 found that preexposure of the test pieces to water at 250–300  C for 24 h
in an autoclave caused a significant improvement in
corrosion resistance, but not at higher flow rates. The
ATR elements are pretreated30 by immersing them in
water for 48 h at 180  C and pH 5.0. In the early days
of HFIR operation, the new fuel elements were often

stored in the reactor pool water for up to 3 months
before being placed into service. This immersion
resulted in the formation of a rather thick, gelatinous,
corrosion product film on the element surfaces.21 In
an attempt to avoid that condition, some of the elements were pretreated by boiling them in deionized
water for 24 h to produce a thin, boehmite film on the
surfaces of the elements before they were placed into
service. When the pretreated elements were used, the
coolant flow rate was found to gradually decrease
and the pressure drop across the elements gradually
increased during the reactor fuel cycle. No significant
damage was caused. Changes in coolant flow rate
and pressure drop were not observed when the reactor was operated with non-pretreated fuel elements.
Metallographic examinations of cross-sections of
the spent fuel plates revealed much thicker corrosion
films on the pretreated plates. Pretreatment of the
HFIR fuel elements was discontinued. Most RRs do
not practice pretreatment of their fuel elements. It is
proposed here that because of the seemingly large
effects of dehydration on retarding subsequent film
growth as discussed earlier, at least one in-reactor
trial should be made of a prefilmed fuel plate with
a dehydration step or a low temperature baking
treatment added. A drying treatment might also be
worthwhile for a newly spent fuel element before it
enters pond storage.
What we really need to learn from corrosion measurements and film thickness data is the thickness of
uncorroded Al cladding remaining on the fuel element at the end of reactor service, and whether that
thickness will be sufficient to continue to seal the
spent fuel through further corrosion expected during

cool-down storage in water basins. That is, we need
reliable corrosion rates pertinent to the particular
application. Corrosion product thickness data are
invaluable in identifying and characterizing the
major factors governing corrosion and the interplay
between them, but they are meaningless to corrosion
rates if a reproducible relationship between film
thickness and corrosion rate is not established. We
need predictability. To that end, efforts are underway
to derive predictive models for film thicknesses40,46
and corrosion rates.42 These models are in their


158

Performance of Aluminum in Research Reactors

infancy. Because they lack a large body of consistent
data to draw on, the authors must make many assumptions, fittings, and correlations to derive constants,
correction factors, adjustment factors, and augmentation factors. With so much flexibility built into
the prediction equations, it is not surprising that the
authors can find good correlations with selected data
from measurements made on spent fuel cladding.
This is not intended as a criticism of the modelers;
it is a reflection of the paucity of input data. Reliable
modeling is essential. But it needs reliable input data.
Data obtained from recycled test coupons should
either be excluded from the models or modeled as a
separate category. To be generically applicable, film
thickness models and corrosion rate models should

attempt to merge in a complementary manner.
In low power RRs where convective flow is
sufficient to take care of cooling and water quality
is adequately controlled, problems from corrosion
films formed on the aluminum cladding and on other
aluminum components elsewhere in the reactor are
uncommon. In HPRRs, the most prominent corrosion
problems were those in the early days of operation that
caused a milky turbidity of the coolant and a white
deposit and increase in surface radioactivity on all
surfaces exposed to the coolant. The turbidity was
identified as a fine suspension of boehmite, and the
g-radioactivity was consistent with decay of 24Na, both
effects attributable to corrosion/erosion of the fuel
cladding. The turbidity is created by increase in the
cladding temperature due to the warming effects of
the hydroxide film. In turn, the temperature of the
coolant in immediate contact with the film is raised.
This increases the solubility of aluminum oxide in the
immediate volume of coolant. When this small volume
moves on and merges with the cooler bulk coolant,
the solubility falls and much of the dissolved film is
released as a particulate suspension. Particles of film
washed directly into the coolant by erosion of the
cladding due to the high coolant flow rate contribute
to the turbidity. Since turbidity ensues when the concentration of aluminum in the bulk water exceeds the
solubility of the aluminum oxide, turbidity problems
are brought under control by tuning demineralization
treatments to remove dissolved aluminum from the
bulk water and by reducing the degree of dissolution

through adjustments in pH to between 5.1 and 5.4
where aluminum oxides have minimum solubility.
In-reactor pitting corrosion and galvanic corrosion have not been serious problems. Pitting of Al,
which is encouraged by the presence of ions of Cu,
halides, and bicarbonates, is more serious in storage

pools where poorer water chemistry and nearly stagnant water conditions may exist, but diligent monitoring and control of water chemistry can mitigate
these concerns. Intergranular corrosion has not been
a problem in RRs, but it could become an issue at
high irradiation temperatures as evidenced by the
AG3-NET cladding described earlier.
Overall, aluminum cladding has given very good
service in water-cooled RRs and continues to do so.
The major variables influencing the corrosion process
(es) and corrosion products are fairly well identified
except for effects of irradiation. More data from spent
fuel elements are needed to guide and refine models
for predicting film thicknesses and corrosion rates.

5.07.6 Radiation Effects
5.07.6.1

Basics

As in other metals, irradiation of Al with neutrons or
charged particles introduces lattice vacancies, selfinterstitial atoms, and transmutation products that
evolve into radiation damage microstructure, which
causes swelling, radiation hardening, and loss of ductility. Radiation damage effects in aluminum differ
from those in most other metals in two respects. One
is that the radiation damage is affected strongly by a

solid transmutation product, silicon, discussed more
in Section 5.07.6.2.3. The other is that Al is much
more tolerant of radiation effects than most other
metals. At least, it is for irradiations conducted at
ambient temperatures. Neutron irradiation of Al
at temperatures between 25 and 100  C does not
induce detectable radiation hardening until the fast
neutron fluence exceeds about 1 Â 1024 n mÀ2, whereas
in Fe and Zr, radiation hardening is detectable at
fluences two to three orders of magnitude less than
that.47 Moreover, even when Al is radiation hardened
at 25–100  C, it still retains significant ductility when
compared with considerably reduced ductilities in Fe
and Zr. This delayed display of radiation hardening
exists despite the fact that the number of atomic
displacements per atom in Al are about twice as
many as in other metals at the same fast fluence,
which is brought about by the lower displacement
threshold energy for Al. The larger part of Al’s better
tolerance of radiation damage is owed to its low
melting temperature, which makes its homologous
temperature high compared with those for Fe and Zr.
At room temperature, the homologous temperature
of aluminum is 0.32Tm, versus 0.175 for austenitic
steel, $0.17 for ferritic steel, and 0.26 for a-Zr if


Performance of Aluminum in Research Reactors

referred to the a ! b-transition temperature of about

860  C, or 0.14 if referred to the m.p. of b-Zr, 1852  C.
In general, noticeable thermally induced movement
of lattice vacancies will occur in metals at homologous temperatures above about 0.3Tm. Because of that
movement at room temperature in Al, there will be a
greater loss of radiation-produced vacancies and of
interstitials to mutual recombination, resulting in less
nucleation of the point defect clusters that are the
seeds of damage microstructure, hence less radiation
hardening.
A feature of radiation damage in polycrystals is
the absence of point defect damage microstructure at
grain boundaries and other incoherent interfaces.
Point defect clusters and voids do not develop on
the boundaries, and damage-free zones are formed
on each side of a grain boundary. This denuding may
be difficult to see in many high melting point metals
irradiated at temperatures below $100  C because
the zones are narrow. In Al, the high Tm allows
development of wide and conspicuous denuded
zones. Incoherent interfaces are comprised of structural dislocation networks and high equilibrium
concentrations of vacancies that make the boundaries deep sinks for absorption and recombination of
freely migrating point defects. They are pulled in
from the near regions of the butting grains, leaving a
volume of matrix straddling the boundary that is
diminished in radiation-produced point defects.
Due to the greater mobility of the interstitials and
the bias of dislocations for absorption of interstitials,
the zone deprived of interstitials is wider than that
for the vacancies. This creates an unbalanced concentration of vacancies at the rim of the denuded
zone. Therefore, vacancy clusters are encouraged to

form in that rim and they become more numerous
and/or larger than those in the grain matrix. Impurities and transmutation products are also drawn
into the grain boundary, but they are not annihilated
there; they accumulate. If they are largely insoluble,
as H, He, and Si are in Al, they will precipitate and
grow there, the gases as bubbles and the Si as particles
or films. Within the grains, the gases will stabilize
embryo clusters of vacancies and facilitate their
growth into voids as long as there is an excess of
vacancies. Some of the Si will attach itself to the
voids. Grown-in dislocations in the grain interiors
are also sinks for point defects.
Diffusion and binding of freely migrating point
defects and solute atoms are important for understanding and analyzing radiation effects. Some useful
parameters for pure Al are:48,49

159

 The threshold energy for atomic displacements is
25 eV compared with 40 eV or more for most other
metals.
 The self-diffusion rate, Dsd ¼ A(ÀQsd/kT), where A is a
constant and the activation energy Qsd is the sum
of the formation energy of a vacancy, Evf , and the
migration energy of the vacancy, Evm .
 For temperatures between 298 and 580 K, A is
$1.6E À 5 m2 sÀ1 and Qsd is $1.3 eV.
 For temperatures between 570 and 923 K, A is
$2.0E À 4 m2 sÀ1 and Qsd is $1.48 eV.
 Evf ¼ $0.6 eV.

 Evm ¼ $0.7 to 0.88 eV, deduced from Evm ¼ Qsd À Evf .
 Eif ¼ the formation energy of an aluminum interstitial atom, >$3 eV.
 Eim ¼ the migration energy of an aluminum interstitial atom, $0.1 eV.
 Evb-s ¼ the binding energy between a vacancy and a
solute atom and is considered to be <0.1 eV for Ag,
Cu, Mg, Zn, and Si.
 Eib-s ¼ the binding energy between an aluminum interstitial atom and a solute atom. No
values are available, but for the solute to reduce
b
needs to
radiation damage microstructure EiÀs
m
m
b
ÀE
Þ
þ
E
.
be >ðEv i
vÀs
Solutes that seem to reduce radiation damage structure most strongly at concentrations of 100 appm
are Cr, Cu, Mn, Ti, V which have the largest negative
lattice misfits, defined as (a-a0)/fa0, where a0 is the
lattice parameter of pure Al, a is the lattice parameter
of the alloy, and f is the atomic fraction of solute.
No relationship is found between degree of radiation damage and thermal diffusion rates of solutes.
5.07.6.2

Microstructures

Examples of the damage microstructure in highpurity Al irradiated to a fast neutron dose of
3.5 Â 1024 n mÀ2 at $50  C are presented in Figure 3.
The two photographs are not the same field. The
one on the left shows typical radiation-produced
dislocation loops. Some of the smaller spots are
particles of radiation-produced Si. The other photograph is tilted to put the dislocations out of contrast and reveal the voids more clearly; they are
facetted. The loops and voids are of order 30 nm
diameter. In ferritic steel, austenitic stainless steel,
and Zircaloy-4 alloys irradiated under similar
temperature and neutron fluence conditions as in
Figure 3, the radiation damage microstructure
is resolvable as 1–2 nm black dots.47 The loops in
Figure 3 are not faulted. Nobody has reported


160

Performance of Aluminum in Research Reactors

(a)

0.1 mm

(b)

0.1 mm

Figure 3 Dislocation loops (a) and voids (b) in high-purity aluminum after irradiation at 50  C to a fluence of
3.5 Â 1024 n mÀ2 (E > 0.1 MeV).

faulted loops or stacking fault tetrahedra in neutron-irradiated Al, presumably because the
very high stacking fault energy (SFE) of Al,50,51
about 160–200 mJ mÀ2, would inhibit faulting. Yet
faulted loops, and multilayered loops, are formed in
aluminum during electron bombardment in a highvoltage electron microscope52 and in thin foils that
are water-quenched from temperatures near the
melting point then aged.53 The occurrence of faulting in these cases is increased with increasing aging
temperature and impurity level. It is possible that
the SFE may have been reduced by contamination
occurring through the foil surfaces.
5.07.6.2.1 Fluence

Experiments54 made to establish the minimum fast
neutron fluence for the onset of visible radiation
damage microstructure in high-purity Al foils irradiated at <60  C indicate a threshold in the vicinity
of 3 Â 1023 n mÀ2, E > 1 MeV. The specimens were in
two conditions, annealed and cold worked, and were
of three thicknesses, $100 nm, 12.7 mm, and 76.2 mm.
None of the 100 nm foils showed damage structure
for exposures up to 5 Â 1022 n mÀ2. At a fluence of
3 Â 1023 n mÀ2, the 12.7 and 76.2 mm foils, both
annealed and cold worked, showed many small
loops, about 25 nm size in low concentration and
spotty distribution. After a dose of 1.6 Â 1024 n mÀ2,
the loop concentration was higher, but there was little
or no change in size. At both of these doses, the
annealed specimens contained loosely tangled
dislocation lines at higher density than is characteristic of well-annealed high-purity Al. These dislocations were kinked, and some of them that moved
while under observation in transmission electron

microscopy (TEM) examination were seen to have

been pinned at the loops. In annealed specimens, the
grain boundaries had well-defined denuded regions
about 0.35 mm wide on each side at a dose of
1.3 Â 1024 n mÀ2, and in the regions next to the
denuded zones many of the loops were large and
there were dislocation segments among them. Some
of these segments spanned the denuded zones. These
denuded regions are wider than the 100 nm thick foils
in which no radiation damage was found, suggesting
that the radiation point defects had migrated from
those foils. For cold-worked specimens, nothing
was said about the cold work dislocations. It is quite
possible that considerable recovery of the coldworked dislocation structure occurred before and
during irradiation. At grain boundaries in the coldworked material after irradiation to 6 Â 1023 n mÀ2,
a few loops and some dislocation segments are present within a distance of about 0.35 mm from the
boundary. In the adjoining regions, loops and dislocation segments are present in high concentrations, and
many of the loops are large, >0.1 mm, and they encircle smaller loops or are kinked by smaller loops.
Evidently, the dislocation segments are portions of
growing loops. The tangled dislocation lines in the
annealed specimens at the lower doses probably arose
from growth of the earliest loops. No radiation voids
were seen in these experiments. It was speculated that
the loops were vacancy-type, growing from collapse of
vacancy clusters produced by the irradiation. French
studies55,56 have corroborated and enlarged on the
heterogeneous nature of evolution of early damage
microstructure and the roles of dislocations in Al.
With increasing fluence at constant irradiation

temperature, the loops evolve into dislocation lines,
and voids and Si precipitates arise, which increase in
concentrations and sizes. The voids are larger and


Performance of Aluminum in Research Reactors

161

0.1 mm

Figure 4 Denuded grain boundary and associated void
enhanced regions in 4–9 purity aluminum after irradiation at
50  C to a fluence of 3.4 Â 1026 n mÀ2 (E > 0.1 MeV).

less numerous than the Si precipitates. The voids are
facetted and so are the larger particles of Si. Most,
if not all, of the voids have a facetted Si particle
attached to the outside of one facet of the void.
The majority of the Si particles are not attached to
voids. The grain boundary denuded regions are not
enlarged, but the voids at their rims are exaggerated,
as illustrated in Figure 4. An unpleasant potential
consequence of these sheets of large voids is that
in the event of an unexpected overload, they may
provide paths for premature failure by a tearingalong-the-dotted-line-type separation.
5.07.6.2.2 Temperature

Raising the irradiation temperature coarsens the
damage microstructure and decreases the degree

of radiation hardening for a given dose. At 150  C,
Figure 5, the dislocation structure almost disappears;
there are fewer voids, but they are larger than
at 50  C and are strongly facetted, and many of
them are very much elongated.57 Particles of
radiation-produced Si are attached to one face of a
void, usually at the narrow end of elongated voids.
Freestanding Si particles are likewise facetted and
elongated, some in ribbon shapes. The denuded
regions straddling grain boundaries are wider, about
1 mm each side. For annealed materials irradiated at
temperatures above 150  C, certainly above 200  C,
no dislocation-type radiation damage microstructure
or voids are produced; coarse Si particles are seen. In
cold rolled pure Al, some large cavities remained58
after irradiation at 220  C. During postirradiation
annealing59 removal of damage microstructure was
slowed by impurity content and by higher doses. For
1 h anneals of 1100Al, void swelling began recovery at
200  C and was almost complete at 300  C where gas

1 mm

Figure 5 Voids in high-purity aluminum irradiated at
150  C to a dose of 2 Â 1025 n mÀ2 (E > 0.1 MeV). Right:
Preinjected with 3 appm He. Reproduced from Farrell, K.;
Wolfenden, A.; King, R. T. Radiat. Eff. 1971, 8, 107–114, with
permission from Taylor and Francis.

bubble swelling intervened. Void ripening preceded

void elimination. Si precipitates ripened and were
prevalent on grain boundaries.
5.07.6.2.3 Transmutation products

The gases helium and hydrogen produced from (n, a)
and (n, p) reactions of fast neutrons with lattice atoms
considerably affect the development and effects of
radiation damage structure by encouraging the nucleation of voids, dislocation loops, and bubbles.57,60,61
Helium is insoluble in Al. It binds strongly with
vacancies and has very limited mobility. Hydrogen
is almost insoluble at ambient temperatures, becoming more soluble with increasing temperature. In the
lattice, it is mobile even at room temperature. The
degree of promotion of voids and loops by the gases
decreases with increasing irradiation temperature,
and the promotion of bubbles increases. In the righthand micrograph of Figure 5, it is evident that even
for a high irradiation temperature of 150  C, the
presence of just 3 appm He implanted at room temperature has boosted nucleation of voids. The string
of voids in the field is not on a grain boundary. It
probably marks the position of a grown-in dislocation
present during implantation that has climbed away
during irradiation. At higher irradiation temperatures58 or during postirradiation anneals,59 the gases
form bubbles. The levels of helium and hydrogen
produced in Al are not widely different from those
in other metals. For the purpose of comparison we
can use the production rate tables of Gabriel et al.62


162

Performance of Aluminum in Research Reactors

to estimate the levels of He and H produced in Al,
Fe, and Zr during a 1-year exposure in the high flux
peripheral target positions of the HFIR core. There,
on the horizontal mid-plane at current operating
conditions, the annual fast neutron fluence will be
2.4 Â 1026 n mÀ2, E > 0.11 MeV, and the thermal fluence will be 4.5 Â 1026 n mÀ2. In Al, this fast fluence
will generate gas concentrations of 11 appm He and
63 appm H, and in Fe there will be 5 appm He
and 95 appm H. In Zr, which is noted for its low
cross-sections, only about 0.25 appm He and 5 appm H
would be produced. The atomic displacement levels
for the three metals can also be calculated. They are
36 dpa for Al, 18 dpa for Fe, and 17 dpa for Zr.
The larger dpa level in Al is due primarily to its low
effective displacement energy, 25 eV versus 40 eV for
Fe and Zr.
There are three interesting reactions with thermal
neutrons that produce gases from the foreign elements Li, B, and Ni, which may be present in some
Al alloys. The spatial distributions of gases from these
sources are each different. Lithium has high solubility in Al and generates a uniform distribution of
gases. Boron and Ni are insoluble and they produce
localized concentrations of gas. Lithium is not a common impurity in Al, but there is a commercial series
of Al–Li alloys developed for their lightweight highstrength properties. Laboratory-made Al–Li alloys
have been used to study radiation hardening, helium
embrittlement, and swelling.63,64 Natural Li contains
7.5% of the 6Li isotope, which has a capture crosssection of about 950b and decomposes to He and
tritium via the reactions 6Li þ nth ! 7Li ! 4He þ 3H.
The present writer65 made an Al–0.052 wt% Li
alloy using 6–9 high-purity Al and Li enriched to

98% with 6Li, and irradiated it to a dose of
5.5 Â 1025 n mÀ2, E < 0.0253 eV and 2.2 Â 1025 n mÀ2,
E > 0.1 MeV at $55  C. About 95–99% of the 6Li was
burnt up to produce about 2200 appm each of He and
tritium. The atomic displacement level was about
3 dpa, not including any displacements from the
recoiled gases. The effects of these high levels of
gases were striking, see Figure 6. The insert is an
enlarged view of the matrix cavities. Compared with
irradiated pure Al control specimens, the concentrations of matrix cavities were increased 1000-fold,
and their sizes decreased tenfold; dislocation densities
were increased tenfold. Most grain boundaries were
crammed with large bubbles, many so interconnected
that it was difficult to obtain thinned foils for TEM
examination because the grain boundaries were eaten
away before much thinning of the grain interiors

0.1 mm

1 mm

Figure 6 Modification of void structure by very high
helium and tritium levels from burnup of 6Li. Reproduced
from Farrell K.; Houston J. T. Combined Effects of
Displacement Damage and High Gas Content in Aluminum,
ORNL-TM-5395; Oak Ridge National Laboratory: Oak
Ridge, TN, May 1976. Also available in Proceedings of
International Conference on Radiation Effects and Tritium
Technology for Fusion Reactors, Gatlinburg, TN, Oct. 1–3,
1975, U.S. Department of Commerce CONF-750989,

Mar 1976; pp. II-209–II-233.

occurred. The grain boundary in Figure 6 is one
with a low concentration of bubbles. Hardness measurements gave a Vickers pyramid hardness (VPH) of
137 MPa for the annealed, unirradiated Al and the
alloy, and 382 and 902 MPa for the irradiated specimens. In bend tests made in air and liquid nitrogen
(LN), the unirradiated materials and the irradiated
pure Al were bent through full circles without rupture. The irradiated alloy broke with an audible crack
and with no detectable plastic strain. Fracture was
accompanied by release of tritium. The fracture surfaces displayed 100% intergranular failure. These
are incredible hardening and embrittling effects of
the gases. Electron microcopy examination of carbon
replicas taken from the fracture surfaces showed huge
irregular interconnected bubble cavities. Failure
occurred by plastic tearing of the small areas of intact
grain boundaries between the cavities. Postirradiation
annealing treatments caused the appearance of a coarse
distribution of large facetted matrix cavities superimposed on the small matrix cavities, and with
no denuding of the surrounding small cavities. These
enlarged cavities were frequently associated with large
silicon particles that grew concurrently during the
anneals. An anneal at 500  C showed incipient disintegration of the specimens and TEM foils could not


Performance of Aluminum in Research Reactors

be obtained. It was postulated that the large cavities
grown during annealing were tritium bubbles.
Al often contains trace quantities of B in the
form of small B4C inclusions. Natural B contains

19.8% 10B, which has a large neutron capture
cross-section of 3835b, producing Li and He via
10
B þ nth ! 11B ! 7Li þ 4He. The range of the
recoiled He atoms is about 5 mm, and the He is
segregated in a well-defined band in a halo around
the parent inclusion. The larger Li atom has a smaller
range and is soluble; it forms a diffuse halo. At low
irradiation temperatures and low doses, the halos are
very prominent because of heavy decoration with
dislocation loops. As the dose increases, the loops
grow and move off leaving dense halos of voids, especially for the He halo. The writer has seen hundreds of
these halos. Most of them were circular or near circular, with an occasional cigar shape, depending on the
shape of the mother particle. Most were isolated
randomly, but some were in groups or were strung
in chains on a grain boundary. One is illustrated in
Figure 7. This particular halo is slightly squashed,
following the elliptical contour of the central particle.
The dark region of the outer halo is actually filled
with small cavities, resolvable at higher magnification. Where a halo intercepts a grain boundary the
voids seem to disappear, but during annealing they
become visible as bubbles on the boundary that grow

1 mm

Figure 7 Damage halos around a suspected B4C particle
in 1100-OAl irradiated to 2.9 Â 1026 n mÀ2 at about 55  C.

163

faster than those elsewhere in the halo. Such highly
heterogeneous distributions of transmutant gas have
been perceived more as a novelty than as a possible
threat to the integrity of the host material. This
attitude may be unwarranted. A highly localized concentration of helium in a patch on a grain boundary
could be a prime site for premature helium embrittlement at stresses and temperatures below the ranges
for normal helium embrittlement elsewhere in the
specimen. A spongy helium halo that intercepts the
surface of Al cladding may provide a potential site
for initiation of local corrosion. For these reasons, it
might not be a good idea to consider placing particles
of B4C burnable poison in single-layer cladding on
monolithic LEU fuel plates; a better location would
be in the inner layer of a two-layer cladding.
Some of those considerations apply to He produced from Ni in Al. It comes from the 59Ni isotope,
which is not found in natural Ni. The 59Ni must
first be created from 58Ni that comprises 68.1% of
natural Ni. The two-step process66 to yield the He is
58
Ni þ nth ! 59Ni; 59Ni þ nth ! 56Fe þ 4He. Helium
generation via this route does not scale linearly with
time. It is slow to start while the 59Ni accrues, then
it increases as the square of the fluence. It is favored
by long-term exposures or strongly thermalized neutron spectra. Only trace quantities of Ni are found
in most Al alloys except the X8001 and AlFeNi-type
alloys, which contain a nominal 1 wt% Ni. These
alloys were developed for cladding because early
laboratory corrosion tests indicated they might have
better corrosion resistance than existing cladding
alloys. Trials of the X8001 have not shown superior

performance. The alloying elements in X8001 are
insoluble in the solid alloy and form intermetallic
inclusions that are malleable and become deformed
and extended into stringers during unidirectional
rolling and extrusion processing. The helium atoms
formed from the Ni in the stringers are recoilimplanted into the near-matrix regions surrounding
the stringers where they accelerate local formation of
voids and dislocations. The recoiled 56Fe atoms cause
extra dpa locally.67 These He-enriched regions are
not as obvious as those around B4C inclusions. They
are indicated by higher concentrations of voids, and
the emergence of more numerous He bubbles during
postirradiation annealing. This localized damage
offers an explanation of a hitherto inexplicable puzzle found in the corrosion response of X8001 alloy.
A characteristic of extruded X8001 tubes undergoing
aqueous corrosion in reactors is that smooth shallow
troughs or discontinuous ruts lying in the direction of


164

Performance of Aluminum in Research Reactors

the tube axis are created in the corroded surfaces.68
Such troughs have not been reported in laboratory
corrosion tests of X8001. It is suggested here that the
troughs are stringer beds left when the Ni-rich stringers are eased out of the surface by selective corrosion/erosion of the more highly damaged He-rich
matrix at the stringers.
Localized enhancement of He at Ni-rich stringers
is also believed to play a major role in the occurrence

of axial cracking of the X8001 cladding on HFIR
long-term isotope target rods.69,70 The target material is a cylindrical compact of actinide oxides in
an Al powder matrix, 6.3 mm diameter  14.5 mm
long, each jacketed in 1100Al. The meat contains
about 10% porosity to accommodate fission gases.
A target rod consists of 35 jacketed capsules stacked
in a tube of X8001 alloy that is hydrostatically compressed around them to form the outer cladding. The
tubes are made by extrusion and have six equally
spaced longitudinal fins. Before the target slugs are
loaded into the tubes, most of the fins are machined
off, leaving short lengths of fins at several locations
along the active length of the rod. The loaded target
rod is slid into an X8001 tubular sheath with hexagonal ends, known as a hex can. The groups of remnant
fins along the length of the rod act as spacers that
centralize the rod in the hex can and maintain
an annular water-cooling channel around the rod.
A bundle of 31 sheathed target rods just fills the
vertical HFIR trap. Cooling water flows inside and
outside the hex cans. Cracks were found in the mid
length, high flux regions of the target rods during a
search for the source of a-contamination detected in
the exiting coolant. Investigation showed that the
cracks were intergranular and were oriented in the
length of the rods at locations where the fins had been
removed. Lengths were up to 66 mm. The cracks
originated in the target rod cladding, but some of
the larger ones had penetrated the 1100Al jackets of
the target slugs. There was no evidence that corrosion
was involved. The hex cans were not cracked. The
exposure history of the rods is that they were first

irradiated for about 1 year in the D2O environment of
the C reactor at Savannah River Nuclear Laboratory
at a temperature of 20  C. They were returned to
ORNL and inserted in the HFIR at 46  C where,
during their fifth fuel cycle, the a-leak was detected.
The summed fluences were $6.9 Â 1026 n mÀ2, thermal, and 1.2 Â 1026 n mÀ2, E > 0.82 MeV. The conclusion from the investigation was that gas swelling of
the target meat had imposed a hoop stress on the
radiation-damaged cladding that had become too stiff

to undergo plastic flow and had cracked instead.
Rupture tests on unirradiated lengths of the cladding
tubes by internal pressure caused failure along the
machined-off fin lines, indicating the lines were weak
regions in the tubes. A possible solution to the cracking problem was to decrease the swelling-related
hoop stresses on the cladding by raising the pore
volume in the target meat from the then current
10% level to 20% and 25%. Trials were successful,
and the cracking no longer occurs. The questions of
why the cracks were located only on the fin lines and
why it was intergranular were not answered, but were
pursued.71 The intergranular nature of the fractures
made helium embrittlement a suspect. However, calculations of helium levels from the 27Al(n, a) reaction
with fast neutrons had given $7 appm, which was
considered inadequate for helium embrittlement at
the low irradiation temperatures experienced by the
rods. But supposing it was occurring, why did it favor
the fin lines? Metallographic and TEM examinations
of pieces of unirradiated cladding tubing showed that
the extrusion process had stretched the inclusion
particles into stringers and forced sheets of them

into the fins from which they extended back into
the tube wall. Removal of the fins left behind an
aggregation of stringers protruding into the tube
wall. Ergo the weakened regions along the fin lines
seen in the rupture tests of the unirradiated tubing.
When the enhanced production of helium from Ni by
thermal neutron capture was announced,66 a connection between aggregates of Ni-rich stringers and
helium embrittlement was discerned. Irradiated pieces
of high-purity Al, 1100Al, and a X8001 hex can were
sent to a specialist laboratory for helium analyses. For
a common thermal fluence of $1 Â 1026 n mÀ2, the
results were 1.8, 4.8, and 9.5 appm, respectively. After
a fluence of $3 Â 1026 n mÀ2, the corresponding
values were 7.2, 18.1, and 145. A piece of hex can
irradiated to 5.8 Â 1026 n mÀ2 yielded 220 appm.
A piece of 6061Al at 13.8 Â 1026 n mÀ2 gave 47 appm,
much less than the hex can at 5.8 Â 1026 n mÀ2. These
results leave no doubt that the presence of Ni in Al
irradiated to high thermal neutron fluences greatly
boosts the helium levels. And since the He will
remain close to the Ni particles, there must be very
large concentrations around the stringers. Any grain
boundaries overlapped by those local helium clouds
will be prime candidates for helium embrittlement
cracking under the influence of hoop stresses. Hence,
the intergranular cracking at the fin lines.
In most metals, the gaseous transmutation products
play a larger role in the development and effects of

Performance of Aluminum in Research Reactors

radiation damage structure than do the nongaseous
transmutants, one reason being that most construction
metals do not produce much nongaseous transmutants. Al is different. Depending on the degree of
thermalization of the neutron spectrum, Al can produce large quantities of silicon from the two-step reaction 27Al þ nth ! 28Al þ g; 28Al ! 28Si þ bÀ. A rough
guide to the quantity expected annually in the HFIR
PTP spectrum can be obtained by multiplying the
thermal neutron fluence by 230mb, the standard thermal neutron (0.0235 eV, 2200 m sÀ1) absorption crosssection for Al. The result is 1.035 at.% Si (1.073 wt%).
The Si is insoluble in Al at temperatures below about
350  C and is usually manifest as a precipitate of
elemental Si.72 This precipitate makes a substantial
contribution to radiation damage in Al, and is the
dominant hardening agent at high thermal neutron
fluences. There is one outstanding qualifier to that
generalization. In the 5xxx-type Al–Mg solid solution
series, the free Mg atoms dissolved in the Al will react
with the atoms of transmutant Si to form a precipitate
of Mg2Si.73 Thus, a 5xxx series alloy will be converted
to a 6xxx-like alloy.73–76 Figure 8 shows the Mg2Si
microstructure formed in irradiated 5052-O alloy.
Because this precipitate occurred at a temperature
below the usual 160  C aging temperature used to
obtain the T6 tempered condition in 6061Al, the
Mg2Si precipitate developed in the 5000 alloy is finer
than in the 6061-T6 alloy. The microstructure of
heavily irradiated 6061-T6 alloy is illustrated in
Figure 9. Since there are usually larger quantities of
Mg in the 5xxx alloys than in the 6xxx alloys, a greater
volume of Mg2Si can be created in the former 5xxx

alloys. Hence, irradiated 5xxx alloys will undergo radiation hardening and precipitation hardening simultaneously, and their overall hardening rate will be larger
than in other Al alloys exposed to the same neutron
fluence. Note that there are no voids in Figure 8.
Sparsely distributed voids are found75 at a higher
fluence of 1.8 Â 1027 n mÀ2. At half of that dose, the
6061-T6 alloy contains many more voids, Figure 9.
The association of the transmutant Si with voids is
interesting. We saw earlier that voids and particles of
Si become visible in the microstructure at about the
same dose. The voids are larger and fewer than the Si
particles. A Si particle is usually attached to one facet
of a void, and that particle is larger than its unattached brethren in the matrix. It is also facetted.
As irradiation continues, a change occurs in the
void-Si relationship. The voids lose their facetted
shape and become rounded.77 They are completely
covered with a thin coating consisting of mostly Si

165

100 nm

200

000

220

020

Figure 8 Precipitates of Mg2Si and excess Si in formerly

5052-OAl irradiated to 5.7 Â 1026 n mÀ2 at $55  C.
Reproduced from Farrell, K. J. Nucl. Mater. 1981, 97, 33–43,
with permission from Elsevier.

0.1 mm

Figure 9 Voids with background precipitate of
radiation-produced silicon and silicon-decorated original
Mg2Si precipitates in 6061-T6Al irradiated to $1027 n mÀ2
at $55  C.


166

Performance of Aluminum in Research Reactors

0.1 mm
Figure 10 Si particles and Si-coated voids on a
carbon extraction replica from 1100-OAl irradiated to
1.4 Â 1027 n mÀ2 (E > 0.1 MeV) and 2.3 Â 1027 n mÀ2
(E < 0.025 eV) at 55  C. Reproduced from Farrell, K.;
Bentley, J.; Braski, D. N. Scripta Metall. 1977, 11,
243–248.

with some Al. The coating is noncrystalline and flexible. The unattached Si particles in the matrix are also
rounded but are crystalline. The coated voids jut out of
the thinned edge of the hole in TEM foils and can be
lifted from the matrix on carbon extraction replicas.
Figure 10 is an example. The larger features with the
dark rims are the coated voids. Four of them have

partially collapsed without breaking, indicating a
highly ductile coating. Many of the Si particles seem
to have a layered structure. The Al–Si system is a
simple eutectic; there are no compounds. It is suspected
that the small amount of Al found in the void coatings
may be from the Al matrix that was not completely
dissolved from the voids during the electrolytic
extraction process. Silicon is obviously involved in
void formation and growth but its specific role is
unclear.

5.07.7 Property Changes
5.07.7.1

Swelling

Radiation swelling is the increase in volume arising
by accumulation of voids from excess vacancies and
by formation of gas bubbles. For Al, there are also
small swelling contributions from build-up of particles of transmuted Si and, in 5xxx alloys, Mg2Si,
which have densities of 2329 and 1990 kg mÀ3,
respectively.78 Gas bubble swelling is not an issue
for Al in RRs because the temperature is too low,
except perhaps in fuel cladding where some pores

found in the cladding may have been formed by the
accumulation of hydrogen. Swelling can be measured
from dimensional changes. More often it is determined from changes in immersion density values.
Swelling in various Al alloys is shown in Figure 11.
These alloys were all irradiated in the core of the

HFIR and they make the most comprehensive and
consistent set of swelling data.79 For reference, the
dotted line is estimated for the swelling from Si alone.
It is evident that the unirradiated chemical compositions and microstructures have major effects on the
degree of radiation swelling. The purest grades, sixnines and four-nines, show swelling earliest in dose
and swell at the highest rates with dose. The rates
decrease above a dose of about 1 Â 1025 n mÀ2.
Swelling in the two-nines grade (1100-O) requires
significantly higher doses, but the swelling rate is
unchanged. The 6061-T6 alloy, with its inherent
Mg2Si phase, starts swelling appreciatively later in
dose than the 1100-O. This is traceable to reduced
nucleation of voids, but its swelling rate is about the
same as the other alloys. The greatest resistance is
in the 5052-O alloy. There, the swelling is less than
for the Si alone. In this alloy, much of the early
swelling is not due to voids; it is caused by the silicon
and the new Mg2Si phase and by the increase in the
original density of the matrix, r0, as Mg is drawn
from solution to create the Mg2Si.
The effects of prior cold work on swelling in
Al agree in general that the presence of cold work
dislocation structure decreases the overall void
swelling but the reduction is not massive; concurrence
of dislocation recovery confuses the details.80–83
5.07.7.2

Mechanical Properties

The major consequences of radiation damage structures on the mechanical properties of Al alloys are

radiation hardening and associated loss in ductility.
There are too many data from too many sources
to be described in detail here. A good source of
compiled data, including the sparse information on
fracture toughness and weldments, is Marchbanks.84
For 6061Al in particular, see Farrell.85 Strengthening and loss of ductility are demonstrated best
in tensile properties. In Figure 12 we can directly
compare the changes in strength and ductility of
different alloys irradiated and tested under the
same conditions.79 The most striking feature is the
relatively rapid hardening displayed by the 5052-O
alloy. As explained earlier, this is caused by the
combined effects of radiation damage and in-reactor


Performance of Aluminum in Research Reactors

wt % Si 0.1

1

dpa 0.1

10

10

10

100

6-9
4-9
2-9 (1100-O)
6061-T6
5052-O

( ri )

%

Ti = 328 K (0.35 Tm)
1
6061-T6

Swelling

ro−ri

1

167

1100-O
5052-O
0.1
Pure aluminum
28

Si

0.01
1024

1025
1026
Fluence (n m-2 > 0.1 MeV)

1027

Figure 11 Radiation-induced swelling in various Al alloys as a function of fast fluence. Reproduced from Farrell, K.
In Proceedings of the Conference on Dimensional Stability and Mechanical Behaviour of Irradiated Metals and Alloys,
Brighton, Apr 11–13, 1983; British Nuclear Energy Society: London, 1983; Vol. 1, pp 73–76, with permission from British
Nuclear Energy Society (now The Nuclear Institute).

formation of a fine precipitate of Mg2Si. In contrast,
the 6061-T6 alloy, which contains Mg2Si before
irradiation, begins radiation hardening at about the
same fluence as the 1100-OAl, and hardens thereafter at the same rate. The 1100-O alloy contains no
Mg2Si before or after irradiation. From which we
deduce that preexisting Mg2Si precipitates play no
role in radiation hardening. This is an interesting
conclusion. It contradicts the expectation that the
precipitates would diminish the degree of radiation
hardening in 6061-T6Al by promoting the recombination of freely migrating vacancies and interstitials.
Perhaps that expectation is wrong. But it seems
satisfactory for explaining the delayed swelling in
the 6061-T6 alloy, where nucleation of voids is
retarded, perhaps until the transmutant gases enable
achievement of critical size cavity nuclei. Alternatively, maybe the radiation-produced Si dominates

the hardening process. It is a mystery. The fluence
for the onset of radiation hardening in the weak
4–9Al is about one order of magnitude less than in

the other alloys, and the subsequent rate of hardening is less than for the others. Here, again, we invoke
the recombination argument. This higher purity
material contains less solutes and inclusions. Thus
there is less trapping and annihilation of freely
migrating point defects, hence more point defect
clusters are formed in the early stages of irradiation.
It is suspected that the reduced rate of hardening is
connected with dynamic recovery of deformation
during the tensile test. It was pointed out in Section
5.07.3.2 that recovery from cold work occurs readily
in high-purity Al at room temperature. Loss in
uniform elongation in all of the alloys is concomitant with increase in strength . . . to a point. At a
fluence of about 1026 n mÀ2 the ductility reaches a
plateau of 3–5% even though the strength continues
to rise. The 1100-OAl has the least ductility in the
plateau region and displays an intergranular-like
fracture mode that may be caused by tearing and
void interconnection in the void-rich regions lying
alongside the grain boundaries.