Temperature monitors for uninstrumented irradiation experiments

Journal of Nuchx

Energy, Vol. 25, pp. 223 to 240. Pergamon Prw 1971. Printed in Northern Ireland

TEMPERATURE MONITORS FOR UNINSTRUMENTED IRRADIATION EXPERIMENTS J. I. BRAMMAN,A. S. FRA!XJR U.K.A.E.A. Dounreay Experimental Reactor Establishment, Thurso, Caithness

W. H. MARTIN* U.K.A.E.A., Reactor MaterialsLaboratory, Culcheth,Lancashire (Received 14 September 1970) Abstract-The

construction, use of and the interpretation of data from temperature monitors developed for incorporation in uninstnuncnted irradiation facilities is surveyed. Depending on !he

choiceof monitor, it is shownthat maximum,mean or end-of-runirradiationtemperaturesinpositioft~ inaccessible to thermocouple leads can be estimated with acceptable accuracy for most practical

purp-* 1. INTRODUCTION THOUGH materials testing reactors, specifically designed to permit the insertion of instrumented irradiation experiments, are widely used for work associated with power reactor development programmes, it is now becoming common to find prototype and on-line power reactors being used in part as test-bed facilities for more advanced fuel types. In general, these reactors were not designed with materials testing work in mind and the irradiation positions are not normally accessible to connecting cables for continuously recording instrumentation. The Dounreay Fast Reactor (DFR) is such a prototype facility. It operates at a thermal power of 60 MW driving a turbo-alternator of 15 MW electrical capacity, and was designed and built by the U.K.A.E.A. to demonstrate the feasibility of the fast reactor system for power production. In addition to fulfilling its original purpose by operating at full power since July 1963, the reactor and its operating schedule have been modified to provide an irradiation facility for the testing of materials for future fast power reactors. In this role, the reactor has for some years provided a unique service. Most of the experiments in DFR have to be accommodated in facilities having the external form of driver charge fuel or breeder elements and are generally described as being of the “replaced fuel element” (R.F.E.) type. Because of inacessibility to instrumentation leads, the experiments usually have to be provided with temperature monitors which can be examined during post-irradiation breakdown to give data indicative of irradiation temperatures. The authors have been directly concerned with the development and evaluation of temperature monitoring devices for uninstiumented DFR experiments, This paper reviews the present status of the various monitors. 2.

TEMPERATURE

MONITORS

A temperature monitor gives information which is only accessible when the experimental facility is broken down at the end of its irradiation period. Seven types have been intensively studied. As these are described it will be seen that each gives one * Present address: C.E.A., Centre d’8tudes Nucl6airesde Saclay, Gif-sur-Yvette,Fiance. 1

223

224

J.

I. BRNWAN,A. S.

specific kind of temperature indication: either

FRAU

and W. H. MARTIN

the temperature defined by the monitor is

(a) the highest temperature reached by the monitor during the irradiation period; (b) the temperature at the end of the irradiation period, or (c) a temperature averaged over the whole irradiation period with a tendency, because of diffusion kinetics, to a high temperature bias. Thus, in choosing a monitor for a particular experiment, the nature of the phenomenon being studied could be a determinant; for example, in studies of a diffusion process, such as creep, there would seem to be an advantage in using a monitor of the integrating type (c) in preference to an indicator of the maximum temperature (a) which would be undesirably (and perhaps irrelevantly) influenced by short duration temperature surges. However, if the temperature indicated by a monitor is either (a) the maximum temperature reached or (b) an end-of-run value, an evaluation of the thermal history throughout the irradiation period can often be made by examination of the reactor operating parameters; as conditions in the majority of our DFR facilities remain sensibly constant during full power operation, it is generally possible to derive a good approximation to the actual thermal history of the monitor in this way. To some extent this offsets the common disadvantage of all the monitors in so far as none of them is intrinsically capable of giving a continuous record of its irradiation temperature. In common with’all temperature indicating devices each monitor gives an indication of its own thermal history. This is an important point, and, as with thermocouples in instrumented irradiation experiments, it is essential that the monitors in uninstrumented experiments be inserted into positions for which heat transfer calculations permit a reliable estimation of temperature differences with respect to the test specimen; there is no virtue in any experiment in seeking a temperature indicator giving data accurate to say f 10 deg C if temperature gradients and the physical size of the device preclude estimation of the difference between the sensor and specimen temperatures with similar or better accuracy. The size of a monitor is therefore an important factor when considering potential applications. In the review of monitor developments which follows, the important questions about a monitor are: (i) what does it measure-is to do? (ii) how big is it-have

this quantity relevant to the experiment you wish

you room for it in an appropriate place in your apparatus?

(iii) what is its indicating range? (iv) what is the accuracy of the indication? It is most convenient to review the monitors in terms of the type of temperature indication. 3. MONITORS

INDICATING

MAXIMUM

TEMPERATURES

Three devices of this type have been studied. These are: monitors using melting points for indication; bi-metal assemblies utilizing a difference in thermal expansion coefficients; and a device resembling a clinical thermometer in which the temperature

Temperature monitors for uninstrumented irradiation experiments

225

is evaluated from a measurement of the volume of a displaced liquid. It will be seen that in their current form, each of these devices has the disadvantage of a large size. This tends to limit the use of maximum temperature indicators to applications requiring an estimate of the temperature in a relatively large uniform temperature zone. There has been little demand for such monitors in the U.K.A.E.A. irradiation programmes and thus there has been little incentive to develop these devices beyond the experimental stage. Fina weld

Lowestmelthgpint

sentine/

initial weld

FIG. 1.-Fusible

alloy monitor.

3.1. Fusible alloy monitors The basic principle of these monitors is simpler than that of any other monitor so far developed. Pure metals or alloys with well defined melting points are used; examination of an originally machined pellet of such a material should give an unequivocal indication of whether the temperature has or has not subsequently been above the melting point. The first experiments on the development of an in-pile fusible monitor were carried out by HUMPHRIIB(1965). A typical monitor consists of a series of sentinels (small machined or cast cylinders of either pure metals or eutectic or solid solution alloys) contained in an evacuated and sealed tube (Fig. 1). Crimping of the tube during assembly holds the pellets apart. It is general practice to select a series of sentinels to cover the temperature range of interest and to load them into the tube with the lower melting point sentinels towards the bottom. The monitor should then be fixed in the irradiation facility so that it lies in the same vertical orientation; it is then not possible for molten sentinel material to seep past a crimp and to run onto an unmelted pellet. The maximum temperature experienced by the monitor may then be inferred to fall

226

J. I. Bm,

A. S. FRASERand W. H. MARTIN

within the range bracketed by the melting points of the highest melted sentinel and the lowest unmelted sentinel. The major problem in the development of this type of monitor is one of compatibility of the sentinels with the containing tube. In his original survey, Humphries considered more than thirty readily available metallic sentinel materials, but use of the majority of these is impracticable because of incompatibility with the tube materials considered. The most satisfactory monitor evolved from this early work uses low TABLE 1.

FUSEILEMONIMRMATERIALS: 199-327°C MONITORS (Data from HUMPHRIES, 1965)

Melting Point “C

Composition wt. 0%

Density g ml-l

199 215 221 232 236 247 26.5 271 290 304 321 327

91Sn-9Zn 85Pb-15Au 965Sn-3.5Ag Sn 79.7Pb-17*7Cd-2.6Sb 87Pb-13Sb 82*5Cd-17.5Zn Bi 95Pb-5Pt 97*5Pb-2*5Ag Cd Pb

7.3 12.6 94 7.3 10.5 10.5 8.3 9.8 11.9 11.3 8.6 11.4

Comment

Limited compatibility Limited compatibility Limited compatibility Limited compatibility

melting point alloys in a O-2 cm dia. stainless steel tube. The temperature intervals between the sentinels are about 10-U) deg C (Table 1) with an overall range of 199327”C, though only the range above 230°C (the liquid metal coolant inlet temperature) is useful for DFR experiments. Some of the sentinels used have long term incompatibility with stainless steel, but the monitors have been satisfactorily used in large numbers for irradiation periods of the order of 3-6 months in DFR. A typical monitor would have a length of about 4 cm for a sentinel range of 100 deg C. The monitors, when removed during post-irradiation breakdown, are intensely radioactive, and data must be evaluated remotely. As all the sentinel materials used in this low temperature range have densities lying between 7.3 and 12.6 g ml-‘, they are generally examined by X-radiography. A more recent development (MOSEDALE, 1968) has been a monitor using solid solution magnesium-cadmium alloys with cadmium contents ranging from one to five atomic per cent, clad in tubes of pure iron. The device indicates in a limited range, 605-64O”C, with intervals between sentinels of about 5 deg C, and was devised especially for in-pile creep experiments, though extension of the range of indication for other applications is currently being considered. The overall size of the monitor is similar to that of the low temperature device developed earlier. Though the alloys are of low density, X-radiography shows a pellet with as little as 1 at. % cadmium quite clearly in the iron tubing used, but the use of cadmium as a solute to depress the solidus for the alloy series offers the possibility of using neutron radiography for postirradiation examination in case of difficulty. It is important with fusible monitors that the pellet should show a clear change of shape on melting. Some materials are particularly prone to surface oxide film formation and out-of-pile trials showed that such films can hold a pellet in its original form

Temperature monitors for uuiustrumented irradiation experiments

227

when molten. This emphasizes the necessity for extreme care in assembly of these monitors and the importance of either evacuating the tube or filling it with a pure inert gas before sealing. Even so, very careful examination of the post-irradiation radiographs may be required before it can be decided which sentinels have melted and which have not. In an extreme case, metallographic examination can be useful in resolving a dispute. The main disadvantage of this type of monitor is that the temperature indication is discontinuous. The sensitivity of the device is restricted by the temperature intervals between the melting points of the sentinels used. In the Dounreay examples described above the intervals are small; at worst the temperature reached is defined within f10 deg C and in the case of the new higher temperature monitor the uncertainty has been reduced to better than &5 deg C. A fusible alloy monitor described by LAURITZEN and COMPRELLI (1967) is much less useful; in the form described, the operating range is 484-66O”C with only two intermediate sentinels. Melting of a fusible alloy sentinel pellet takes place if the temperature rises to the melting point, and this can be disadvantageous in practice. A high temperature excursion of very short duration can be recorded, so giving a misleading impression of the running temperature. As noted earlier, however, DFR is very stable in operation and we have no evidence that temperature excursions require serious consideration in any of our applications. A further problem can arise if sentinels held near but just below their solidus temperatures for prolonged periods give bogus melting indications as a result of creep collapse. There is no evidence that this is likely with the small pellets used in the monitors described. Lauritzen and Comprelli placed a tantalum slug on top of each of their sentinels so that post-irradiation data could be extracted by gamma-scanning; the loading of the sentinel by this small slug might increase the risk of creep collapse. The disadvantages of the fusible monitor must be offset against its simplicity. There is no evidence at all to suggest that the melting points of the sentinel materials are likely to be affected by prolonged irradiation, particularly in a fast reactor where sentinel materials can be selected so that rates of impurity build-up by transmutation are slow. With the small diameter monitors described there is little effect from nuclear heating in the monitor itself, and it is our belief that these devices respond sensitively to the maximum temperature reached in their environment. The principal limitation on their use is their overall size; this is much better suited to applications in which an indication of temperature in a relatively large volume is required than for indication at a particular point in a region where temperature gradients are present. 3.2. The bimetal thermometer This device makes use of the difference between the thermal expansion coefficients of two dissimilar metals or alloys as in a bimetal strip assembly. There are two possible modes of use of a normal bimetal strip: it may be used either within the elastic limits of the component materials or in a manner causing permanent deformation. In the first mode, a “tell-tale” or marker must be used to record the degree of movement at the maximum temperature reached. In the second, the observed degree of deformation is itself a measure of the movement and hence of the maximum temperature. In both cases, under prolonged exposure at elevated temperatures, creep is a factor which has to be taken into account. The effects of creep, however, which process might be

228

J. I. BRAMMN, A. S. FRASSR and W. H. MARTIN

influenced by irradiation, can be avoided by using two metals or alloys which are only loosely connected; that is, by arranging that the two components move independently, with one acting as a reference only. A monitor operating on this principle has undergone out-of-pile tests (WILSONand EDMONDSON, 1969). In its latest form-a molybdenum rod surrounded by a stainless steel tube, the two being pegged together at one end with the other end of the tube Molybdenum rod

Mnonic alloy cl@ rtell-tale?

Fro. 2.-Developmental bimetalthermometer. moving freely to drive a nickel alloy tell-tale up the rod (Fig. 2)-a temperature between 100 and 700°C can be inferred with an accuracy of f-10 deg C, from projection microscope measurements of the distance between the tell-tale and the unpegged end of the tube. Though the device has not yet been tested in-pile there is no reason to suppose that it will not operate with similar sensitivity to that noted in laboratory trials. Like the fusible sentinel monitor it is sensitive to high temperature excursions, but it does have a distinct advantage in being a continuously indicating device. Its principal limitation is its size: in its present form, the bimetal monitor is about 10 cm in length and O-6 cm dia. There is little prospect of reducing the overall length without loss of sensitivity: the present materials of construction are particularly attractive because of their compatibility with a sodium environment, and though the device could be compacted in length by using a concentric alternating series of tubes this would be at the expense of an increase in overall diameter. It is, in fact, the diameter which imposes the severest restriction on the use of this monitor: most DFR experimental rigs can accept the length but not the diameter, and it is in the radial direction that thermal gradients in in-pile facilities are generally most severe. Thus the bimetal monitor, like the fusible sentinel device, is best suited to use in a uniform temperature environment of relatively large volume. For this type of application, however, it has one important advantage

Temperaturimonitors for uninstrumentedirradiation experiments

229

over the fusible alloy monitor: if, for some reason, the maximum temperature experienced lies outside the range covered by the melting points of its sentinels, a fusible alloy monitor will only give either an upper or a lower temperature indication with no information as to how much this deviates from the actual temperature experienced; a bimetal monitor, on the other hand, will give an indication whatever the temperature, the only intrinsic limitation being set by melting of one of its component materials. 3.3. The clinical or ejection thermometer Two devices depending on the measurement of the amount of liquid displaced have been tested out-of-pile (WILSON and EDMONJXON, 1969). The first device was an ejection thermometer in which the amount of liquid displaced from a bulb at temperature and ejected from a capillary is measured by before alloy and after weighing. The laboratory devices used a bismuth-lead-indium (melting point 73°C) in tubes of either quartz glass or stainless steel; there is incompatibility between the alloy and steel but not sufficiently serious to afIect a short term test. The two major practical problems concerned, firstly, the complete filling of the bulb and capillary with exclusion of cavities or bubbles, and secondly, the development of a suitable outlet from the capillary so that alloy, once ejected, could not be sucked back when the temperature began to fall. The filling problem was satisfactorily solved but the ejection problem gave trouble throughout the experimental programme : there was always a possibility that liquid from a droplet of a size equivalent to the amount ejected by a 100 deg C temperature rise could be sucked back, giving erratic temperature indications. This led to a termination of further development work. The second device was a sealed clinical thermometer which would not be prone to the droplet problem. In the experimental form these were fabricated from stainless steel and contained a gold-lead alloy (melting at 215°C). Trials gave erratic results due to irregular solidification behaviour of the alloy-an alloy which solidified after removal from the reactor was selected in preference to one remaining liquid (for example, a binary based on gallium) to reduce the risk of erroneous indications due to mechanical shock. Though the clinical thermometer can be made small and compact, these out-of-pile trials did not encourage the belief that the device can be made to give a sensitivity comparable with that of the bimetal monitor previously described. Precalibration is not possible because, at temperature, the indicating alloy wets the walls of and thus fills the hypodermic tubing making it impossible to return most of the alloy to the bulb. Our present view is that ejection and clinical thermometer types of in-pile monitors, whilst feasible, are so beset by practical problems that there is little potential for further development. 4. MONITORS

INDICATING

AN END-OF-RUN

TEMPERATURE

Two methods of deriving an end-of-run temperature have been the subject of considerable experimentation and both have established themselves as reliable techniques in their particular applications. The first of these utilizes the irradiationinduced dimensional change in silicon carbide and is of general applicability; the other makes use of the release of krypton-85 during post-irradiation annealing as an indicator of the temperature of fuel pin cladding.

230

J. I. BRAMMAN, A. S. FRASBR and W. H. MARTIN

4.1. Silicon carbide monitors

During an investigation of possible sleeving materials for advanced gas-cooled reactors, THORNE, HOWARDand HOPE (1967) determined the dimensional change behaviour of silicon carbide in the temperature range 250-750°C at equivalent fission neutron doses up to 1.8 x lOa n cm-s in experiments performed in the Dounreay Materials Testing Reactor (DMTR). The dimensions of pellets of /?-Sic ((‘Crusilite”; Morganite Electroheat Ltd., London, S.W.18) were found to increase rapidly under fast neutron irradiation up to a dose of about 3 x lOaon cm-a at all temperatures. Above this dose, the dimensions remained constant up to the peak dose of 1.8 x 102l n cm-a. The saturation level depends on temperature, growths of about @7 per cent at 250°C and about O-3 per cent at 700°C being observed. As early as 1961, PRAVDYUK et al. reporting similar lattice expansion studies on a-Sic in the temperature range 12042O”C, had suggested that the material might be used as an indicator of irradiation temperature (and of neutron dose at doses below the saturation level). The THORNEet al. data on /I-Sic showed a large scatter, and a measurement of the irradiation-induced dimensional change of saturation wouldgive only a crude estimate of the irradiation temperature. On isochronally annealing the irradiated silicon carbide, however, THORNE et al. showed that the dimensions were unaffected until the annealing temperature exceeded the irradiation temperature, after which the dimensional change decreased linearly with increasing temperature. For the specimens irradiated in the DMTR experiments, it was found that the temperatures at which annealing of the irradiation-induced dimensional changes commenced were the irradiation temperatures as measured by thermocouples incorporated in the experiments. The first experiments in DFR in which silicon carbide was used to measure the irradiation temperature were described by MARTINand PRICE(1967). Data extraction was by isochronal annealing of irradiation-induced length changes. Since 1967, large numbers of silicon carbide pellets have been irradiated in instrumented facilities in DFR and have been shown to give good agreement with thermocouple temperatures. Figure 3 shows the agreement between silicon carbide and thermocouple temperatures from recent instrumented DFR experiments. With the accumulation of evidence of the fidelity with which the length changes in silicon carbide pellets can be used as indicators of irradiation temperature, a decision was made to use these monitors in the majority of DFR experiments. To cope with large numbers of monitors, trials of a more rapid data extraction method using a continuous differential dilatometer have been made. It would be expected that the annealing behaviour might be affected by the rate of temperature rise in the dilatometer and experiments were conducted over a wide range of heating rates. Even in the best result obtained, the onset of annealing was ill-defined and the isochronal annealing technique, though more time-consuming, is now accepted as a more sensitive method of estimating the annealing temperature. Recent evaluations using X-ray diffraction measurements of the lattice parameter instead of pellet length measurements seem to give some further improvement in sensitivity. In addition, the use in this method of analysis of only small bits of silicon carbide has made it possible to obtain data both from powder samples, used in experiments where space limitations make the use of pellets impracticable, and from pellets broken during post-irradiation rig breakdown.

Temperature monitors for uninstrumented irradiation experiments

231

100

0

FIG. 3.-Derived

silicon carbide monitor temperatures from monitors irradiated in close proximity to thermocouples.

The typical pellet size used is 1.2 cm long x 0.3 cm dia. As silicon carbide is incompatible with the reactor coolant, the pellets are generally sealed in capsules usually of stainless steel at the lower temperatures, but above 650°C a molybdenum capsule (or a steel capsule with a molybdenum liner) would be used to evade a minor incompatibility problem causing the pellet to stick to a steel surface. This could be troublesome when recovering monitors after irradiation but there is no evidence that it would have any repercussions on the usefulness of the pellets as temperature monitors. (Radial gaps around a pellet inside the capsule would not lead to errors of more than f3 deg C due to nuclear heating.) Using the isochronal annealing/length measurement method of data extraction, irradiation temperatures are believed to be determined with uncertainties of f 15-30 deg C at temperatures below 75O”C, but the errors increase at higher temperatures and could reach f 100 deg C at 1000°C. With the isochronal annealing/X-ray lattice parameter method, smaller errors are claimed: typical values are f15 deg C at 700°C rising to &25 deg C at 1000°C. Figure 4 shows a typical result. In both cases the useful range over which the monitors have been used is 300-l 150°C. On evidence to date, silicon carbide pellets provide the most versatile monitor so far developed, though recent assessments of data extracted by the X-ray technique (SHARPE, 1969) indicate that the material becomes amorphous and unsuitable for lattice parameter determinations at high doses. On the basis of these assessments it would appear that examination by X-ray methods cannot be used for material irradiated to dose levels in excess of about lOaan cm-s, though the evidence accumulated so far is insticient to permit a complete evaluation of the inter-related effects of total dose, dose rate and irradiation temperature. No difficulty has yet been

232

J. I. BRA~I~UN,A. S. Frusa~ and W. H. MARTIN

3 E s 2 lu i? 2c

4-3760

Pellet

1561’22

JX

4.380

4.3720

4.3660

0

100

FOG.4.-Isochronal

200

.m

400

500

600

700

aw ANNEALING

900 xIw TEIWWWWE~CI

annealing behaviour of an irradiated silicon carbide monitor.

experienced with data retrieval by length measurement techniques from material irradiated to similarly high doses. Accepting this limitation on data extraction, the useful temperature range of application of silicon carbide monitors is wide and the monitor is small-indeed, in the conditions under which the X-ray method can be used, pellets as small as 0.3cm long x O-3 cm dia. and 1.2 cm long x O-1 cm dia. (or, in certain cases, equivalent volumes of powder) have been successfully examined. The silicon carbide monitors can therefore be used in rigs with more confined geometries than any of the other devices so far considered in this paper. Because the saturation level of irradiation induced dimensional change is reached for a total fast neutron dose of about 3 x lOao n cm-a and annealing at any temperature is sensibly complete after a few hours, the dimensional change settles rapidly at a level appropriate to the irradiation temperature. Thus, the temperature indicated by a silicon carbide monitor will correspond to the approximate mean temperature during the latter part of the irradiation period when the last saturation dose increment of total dose was being accumulated. The monitor is for this reason said to indicate an end-of-run temperature. A consequence is that a slow reactor shut-down leads to the observation of multiple annealing stages and might result in all evidence of an annealing stage corresponding to original full-power operation being lost. The successful use of silicon carbide monitors thus demands careful attention to the manner in which the reactor is operated during the last few hours of the irradiation period. 4.2. Krypton-85 release The use of krypton-85 as a universal tracer has been extensively developed by CHLBCKand his co-workers (1963). Krypton is introduced into a solid specimen either by ion bombardment or by heating in a high pressure krypton atmosphere and quenching. In typical applications, s6Kr is then released for counting when a solid which has been impregnated with krypton containing the radioisotope reacts with its environment. The principal attraction of using this tracer method lies in its lack of disturbance of the physical or chemical state of the specimen. The use of surfaces impregnated with radioactive inert gas to find the surface temperature depends on the observation (K~uY, 1961) that when an impregnated

Temperature monitors for uninstrumentedirradiation experiments

233

specimen is heated to a particular temperature, it loses some of its initial activity but that there is then no further loss of activity during further heating at or below this temperature. A temperature previously experienced by a krypton impregnatedsurface can therefore be determined by re-heating the specimen in a controlled rising temperature and monitoring the onset of further a5Kr release. A straightforward in-pile application would seem to be to impregnate the surfaces of a specimen with ssKr before irradiation and then subsequently to monitor the irradiation temperature by annealing and counting techniques. Unfortunately, it has been found that surface impregnation will not withstand the effects of fast flux irradiation; pre-impregnated graphite specimens, which, in laboratory tests showed smooth release up to temperatures in excess of 8OO”C,were found to have lost all their 85Kr after irradiation at 600°C. However, reactor fuel cans are subjected to bombardment by fission products including *5Kr in addition to being subjected to neutron irradiation while holding an increasing pressure of fission gas. It would be expected, therefore, that can material would experience continuous impregnation of the inner surface throughout irradiation. Work at Dounreay by SWANSON, BUTLERand CLARKE (1969) has shown that this can be used to determine, by post-irradiation annealing and counting, a temperature for small samples of fuel-pin cladding. This temperature is thought to be the maximum temperature of the sample during the last period of irradiation where this period is of the order of the time for 50 per cent of the atoms in the specimen to be displaced by the fast neutron flux in the reactor. For DFR this time can be as short as 3 hr, and thus the krypton release method permits monitoring of the end-of-run temperature. The method has been used over the range 270-650°C with an apparent sensitivity of f25-30 deg C. Aside from the special case of fuel pin cladding, the use of adjacent fuel to continuously regenerate the surface impregnation with asKr seems to offer potential for the development of a more versatile monitor. A tensile test specimen of stainless steel which had been irradiated in DMTR in close contact with a foil of enriched uranium and which had been shown by thermocouple measurements to have been irradiated at 395’C was found by the krypton release method to begin krypton emission at 397’C. It would seem therefore that a foil sandwich of stainless steel and enriched uranium might be used to monitor the temperature of a reactor environment. It is possible that such a sandwich could be made with small enough dimensions to permit its insertion in a wide variety of experimental facilities. 5. MONITORS INDICATING AN AVERAGED TEMPERATURE Two monitors have been tested in which the property change from which the irradiation temperature may be derived is diffusion controlled. These monitors therefore indicate a temperature averaged over the whole irradiation period, though in the case of a varying temperature it would be expected that the average would show a high temperature bias as a consequence of the kinetics of diffusion processes. The two monitors are “Templugs” (Shell Research Ltd., Chester), and samples of a precipitation-hardening alloy. 5.1. Templugs Templugs were originally developed by Shell Research Ltd. as a means of assessing temperatures in experimental internal combustion engines, where conventional means

234

J. I. BRNGUN, A. S. F~USBRand W. H. MARTIN

of temperature measurement were impossible. The concept depends on the structural changes which occur when a fully hardened carbon steel is tempered by annealing at elevated temperature, the effect of tempering being reflected as a change in hardness. The only reliable Templug materials so far developed in laboratory trials have been steels of near eutectoid composition, the first a silver steel to British Standard 1409, claimed from trial results to give satisfactory data over the temperature range 7872O”C, and the second to British Standard EN52, suggested as suitable for use at

3

26 24

#O-

s 2

5w iO%

7w -

if!

s600 500-

400 3w2w100 ’

100

200.

FIG.K-Tempering

300

400

500

700 600 TEMFERfhG TEMPERAlURE M

curves for B.S. 1409 silver steel Templug material.

temperatures between about 560 and 850°C. In the original Shell work, the maximum tempering times of interest were of the order of 1000 hr, although in the engine test work times in excess of 500 hr were rarely achieved. Templug temperatures are estimated by comparing hardness figures with families of prepared calibration curves of hardness against temperature for a range of tempering times (Fig. 5) or the derived curves of hardness against logarithm of time for specific temperatures (Fig. 6). The device appeared to offer certain attractions for use in irradiation facilities; the Templugs could be made sufficiently small to permit easy insertion into experiments where space restrictions were severe, and post-irradiation hardness was a relatively easily measured parameter. It was recognised that use as a temperature monitoring device in irradiation experiments would necessitate considerably longer tempering times than had been studied in the laboratory, and that difficulties could be encountered in the recovery and handling of small irradiated pellets. The effect of fast neutron irradiation on the tempering process was a matter for conjecture until experimental trials had been conducted. The Templugs selected for test and subsequent use in irradiation experiments were right cylinders of silver steel to BS 1409, 0.23 cm dia. and O-255 cm in length, heat treated to develop a fully hardened structure.

Temperature monitors for uninstrumented irradiation experiments

loo

’

lo

MO

m TIME AT TEtWFRAlURE

FIG.6.-Derived

235

mm lhovsl

hardness/time/temperatimeltemperature wrves for B.S. 1409 silver steel Templugs.

Trials (FRASERand SINCLAIR,1968) showed that for use in any environment, Templug material should be drawn from the same billet and that zones of surface decarburization left by the drawing process should be removed by grinding. Consistency of stock material is a pre-requisite for the successful exploitation of the monitors, and must be ensured by frequent sampling for chemical analysis and metallography. The hardening and tempering characteristics of the alloy are profoundly affected by chromium content; in the specification, chromium up to 0.5 wt. % is optional, but in view of its role in eliminating graphitisation during high temperature annealing by stabilising carbides, a chromium content of O-5per cent was specified for our alloy. The optimum heat treatment for the chosen material involved a 30 min soak at 81O”C, quenching into water and tempering for 15-20 min in boiling water to remove retained austenite. Careful control of all the heat treatment processes is essential. Templugs developing cracks during heat treatment should be rejected; a 100 per cent fluorescent dye penetration test is necessary. Hardness values within the batches varied and any Templugs with values lying outside set acceptance limits (in our case 9204029 V.P.N.) should be discarded. Control data were extended to tempering times of 5,000 hr in the temperature range 250-800°C. All tempering curves displayed a minimum at ~720°C (the AC, temperature) so that a range exists within which a particular hardness value could be interpreted as either of two temperatures, though differing structural characteristics permit a distinction to be made by metallographic examination. The rate of cooling from the tempering temperature was unimportant at temperatures below 72O”C, but above this, the slower the cooling rate, the higher the suggested temperature. The deleterious effects of decarburization necessitate the encapsulation of Ternplugs in any application where contact with liquid sodium is possible. Capsules for Templugs are made with an internal diameter giving a radial gap of no more than 0.012 cm, so that nuclear heating effects give less than 5 deg C error in the Templug temperature. When sealing the capsules by welding care must be taken to avoid heating the Templugs and so prematurely tempering them. Ternplugs have been the subject of intensive out-of-pile and in-pile study in an

236

J. I. B-,

A. S. -

and W. H. Mmm

effort to extend the known tempering range beyond the original 500-1000 hr limit and to investigate their reliability in a reactor environment. The results have shown that Templugs are reliable indicators of irradiation temperatures only in the 450650°C range and that an irradiation hardening effect at temperatures below 450°C leads to errors of, for example, 80 deg C at 350°C, rising to 100 deg C at 230°C. Within the temperature range 450-6OO”C, Templugs appear to be accurate to within f15 deg C for exposure times up to 5000 hr. The accuracy is quoted as f25 deg C for times not exceeding 2000 hours in the 600650°C temperature range, but Templugs indicating temperatures in excess of 600°C should be metallographically examined. Templugs tempered for times in excess of the stated limits are less reliable as discrimination is more difficult because of converging hardness/time/temperature curves. Within their limits, Templugs do appear to give satisfactory performance in irradiation facilities, provided that the rigorous requirements of characterisation, preparation and encapsulation are met. For many applications the small size is a major factor in recommending their use, and insertion in duplicate is relatively easy in most monitoring positions. Apart from the low temperature errors, radiation damage effects have been less serious than might have been expected, and it appears that the thermal tempering process swamps any irradiation effects in the useful temperature range. Templugs give an average value of temperature with a tendency to high temperature bias, though under normal operating conditions in DFR the bias would be negligible. 5.2. Ageing phase growth in a precipitation hardening alloy A monitoring technique, currently at an experimental stage of development, has been based on the growth of ageing phase particles in Nimonic PE16 (FRASER and FULTON, 1969). This alloy contains between 1.1 and 1.4 per cent of both titanium and aluminium, and controlled precipitation of Nis(A1, Ti) ageing phase occurs during manufacture. The precipitated particles, which can be observed by high-resolution dark-field electron microscopy of a thin foil sample, have been found to be spherical after most extended thermal ageing treatments, although some signs of cubic symmetry become apparent after longer ageing times at temperatures near the solvus temperature (860°C). Laboratory studies of the growth kinetics of the ageing phase have shown nucleation takes place easily, establishing an equilibrium concentration from the earliest times and that ageing phase particles grow according to a (time)lj3 law as theory (WAGNER, 1961) predicts. The rate of coarsening of the particles is determined by the rate of diffusion of aluminium and/or titanium and by the energy of the particle/matrix interface. The near identity of the lattice parameters of matrix and particle means that the interfacial energy is low, so that the rate of particle growth is slow. The ageing phase precipitates relatively uniformly throughout the matrix and, whilst the sizes of particles vary in any particular sample, it has been demonstrated that they follow a normal distribution; a mean particle diameter may thus be computed from a statistical sample. Using a specific cast of Nimonic PE16 in a standard heat treatment condition (either solution treated for 2O.min at 1040°C and air cooled, or solution treated for 20 min at 104O”C, air cooled, aged for 16 hr at 700°C and finally air cooled again) a series of calibration curves of mean particle diameter plotted on a log-log basis against

Temperature monitors for uninstrumented irradiation experiments

237

time for various ageing temperatures can be developed. Examples are shown in Fig. 7. These curves, together with derived curves of mean particle diameter vs. temperature for a range of specific ageing times, may then be used to derive a temperature for a sample (from the same cast of material) which has been aged for a known time to give an observed mean particle sire.

_-

Fn7.-Ageing-phase

Tempemture _--Boo%

particle size~time/temperature curves for solution treated and air-cooled Niionic PE 16.

The use of ageing phase growth as a temperature monitoring device for nuclear environments demands that the process is not affected by irradiation so that the calibration curves derived from out-of-pile data are not invalidated. The phenomenon of radiation-enhanced diffusion at relatively low temperatures is well established (BILLINGTON and ~&GEL, 1950; MURRAYand TAYLOR,1954) and diffusion processes in, for example, nickel-beryllium and copper-beryllium, are known to be affected by irradiation. Our evidence from DFR irradiations in both instrumented and uninstrumented experiments (where cross check information was available from, respectively, thermocouples and other monitors), suggests, however, that particle growth rates in Nimonic PE16 are not influenced to any detectable extent by irradiation at temperatures in excess of 550°C and confirms the inference from a theoretical model that enhancement of growth rates is unlikely in DFR fluxes at temperatures above 600°C. The present indications are that PE16 monitors might be used to determine temperatures in the 5504300°C range. Within the range 600-75O”C, the equilibrium volume fraction of ageing phase is relatively constant and the monitors should be capable of defining a temperature with a sensitivity of better than &25 deg C. The ageing phase equilibrium volume fraction changes at temperatures above about 75O”C, but preliminary work suggests that the monitors may be used up to 800°C with perhaps

J. I. BRNWAN,A. S. FRASERand W. H. MARTIN

238

a slight lessening of confidence in discrimination. When monitors in the solution treated and air cooled condition are used, temperatures in the 550400°C can be determined, again with some slight decrease in sensitivity. The size of a monitor is small; a disc O-23 cm dia. and 0.025-O-050 cm thick is acceptable. In addition, Nimonic PE16 is compatible with the DFR coolant in the temperature range of interest so that no cladding is necessary. A minor drawback lies in the high activity levels of irradiated samples, making handling difficult. As the controlling process in this monitor is diffusion, the indicated temperature is an average value, again with a high temperature bias, though, as with Templugs, this would not be a matter of concern in a typical DFR experiment because of stability of operation. 6. DISCUSSION

The important features of all the monitors considered in this paper are summarised in Table 2. Using this summary of information it is possible to select, from those regarded as reliable within the stated limits, the most satisfactory monitor for any particular application, noting, as mentioned in the introductory remarks, that the physical size of the monitor and the relevance of the temperature indicated to the experiment are important restrictions on the choice in many applications. The apparently low accuracies of indication of the various monitors are a matter of more concern to experiment sponsors than they need be. The measurement of temperatures by thermocouple techniques does not guarantee high accuracy unless very stringent precautions are taken in use, and, in this light, the inability of any of our monitors to indicate a temperature in the 650-700°C range with an accuracy of better than f lO-15 deg C is less serious than casual consideration might suggest. A further criticism of these monitors, mentioned earlier, arises because a monitor only indicates its own temperature, which is not necessarily that of the specimen or environment of interest. This, however, is a limitation of most temperature measuring devices; for instance, the temperature indicated by a thermocouple is the temperature of its junction. The successful use of any sensor requires careful design to get the sensor TABLE2. TEMPERATURE ~oNn.0~ FOR Monitor type

Typical size

Fusible alloy monitors

4 cm long x 0*3cm dia.

Bimetal thermometer All-metal clinical thermometer

10 cm long x 0.6 cm dia. 10 cm long x 0.6 cm dia. in trial form 1.2 cm long x @3 cm dia.

Silicon carbide ton 85 release

Temperature range Specially assembled for limited ranges, e.g. 230-327°C 605-640°C Possibly lOO-800°C Not ascertained 300-l 150°C

7for fuel pin cladding)

@6cm x 2cmarea

270-65O”C

Templugs

0.25 cm long x 0.25 cm dia.

Ageing phase growth in Nimonic PE 16

0.23 cm dia. x 0.025 cm thick

450-6OO”C 600-650°C 550-800°C

Temperature

monitors for uninstrumented

239

irradiation experiments

into such a position that it will be in thermal equilibrium with the sample or environment. In the confined spaces of a typical in-pile experiment this is a severe problem, which highlights the necessity for taking the monitor into consideration from the start of rig design rather than inserting monitors as an afterthought into available sptces. A more serious criticism of the monitors is that only one temperature is indicated, and there has been discussion from time to time of the possibility of developing a continuously-recording device for incorporation in DFR experiments. A recorder of this type was developed at the Authority’s Windscale Laboratories by THrJBLBECK (1964); a complete temperature record is made available for post-irradiation analysis. The sensing element is a bimetallic strip which has a recording stylus at one end and is fixed to a worm driven carriage at the other. Fixed to the same carriage, which moves parallel to the axis of a revolving drum, is a reference stylus. The drum and carriage are clockwork driven by the same mechanism and the two styli trace a continuous record on the polished surface of the drum. The distance between the two traces is a measure of the temperature of the bimetallic strip. In its original form the recorder is about 12 cm long and 4 cm dia. and had an accuracy of f5 deg C. Miniaturization of this device for incorporation in DFR experiments could be undertaken. It is, however, unfair to condemn the monitors on the grounds that only one temperature is indicated. Given the one temperature, and a knowledge of to which part of the irradiation period it refers (i.e. mean, end-of-run or maximum), it is possible from a knowledge of the reactor operating parameters to reconstruct for an experiment in any given reactor position a fair approximation to the complete irradiation temperature history. It is a measure of the need for a continuously recording device, as opposed to the desire for one (which is quite a different thing), that the authors, who, during the past three years have supplied to experiment sponsors many hundreds of individual monitor temperatures, have no knowledge of any use of their data to derive a complete history. It is our experience that the monitors developed are adequate for the majority of DFR experiments, though it must be admitted that there is not currently available a wide temperature range monitor which gives data accurate enough for use in in-pile UNINSTRUMENTED

IRRADIATION

EXPERIMENTS

Accuracy

Parameter measured Highest temperature

reached

f15 deg C f5degC 510 deg C Not ascertained

Reliable to times <6 months Experimental Highest temperature

reached

Experimental

out-of-pile

Highest temperature

reached

Development

diintinued

300-750°C f 15-30 deg C 800-l 150°C f 25-100 deg C

End of run temperature

Reliable

i 25-30deg C

End of run temperature

Reliable

Mean temperature

Reliable

Mean temperature

Experimental

f15degCfor <5OOOhr 4~25 deg C for ~2000 hr &25”C

2

Present status

240

J. I. Bm,

A. S. Fraser and W. H. MARTIN

creep experiments. However, fusible alloy monitors capable of the desired degree of sensitivity can be developed for use in limited temperature ranges (though, as noted earlier, the information from a maximum temperature indicator may not be immediately relevant to irradiation experiments involving diffusion-controlled processes). Nor is there a monitor available for temperatures in excess of 1100°C. Irradiationinduced lattice expansion changes in diamond (PRAVDYUK et al. 1961) might, on consideration of the melting point, be expected to anneal out at higher temperatures than similar changes in silicon carbide, and it has therefore been suggested that diamond might be used to monitor temperatures in excess of 1000°C. Unfortunately, graphitisation occurs at temperatures of 100&1200°C (NIKOLAENKO et al. 1968) so that high temperature use is not possible. Nikolaenko and his co-workers have, however, used 10 mg samples of natural diamond powder to monitor temperatures up to 600°C. The volume expansions are large (~1.6 per cent at 600°C for a dose of 4.3 x 10Bon/cm8) so that in the usable range there is a possibility that temperatures might be derived with better accuracy than those obtained with silicon carbide. Nikolaenko has found that the dimensional changes did not saturate at doses up to 2 x 10n n/cma so that the monitor can also be used to monitor dose and nuclear heating level simultaneously with temperature : more important, the monitor has a memory and “remembers” higher temperature peaks experienced during irradiation. With the two exceptions just noted, however, it is our opinion that the monitors discussed in this paper are adequate for present requirements. AcknowZe&e~ex?s-The authors must express their indebtedness to their many colleagues with whom they have been associated in the task of developing the monitors. In particular, mention must be made of D. Mosedale, J. Humphries, K. J. Wilson, W. Edmondson, W. D. J. Sinclair, W. B. Bremner. E. J. F&on, R. M. Sharpe and T. Berry, all of the Doumeay establishment who kindly permitted quotation from their internal memoranda and unpublished data. REFERENCES BILLINO~Y~N D. S. and SI~GBLS. (1950) Me&ZProgress 58, 847. CHLJICKD., MABHLR. and C~J~~HIARA 0. (1963) Nucleonics 21,53. Cruxcx D., Mm R., CUCCHIARA 0. and CARNBVALB E. (1963) Znt. J. Appl. Z&ad.Zsofopes 14,581. CHLECK D. and MABHLR. (1963) Znr. J. Appl. Rad. Isotopes 14,593. FRASERA. S. and FUETON E. J. (1969) U.K.A.E.A. internal document. FRASERA. S. and SINCLAIRW. D. J. (1968) U.K.A.E.A. internal document. HUMPHR~ES J. (1965) U.K.A.E.A. internal document. KELLY R. (1961) Can. J. Chem. 39,241l. LAURITZENT. and C~MP~~LL.I F. A. (1967) NucZ. Applications 3,390. MARTIN W. H. and Parox A. M. (1967) J. nucl. Energy 21, 390. MOSEDAU D. (1968) U.K.A.E.A. internal document. MURRAYG. T. and TAYLOR W. W. (1954) Acta Met. 2,52. NIKOLAENKO V. A., KARpucm~ V. I. and ALEX~EV S. I. (1968) Kurehatov Institute Report IAZ-1650, Pa~vuvux N. F., NJKOLAENKO V. A., KARPUCHINV. I. and Kuz~~rsov V. N. (1962) Properties of Reactor Materials and the Effects of Radiation Damage, Edited by D. J. L~TIXER,p. 57, Butterworths, London. SHARPBR. M. (U.K.A.E.A., Dounreay) (1969) Private communication. SWANSON K. M., BUTLER J. K. and CLARKBD. (1969) J. nucl. Mat. 33,302. THORNBR. P., HOWARDV. C. and HOPEB. (1967) Proc. Brit. Cerum. Sec. Q, 449. THURLBECK A. (1964) British Patent 1028192. WAGNERC. (1961) 2. Efectrochemie 65,581. WI~N K. J. and EDMONDSON W. (1969) U.K.A.E.A. internal document.