- Email: [email protected]
Deposition behavior of ferrous alloys in molten lead-lithium
Fusion Engineering and Design 14 (1991) 335-345 North-Holland
335
Deposition behavior of ferrous alloys in molten lead-lithium P.F. Tortorelli Oak Ridge National Laboratory, P.O. Box 2008, Oak Ridge, Tennessee 37831, USA Submitted May 1990; accepted 1 August 1990 Handling Editor: G. Casini
The mass transfer of type 316 stainless steel in Pb-17 at% Li was studied using a thermal convection loop operating at a maximum temperature of 500 °C to generate mass change and surface composition data as a function of time and loop position. Data analysis indicated that particles suspended in the flowing liquid metal (particularly those containing nickel) probably played a significant role in overall transport and deposition. There was also some evidence of physical detachment of deposits. The deposition of chromium (but not nickel) correlated with the temperature dependence of solubility, as did previous weight change results from a study of ferritic (Fe-Cr) steels in nonisothermal Pb-17 at% Li. Due to the influence of particulate matter in the liquid metal and deposit detachment, mass transfer prediction for austenitic (Fe-Cr-Ni) steels in Pb-17 at% should be more complicated than that for Fe-Cr steels.
1. Introduction The corrosion and mass transfer of structural and containment materials by molten Pb-17 at% Li (hereafter denoted as "Pb-17Li") is an important technical issue for fusion reactors using this liquid metal as a tritium-breeding fluid a n d / o r coolant. This concern has led to investigations of corrosion by Pb-17Li within the past ten years (see, for example, refs. [1-11]). Most of the initial efforts involved measurements associated with dissolution reactions and focused on understanding the factors that were most influential in determining dissolution rates. However, more recently, greater attention has been given to deposition processes in Pb-17Li [8-12]. With all liquid metals under nonisothermal conditions, deposition constitutes at least as serious a conceru as material thinning. The accumulation of deposits can lead to flow restrictions that increase pumping power requirements and decrease heat transport. In nuclear applications, deposition of radioactive species outside the primary reactor area can necessitate increased shielding and remote maintenance. This paper presents deposition data for type 316 stainless steel (as a model austenitic alloy) exposed in a loop circulating Pb-17Li between 500 and 370-400 o C. It describes the amount and nature of deposits as a function of time, includes an analysis of this and some previously pub-
lished data, and briefly compares the results for austenitic steel with what is known about deposition in the F e - C r alloy/Pb-17Li loop system.
2. Experimental procedures The exposures of annealed and cold rolled electropolished type 316 stainless steel ( F e - 1 7 C r - 1 2 N i - 2 M o 2Mn-ISi-0.08C, wt%) to Pb-17L were conducted in a thermal convection loop (TCL) that is schematically illustrated in fig. 1. The TCL was constructed of type 316 stainless steel tubing which had previously circulated lithium for over 10,000 h. Type 316 stainless steel coupons (1.9 x 0.8 × 0.2 cm) were exposed at the indicated positions in the circuit. (Four specimens were grouped at hottest specimen position for studies of dissolution kinetics.) The ratios of surface area to volume of flowing liquid metal were 0.02 and 0.24 cm -1 for the coupons and total exposed steel, respectively. The maximum loop temperature was 500 ° C. The temperature differential ( A T ) between the hottest and coldest points initially was 130 ° C but decreased with loop operating time. With the exception of those specimens at the maximum temperature, each set of coupons was exposed for a fixed amount of time and then removed in its entirety. In this way, weight change and microstruct-
0 9 2 0 - 3 7 9 6 / 9 1 / $ 0 3 . 5 0 © 1991 - Elsevier Science Publishers B.V. ( N o r t h - H o l l a n d )
P.F. Tortorelli / Deposition behavior of ferrous alloys
336
~STANDPIPE
~_~
;l
J1
I K.~ TWINBALL i]::l VALVES
H1 H2 H3
C1
The preparation of the Pb-17Li has been described in a prior paper [3]. After specimens were exposed in the TCL, low temperature (250-300 o C) molten lithium was used to remove residual lead-lithium from them. This method was effective, but, as discussed in detail previously [3], can lead to partial or complete detachment of the corrosion layer on type 316 stainless steel. If a specimen cleaned in this way was again exposed, corrosion would be greater than if the layer had not been disturbed [3]. Therefore, the weight change data reported below represent measurements from individual coupons, which were cleaned after loop exposure and not re-exposed.
C2 3. Experimental results
C3 H4 C4
H5 76cm H6
,[
H7 H8
0 I
I
/K..~ HEATERS ~INSULATION CORROSION SPECIMENS'~
C5
C6
Net weight changes for type 316 stainless steel coupons positioned around the interior of a Pb-17Li thermal convection loop are shown in fig. 2 for three different specimen sets exposed for 698, 4500, and 10,000 h (the 10,000 h weight change data were reported previously) [10]. In all cases, the mass of material dissolved from the coupons was much greater than the amount deposited on specimens and the total transported mass increased with increasing time. However, compared to the weight losses, there was little increase in weight gains between 4500 and 10,000 h for the coupons in the vertical cooled ("cold") leg of the TCL.
C7
FLOW DIRECTION C8 1 5 ~ I
type 316 stainless steel, Pb-17Li
200
~---"Hot" Leg~
~
O;
"Cold"Leg
. . . .
-200
cm
~'~~IZ EVALVE DUMP TANK Fig. 1. Schematic drawing of the type 316 stainless steel thermal convection loop. Coupons were not located at positions H1, H2, C1, C2, and C3 in the present study.
-800 -1000
I 1
Distance From Bottom Of "Hot" Leg (m)
ural data as a function of loop position were obtained for specimen sets exposed for 698, 4488 (hereafter referred to as 4500), and 10,008 (10,000) h.
Fig. 2. Weight change versus loop position for type 316 stainless steel specimens in a P b - 1 7 at% Li thermal convection loop operating between 500 and 370 to 400 o C. (Distance is measured in direction of liquid metal flow.)
P.F. Tortorelli / Deposition behavior of ferrous alloys type 316 stainless steel, Pb-17Li 60
50
E
40 e•~
30
20
10 0 1.3
L
~ 1.4
.t 1.5
,
_.
_~
,
1.6
1.7
l.S
1.9
~
2.(I
Distance F r o m Bottom O f " H o t " L e g (m)
Fig. 3. Weight gain versus position in the cooled vertical leg for type 316 stainless steel specimens in a Pb-17 at% Li thermal convection loop operating between 500 and 370 to 400 o C. (Distance is measured in direction of liquid metal flow.)
This is shown more clearly in fig. 3, which focuses on the cold leg portions of the mass transfer profiles shown in fig. 2. As noted in Section 2, the A T of the type 316 stainless steel loop decreased during the course of these experiments, presumably due to localized accumulation
80
,
,
Pb-17Li ,
,
,
420
440
1"1 4.5K h, type 316SS [@ 10K h, type 316SS
•=- 4o
0 340
360
380
400
46(1
T e m p e r a t u r e (deg C) Fig. 4. W e i g h t gain v e r s u s t e m p e r a t u r e for type 316 stainless steel s p e c i m e n s in the cooled vertical leg of a P b - 1 7 at% Li t h e r m a l c o n v e c t i o n loop o p e r a t i n g b e t w e e n 500 a n d 370 to 400 ° C .
337
of deposits that impeded the flow of the liquid metal. Therefore, the comparisons of the longer term data in figs. 2 and 3 reflect a changing cold leg temperature profile. Accordingly, fig. 4 shows the 4500 and 10,000 h weight change data of fig. 3 plotted against cold leg specimen temperature rather than position in the TCL. For further comparison, fig. 4 includes previously reported 10,000 h data [10] for F e - 1 2 C r - 1 M o V W steel exposed to Pb-17Li in a ferritic steel TCL under similar conditions. The nature of the process(es) that led to net weight gains was characterized by microscopic examination of the surfaces of cold leg specimens exposed for several thousand hours. Figure 5 contains representative scanning electron micrographs of polished cross sections of coupons exposed at the five cold leg loop positions for 4500 and 10,000 h. Deposition was not uniform across the exposed surfaces: individual or clustered deposits were sometimes observed and the deposit number and size densities were not always constant across a particular coupon. In several cases, "detachment" of deposits was noted (see figs. 5 and 6), particularly for the 10,000 h specimens, which had large deposit-free areas at the lower temperatures [10]. As shown in fig. 2, weight gains were also measured for the specimens positioned at the bottom of the TCL's heated vertical (hot) leg and deposits were observed on the 4500 h specimen in this location. (The equivalent 10,000 h specimen was not examined). Relative to those observed on the cold leg coupons, these deposits were small and less numerous, in agreement with the small weight increase measured for this specimen (fig. 2). Energy dispersive X-ray (EDX) analysis was used in conjunction with the scanning electron microscopy to determine the composition of the observed deposition products. Results for the cold leg specimens exposed for 4500 and 10,000 h are summarized in table 1 which contains average compositions determined from a number of deposits of a particular type observed in cross section (such as those shown in fig. 5). A distinct dependence of deposit composition on loop position, as previously reported based only on the 10,000 h data [10], was observed for both exposute conditions. The agreement between the deposit compositions measured for coupons exposed for 4500 h and those for the 10,000 h surfaces was excellent for the C4 and C5 specimen positions: in all cases the average composition was F e - 7 C r - ( 2 - 3 ) N i (wt%). At lower temperatures, the deposits were a combination of two or more of F e 7 C r - ( 2 - 3 ) N i (A, in table 1), F e - ( 3 2 - 5 0 ) N i - ( l l - 2 5 ) C r (B), F e - ( 3 3 - 3 5 ) C r - - 2 N i (C), and F e - ( 5 0 - 7 0 ) C r - - 2 N i (D), as shown in table 1. (When multiple compositions
338
P.F. Tortorelli / Deposition behavior of ferrous alloys
I
Fig. 5. Scanning electron micrographs of polished cross sections of type 316 stainless steel specimens exposed in the cooled vertical leg of a Pb-17 at% Li thermal convection loop operating between 500 and 370 to 400°C (the temperature decreases from top to bottom); (a) 4488 h, (b) 10008 h.
339
P.F. Tortorelli / Deposition behavior of ferrous alloys
Table 1 Average composition of mass transfer deposits on type 316 stainless steel coupons in cold leg of Pb-17 at% thermal convection loop for two exposure periods Loop
4500 h
position
Temperature a
10008 h Composition b
Temperature a
(*c) CA C5 C6 C7 C8
Fig. 6. Scanning electron micrograph of the surface of a type 316 stainless steel specimen exposed at 430°C in the cooled vertical leg of a Pb-17 at% Li thermal convection loop operating at a maximum temperature of 500 * C.
are shown, they are listed in probable order of abundance). Although there were some compositional differences between the two exposure times, a predominance of Cr-rich deposits were observed at lower temperatures for both specimen sets. Type A deposits were found on the 4500 h specimens in the C6 and C7 positions (see table 1), but such products were completely absent from the surfaces of the corresponding 10,000 h specimens (even when such were analyzed
460 440 420 395 370
(*c) A A A, C, D (A, C, D) ¢ D, C, B
wt~.
normal to the surface rather than in cross section [10]). Another difference between the two sets of results was the point at which nickel-enriched deposits (type B)
Pb-17Li
|
100
|
ll**..~t
Fe
Cr
/'%.. m=
I------
I
~'e= =
art
O ",if,
Ni Start
~,
%_
"L.
A A B, D D, B D, B
A: Fe-(7-10)Cr-(2-3)Ni B: Fe-(32-50)Ni-(11-25)Cr C: Fe-(33-35)Cr-2Ni D: Fe-(50-70)Cr--2Ni c Approximately equal abundance.
80
=
460 445 430 415 400
a Approximate. b Compositions determined by energy dispersive X-ray analysis. Multiple average compositions shown when distinct types of deposits were observed. Listed in decreasing order of apparent abundance when multiple compositions are reported except where noted otherwise. Concentrations are in
Pb-17Li 1O0
Composition b
"
60
40
.
20 ............... ~ " 0 360
r
Ni 410 460 Temperature(degC) ,~l"
--' 0 " ~
"'""
I
0' 360
410
460
Temperature (deg C)
Fig. 7. Elemental concentration versus position in the cooled vertical leg for type 316 stainless steel specimens in a Pb-17 at% Li thermal convection loop operating between 500 and 370 to 400 ° C for 4488 h. Concentrations determined by energy dispersive X-ray analysis; (a) 4488 h, (b) 10 008 (horizontal lines represent starting concentrations of chromium and nickel).
340
P.F. Tortorelli / Deposition behavior of ferrous alloys
Pb-17Li 100
!
!
!
I
Pb-17Li !
100
|
i
&
80
80
!
|
AL
.L
Fe
60 I ]
O
:1
O
i
q~
.......... ~ --
40
Cr Sta~ Ni Start
I
O ~J
20
0 380
Cr
20
D::I./. . . . . . \ " ~ . . . . . . . . . . . . . . . . . . . . . . . . . . . ~ . . . . ~. --. •- . / ' , a . l n . . . . - ."'"W,..-.n,.~w,,.,~|
===========tm====a==========...===l==l=ll=l===l====== m
Cr . _ . _ ~ o , - - # l . m o l l ~ R . . o t ~ . ~ . .
'wr
I
I
I
I
I
400
420
440
460
480
500
520
0 400
i 420
i 440
i 460
Ni
480
T e m p e r a t u r e (deg C) Temperature (deg C) Fig. 8. Elemental concentration versus position in the heated vertical leg for type 316 stainless steel specimens in a Pb-17 at% Li thermal convection loop operating between 500 and 370 to 400 °C for 10008 h. Concentrations determined by energy dispersive X-ray analysis; (a) 4488 h; (b) 10008 h (horizontal lines represent starting concentrations of chromium and nickel).
appeared: such deposits were observed farther up the cold leg (at a higher temperature) in the longer term loop experiment. This is also evident in fig. 7, where surface-averaged concentrations of Fe, Cr, and Ni are plotted versus cold leg loop position. With the caveat that these average values could be somewhat misleading because more than one type of deposit was found on some of the surfaces, these plots clearly illustrate the general trend of increasing C r / F e with decreasing temperature and the nickel enrichment at intermediate cold leg temperatures. There was no apparent clustering of deposits of like compositions in the cases where there was more than one deposit type. Different deposits existed in close proximity to each other. In the case of the 4500 h C7 specimen, there seemed to be a tendency for the Fe-9Cr (type A) deposits to be situated above the C and D (high Cr) types, which were nearer the original liquid/solid interface. For the C8 specimen exposed for 4500 h, there were some indications that the deposits furthest from the substrate tended to be Fe-Ni, while the ones next to the starting coupon surface were again relatively high in Cr. A similar tendency for underlying deposits to be more concentrated in chromium was also found for the C6 and C7 specimens of the 10,000 h set. The compositions of selected specimen surfaces along part of the heated loop section were also determined by EDX analysis. Surface-averaged compositions for these specimens are shown in fig. 8. The deposits formed on
the specimen exposed for 4500 h at the bottom of the heated vertical ("hot") leg (net weight gain) were A type: their average composition was F e - 8 C r - 2 N i - l S i . As reported previously [3,10], specimens at the maximum temperature position (500°C) suffered preferential dissolution of nickel and chromium. In fact, as shown in fig. 8, many of the hot leg specimens that experienced weight losses were depleted in these dements relative to the starting concentrations of the steel. However, the plot for the 4500 h set (fig. 8a) showed a maximum in surface nickel concentration at an intermediate temperature, while the specimens between 460 and 4200 ° C exhibited surface nodules of pure Mo.
4. Discussion
Under equilibrium, steady-state conditions, the net flux, J/, of element i from or to a specimen exposed to molten Pb-17Li at loop position r can be expressed as J~(r) = k ; ( r ) [ C i ° ( r ) - C , ( r ) ] ,
(1)
where k~ is the effective rate constant for dissolution ( z = s ) or deposition ( z = p ) of element i, C° is the solubility of this element in the Pb-17Li, and Ci is the local concentration of element i in the liquid metal. (C° and k~ are dependent on temperature, which is a function of r.) Normally, in the higher temperature region of the loop c O > Ci and dissolution occurs ( J / > 0 ,
P.F. Tortorelli / Deposition behavior of ferrous alloys
weight losses are measured), while in the colder part of the circuit C° < Ci and weight is gained from deposition on the solid surfaces. Such solubility-driven transport can be monitored by measurements of weight changes (E f J / d t ) of coupons arrayed around the loop, which typically produce mass transfer profiles such as those shown in fig. 2. While eq. (1) yields a qualitative understanding of corrosion and mass transfer in this and other liquid metal systems, it cannot, in many cases, explain the quantitative aspects of deposition caused by additional kinetic and thermodynamic factors, such as interactions among solutes and reactions with impurities in the liquid metal or solid. As shown below, this solubility-based approach does not predict many of the features of the present mass transfer profiles. Figure 2 shows that while substantial mass was lost from the hot zone specimens during each loop experiment, much less material per unit area was deposited on the cold leg specimens. Furthermore, this large imbalance between the measured weight losses (Am s) and gains (Am p) increased as a function of exposure time. (The total coupons weight gains and losses for the three exposure periods are listed in table 2.) This discrepancy between A m s and AmP could be explained if deposition in the TCL was preferentially occurring on loop walls or in parts of the loop net containing specimens. However, it is interesting to note that in a recent mass transfer study of austenitic steel in Pb-17Li [11], in which the total amount of deposited mass was determined by destructive examination of loops and physical removal of deposition products from all surfaces, the total dissolved mass was still significantly greater (by a factor of 1.5-2) than the total mass gain from deposition. Such findings tend to indicate that the present observation of an imbalance between Am s and Am p may not be just due to the failure of cold leg coupons to accurately reflect the appropriate amount of deposited material. Instead, it seems likely that there might be particles of solutes in the Pb-17Li. In support of this argument, Borgstedt [12] has reported concentrations of Fe, Cr, and Ni in Pb-17Li above their respective solubility limits. Suspended particles can continue to remain in the liquid metal until they agglomerate to a size where plugging may occur. Therefore, indications of a flow restriction (decreasing AT with time) in the type 316 stainless steel Pb-17Li TCL of the present study do not necessarily indicate that solubility-driven deposition was occurring outside of the area covered by the coupons. The formation of particles in the liquid metal may well be related to an inefficient a n d / o r sluggish deposi-
341
tion process(es). The growth of deposits requires registry with the underlying surface and if such is not possible, or is somehow disrupted during the initial stages of deposition, supersaturation a n d / o r particle formation can occur and be maintained [13]. Previously reported observations of small deposits in definite patterns on some of the 10,000 h surfaces [10] may indicate that deposition on the coupons was retarded in this way. While both supersaturation and particle formation would lead to reduced deposition fluxes [based on eq. (1)], only the former condition would also inhibit dissolution unless the particles dissolved in the hot zone. Therefore, on the basis of the large imbalance in weight gains and losses and the relatively large dissolution area, it can be concluded that the liquid metal was not just greatly supersaturated in the present case. Poor deposit adhesion can also contribute to the observed discrepancy between Am s and Am p. Detachment of deposition products can occur during exposure to the liquid metal (a source of particulate matter), during cooling after specimen removal from the loop, a n d / o r during the specimen cleaning process [3]. Observations such as that of the "delamination" of the deposit layer shown in fig. 6 lend some validity to a contribution of deposit detachment to lower weight gains. Furthermore, the significant amount of deposit-free surface areas on the lower t0mperature coupons of the 10,000 h specimen set is indicative of poor adhesion - a phenomenon that should be more prevalent at longer times when deposited layers become relatively thick. The above consideration of the overall weight change data and general surface morphologies indicate three possible contributions to the discrepancy between the amounts of mass dissolved and deposited: heterogeneous, localized deposition in loop areas not covered by coupons, homogeneous particle nucleation in the molten Pb-17Li, or physical removal of deposits from cold leg coupons. Allowing for all these possibilities, a mass balance for the loop would yield Am s = Am p + Am r + Am d + Am r + Am n,
(2)
where Am s and Am p are as defined above, Am f is the mass of particles that formed and stayed suspended in the Pb-17Li due to solute released from coupons, Am d is that of detached deposits suspended in the liquid metal, Am r is the contributiuon of deposits removed during cooling or specimen cleaning, and Am a is the mass associated with nonproportional deposition in areas of the loop not monitored by coupons. In order to gain a better understanding as to the relative importance of these processes in controlling mass transfer and
342
P.F. Tortorelli / Deposition behavior of ferrous alloys
deposition, the cold leg weight change profiles and surface chemistry data is analyzed in some detail below. As can be seen in figs. 3 and 4, the weight gain profile for the 10,000 h specimen set is distinctly different from that after 4500 h. In the dissolution regime, C°-Ci > 0 , and, for a given time, Ji and measured weight losses should normally increase with increasing temperature, as can be noted in fig. 2 for the three exposure periods. When C°-Ci < 0, deposition should occur. In many of these cases, weight gains increase as the temperature decreases and a weight change profile like that shown in fig. 3 for the 4500 h specimen set is measured. However, if the two functions that determine the magnitude of J,( I Ci° - C, I and k r ) vary oppositely with respect to temperature, it is also possible that a maximum in the weight gain versus loop temperature curve, such as shown in fig. 3 for the 10,000 h set, would result. In view of the 4500 h data, this possibility is considered highly unlikely - the functional dependence of I C ° - Cil or k/p on temperature should not be different for the two loop experiments. Therefore, it appears that the 10,000 h data cannot be explained on the basis of such competition between factors that influence J~, which was one of several possibilities suggested when only the 10,000 h data was considered [10]. If the physical removal of deposits from cold leg surfaces occurred, it could also explain the differences between the 4500 and 10,000 h deposition data. The shape of the cold leg profile for the 10,000 h coupons (see figs. 3 and 4) could have resulted from a loss of previously deposited material that reached a critical thickness for detachment. The observation that the cold leg coupon surfaces in the two extended loop experiments showed similarities in surface composition trends is consistent with a process of initially equivalent deposition profiles: in general, as shown in table 1 and fig. 7, iron was the principal deposited element at higher temperatures of the cold legs, and chromium was dominant at the coldest loop positions. Some of differences in the composition data between the two experiments - f o r example, the appearance of F e - N i nodules at a higher temperature for the 10,000 h set - could possibly be explained by detachment of such deposits at lower temperatures, but the observation of type A (Fe-low Cr) deposits on lower temperature coupons in the 4500 h experiment is not consistent with such an argument. On the other hand, if suspended particles contributed to the mass transfer process, it is entirely possible to have deposition of particles rich in Ni anywhere in the loop [14], particularly at longer times when their density and size in the liquid would be much greater. Further evidence for such deposition was revealed by the hot leg
surface composition data for the 4500 h coupons (see fig. 8a): a maximum in nickel concentration was also observed despite an overall decrease in weight. If these increases in surface nickel concentrations are due to the influence of Ni-containing particles in the Pb-Li, the deposition rate is not governed by solubility considerations, but rather by hydrodynamic factors and sticking coefficients. The EDX data appear to be consistent with the role of particulate matter in the transport of Ni around the loop and thus affirm a contribution of such particles to the discrepancy between measured weight gains and losses. As noted in the Experimental Procedures section, the same TCL was used for the exposures of the three type 316 stainless steel specimen sets to Pb-17Li and, previously, had circulated lithium for over 10,000 h. Therefore, a large fraction of the loop's heated section, originally type 316 stainless steel, was already significantly depleted in nickel when fresh specimens were inserted. The presence of a low-nickel loop surface relative to the composition of the coupons created a possible driving force for transport of dissolved nickel to the loop wall of the hot leg. Such a process could increase the contribution of Arn n [eq. (2)] to the observed imbalance in measured weight losses and gains. However, this situation never seemed to have an effect on dissolution rates in lithium or Pb-17Li [3,15]. Furthermore, as shown in fig. 8, nickel was deposited on some hot leg coupons after 4500 h and also was found on cold leg coupons after 10,000 h (fig. 7, table 1). Thus, it does not appear that the concentration gradient between hot specimens and loop wall had a dominant influence on mass transport. A more important factor may be that, as successive specimens sets were exposed to the Pb-17Li, deposits were already present on the loop wall in the colder Part of the TCL. Therefore, deposit nucleation on this surface could have been favored over that on the fresh coupons during exposure of a later specimen set [10]. If this was the case, a nonproportional partitioning of deposits to the loop wall (Am n) could contribute to the mass imbalance and, as observed in the present experiments, this difference would increase with increasing time as deposition in the cold zone increased with successive exposures. While nickel transfer around the loop appeared to be "irregular", the surface-averaged cold leg composition data (fig. 7) revealed that the chromium concentration increased monotonically with decreasing temperature. In both the 4500 h and 10,000 h experiments, the chromium concentration profiles appeared to show behavior expected for a solubility-driven deposition process since, at a given time, the amount of deposited
P.F. Tortorelli / Deposition behavior of ferrous alloys Table 2 Total measured weight changes for specimens exposed in Pb17Li thermal convection loops Measured wt loss (mg)
Measured wt gain (rag)
316SS, 700 h 0 42 316SS, 4488 h 1051 316SS, 10008 h 2673
1 65 74
chromium, Am~r, increased with decreasing CC0r as predicted by eq. (1):
Am~r= fJc,dt= fk~,[IC~,-CCrl]dr,
(3)
assuming Ccr is constant around the loop. In view of the monotonic behavior of the cold leg chromium concentration with temperature and the preferential dissolution of chromium in the hot leg [10], two mass transfer balance points can be defined as being those loop positions where Jcr = 0. These points delimit the area of chromium dissolution relative to that for its deposition. The balance points can be determined from the data shown in figs. 7 and 8 by noting the loop position (temperature) at which the surface composition curves intersect the horizontal line representing the starting surface concentration of Cr in the alloy [16]. For the 4500 h experiment, they are about 415 and 425 ° C for the hot and cold legs, respectively. Under the assumption that the elemental fluxes (J~) are independent of each other (that is, the activity of chromium at the interface is not strongly affected by changes in the relative concentrations of the other elements), Jc,(Tb) = 0 =
k~r(Tb)[C_~r(Tb)CCr] -
(4)
and, therefore,
CCr = C~r(Tb),
(5)
where Tb is the loop temperature at the balance point in either the hot (Th) or the cold leg (T~). Because solubility is normally a single-valued function of temperature, eq. (5) indicates that if the balance point temperatures are unequal, Ccr cannot be constant around the loop. However, from the 4500 h data, T h -- T~ within the accuracies of the surface composition and temperature measurements. These results are therefore consistent with a constant Ccr and, as mentioned immediately above, with solubility-driven chromium de-
343
position as described by eq. (1). Furthermore, referring to fig. 7, it can be noted that T~ is approximately the same (within 10 o C) for the 4500 and 10,000 h profiles. In view of eq. (5), this finding implies that CCr has reached an equilibrium, steady-state value, which is characteristic of solubility-driven transport as described by eq. (1). (However, a constant Ccr can also be possible with particle transport.) Based on the above considerations, it can be concluded that there is an important homogeneous particle nucleation contribution to overall mass transfer of type 316 stainless steel in Pb-17Li thermal convection loops. The influence of particle formation in nonisothermal liquid metal systems has been recognized for many years [14], and as already noted, a role of particulate matter in determining mass transport in Pb-17Li has been suggested or implied by the results of others [11,12]. While the present results indicate that the particles contain substantial amounts of nickel, it is not necess~-y that all of them are nickel-rich or F e - N i . Some may contain chromium in significant concentrations, but, because such particles neither redissolved (as evidenced by continual chromium depletion for the hot zone specimens and Th ~-T~, see above) nor attached themselves to loop coupons (no irregularities in the chromium profiles), they do not influence the measured quantities except for their contribution to the imbalance in overall weight losses and gains. On the other hand, the nonmonotonic mass transfer profiles for Ni (figs. 7 and 8) may mask some solubility-driven deposition of this element. However, the relatively low concentration of nickel found on many of the cold leg surfaces and the observations that nickel-rich deposits tended to deposit over chromium-enriched ones (that is, chromium came out of solution earlier than nickel) indicate that this type of transport for nickel was quite limited. The molybdenum nodules on some of the 4500 h hot leg coupons may also be an example of particles carried by the liquid metal. Alternatively, because of molybdenum's probable low solubility in Pb-17Li (it has low solubilities in lithium and lead), the nodules may be dissolution-resistant areas that developed into protuberances due to shielding of underlying material as the surface recedes. Their absence from the surface at the hottest loop position and from any 10,000 h coupons may be explained by undercutting and release of these nodules when surface recession exceeded a certain value. Referring to eq. (2), the above analyses of the cold leg coupon data for the 4500 and 10,000 h specimen sets cannot definitely show that any of the possible contributions to the mass imbalance (panicles, detachment, nonproportional deposition) are negligible. Despite the
344
P.F. Tortorelli / Deposition behavior of ferrous alloys
complication of a possible concentration dricing force for preferential deposition on the loop wall in the extended exposure experiments, the present data and the results of others [11,12] strongly support a role of particle transport in this Pb-17Li system. This finding is particularly compelling in view of the work of Flament et al. [11] in which Am n-- 0, and a definite mass imbalance was still measured. There was also direct evidence of some particle detachment (for example, fig. 6) either during exposure or after specimen removal. As mentioned above, in the former case, such release of deposits can then be treated as an additional source term for particles in the liquid metal. Previous work [10] concluded that, in contrast to the type 316 stainless system, solubility-driven deposition was the most important factor for mass transfer of a F e - 1 2 C r - I M o V W steel exposed to Pb-17Li in a ferritic steel TCL under similar conditions. Figure 4 shows the differences in cold leg mass gain profiles between these two steel systems: increasing weight gains with decreasing temperature were measured in the F e - 1 2 C r 1MoVW steel experiment (10,076 h). There was some discrepancy between measured weight gains and losses, but it was much smaller than what was observed for type 316 stainless steel. Cold leg deposits in the ferritic steel loop were type A (Fe-8Cr). This mass transfer behavior of the F e - 1 2 C r - I M o V W steel is similar to much of what was observed in the 4500 h austenitic steel experiment with respect to chromium transport (compare the cold leg mass change profiles shown in fig. 4). In this way, the results for the ferritic system support many of the present findings. Furthermore, the observation of a mass imbalance in the F e - C r experiment (also reported by Sannier et al. [11] for a ferritic s t e e l / P b 17Li loop system) may indicate that, as discussed above, particles involving chromium may form, but are not manifested in a way that is amenable to the type of measurements and analyses of the previous and present studies. If this is the case, mass transfer prediction for ferritic steels in Pb-17Li may not be as straightforward as it would normally seem when a solubility-controlled process is dominant. (Indeed, for both steels, if particles could "float" in Pb-17Li, mass transport becomes increasingiy complex [17].) However, the present results do show that, even apart from the complications of preferential dissolution and nonuniform surface recession [6,10] (not observed for the ferritic steels), predictive capability for modeling, and control of, mass transport in lead-lithium will be more difficult for austenitic ( F e - N i - C r ) alloys than for F e - C r steels.
5. Summary Mass transfer data for type 316 stainless steel in thermally convective Pb-17Li (maximum temperature of 500 ° C) revealed much larger weight losses than weight gains and unusual weight cha~ge and nickel deposition profiles. Analysis of weight change and surface composition and morphological data led to the following conclusions: Particles played a role in overall transport. The irregular movement of nickel around the loop was associated with particle formation and attachment to solid surfaces. While some chromium transport may have involved particles, the surface composition data showed that deposition of chromium was primarily controlled by a solubility driving force. This finding agreed with results from a previous study of a F e - C r steel in nonisothermal Pb-17Li. A balance point analysis indicated that a steady-state chromium concentration in the Pb-17Li was achieved. Physical detachment of deposits (during or after exposure) contributed to the significant differences in measurements of dissolved and deposited masses. The analysis was complicated by the possible influence of preferential deposition on the thermal convection loop wall due to differences in composition between it and the specimens. Due to the influence of particles and lack of good deposit adhesion, as well as the effects of preferential dissolution, mass transfer prediction for austenitic ( F e C r - N i ) steels in Pb-17Li should be more complicated than that for ferritic ( F e - C r ) steels.
Acknowledgement Research sponsored by the Office of Fusion Energy, U.S. Department of Energy under contract DE-AC0584OR21400 with Martin Marietta Energy Systems, Inc.
References [1] P. Fauvet, J. Sannier, and G. Santarini, in: Proc. 13th Symp. on Fusion Technology, Vol. 2, Commission of the European Communities (Pergamon Press, 1984), pp. 1003-1009. [2] V. Coen et al., in: Proc. 3rd Int'l. Conf. on Liquid Metal Engineering and Technology, Vol. 1 (The British Nuclear Energy Society, 1984), pp. 347-354.
P.F. Tortorelli / Deposition behavior of ferrous alloys [3] P.F. Tortorelli and J.H. DeVan, J. Nucl. Mater. 141-143 (1986) 592-598. [4] O.K. Chopra and D.L. Smith, J. Nucl. Mater. 141-143 (1986) 566-570. [5] G.M. Gryaznov et al., Sov. At. Energy 59 (1985) 918-921. [6] H. Tas et al., J. Nucl. Mater. 141-143 (1986) 571-578. 7 H.U. Borgstedt et al., J. Nucl. Mater. 155-157 (1988) 728-731. [8] Z. Peric, G. Drechsler, G. Frees, and H.U. Borgstedt, in: Proc. Fourth Int'l. Conf. in Liq. Metal Engineering and Technology, Vol. 3 (Soci6t6 Fran$aise d'Energie Nucl6aire, 1988), pp. 522-1 - 522-7. [9] M. Broc, P. Fauvet, T. Flament, A. Terlain, and J. Sannier, in: Proc. Fourth Int'l Conf. on Liq. Metal Engineering and Technology, Vol. 3 (Soci~t6 Fran$aise d'Energie Nucl~aire, 1988), pp. 527-1-527-10. [10] P.F. Tortorelli, in: Proc. Fourth Int'l Conf. on Liq. Metal Engineering and Technology, Vol. 3 (Soci6t~ Fran~aise d'Energie Nucl6aire, 1988), pp. 528-1 - 528-10.
345
[111 J. Sannier, M. Broc, T. Flament and A. Terlain, Fusion Engrg. Des. 14 (1991) 299-307, in this issue. [121 H.U. Borgstedt, KfK, private communication (Sept. 1989) [13] O.F. Kammerer et al., Trans. AIME 212 (1958) 20-25. [14] L.F. Epstein, Static and dynamic corrosion and mass transfer in liquid metal systems, Liquid Metals Technology, Chem. Eng. Prog. Syrup. Ser. 20, 53 (1957) 67-81. [15] P.F. Tortorelli and J.D. DeVan, in: Proc. 3rd Int'l Conf. on Liquid Metal Engineering and Technology, Vol. 3 (The British Nuclear Energy Society, 1985), pp. 81-88. [16] P.F. Tortorelli, J.H. DeVan, and J.E. Selle, in: Proc. Second Int'l Conf. on Liquid Metal Technology in Energy Production, U.S. Dept. of Energy report, CONF-800401P2 (1980). [17] J.H. DeVan, Oak Ridge National Laboratory, private communication to P.F. Tortorelli (January 1990).
335
Deposition behavior of ferrous alloys in molten lead-lithium P.F. Tortorelli Oak Ridge National Laboratory, P.O. Box 2008, Oak Ridge, Tennessee 37831, USA Submitted May 1990; accepted 1 August 1990 Handling Editor: G. Casini
The mass transfer of type 316 stainless steel in Pb-17 at% Li was studied using a thermal convection loop operating at a maximum temperature of 500 °C to generate mass change and surface composition data as a function of time and loop position. Data analysis indicated that particles suspended in the flowing liquid metal (particularly those containing nickel) probably played a significant role in overall transport and deposition. There was also some evidence of physical detachment of deposits. The deposition of chromium (but not nickel) correlated with the temperature dependence of solubility, as did previous weight change results from a study of ferritic (Fe-Cr) steels in nonisothermal Pb-17 at% Li. Due to the influence of particulate matter in the liquid metal and deposit detachment, mass transfer prediction for austenitic (Fe-Cr-Ni) steels in Pb-17 at% should be more complicated than that for Fe-Cr steels.
1. Introduction The corrosion and mass transfer of structural and containment materials by molten Pb-17 at% Li (hereafter denoted as "Pb-17Li") is an important technical issue for fusion reactors using this liquid metal as a tritium-breeding fluid a n d / o r coolant. This concern has led to investigations of corrosion by Pb-17Li within the past ten years (see, for example, refs. [1-11]). Most of the initial efforts involved measurements associated with dissolution reactions and focused on understanding the factors that were most influential in determining dissolution rates. However, more recently, greater attention has been given to deposition processes in Pb-17Li [8-12]. With all liquid metals under nonisothermal conditions, deposition constitutes at least as serious a conceru as material thinning. The accumulation of deposits can lead to flow restrictions that increase pumping power requirements and decrease heat transport. In nuclear applications, deposition of radioactive species outside the primary reactor area can necessitate increased shielding and remote maintenance. This paper presents deposition data for type 316 stainless steel (as a model austenitic alloy) exposed in a loop circulating Pb-17Li between 500 and 370-400 o C. It describes the amount and nature of deposits as a function of time, includes an analysis of this and some previously pub-
lished data, and briefly compares the results for austenitic steel with what is known about deposition in the F e - C r alloy/Pb-17Li loop system.
2. Experimental procedures The exposures of annealed and cold rolled electropolished type 316 stainless steel ( F e - 1 7 C r - 1 2 N i - 2 M o 2Mn-ISi-0.08C, wt%) to Pb-17L were conducted in a thermal convection loop (TCL) that is schematically illustrated in fig. 1. The TCL was constructed of type 316 stainless steel tubing which had previously circulated lithium for over 10,000 h. Type 316 stainless steel coupons (1.9 x 0.8 × 0.2 cm) were exposed at the indicated positions in the circuit. (Four specimens were grouped at hottest specimen position for studies of dissolution kinetics.) The ratios of surface area to volume of flowing liquid metal were 0.02 and 0.24 cm -1 for the coupons and total exposed steel, respectively. The maximum loop temperature was 500 ° C. The temperature differential ( A T ) between the hottest and coldest points initially was 130 ° C but decreased with loop operating time. With the exception of those specimens at the maximum temperature, each set of coupons was exposed for a fixed amount of time and then removed in its entirety. In this way, weight change and microstruct-
0 9 2 0 - 3 7 9 6 / 9 1 / $ 0 3 . 5 0 © 1991 - Elsevier Science Publishers B.V. ( N o r t h - H o l l a n d )
P.F. Tortorelli / Deposition behavior of ferrous alloys
336
~STANDPIPE
~_~
;l
J1
I K.~ TWINBALL i]::l VALVES
H1 H2 H3
C1
The preparation of the Pb-17Li has been described in a prior paper [3]. After specimens were exposed in the TCL, low temperature (250-300 o C) molten lithium was used to remove residual lead-lithium from them. This method was effective, but, as discussed in detail previously [3], can lead to partial or complete detachment of the corrosion layer on type 316 stainless steel. If a specimen cleaned in this way was again exposed, corrosion would be greater than if the layer had not been disturbed [3]. Therefore, the weight change data reported below represent measurements from individual coupons, which were cleaned after loop exposure and not re-exposed.
C2 3. Experimental results
C3 H4 C4
H5 76cm H6
,[
H7 H8
0 I
I
/K..~ HEATERS ~INSULATION CORROSION SPECIMENS'~
C5
C6
Net weight changes for type 316 stainless steel coupons positioned around the interior of a Pb-17Li thermal convection loop are shown in fig. 2 for three different specimen sets exposed for 698, 4500, and 10,000 h (the 10,000 h weight change data were reported previously) [10]. In all cases, the mass of material dissolved from the coupons was much greater than the amount deposited on specimens and the total transported mass increased with increasing time. However, compared to the weight losses, there was little increase in weight gains between 4500 and 10,000 h for the coupons in the vertical cooled ("cold") leg of the TCL.
C7
FLOW DIRECTION C8 1 5 ~ I
type 316 stainless steel, Pb-17Li
200
~---"Hot" Leg~
~
O;
"Cold"Leg
. . . .
-200
cm
~'~~IZ EVALVE DUMP TANK Fig. 1. Schematic drawing of the type 316 stainless steel thermal convection loop. Coupons were not located at positions H1, H2, C1, C2, and C3 in the present study.
-800 -1000
I 1
Distance From Bottom Of "Hot" Leg (m)
ural data as a function of loop position were obtained for specimen sets exposed for 698, 4488 (hereafter referred to as 4500), and 10,008 (10,000) h.
Fig. 2. Weight change versus loop position for type 316 stainless steel specimens in a P b - 1 7 at% Li thermal convection loop operating between 500 and 370 to 400 o C. (Distance is measured in direction of liquid metal flow.)
P.F. Tortorelli / Deposition behavior of ferrous alloys type 316 stainless steel, Pb-17Li 60
50
E
40 e•~
30
20
10 0 1.3
L
~ 1.4
.t 1.5
,
_.
_~
,
1.6
1.7
l.S
1.9
~
2.(I
Distance F r o m Bottom O f " H o t " L e g (m)
Fig. 3. Weight gain versus position in the cooled vertical leg for type 316 stainless steel specimens in a Pb-17 at% Li thermal convection loop operating between 500 and 370 to 400 o C. (Distance is measured in direction of liquid metal flow.)
This is shown more clearly in fig. 3, which focuses on the cold leg portions of the mass transfer profiles shown in fig. 2. As noted in Section 2, the A T of the type 316 stainless steel loop decreased during the course of these experiments, presumably due to localized accumulation
80
,
,
Pb-17Li ,
,
,
420
440
1"1 4.5K h, type 316SS [@ 10K h, type 316SS
•=- 4o
0 340
360
380
400
46(1
T e m p e r a t u r e (deg C) Fig. 4. W e i g h t gain v e r s u s t e m p e r a t u r e for type 316 stainless steel s p e c i m e n s in the cooled vertical leg of a P b - 1 7 at% Li t h e r m a l c o n v e c t i o n loop o p e r a t i n g b e t w e e n 500 a n d 370 to 400 ° C .
337
of deposits that impeded the flow of the liquid metal. Therefore, the comparisons of the longer term data in figs. 2 and 3 reflect a changing cold leg temperature profile. Accordingly, fig. 4 shows the 4500 and 10,000 h weight change data of fig. 3 plotted against cold leg specimen temperature rather than position in the TCL. For further comparison, fig. 4 includes previously reported 10,000 h data [10] for F e - 1 2 C r - 1 M o V W steel exposed to Pb-17Li in a ferritic steel TCL under similar conditions. The nature of the process(es) that led to net weight gains was characterized by microscopic examination of the surfaces of cold leg specimens exposed for several thousand hours. Figure 5 contains representative scanning electron micrographs of polished cross sections of coupons exposed at the five cold leg loop positions for 4500 and 10,000 h. Deposition was not uniform across the exposed surfaces: individual or clustered deposits were sometimes observed and the deposit number and size densities were not always constant across a particular coupon. In several cases, "detachment" of deposits was noted (see figs. 5 and 6), particularly for the 10,000 h specimens, which had large deposit-free areas at the lower temperatures [10]. As shown in fig. 2, weight gains were also measured for the specimens positioned at the bottom of the TCL's heated vertical (hot) leg and deposits were observed on the 4500 h specimen in this location. (The equivalent 10,000 h specimen was not examined). Relative to those observed on the cold leg coupons, these deposits were small and less numerous, in agreement with the small weight increase measured for this specimen (fig. 2). Energy dispersive X-ray (EDX) analysis was used in conjunction with the scanning electron microscopy to determine the composition of the observed deposition products. Results for the cold leg specimens exposed for 4500 and 10,000 h are summarized in table 1 which contains average compositions determined from a number of deposits of a particular type observed in cross section (such as those shown in fig. 5). A distinct dependence of deposit composition on loop position, as previously reported based only on the 10,000 h data [10], was observed for both exposute conditions. The agreement between the deposit compositions measured for coupons exposed for 4500 h and those for the 10,000 h surfaces was excellent for the C4 and C5 specimen positions: in all cases the average composition was F e - 7 C r - ( 2 - 3 ) N i (wt%). At lower temperatures, the deposits were a combination of two or more of F e 7 C r - ( 2 - 3 ) N i (A, in table 1), F e - ( 3 2 - 5 0 ) N i - ( l l - 2 5 ) C r (B), F e - ( 3 3 - 3 5 ) C r - - 2 N i (C), and F e - ( 5 0 - 7 0 ) C r - - 2 N i (D), as shown in table 1. (When multiple compositions
338
P.F. Tortorelli / Deposition behavior of ferrous alloys
I
Fig. 5. Scanning electron micrographs of polished cross sections of type 316 stainless steel specimens exposed in the cooled vertical leg of a Pb-17 at% Li thermal convection loop operating between 500 and 370 to 400°C (the temperature decreases from top to bottom); (a) 4488 h, (b) 10008 h.
339
P.F. Tortorelli / Deposition behavior of ferrous alloys
Table 1 Average composition of mass transfer deposits on type 316 stainless steel coupons in cold leg of Pb-17 at% thermal convection loop for two exposure periods Loop
4500 h
position
Temperature a
10008 h Composition b
Temperature a
(*c) CA C5 C6 C7 C8
Fig. 6. Scanning electron micrograph of the surface of a type 316 stainless steel specimen exposed at 430°C in the cooled vertical leg of a Pb-17 at% Li thermal convection loop operating at a maximum temperature of 500 * C.
are shown, they are listed in probable order of abundance). Although there were some compositional differences between the two exposure times, a predominance of Cr-rich deposits were observed at lower temperatures for both specimen sets. Type A deposits were found on the 4500 h specimens in the C6 and C7 positions (see table 1), but such products were completely absent from the surfaces of the corresponding 10,000 h specimens (even when such were analyzed
460 440 420 395 370
(*c) A A A, C, D (A, C, D) ¢ D, C, B
wt~.
normal to the surface rather than in cross section [10]). Another difference between the two sets of results was the point at which nickel-enriched deposits (type B)
Pb-17Li
|
100
|
ll**..~t
Fe
Cr
/'%.. m=
I------
I
~'e= =
art
O ",if,
Ni Start
~,
%_
"L.
A A B, D D, B D, B
A: Fe-(7-10)Cr-(2-3)Ni B: Fe-(32-50)Ni-(11-25)Cr C: Fe-(33-35)Cr-2Ni D: Fe-(50-70)Cr--2Ni c Approximately equal abundance.
80
=
460 445 430 415 400
a Approximate. b Compositions determined by energy dispersive X-ray analysis. Multiple average compositions shown when distinct types of deposits were observed. Listed in decreasing order of apparent abundance when multiple compositions are reported except where noted otherwise. Concentrations are in
Pb-17Li 1O0
Composition b
"
60
40
.
20 ............... ~ " 0 360
r
Ni 410 460 Temperature(degC) ,~l"
--' 0 " ~
"'""
I
0' 360
410
460
Temperature (deg C)
Fig. 7. Elemental concentration versus position in the cooled vertical leg for type 316 stainless steel specimens in a Pb-17 at% Li thermal convection loop operating between 500 and 370 to 400 ° C for 4488 h. Concentrations determined by energy dispersive X-ray analysis; (a) 4488 h, (b) 10 008 (horizontal lines represent starting concentrations of chromium and nickel).
340
P.F. Tortorelli / Deposition behavior of ferrous alloys
Pb-17Li 100
!
!
!
I
Pb-17Li !
100
|
i
&
80
80
!
|
AL
.L
Fe
60 I ]
O
:1
O
i
q~
.......... ~ --
40
Cr Sta~ Ni Start
I
O ~J
20
0 380
Cr
20
D::I./. . . . . . \ " ~ . . . . . . . . . . . . . . . . . . . . . . . . . . . ~ . . . . ~. --. •- . / ' , a . l n . . . . - ."'"W,..-.n,.~w,,.,~|
===========tm====a==========...===l==l=ll=l===l====== m
Cr . _ . _ ~ o , - - # l . m o l l ~ R . . o t ~ . ~ . .
'wr
I
I
I
I
I
400
420
440
460
480
500
520
0 400
i 420
i 440
i 460
Ni
480
T e m p e r a t u r e (deg C) Temperature (deg C) Fig. 8. Elemental concentration versus position in the heated vertical leg for type 316 stainless steel specimens in a Pb-17 at% Li thermal convection loop operating between 500 and 370 to 400 °C for 10008 h. Concentrations determined by energy dispersive X-ray analysis; (a) 4488 h; (b) 10008 h (horizontal lines represent starting concentrations of chromium and nickel).
appeared: such deposits were observed farther up the cold leg (at a higher temperature) in the longer term loop experiment. This is also evident in fig. 7, where surface-averaged concentrations of Fe, Cr, and Ni are plotted versus cold leg loop position. With the caveat that these average values could be somewhat misleading because more than one type of deposit was found on some of the surfaces, these plots clearly illustrate the general trend of increasing C r / F e with decreasing temperature and the nickel enrichment at intermediate cold leg temperatures. There was no apparent clustering of deposits of like compositions in the cases where there was more than one deposit type. Different deposits existed in close proximity to each other. In the case of the 4500 h C7 specimen, there seemed to be a tendency for the Fe-9Cr (type A) deposits to be situated above the C and D (high Cr) types, which were nearer the original liquid/solid interface. For the C8 specimen exposed for 4500 h, there were some indications that the deposits furthest from the substrate tended to be Fe-Ni, while the ones next to the starting coupon surface were again relatively high in Cr. A similar tendency for underlying deposits to be more concentrated in chromium was also found for the C6 and C7 specimens of the 10,000 h set. The compositions of selected specimen surfaces along part of the heated loop section were also determined by EDX analysis. Surface-averaged compositions for these specimens are shown in fig. 8. The deposits formed on
the specimen exposed for 4500 h at the bottom of the heated vertical ("hot") leg (net weight gain) were A type: their average composition was F e - 8 C r - 2 N i - l S i . As reported previously [3,10], specimens at the maximum temperature position (500°C) suffered preferential dissolution of nickel and chromium. In fact, as shown in fig. 8, many of the hot leg specimens that experienced weight losses were depleted in these dements relative to the starting concentrations of the steel. However, the plot for the 4500 h set (fig. 8a) showed a maximum in surface nickel concentration at an intermediate temperature, while the specimens between 460 and 4200 ° C exhibited surface nodules of pure Mo.
4. Discussion
Under equilibrium, steady-state conditions, the net flux, J/, of element i from or to a specimen exposed to molten Pb-17Li at loop position r can be expressed as J~(r) = k ; ( r ) [ C i ° ( r ) - C , ( r ) ] ,
(1)
where k~ is the effective rate constant for dissolution ( z = s ) or deposition ( z = p ) of element i, C° is the solubility of this element in the Pb-17Li, and Ci is the local concentration of element i in the liquid metal. (C° and k~ are dependent on temperature, which is a function of r.) Normally, in the higher temperature region of the loop c O > Ci and dissolution occurs ( J / > 0 ,
P.F. Tortorelli / Deposition behavior of ferrous alloys
weight losses are measured), while in the colder part of the circuit C° < Ci and weight is gained from deposition on the solid surfaces. Such solubility-driven transport can be monitored by measurements of weight changes (E f J / d t ) of coupons arrayed around the loop, which typically produce mass transfer profiles such as those shown in fig. 2. While eq. (1) yields a qualitative understanding of corrosion and mass transfer in this and other liquid metal systems, it cannot, in many cases, explain the quantitative aspects of deposition caused by additional kinetic and thermodynamic factors, such as interactions among solutes and reactions with impurities in the liquid metal or solid. As shown below, this solubility-based approach does not predict many of the features of the present mass transfer profiles. Figure 2 shows that while substantial mass was lost from the hot zone specimens during each loop experiment, much less material per unit area was deposited on the cold leg specimens. Furthermore, this large imbalance between the measured weight losses (Am s) and gains (Am p) increased as a function of exposure time. (The total coupons weight gains and losses for the three exposure periods are listed in table 2.) This discrepancy between A m s and AmP could be explained if deposition in the TCL was preferentially occurring on loop walls or in parts of the loop net containing specimens. However, it is interesting to note that in a recent mass transfer study of austenitic steel in Pb-17Li [11], in which the total amount of deposited mass was determined by destructive examination of loops and physical removal of deposition products from all surfaces, the total dissolved mass was still significantly greater (by a factor of 1.5-2) than the total mass gain from deposition. Such findings tend to indicate that the present observation of an imbalance between Am s and Am p may not be just due to the failure of cold leg coupons to accurately reflect the appropriate amount of deposited material. Instead, it seems likely that there might be particles of solutes in the Pb-17Li. In support of this argument, Borgstedt [12] has reported concentrations of Fe, Cr, and Ni in Pb-17Li above their respective solubility limits. Suspended particles can continue to remain in the liquid metal until they agglomerate to a size where plugging may occur. Therefore, indications of a flow restriction (decreasing AT with time) in the type 316 stainless steel Pb-17Li TCL of the present study do not necessarily indicate that solubility-driven deposition was occurring outside of the area covered by the coupons. The formation of particles in the liquid metal may well be related to an inefficient a n d / o r sluggish deposi-
341
tion process(es). The growth of deposits requires registry with the underlying surface and if such is not possible, or is somehow disrupted during the initial stages of deposition, supersaturation a n d / o r particle formation can occur and be maintained [13]. Previously reported observations of small deposits in definite patterns on some of the 10,000 h surfaces [10] may indicate that deposition on the coupons was retarded in this way. While both supersaturation and particle formation would lead to reduced deposition fluxes [based on eq. (1)], only the former condition would also inhibit dissolution unless the particles dissolved in the hot zone. Therefore, on the basis of the large imbalance in weight gains and losses and the relatively large dissolution area, it can be concluded that the liquid metal was not just greatly supersaturated in the present case. Poor deposit adhesion can also contribute to the observed discrepancy between Am s and Am p. Detachment of deposition products can occur during exposure to the liquid metal (a source of particulate matter), during cooling after specimen removal from the loop, a n d / o r during the specimen cleaning process [3]. Observations such as that of the "delamination" of the deposit layer shown in fig. 6 lend some validity to a contribution of deposit detachment to lower weight gains. Furthermore, the significant amount of deposit-free surface areas on the lower t0mperature coupons of the 10,000 h specimen set is indicative of poor adhesion - a phenomenon that should be more prevalent at longer times when deposited layers become relatively thick. The above consideration of the overall weight change data and general surface morphologies indicate three possible contributions to the discrepancy between the amounts of mass dissolved and deposited: heterogeneous, localized deposition in loop areas not covered by coupons, homogeneous particle nucleation in the molten Pb-17Li, or physical removal of deposits from cold leg coupons. Allowing for all these possibilities, a mass balance for the loop would yield Am s = Am p + Am r + Am d + Am r + Am n,
(2)
where Am s and Am p are as defined above, Am f is the mass of particles that formed and stayed suspended in the Pb-17Li due to solute released from coupons, Am d is that of detached deposits suspended in the liquid metal, Am r is the contributiuon of deposits removed during cooling or specimen cleaning, and Am a is the mass associated with nonproportional deposition in areas of the loop not monitored by coupons. In order to gain a better understanding as to the relative importance of these processes in controlling mass transfer and
342
P.F. Tortorelli / Deposition behavior of ferrous alloys
deposition, the cold leg weight change profiles and surface chemistry data is analyzed in some detail below. As can be seen in figs. 3 and 4, the weight gain profile for the 10,000 h specimen set is distinctly different from that after 4500 h. In the dissolution regime, C°-Ci > 0 , and, for a given time, Ji and measured weight losses should normally increase with increasing temperature, as can be noted in fig. 2 for the three exposure periods. When C°-Ci < 0, deposition should occur. In many of these cases, weight gains increase as the temperature decreases and a weight change profile like that shown in fig. 3 for the 4500 h specimen set is measured. However, if the two functions that determine the magnitude of J,( I Ci° - C, I and k r ) vary oppositely with respect to temperature, it is also possible that a maximum in the weight gain versus loop temperature curve, such as shown in fig. 3 for the 10,000 h set, would result. In view of the 4500 h data, this possibility is considered highly unlikely - the functional dependence of I C ° - Cil or k/p on temperature should not be different for the two loop experiments. Therefore, it appears that the 10,000 h data cannot be explained on the basis of such competition between factors that influence J~, which was one of several possibilities suggested when only the 10,000 h data was considered [10]. If the physical removal of deposits from cold leg surfaces occurred, it could also explain the differences between the 4500 and 10,000 h deposition data. The shape of the cold leg profile for the 10,000 h coupons (see figs. 3 and 4) could have resulted from a loss of previously deposited material that reached a critical thickness for detachment. The observation that the cold leg coupon surfaces in the two extended loop experiments showed similarities in surface composition trends is consistent with a process of initially equivalent deposition profiles: in general, as shown in table 1 and fig. 7, iron was the principal deposited element at higher temperatures of the cold legs, and chromium was dominant at the coldest loop positions. Some of differences in the composition data between the two experiments - f o r example, the appearance of F e - N i nodules at a higher temperature for the 10,000 h set - could possibly be explained by detachment of such deposits at lower temperatures, but the observation of type A (Fe-low Cr) deposits on lower temperature coupons in the 4500 h experiment is not consistent with such an argument. On the other hand, if suspended particles contributed to the mass transfer process, it is entirely possible to have deposition of particles rich in Ni anywhere in the loop [14], particularly at longer times when their density and size in the liquid would be much greater. Further evidence for such deposition was revealed by the hot leg
surface composition data for the 4500 h coupons (see fig. 8a): a maximum in nickel concentration was also observed despite an overall decrease in weight. If these increases in surface nickel concentrations are due to the influence of Ni-containing particles in the Pb-Li, the deposition rate is not governed by solubility considerations, but rather by hydrodynamic factors and sticking coefficients. The EDX data appear to be consistent with the role of particulate matter in the transport of Ni around the loop and thus affirm a contribution of such particles to the discrepancy between measured weight gains and losses. As noted in the Experimental Procedures section, the same TCL was used for the exposures of the three type 316 stainless steel specimen sets to Pb-17Li and, previously, had circulated lithium for over 10,000 h. Therefore, a large fraction of the loop's heated section, originally type 316 stainless steel, was already significantly depleted in nickel when fresh specimens were inserted. The presence of a low-nickel loop surface relative to the composition of the coupons created a possible driving force for transport of dissolved nickel to the loop wall of the hot leg. Such a process could increase the contribution of Arn n [eq. (2)] to the observed imbalance in measured weight losses and gains. However, this situation never seemed to have an effect on dissolution rates in lithium or Pb-17Li [3,15]. Furthermore, as shown in fig. 8, nickel was deposited on some hot leg coupons after 4500 h and also was found on cold leg coupons after 10,000 h (fig. 7, table 1). Thus, it does not appear that the concentration gradient between hot specimens and loop wall had a dominant influence on mass transport. A more important factor may be that, as successive specimens sets were exposed to the Pb-17Li, deposits were already present on the loop wall in the colder Part of the TCL. Therefore, deposit nucleation on this surface could have been favored over that on the fresh coupons during exposure of a later specimen set [10]. If this was the case, a nonproportional partitioning of deposits to the loop wall (Am n) could contribute to the mass imbalance and, as observed in the present experiments, this difference would increase with increasing time as deposition in the cold zone increased with successive exposures. While nickel transfer around the loop appeared to be "irregular", the surface-averaged cold leg composition data (fig. 7) revealed that the chromium concentration increased monotonically with decreasing temperature. In both the 4500 h and 10,000 h experiments, the chromium concentration profiles appeared to show behavior expected for a solubility-driven deposition process since, at a given time, the amount of deposited
P.F. Tortorelli / Deposition behavior of ferrous alloys Table 2 Total measured weight changes for specimens exposed in Pb17Li thermal convection loops Measured wt loss (mg)
Measured wt gain (rag)
316SS, 700 h 0 42 316SS, 4488 h 1051 316SS, 10008 h 2673
1 65 74
chromium, Am~r, increased with decreasing CC0r as predicted by eq. (1):
Am~r= fJc,dt= fk~,[IC~,-CCrl]dr,
(3)
assuming Ccr is constant around the loop. In view of the monotonic behavior of the cold leg chromium concentration with temperature and the preferential dissolution of chromium in the hot leg [10], two mass transfer balance points can be defined as being those loop positions where Jcr = 0. These points delimit the area of chromium dissolution relative to that for its deposition. The balance points can be determined from the data shown in figs. 7 and 8 by noting the loop position (temperature) at which the surface composition curves intersect the horizontal line representing the starting surface concentration of Cr in the alloy [16]. For the 4500 h experiment, they are about 415 and 425 ° C for the hot and cold legs, respectively. Under the assumption that the elemental fluxes (J~) are independent of each other (that is, the activity of chromium at the interface is not strongly affected by changes in the relative concentrations of the other elements), Jc,(Tb) = 0 =
k~r(Tb)[C_~r(Tb)CCr] -
(4)
and, therefore,
CCr = C~r(Tb),
(5)
where Tb is the loop temperature at the balance point in either the hot (Th) or the cold leg (T~). Because solubility is normally a single-valued function of temperature, eq. (5) indicates that if the balance point temperatures are unequal, Ccr cannot be constant around the loop. However, from the 4500 h data, T h -- T~ within the accuracies of the surface composition and temperature measurements. These results are therefore consistent with a constant Ccr and, as mentioned immediately above, with solubility-driven chromium de-
343
position as described by eq. (1). Furthermore, referring to fig. 7, it can be noted that T~ is approximately the same (within 10 o C) for the 4500 and 10,000 h profiles. In view of eq. (5), this finding implies that CCr has reached an equilibrium, steady-state value, which is characteristic of solubility-driven transport as described by eq. (1). (However, a constant Ccr can also be possible with particle transport.) Based on the above considerations, it can be concluded that there is an important homogeneous particle nucleation contribution to overall mass transfer of type 316 stainless steel in Pb-17Li thermal convection loops. The influence of particle formation in nonisothermal liquid metal systems has been recognized for many years [14], and as already noted, a role of particulate matter in determining mass transport in Pb-17Li has been suggested or implied by the results of others [11,12]. While the present results indicate that the particles contain substantial amounts of nickel, it is not necess~-y that all of them are nickel-rich or F e - N i . Some may contain chromium in significant concentrations, but, because such particles neither redissolved (as evidenced by continual chromium depletion for the hot zone specimens and Th ~-T~, see above) nor attached themselves to loop coupons (no irregularities in the chromium profiles), they do not influence the measured quantities except for their contribution to the imbalance in overall weight losses and gains. On the other hand, the nonmonotonic mass transfer profiles for Ni (figs. 7 and 8) may mask some solubility-driven deposition of this element. However, the relatively low concentration of nickel found on many of the cold leg surfaces and the observations that nickel-rich deposits tended to deposit over chromium-enriched ones (that is, chromium came out of solution earlier than nickel) indicate that this type of transport for nickel was quite limited. The molybdenum nodules on some of the 4500 h hot leg coupons may also be an example of particles carried by the liquid metal. Alternatively, because of molybdenum's probable low solubility in Pb-17Li (it has low solubilities in lithium and lead), the nodules may be dissolution-resistant areas that developed into protuberances due to shielding of underlying material as the surface recedes. Their absence from the surface at the hottest loop position and from any 10,000 h coupons may be explained by undercutting and release of these nodules when surface recession exceeded a certain value. Referring to eq. (2), the above analyses of the cold leg coupon data for the 4500 and 10,000 h specimen sets cannot definitely show that any of the possible contributions to the mass imbalance (panicles, detachment, nonproportional deposition) are negligible. Despite the
344
P.F. Tortorelli / Deposition behavior of ferrous alloys
complication of a possible concentration dricing force for preferential deposition on the loop wall in the extended exposure experiments, the present data and the results of others [11,12] strongly support a role of particle transport in this Pb-17Li system. This finding is particularly compelling in view of the work of Flament et al. [11] in which Am n-- 0, and a definite mass imbalance was still measured. There was also direct evidence of some particle detachment (for example, fig. 6) either during exposure or after specimen removal. As mentioned above, in the former case, such release of deposits can then be treated as an additional source term for particles in the liquid metal. Previous work [10] concluded that, in contrast to the type 316 stainless system, solubility-driven deposition was the most important factor for mass transfer of a F e - 1 2 C r - I M o V W steel exposed to Pb-17Li in a ferritic steel TCL under similar conditions. Figure 4 shows the differences in cold leg mass gain profiles between these two steel systems: increasing weight gains with decreasing temperature were measured in the F e - 1 2 C r 1MoVW steel experiment (10,076 h). There was some discrepancy between measured weight gains and losses, but it was much smaller than what was observed for type 316 stainless steel. Cold leg deposits in the ferritic steel loop were type A (Fe-8Cr). This mass transfer behavior of the F e - 1 2 C r - I M o V W steel is similar to much of what was observed in the 4500 h austenitic steel experiment with respect to chromium transport (compare the cold leg mass change profiles shown in fig. 4). In this way, the results for the ferritic system support many of the present findings. Furthermore, the observation of a mass imbalance in the F e - C r experiment (also reported by Sannier et al. [11] for a ferritic s t e e l / P b 17Li loop system) may indicate that, as discussed above, particles involving chromium may form, but are not manifested in a way that is amenable to the type of measurements and analyses of the previous and present studies. If this is the case, mass transfer prediction for ferritic steels in Pb-17Li may not be as straightforward as it would normally seem when a solubility-controlled process is dominant. (Indeed, for both steels, if particles could "float" in Pb-17Li, mass transport becomes increasingiy complex [17].) However, the present results do show that, even apart from the complications of preferential dissolution and nonuniform surface recession [6,10] (not observed for the ferritic steels), predictive capability for modeling, and control of, mass transport in lead-lithium will be more difficult for austenitic ( F e - N i - C r ) alloys than for F e - C r steels.
5. Summary Mass transfer data for type 316 stainless steel in thermally convective Pb-17Li (maximum temperature of 500 ° C) revealed much larger weight losses than weight gains and unusual weight cha~ge and nickel deposition profiles. Analysis of weight change and surface composition and morphological data led to the following conclusions: Particles played a role in overall transport. The irregular movement of nickel around the loop was associated with particle formation and attachment to solid surfaces. While some chromium transport may have involved particles, the surface composition data showed that deposition of chromium was primarily controlled by a solubility driving force. This finding agreed with results from a previous study of a F e - C r steel in nonisothermal Pb-17Li. A balance point analysis indicated that a steady-state chromium concentration in the Pb-17Li was achieved. Physical detachment of deposits (during or after exposure) contributed to the significant differences in measurements of dissolved and deposited masses. The analysis was complicated by the possible influence of preferential deposition on the thermal convection loop wall due to differences in composition between it and the specimens. Due to the influence of particles and lack of good deposit adhesion, as well as the effects of preferential dissolution, mass transfer prediction for austenitic ( F e C r - N i ) steels in Pb-17Li should be more complicated than that for ferritic ( F e - C r ) steels.
Acknowledgement Research sponsored by the Office of Fusion Energy, U.S. Department of Energy under contract DE-AC0584OR21400 with Martin Marietta Energy Systems, Inc.
References [1] P. Fauvet, J. Sannier, and G. Santarini, in: Proc. 13th Symp. on Fusion Technology, Vol. 2, Commission of the European Communities (Pergamon Press, 1984), pp. 1003-1009. [2] V. Coen et al., in: Proc. 3rd Int'l. Conf. on Liquid Metal Engineering and Technology, Vol. 1 (The British Nuclear Energy Society, 1984), pp. 347-354.
P.F. Tortorelli / Deposition behavior of ferrous alloys [3] P.F. Tortorelli and J.H. DeVan, J. Nucl. Mater. 141-143 (1986) 592-598. [4] O.K. Chopra and D.L. Smith, J. Nucl. Mater. 141-143 (1986) 566-570. [5] G.M. Gryaznov et al., Sov. At. Energy 59 (1985) 918-921. [6] H. Tas et al., J. Nucl. Mater. 141-143 (1986) 571-578. 7 H.U. Borgstedt et al., J. Nucl. Mater. 155-157 (1988) 728-731. [8] Z. Peric, G. Drechsler, G. Frees, and H.U. Borgstedt, in: Proc. Fourth Int'l. Conf. in Liq. Metal Engineering and Technology, Vol. 3 (Soci6t6 Fran$aise d'Energie Nucl6aire, 1988), pp. 522-1 - 522-7. [9] M. Broc, P. Fauvet, T. Flament, A. Terlain, and J. Sannier, in: Proc. Fourth Int'l Conf. on Liq. Metal Engineering and Technology, Vol. 3 (Soci~t6 Fran$aise d'Energie Nucl~aire, 1988), pp. 527-1-527-10. [10] P.F. Tortorelli, in: Proc. Fourth Int'l Conf. on Liq. Metal Engineering and Technology, Vol. 3 (Soci6t~ Fran~aise d'Energie Nucl6aire, 1988), pp. 528-1 - 528-10.
345
[111 J. Sannier, M. Broc, T. Flament and A. Terlain, Fusion Engrg. Des. 14 (1991) 299-307, in this issue. [121 H.U. Borgstedt, KfK, private communication (Sept. 1989) [13] O.F. Kammerer et al., Trans. AIME 212 (1958) 20-25. [14] L.F. Epstein, Static and dynamic corrosion and mass transfer in liquid metal systems, Liquid Metals Technology, Chem. Eng. Prog. Syrup. Ser. 20, 53 (1957) 67-81. [15] P.F. Tortorelli and J.D. DeVan, in: Proc. 3rd Int'l Conf. on Liquid Metal Engineering and Technology, Vol. 3 (The British Nuclear Energy Society, 1985), pp. 81-88. [16] P.F. Tortorelli, J.H. DeVan, and J.E. Selle, in: Proc. Second Int'l Conf. on Liquid Metal Technology in Energy Production, U.S. Dept. of Energy report, CONF-800401P2 (1980). [17] J.H. DeVan, Oak Ridge National Laboratory, private communication to P.F. Tortorelli (January 1990).