.. r | OF I ORNL P 2642 2 . . . . . V- . . S . : . . . • . EEEEEEEE Bo ... Lt MICROCOPY RESOLUTION TEST CHART NATIONAL BUREAU OF STANDARDS – 1963 ORNKiPa ni en many m ..?. K.C. $.00; MN 588 NOV 2 9 1966 INVESTIGATION OF MATERIAL FAJLURES IN DCX-1 LINER* W. J. Leonard Oak Ridge National Laboratory Cak Ridge, Tennessee RELEASED FOR ANNOUNCEMFNT IN NUCLLAR SCIENCE ABSTRACTS This paper covers the investigations of recent material failures occurring in the DCX-l liners during operation. Figure 1 is a schematic drawing illustrating the DCX-l vacuum system. The outermost tank is constructed of austenitic stainless steel with many access openings sealed and constructed in a conventional manner. The first liner consists of an OFHC copper vessel (1/8 in. wall), which is of slightly smaller volume than the stainless steel containment tank. The interior liner, also constructed of copper is of much smaller volume than the outer liner. The outer surface of the stainless steel vacuum tank and each inner liner are jacketed with 1/2 inch diameter, 065 in. wall OFHC copper tubing. Figure 2 shows one version of an inner liner used in an operational run in DCX-1. At present three distinct vacuum systems exist in the DCX-1 machine: (1) The volume between the inner surface of the containment vessel and outer copper liner (designated "barrier liner"), (2) the volume between inner surface of the barrier liner and the outer surface of the inner liner, and (3) the volume enclosed within the inner liner. The inner liner is often removed and varied in configuration, depending on experimental requirements. The copper tubing is formed so as to contact the maximum liner surface practicable, as shown in Figure 2. Usually two parallel copper tubing circuits are used on a liner to achieve maximum heat transfer between the liner and jacketed tubing. The tubing is braze-welded to the liner using an inert gas arc process. The filler metal used is a 60% silver, 30% copper, 10% tin alloy, which is a brazing alloy having solidus-liquidus temperature range of approxi- mately 1100°- 1330°F. The copper tubing terminals are brazed with the above Хе. *Research sponsored by the U. 5. Atomic Energy Commission under contract with : the Union Carbide Corporation. ......... nr. ........... .. -2. . A filler metal into an austenitic stainless steel header system, using a socket-joint design and an oxyacetylene torch as the heat source. The liner provides a differentially pumped plasma region wherein pressures of the order of 1 x 10-? torr are attained. After evaluation with oil difíusion pumps in the back-up region, the liner is taken through a conditioning cycle involving a steam bakeout at 450°C followed by cooling to liquid nitrogen temperature. While at the low temperature titanium is evaporated in the liner, then the liner is cycled a few degrees by turning the liquid nitrogen off and on. Over the years of operation, the copper tubing kas intermittently developed pin hole defects in the tubing walls, necessitating shutdown and iubing replacement. The causes of these leaks were the subject of exhaustive investigations. ,592,4 Material quality, composition, joint design, improper brazing and welding techniques, and mecbanical damage during fabrication were some of the facets of this problem initially investigated. Scule past failures were found to be attributable to some of these factors. Latter investigations have included ultrasonic inspection of tubing, tube burst testing at pressures and times much greater than service requirements, and steam analysis. Tests conducted in test stands simulating the DCX service condition did not result in a logical tie-in with observed operational failures.' FAILURE ANALYSIS Failures have occurred most frequently at the torch brazed socket joint of the inner liner copper tubing to the stainless steel header tube on the steam inlet side. An identical joint exists on the steam éxit side. :Defects have occurred in two distinct locations at the steam inlet joint; in the braze metal filling the crevice of the socket joint, and in the copper tube wall within an inch of the joint where the surface is "wet" with braze metal. Less frequently, pin hole defects have developed in the copper tubing as much as five feet down-stream from the joint. These latter failures usually occur after several failures inay have occurred and been repaired at the brazed inlet joint. Thus, failures in tubing down-stream from the steam entrance joint appear to take much longer to develop than tubing at or in close proximity to the brazed joint. Failures at the exit joint have occurred but quite infrequently. l'he steam used is plant steam which is passed through a superheater at 600°C and 90 psig in a stainless steel piping system. Properties of the steam Es are given in Appendix A. The temperature, of the steam entering and leaving the copper tubing is, of course, dependent on throttling. Maximum temperatures of 460°c may be experienced at the stainless steel-to-copper entrance joint on the inner liner and 340°C at the exit joint. The inlet and outlet tempera- ture, particularly the latter, vary with various inner liner design, but 450°C has been the average temperature at the tubing inlet. The steam pressure at the liner tubing inlet joint averages 75 psig; the outlet averages 15 psig. Thermodynamic calculations for the system under maximum temperature conditions indicate an inlet steam velocity of 513 fps, outlet velocity of 1386 fps, and average velocity of 950 fps through the tubing. The tubing on the barrier liner is fed by a parallel steam line with inlet temperature often as much as 80°c lower than that of the inner liner. . .. The stress experienced in the copper tubing walls under the flow and . thermodynamic state in the system are discussed in detail in Appendix B. It will suffice to say that the calculated stress in the copper tube walls at the most severe operational conditions is well below a level which would induce plastic flow or rupture in the tubing. No cases of tubing rupture have occurred in service. . . ... . .. .. ... .. . . .. TAL ". . . - . . . "," . . ... .. . . . . . .. . . . . . . In operation a liner bakeout may last 25 to 45 hours. On completion. of an experiment this liner may be removed and replaced by a different liner for the next experiment. Usually, however, the same liner remains in for seversal bakeout cycles. Thus, some liners may experience several hundred hours of service at elevated temperatures. In other cases, the liner tubing may develop a defect within 25 hours during the first bakeout, usually at the inlet joint. In this case, the copper tubing ends are withdrawn from the stainless steel socket (by heating), the tube cut just down-stream from the defect and then re-brazed in the socket. There are several reasons why 'a re-brazed joint can result in a lowered corrosion resistance in the OFHC copper tubing. First, even though tiae copper base metal is protected from air oxidation by a flux and reducing torch atmos- phere, field torch brazing always poses the problem of oxidation of the copper base metal. The greater number of times these copper tube ends go through a brazing process, the greater this possibility. Therefore, if tube jeaks are repaired and leaks re-develop shortly, the liner is removed from the machine, the tubing jacket rerdoved, and a new tubing jacket is placed on the liner. There are metallurgical considerations which will be discussed in more detail later, which make re-brazing unattractive from the viewpoint of ccarosion resistance. If during operation a liner fails, it is removed and pressurized with nitrogen or air, using a high viscosity liquid on the tube for defect detection. Often 'a network of very fine leaks is detected in a line type distribution within a small, area. At other times only one distinct pin hole leak is found. Regions of tubing containing a leak are sectioned and carefully ground and polished for metallographic examination of the defect area. NATURE OF DEFECTS A brief discussion of the metallurgice). con iderations involved 18 necessary before insight can be gained into the corrosion kinetics involved.. A copper-oxygen phase diagram will show that the maximum percentage of oxygen that may be held in solution at 1065°C (the solidification temperature of copper) is only 0.0035 (35 ppm 0). On cooling to room temperature the maxi- mum solubility of oxygen is only 4-5 ppm. Thus, copper of an oxygen content greater than this will consist of a pure copper phase (alpha), plus a mu-Cuo phase designated beta; which latter phase will be upproximately 0.4 percent oxygen by weight, the eutectic composition. Thus, the microstructure of copper normally appears as a copper grain, with a grain boundary en ve.Lope consisting of eutectic composition. However, OFHC copper is an exception, appearing as a single phase structure, with discernable grain boundarjes due in general to high angle orientation differences between distinct copper . grains. OFHC copper generally contains less than 2 ppm oxygen, whereas electrolytic tough pitch and elemental deoxidized coppers may contain anywhere from 200 to 1000 ppm oxygen. The conductivity (both electrical and the:mal) is lowered by the presence of the eutectic phase. Likewise, the corrosion resistance of copper in many corrosive media is similarly lowered. Hydroger or hydrogen ions vigorously attack the eutectic phase at elevated temperatures. In a typical defect area of copper tubing down-stream from the joint, the first event noted in a microscopic examination is that the eutectic phase begins to appear in grain boundaries. The defect path, when found, will appear to follow grain boundaries which have suffered rupturing on a micro scale. In more gross defects, macro-ruptures may often appear in the grain boundaries with gas evolution leaving fine bubble voids.. The leak path is never a single L 11.12.17 Uhren ' . void proceeding from the inner to outer wall surface; rather it is a devicus network of grain boundary microfissures meandering from the inner surface and eventually finding access to the citer surface, commonly called intergranular corrosion. The defects observed at the brazed joint are quite similar in nature except more grossly occurring. Figure 3 is a photomicrograph in an area which contains the leak path. The photomicrograph is of metal in the unetched condition. The voids formed from internal gas pressure are clearly visible. Figure 4 is a photomicrograph of the sane area etched to reveal the microscopic constituents. The inter- granular nature of the defect is obvious. These two figures also illustrate the usefulness of etched and wetched techniques to differentiate actual voids from coarse eutectic structure, Figure 5 is a composite photograph of a defect in the copper tubing at the socket joint. The defect evidenced itself as a distinct pinhole leak op the tube svirface approximately 1/4 inch from the entrance into the socket. The first photograph labelled o denth was taken at a position where the defect was first discernable on grinding into the defect area. The 1.2 O'clock position shows a void at the surface, extending a very slight depth in a plane perpen- dicular to the tube axis. In following this defect in an axial direction toward the joint, it was seen that the defect is open to the outer surface for a distance of 15 mils and moves in a spiral direction into the tube wall. The defect encounters the inner tube surfece at approximately 20 mils and persisted to a depth of 62 mils, moving around to a circumferential position approximately 90 degrees from the location of the outer surface emergence of the defect. On grinding further along the axial length, the defect appeared to spiral back toward the 12 O'clock position, and although not shown in this figure, appears to emerge to the surface at approximately 125 mil depth. The tube surface was covered with braze filler metal at this position, so no surface leak was : : . .. ... . .. ... discemable. Also observed at 72 mils distance was the appearance cf defects at 3 o'clock position at a midwall location. These defects remained at this rem approximate position moving slightly toward the inner surface at 125 mil depth. The significance of this latter observation is that several concurrent defects are developing from the inner surface to the outer wall surface, and they may in some cases emerge more or less simultaneously and indicate numerous tube surface leaks on testing; or one defect path may progress more rapidly than the several others and indicate only one surface leak. DISCUSSION The basic question is: What is the sequential mechanisms' that permit OFHC copper corrosion by the steam used in DCX-1? The first requirement is for oxidation of the OFHC copper to permit formation of an intergranular copper-copper oxide eutectic. This can be explained at the joint where copper temperatures may reach 1500°F during a field brazing operation. By proper brazing techniques, this can be eliminated as a cause. However, the tube down stream from joints never exceeds 460°c, at which temperature steam would not be expected to create, from thermodynamic-diffusion data, an oxide diffusion problem. Tubing removed from liners after several hundred hours of service do exhibit a thin black copper oxide film on the inner wall. One can only conclude that, at the steam velocity, temperature, steam impurities, wall temperature gradient, and wall strain, a corrosion-erosion condition exists that permits oxygen diffusion into the copper wall. The second reaction required is a reaction between the intergranular eutectic formed and a reducing gas. At the temperatures under consideration hydrogen gas will readily react with eutectic to form a spongy copper and H,O gas. Iron and some other elements will react with steam at elevated temperature to produce some amount of free hydrogen. Thus, hydrogen or hydrogen ions could THE 71 m . , . . , 4. # CILE - .. . .- . .... .... ... . .... . . . . . . ... .... . ... . .... . . . .. . - - - - - -- S .Y"1. TIT T ' 'n ... J ' . . . . . - - - - oder der - .. . ., -8 exist in the steam delivery system. Figures 3 and 4 ceguly show that an . extremely high pressure reaction occurred at the graiu bundaries, as quasi- spherical voids can be observed at this location. Gas voids have been observed in specimens in which the outer wall was still intact, which would eliminate high velocity steam erosion as the agent for creating these voids. The observed microscopic reactions in DCX tubing is identical in nature to those observed in controlled hydrogen embriittlement experiments of ETP or other higher cxygen content copper. This f'act leads vs to believe that this is the reaction we are observing. In the barrier liner tubing the steam inlet temperature seldom exceeds 425°C. Thus, steam velocity conditions are likewise not as severe. However, the hydrogen and other impurities are the same, as this is in the same steam system. Intergranular corrosion is not experienced in this tubing system. Thus, one concludes that the energetics of this system are such that OFHC copper does not unäergo intergranular oxidation. LEGAL NOTICE This report was prepared as an account or Lovorament sponsored work. Neither the United Statos, nor the Commission, nor any person acung on behalf of the Commission: A. Makes any warranty or representation, expressed or implied, with respect to the accu. racy, completeness, or usefulness of the information contained in this report, or that the use of any Information, apparatus, method, or process disclosed in this report may not infringe privately owned righato; or B. Assumes any liabilities with respect to the use of, or for damages rosulting from the use of any information, apparatus, method, or pro:ess disclosed in this report. As used in the abovo, "person acting on baball of the Commission" includes any om- ployoo or contractor of the Commission, or employee of such contractor, to the extent that such omplcyoe or contractor of the Commission, or employee of the contractor properes, dissominatos, or provides access to, any information pursuant to his employment or contract with the Commission, or his employment with such contractor. REFERENCES 1. ORNL-2802, Thermonuclear Div. Semiann. Progr. Rept, . for pericd ending Jan. 31, 1959, p. 93. ORNL CF Memo 59-1-12, R. E. Clausing, "Progress Report on the Investiga- tion of the Development of Leaks in Copper Tubing used on the DCX Liner." 3. R. E. Clausing, Private Communication. 4. ORNL-3989, Thermonuclear Div. Semiann. Progr. Rept. for period ending April 30, 1966, pp. 136-139. - - - - . .. .. . . . . 1 - ... 11.. APPENDIX A PROPERTIES OF SOURCE STEAM Plant steam which is used as a soruce for the superheater is saturated steam at 250 psig. Analysis on this steam indicates the following acidity and impurity content: - pH 6.2 (average) NHz .07 ppm C11 ppm 924 1 ppm The source steam is heated by a superheater which is located on floor level directiy below the DCX-i equipment. During bukeout operation, the superheated steam is withdrawn from the top of the superheater and flows approximately 20 Teet through an insulated piping system to the stainless. steel header in the vacuum system. The isteam flow from the superheater .system is so controlled that a temperature of 600°c and pressure of 100 psig is maintained at the superheater outlet. APPENDIX B STRESS CALCULATIONS IN COPPER TUBING The following calculations on 1/2 inch diameter .065 in. wall OFHC copper tubing are based on the ASA Code for Pressure Piping, with additional modifications based on mechanical property data developed at ORNL. The basic maximum hoop stress S = PD/2t + .4 P where t = design thickness of tube wall P = maximum internal design pressure in psi D = outside diameter of tube in inches. S = Max. hoop stress developed The stress at 100 psi internal pressure for the DCX-1 copper tubing is calcu- lated to be 410 psi. The following mechanical property data on OFHC copper at 800°F (the maximum test temperature for ORNL data) are given: Stress Level for Tensile Strength Yiela Strength (.2%) Young's Modules 10-4%/hr creep_mate 13,000 (psi) 5,100 psi 15.9 x 106 125 psi These data would indicate an 0.1% elongation in OFHC copper tubing at 800°F after 1000 hours at 125 psi stress. At 420 psi our creep data would indicate a maximum temperature of 670°F for 1000 hours life (based on 0.1% elongation) or 520°F based on 10,000 hours life. One hundred hour life maximum operation tempera- ture using these seme criterion would be approximately 830°F. Tube burst tests for 1000 hours duration were conducted on 3/8 in. diameter .032 wall OFHC copper (the initial size tube used in DCx) at 100 psi, and 170 psi test pressure; which produced a fiber stress of 603 and 990 psi respectively. Test temperature was 752°F. No ruptures occurred and no change in dimensions occurred, indicating no significant plastic flow. From theoretical calculations LORNL-IM-102 dated Jan. 5, 1962, Instantaneous and Time Dependent Collapse of OFHC Copper Vessels by External Pressure, C. R. Kennedy. at these stresses (from uniaxial creep data) the 100 psi test should have experienced a creep strain of 0.6% and the 170 psi test a creep strain of 1%. Thus, in our present tubing system, in which an inlet temperature of 880°F. may be experienced for a maximum of 200 hours service, we do not expect any measurable plastic deformation. . .--.- . . ... . Hat BEAM ORNL-DWG 65-1505R ; TITANIUM EVAPORATOR E PROBE LINER REGION ROTATING BEAM TARGET PROTON EQUILIBRIUM ORBIT NEUTRAL PARTICLE DETECTORS - PROBES Figure 1 - DCX-1 Vacuum System Schematic IN " . 17 France IN 1 r. --. -... .: ir... mannen min, Figure 2 - Photograph of a DCX Inner Liner with Jacketed Tubing -... -- ---.. - - ; - -.. --. Copper Tubing (Unetched). Figure 3 - Photomicrograph Illustrating Defects in Outer Wall of OFHC ti - 0.035 INCHES - in 100X .- .-.- . - - - - - - -- - - --- - - - - - - - . . . . . . . .: . 1 -0.035 INCHES IN 100X . 14 - ... • . . -. ... ... ... . . . . - .. Figure 4 - Photomicrograph Illustrating Defects in Same Area as Figure 3 (Etched). ----...--..-............... . ..o. ... M n o'clock Pontian Aniel Depth Müll 101 JOX JOX Photomicrograph of Pinhole Defect in OFHC Copper Tube; 0.065 in. Wall-Transverse Section Figure 5 - Composit Photomicrograph Illustrating Defect Path in 1/2 in., .065 in. Wall, OFHC Copper Tubing. - . .. - -. END DATE FILMED 12/ 21 / 66 .. . . 21 . 2 PIL . .