Document zZYdo1gZ1jk5dJDxmE1aRN87

f\ V \n C It rtf ' /AoAemW c4- CcrvsHucVibr) DOW CHEMICAL U.S.A. February 12, 1988 -"PTf,JT'BiTello"r'B=3601 C. W. Brannan, B-2601 LOUISIANA DIVISION p. o. box iso PLAQUEMINE. LOUISIANA 70766-0150 604 389-8000 cc: W. T. Austin, B-3701 H. H. Bell, B-4701 J. Y. Pennington, B-2510 PASSIVATION OF TITANIUM IN CHLORINE The attached article is the culmination of a literature search by Rich Chauviere of the Chlor-Alkali Tech Center. Nothing. completely new here, but an excellent summary. Conclusions: 1. As the temperature of chlorine increases, more water is required for passivation. About 1.5% water is required around 400 degrees F. I have not seen any data at higher temperatures. 2. Flowing conditions require less water for passivation than static conditions. 3. Surface condition of the titanium has an effect - a rough or scratched surface is more likely to react than a smooth one. 4. Pressure is thought to have an effect, but this article did not address that variable. 5. You will notice that the data is somewhat erratic, i.e., not all the specimens reacted at a given set of conditions. 6. It is pointed out that once a reaction begins it is self sustain ing, regardless of water content in the chlorine, unless there is a film of water on the metal. In short, it must be recognized that the water content in chlorine required to prevent a reaction with titanium is not a constant. Utilize this article for planning safely. R\ W/'Cochran Safety/Loss Prevention/Security Enc. am AN OPERATING UNIT OF THE DOW CHEMICAL COMPANY CO 073035 CONFIDENTIAL Factors Affecting Water Content Needed to Passivate Titanium in Chlorine* By E.E.MILLAWAY and M.H. KLEINMAN* INTRODUCTION Titanium is known to react in dry chlorine but to be passive in wet chlorine. Titanium specimens ignited and burned*1 after 139 days in 97 percent chlorine, balance carbon dioxide, hydrogen, air, etc., 0.0005 percent water at normal temperature. In a 99-day test in saturated chlorine in a cell header evaporator space at 206.6 F (97 C) the average corrosion rate of titanium was 0.0003-in per year.1 The amount of water required to stop the attack by chlorine on titanium has been reported as 0.005 percent, as 0.1 percent3and 0.35 percent. The 0.35 percent was obtained by scratching a titanium surface in an atmosphere of chlorine.4 This investigation was made to develop more information on the amount of water necessary to inhibit the reaction of titanium with chlorine. Because some prior testing had shown erratic results when titanium was exposed to an atmos phere of chlorine, static exposure of titanium was not used as a primary test method. Instead, scratching the surface to expose fresh titanium to the chlorine was preferred. Static exposure tests, however, were used in an attempt to verify some of the data obtained by exposing a fresh titanium surface to chlorine. Some work was done also to find a more reliable test method than the ones used previously. One series of tests was made with relatively high purity bottled chlorine; another was made with dried, crude Hooker cell chlorine so results would be more consistent with industrial plant exposure conditions, EXPERIMENTAL PROCEDURES Static Tests Static exposure tests were made by placing titanium specimens in gaseous chlorine of various water contents at several temperatures for * prr.f'MrL'A ; the Hrttf Conference, National Aaaocianon ol Corrosion Engineer*. Milmi Beach, Florida, April J960, 1 itanium MrtaU Corp. of America, Hvfwitrfon. Nev6a. ABSTRACT Tlu*fum generally la reaertve to dry, g*aua chtorlr* a mil} amount of water in gaaeoua cMorlne will inMbic th.i reaction A (rating program *bo**4 the amount of wafr require* for inhibition varteti with txpoaure condition* fr-rr about 0.20 to 1.1 over the range et about 75 ro 175 C (77-Ji? F>. When water content in the gaaeoua chlorine la uffteirv titanium can be considered paaatve. Para were obtained moatiy by scratching ft> expoae fr**h titanium aurfacea to gaaeoua chlorine and observing reaction#. Under acme condition* a definite reaction sequence often took place, which could be observed from initiation to Ipmti&n of a specimen. Some data were obtained from exposure of un disturbed titanium specimens la gaseous chlori;<e. The water content required to passivate titanium in chlorine gas was fourd to be dependent upon temperarur*, gai rate and chlorine presaure. It was affecr*3 also by uranium surface condition, some roughened surfaces showing increased reactivity. Under static conditions tnvolvfrg chlorine of ht?ft purity, water required was greater than undrr fluw londiurnf. In dried, crude Honker ceil chlorine. Undiatufbed cita.nu n surfaces exposed st I9S C (3*2 F) fer over a year showed nc reaction when chlorine water content was 1*5 percent nr more. extended periods. The exposure tubes used shown in Figure 1. Test specimens w:.-r 1/4 x 1 1/2 x 0.035-in sheet of ASTM 26s-_.r Grade 2. Specimens were cut, stamped with r identification number, scrubbed in hot water ~ an abrasive cleaner, rinsed, pickled for a." one minute in a 10.5 HN03-1.5 HP (vol.^soluuci rinsed in distilled water, dried and wc /.vDuring assembly, specimens were place.l ir end of the glass tubes and cooled with fL`*u* helium until the opposite end was sealed. 7..?` the tubes were evacuated, a Cla-n0 miAcrr added and the stems sealed off. Quantiti:.- * Cla + HgO were such as to give just slightly atmospheric pressure at the exposure tem tures. For high water content tests the require water was added by volume measurement a 50-microliter Hamilton^ syringe, frozen *-5 dry ice and acetone, the tubes evacuated, zsi warmed slightly, and dry Cl- admitted an- tz* stems sealed off. The tubes were placed in c at the desired temperatures and inspect periodically. Non-Flow Scratch Tests For making tests of fresh titanium sun-c-* exposed to non-flowing chlorine the appnr-z-* shown schematically in Figure 2 was used. > Co., 1\ O. Do. 307, Whittm. Clllfornu. rs tV 88 DO 073036> . I until liiftri Wn.cd amount of water could be added to a in volume of dry chlorine to obtain a known `,'...r content. Details of the specimen exposure are shown schematically in Figure 3. Higher -;er3tures were obtained by wrapping the -cirr.en exposure tube with heating tape. For .. i non-flow scratch and static tests commercial , nie- chlorine of 99.5 percent purity was used, j; miens used in non-flow scratch tests were , x 3 x 0.035-in sheet of ASTM 265-52T Grade 2 a 1/8-in diameter hole drilled near one -*. They were scrubbed in hot water with an ;::is:ve cleaner, rinsed in both tap and dis- water, air dried and stored in a covered . container until used. iter a specimen was placed in the holder - tr.fc exposure tube, the tube was assembled ;,-.i evacuated to a few microns and the chlorine- mixture introduced into it. The specimen ;.= then scratched and observed for reaction. r.o reaction occurred in about 5 minutes the i;.'.-imen was scratched again. This was repeated about an hour. . Scratch Tests The apparatus shown schematically in figure 4 45 used for tests under flowing conditions. Water j5 introduced by mixing streams of nearlyiiterated and of dry, crude Hooker cell chlorine then computing the water content of the - irure, assuming saturation of the stream that passed through water. For most flowing s dried, crude Hooker cell chlorine that had teen compressed and revaporized was used. Figure I - A. Sealed tubes for low water content chlorine test ing of titanium. 0. to 0.93 percent H,0. B. Sealed tubes for high water content chlorine testing of titanium. 1.5 to 6.9 percent HjO. TO 10G7 89 Figure 3 - Specimen exposure tube of 25mm heat resistant glass tubing and specimen handling arrangement- Heating tnpe not shown. was observed for reactions for periods vt from a few seconds to 10 minutes. If no re; occurred, the surface was scratched again, process was repeated for periods from 5 m: to 4 hours until it appeared certain that a re; was not going to take place. A reactior considered to take place when any visible si change occurred. Most of the scratching was done with a shar Ti-6A1-4V alloy rod. Some scratching was with a sharpened stainless steel rod, a p< glass rod or a diamond-tipped device. vt Unless otherwise noted, the tests were at close to atmospheric pressure. The che composition of ASTM Designation B265-52T g are given in Table 1. Table 1 - Chemical Composition, Percent*11 Element Grade Number 1 23 Figure 4 - Setup used for flowing chlorine studies. Titanium, min. Nitrogen, max. 99.3 99.2 99.0 95 0.08 0.10 0.15 0. Carbon, max. 0.05 0.20 0.20 0. Specimens used in these scratching tests were 3/8 x 3 x 0.025-in sheets of ASTM 265-52T Grade 4. Preparation was the same as for other Iron, max. Tungsten, max. . Oxygen, max. 0.12 0.25 0.08 0.02 0.15 0.20 0.25 0.02 0.25 0. 0. 0. scratch' tests. ! Chromium, magnesium, aluminum, mangas nickel, copper, lead, molybdenum, vanad Specimens of unalloyed titanium were put into and other elements not specifically mentic the exposure tubes and when the desired con shall nor total over 0.25 percent and no ditions were reached the surface was scratched element not specifically mentioned shall exc to expose fresh titanium. The scratched surface 0.10 percent. 90 CORROSION - K, 00 07303a CONFTDENTrAI. RESULTS ic Tests vsults of static exposures are shown in .jv 5. Detailed results are tabulated in Table 2, reactions occurred in 17 months' static ?urt- tests when titanium was exposed to rine that contained 0.71 and 0.93 percent -r at room temperature. The chlorine to jh 0.93 percent water had been added was bally supersaturated some of the time. ;r, Fieure 5 are shown percentages of specimens groups of five that had reacted after 17 months : t/Nposure to chlorine with several levels of i:cr content and at four temperatures. It should be noted that reactions did not occur at 50 C (122 F) or in dry chlorine at 135 C (275 F) while reactions did occur at room temperature. One reason for this is that specimens were placed in their specific chlorine atmospheres before ovens were available and were thus exposed at room temperature prior to the higher tempera ture exposure. Thus specimens that had a tendency to react did so at room temperature. At tempera tures of about 195 C (3S3 F), 0.93 percent water was insufficient for passivation. Croups of five specimens each tested at 195 C with higher water contents of 1.5, 2.9, 4.3, 5.6 and 6.5 percent showed no reactions after 21 months' exposure. Some temperatures in Figure 5 are five degrees below oven settings and were so plotted because of temperature variations. Table 2 - Specimens Exposed to Chlorine in Scaled Tubes and Number of Reactions that Occurred In 17 Months Chlorine Water Content % nl! 0.12 0.14 0.24 0.40 w. 7 J 0.93 (sat.) Used 26 18 9 25 23 24 24 Reacting on Loading 3<i) 2 (2) 0 A (3) 3(4) 0 0 NUMBER O F SPECIMENS Placed at Reacting at 190-200 C 190-200 C Placed at Reacting at 130-140 C 130-140 C 5 1 5.(<6s)> 4 5 45,(Ce7)s 6 69\ 5 5 =(i) 5U D 50 0 5 50aa) 5 5 13) 5 5 23C1i *8) 50 Placed at 100-110 C 5 5 0 5 4 5 5 hLrinc 'V.ir-T t "nuru ail 0.12 ", 14 :t ".24 a! ".4;> .`h "93 (sat.) NUMBER OF SPECIMENS Reacting at 100-110 c (16) *(16) 0, *(16) *(16) *(16) *(15) Placed at Reacting at 50-55 C 50-55 C 50 50 00 50 50 50 50 Left at Room Temp. 3 5 0 1 0 4 4 Reacting at Room Temp Test Total ,(i7) jfao) 0 0 0 0 0 4 2 - 4 3 0 0 ''specimens lost when oven temperature inadvertently was raised to approximately 250 C. !iy loading is meant placing titanium specimens and chlorine gas in tubes. Reieted ipproiimately 1 minute to 30 minutes liter loading.** [*{ Reacted about 5 minute* After loading, Reacted in about I minute to 1 day alter loading. J*J'4' Reacted 5 day* alter loading. Reacted in from 1 week to 2 months. jj Reacted In 4 hours. Reacted In from l day to t month*. Reacted in from | day to 2 month*. '* Reacted n from 1 week to 2 montha J1 Reacted In from 1 to 9 flays, j1** All reacted in 2 months when container* taken from oven and examined. " All reacted in 2 montha I week. * a* Reacted in from 1 to 3 months Rented in from 2 to 5 months. Specimen* lost atx>ui 3 1/2 m the after placing tn 105-110 C even when temperature inadvertently raided to approximately 250 C. One specimen survived the 250 C in the 105-110 C oven and wit placed in 100-200 C oven whrrc it reacted approximately 1 month later. : 'Specimen reacted befween 4th and 5th month, Slo* reaction. Specimen reacted 5 1/2 months alter placing in oven. : Specimen* tested plus those that reacted on loading. ' Found reacted alter 17 montha. DO 073039 CONFIDENTTAl Figure 5 - Results of static tests of unalloyed titanium exposed to Cla gas at varying water contents and varying temper atures. Non-Flow Scratch Tests Summarized results of scratching tests made under non-flow conditions using bottled chlorine are shown in Table 3. In these tests reactions were obtained in chlorine that contained about 0,85 percent water, near saturation at 26 C (79 F). When less than 0,85 percent water was present, reactions occurred readily with sera tdtir; on all but a vacuum annealed specimen and when the chlorine pressure was 16-in of Hg or less. No reaction occurred when liquid water was added to the reactor tube to assure saturation. Table 3 - Results of Tests of Unalloyed Titanium Sheet Scratched in Bottled Chlorine Test No. % HaO Test C Test Type No. Pressure Tests Results 1 2,,. 2^) 3 4 5 6 7 7 8 8 9 10 12(3) Dry Dry Dry .27 .59 80 Abt . 85(3) Dry Dry Dry Dry Dry Dry Dry Abt. 85C3) 13(4) 14 Abt. 6S(3> Sat'd. 27 26 92 (3) 26 26 26 26 26 26 26 26 26 26 26 26 Non-Flow Atm. m m m m tt Flow(s) Atm. Atm. Atm. Atm. Atm. Atm. 14" Raised to 20" He 16" Hg To 18" Hg 20* 22* 3-4 psig Atm. 26 Flow(s) Atm. 26 Non-Flow Atm. 1 Color appeared after 23 min. of gentle scratching. Specimen ignited after 31 min. 1 No reaction when scratched. 1 When temp, raised to 92 reaction started. 1 Reaction started after 10 min. 1 Color started in 13 min. 2 Reacted after 1 and 47 min. 4 Three reacted. One did not. No reaction after scratching. 3(6) Reacted. 1 No reaction. Reacted after 12 min. 1 No reaction. 1 Reacted 14 min. after scratching. 2 Light scratching caused immediate reaction. 4 Three specimens. No reaction after considerable scratching. One specimen-reaction on un scratched titanium bolding strip. 2 No reaction. 1 No reaction after four hours of scratching. f1* Specimen was vacuum annealed. J3' Heating time was 70 minutes. J3) Near saturated. :4' A specimen with an electrolytically plus hydrochloric acid roughened surface. '5) At a low flow rate that was not measured. Pressure raised in steps of about 2* of Hg and specimen scratched at each pressure. c ' To assure saturation, liquid HaO and then CI3 were added to evacuated reactor tube. Possible supersaturatlon. DO 073040 OONFTnpMTT 1 utr. NTIAL 92 CORROSION - NACE When a fresh titanium surface is exposed to chlorine gas a very rapid reaction can result. Reactions initiated in non-flow,,ing, bottled chlorine by scratching had the following general charac teristics: (a) A rainbow coloration would start at the point of initiation and then spread. Sometimes the color spread over the entire specimen surface. The time between the beginning color formation and the scratching varied from less than a minute to nearly an hour. (b) A brown, liquid material would start to form in the scratch after color formation began and would continue to build up. (c) A wet appearing, brownish film would start to form at the scratch and then spread gradually and tend to obscure the color formation. (d) Sometimes surface pitting, generally near the scratch mark, occurred after the color spread and formation of brown material in the scratch mark was well advanced. (e) After the color spread was well advanced and a considerable amount of brown material had formed in the scratch, the specimen would sometimes ignite. The product of the ignition reaction was titanium tetrachloride. Due to the small amount of chlorine available, the specimen was not destroyed. Figure 6 shows a sequence of such reactions up to the time of ignition. When sufficient water was present this reaction sequence did not occur. The colored phase did not always start at the scratch. Sometimes it was at an edge or at the point of contact of the specimen and specimen holder. Under flow conditions the reaction tended to stop after initiation and there was a tendency for a white solid to form at reaction sites when there was an appreciable amount of water in the chlorine. water in the chlorine no reaction occurred \ the surface of the specimen was scratched. A1 155 C (311 F) was required before any reac: was observed. In dry chlorine the reaction star readily with scratching. It appears thettherem' of hydiogen increased resistance to chlorine att; Flow Scratch Tests Flow scratch test reactions were not as dram as those in non-flow bottled chlorine. Often c small spots of reaction occurred. Some typ: reaction results are shown on specimens in Figu. which also shows a specimen of unalloyed titan that withstood exposure to dry, static hot: chlorine for 17 months with no evidence of react: In Table 3 are shown some data for flov bottled chlorine. In flowing, high purity ehlo: of about 0.85 percent water content one rtac occurred in six tests. This reaction starter' an unscratched titanium holder not on a i specimen. Two specimens that did not react electrolytically roughened surfaces that reacted readily without scratching in non-flov. chlorine of 0.11 percent water content. Th results show that because a reaction is 1 likely to occur in flowing chlorine, less wis required for passivation. Miscellaneous Tests When 0.030-in diameter unalloyed titanium w was twisted until broken in the presence chlorine, either dry or saturated with wate: There was a difference in the makeup of the colored surface material when bottled chlorine was used and when dried, crude Hooker cell chlorine was used. Bottled chlorine produced a colored surface material that generally could be wiped off. Dried, crude Hooker cell chlorine, produced a colored material that was either more tenacious or could not be wiped off at all. f3 While investigating possible ways of accelerating the reaction between titanium and wet chlorine some observations were made. Specimens that were deeply pitted eleccrolytically and thenfurther roughened by exposure to hydrochloric acid, reacted quite readily to chlorine that contained 0.11 percent water at room' temperature in non flow tests. No surface disturbance was required. When the hydrogen was removed from such a roughened surface by vacuum annealing at 1000 C (1832 F) it became much more resistant to chlorine. At room temperature with about 0.11 percent Figure 7 - A. Specimen of unalloyed titanium that had been t posed to dry, static chlorine at room temperature for 17 months without reacting. D, Specimens of unalloyed titanium that had been scratched in flowidried, crude Ilooker cell chlorine and showed van degrees of attack from spot color formation to he:, attack. 94 CORROSION - RAC tests at 195 C (383 F) indicated that between 0.93 and 1.5 percent water was required. Tests at low flow rates and 175 C (347 F) indicated that about 1.1 percent water was required. .!r* ;i jet of dry chlorine at room temperature :: o paig was directed onto an unalloyed surface, no reaction occurred unless rtf ire tip touched the titanium surface. When .Morine was water-saturated no reaction red even when the surface was rubbed with - or the glass orifice. RESULTS OF SCRATCH TESTS IN HOOKER CHLORINE -jits of a large number of scratch tests flowing, dried, crude Hooker cell chlorine ihown in Figures 8, 9, 10 and 11, Chlorine me had some effect in reducing the water rements for inhibition of the reaction. This mosc pronounced at room temperature and 5 C (346 F). There was less effect at 75 C 25 C (157-257 F), At higher temperatures water was very definitely required for non of titanium in chlorine. In general the .r tests showed considerably less, water required for inhibition than was necessary mm-flow tests in purer chlorine, dts of static and the flowing chlorine tests nparable. Static tests indicated between 2 ('.71 percent water was required at temperature, while at low flow rates the red content was about 0.4 percent. Static The small amount of oxygen in the Hooker Cell chlorine probably had an inhibiting effect beyond mere dilution but not in the same order as the effect of water. Because the standard free energy of formation of Ti02 is -203.8 Kcal and that of TiCl4 fliq.> is -161.2 Kcal at 25 C it is likely that any reaction involving oxygen is directly with titanium, viz. Ti + Oa -* TiOg. The Os and Cl3 are merely a mixture. With water and chlorine, however, there is probably some degree of chemical combination, such as the formation of hypochlorous acid, H30 + Cl3 - HC1 - HOC1. It may be assumed that the brown liquid observed in some of the scratching tests was chlorine trioxide, C103, which is a brown viscous liquid. This would require some degree of combination of water with chlorine. However, it is more likely that the brown material was wet titanium dichloride. Among the scratching devices, Ti-6A1-4V alloy was the most satisfactory material. Stainless steel reacted to form a wet, green coating, probably of ferrous chloride. The diamond device chipped so that the cutting edge did not remain sharp. Glass rods broke. The Ti-6A1-4V alloy rod could be sharpened readily to produce a good scratch. Chlorine Flow Rate, Feet Per Second 1-2, Figure 9 - Reactions occurring at 75 C in dried, crude Hooker cell chlorine. 0.7 Chlorine Flow Rote, Feet Per Second 96 CORROSION - NACE Figure 11 - Reactions occurring at 175 C in dried, crude Hooker cell chlorine. To check the influence of surface preparation tests were made, on specimens with no . preparation, specimens that were heavily .-.-rprintod, some that were scrubbed with an ? ivc- cleaner and some that were scrubbed then pickled in 10 percent HN03-2 percent HF -non. The cleaner surfaces appeared to be "'.what more reactive, but the difference was ^Treciable. At higher temperatures to about 200 C (392 F) about 1.5 percent water is sufficient. Flowing dried, crude Hooker cell chlorine did not require as much water for inhibition. At room temperature about 0.4 percent water was sufficient. At about 175 C (347 F), the amount of water required for passivation was about 1.1 percent or less, depending on the flow rate. 'hat appeared to be an important variable was ntch depth. A definite cutting action often -Ited in reactions occurring under conditions * none could be obtained with a dull scratching ::e. CONCLUSIONS Although titanium is very sensitive to chlorine *i, a small amount of water will inhibit attack. * amount of water required is a variable ending upon temperature, gas movement and '-'sure. Pure (99.5 percent) chlorine under '-c conditions appears to require about 0.93 3'tnt water at room temperature. Somewhat water is required if the chlorine is flowing. The amount of water required for protection of titanium is considerably higher than has been considered sufficient heretofore. Saturation may not be sufficient if the saturation value is below the required amount for a given set of conditions unless a film of water is present on the titanium surface. Precise values for the degree of inhibition provided by a given amount of water are indeterminate because a reaction once started tends to be catastrophic. There is no easily measurable rate of corrosion. The water either does or does not prevent a reaction. It does not necessarily limit the rate of attack once a reaction has started unless a film of liquid water is present. REFERENCES 00 073044 CONFIDENTIAL 'T Titanium Summary. International Nickel Company, ;,* Oil if. Corrosion and Corrosion Control. John Wiley ' "na, New York, New York, 1963, , `tch. Assoc, Pulp and Paper IncJ,, Vol. 44, 1B3A-8A ^i) February. pfiIL, 1967 4. Private communication from another laboratory. 5, ASTM Standards, 1932. Part 2, Non-Ferrous Metals. American Society for Testier, Materials, Philadelphia, Pa., p. 749. 97