Document LJVGGnodvD9Ydm6OaG7zvp4dw

u iToiuv hitwiru the values is miuwii to be small. Although tst's, the difference appears to be .ig metal for which measurements it ion which provides an estimate .be of assistance in providing an mted by Taylor for the surface n Fig. 0 plotted in the form of a >f sublimation. The corresponding ace tensions, those for tungsten* figure. The straight line relationnent with these two values. This does provide a way of obtaining ctory metals. relation exists: that between the tures. Their suggested correlation or the refractory metals and the is apparent that the Siuta and i estimating surface tension data. y appreciated. The skill and peroperating the apparatus is grate- Laboratory, Wright Air DevelopVF '''ntract No. AF33(6i6)-626g. II, New York, 1957. V Thomas, Fracture, proceedings of s Wiley, New York, 1959. ,c. Metals, Cleveland, Ohio. -2-H. (i7h . Can. J Cheni , 34 (1956) 1553 185 (1949) 185 91 (1951) 1406. E, 191 (195U 1409 `s, 88 (i960) 217. Metals, Cleveland. Ohio. If'.-/ DC. Tech. Kept. S7~U-l. (`9S9>- 951) -U5 1531 '15. I 1 Less-Common Metals. 3 (1961) 360-366 JnlKNAI, i>I- I'HE I.K^-CIIMMIIN METALS J<>7 A STUDY OF METAL IGNITIONS I. THE SPONTANEOUS IGNITION OF TITANIUM* 1-. E. UTTMAN, I'. M. CHURCH and E. M. KINDERMAN Stanford Kesearch Institute, Menlo Park, Calif, (U.S.A.) (Received June 23rd, 1961) SUMMARY Spontaneous ignition of massive shapes of titanium can occur at room temperature when a fresh, oxide-free metal surface is exposed to oxygen under pressure. Such a surface can be produced by the rupture of a titanium specimen under tension. If these surfaces are exposed to oxygen pressure in excess of 350 p.s.i.g. spontaneous ignition of the sample occurs. With dilute oxygen, using helium or steam as a diluent, higher pressures are required for ignition. Spontaneous ignition at various concentrations of oxygen was investigated and-the effect of temperature on ignition limits was also determined. The hypothesis is suggested that only those metals whose oxides are soluble in the metal will ignite spontaneously and ignition will occur only if the initial reaction is vigorous enough to raise the surface temperature to the melting point of the metal or an eutectic. INTRODUCTION Spontaneous ignition of massive titanium was investigated because of concern over inexplicable failures of titanium parts used in test loops in the Homogeneous Reactor Program at Oak Ridge1-2. The damage had the appearance of a bum, and metallographic examination suggested that the x-/3 transformation temperature (85oC) had been exceeded. The presence of titanium oxide further substantiated the suspicion that the failures were due to a spontaneous ignition of the metal in contact with either hot aqueous solutions or the atmosphere above it, consisting mostly of oxygen under nressure. Since these reactions occurred spontaneously and appeared to be self-sustaining, they constituted a considerable potential hazard. A project was therefore initiated at Stanford Research Institute to answer the following questions: (1) How are such processes initiated and (2) once initiated, how are they sustained? Equipment spontaneous ignition of titanium bars and its limits Suspecting that ignitions occurred when the protective oxide film normally covering the surface of titanium is disturbed, an apparatus was designed which would make possible exposure of oxide-free surfaces to various environments. A stainless steel high pressure vessel was equipped with a sliding bar, which was in ' This work was performed under Contract No. W-7045-ENG-26, S' bcontract No. 1088 for nion Carbide Nuclear Corporation, Oak Ridge, Tenn. J. Less-Common Metals, 3 (1961) 367-378 . v.prS^i J|><S l\ K. HUMAN. !. M. t'Hl'UCH, K. M. KIN IMiRMAN Fig. i. Reactor assembly. Fig. i. Reactions of titanium with helium-oxygen and steam-oxygen mixtures. Static tests. J. [-rss-Cinwit Metals, 3 (19A1) 367-378 M. KINMKKMAN iblv. 'l,nS'T.\Ni:i>r> KiNITION Dl; T1 TANH'M 3<*9 turn connected to a hydraulic ram. Titanium bars, fabricated in the shape of a tensile test specimen, were connected to the sliding bar ami broken in tension, thus exposing a freslt surface to tlu* atmosphere. The composition and pressure of the atmosphere surrounding the sample could he varied over wide limits. This equipment is shown in Tig. I. Static tests The procedure used consisted of placing a sample in the reactor, securing the reactor in the steel frame located in an explosion test pit, and pressurizing to the desired pressures. When proper conditions were attained, the sample was broken in tension bv the hydraulic pull-ram. The occurrence of an ignition reaction was indicated by a sudden decrease of the gas pressure. The reactor was then vented and opened up for visual inspection of the contents. The lower ignition limits were established by breaking a series of titanium rods. The pressure for a given oxygen and diluent composition was increased from run to run, until ignition occurred. Sometimes "initial" ignition was noted: that is, the test sample would show burned spots, usually on the edges of the fresh surface, but the ignition did not propagate. The "initial" ignition results were observed under con ditions close to the borderline shown in Fig. 2. The results of tests with heliumoxygen mixtures are shown in Table I. Table II shows results obtained under other wise similar conditions, but using 1/2-in. strips of 12-mil titanium foil in place of the i.'4-in. rod. A third set of results is shown in Table III, where steam was used as a diluent in place of helium. All of these data are shown graphically in Fig. 2. The line drawn through these points indicates the boundary between the go and no-go regions. There seems to be no significant difference between the behavior of titanium foil and rod, nor between ; helium and steam as a diluent. Most of the samples of titanium used in the static experiments consisted of 5-in. lengths of 1/4-in.-diameter rod, reduced to a 1/8-in.-diameter cross section near the middle, with a I/2-in. taper. Several runs were made with differently shaped speci mens. Some rods were reduced to i/S and 1/16 in., respectively, by a square-cut groove I ti TM too steam-oxygen mixtures. Static tests. /,* ss-Conniiuit Metals, 3 (19^1) 367--37S TABLE I REACTIONS OF 1,4-in. TITANIUM ROD WITH HELIUM-OXYGEN MIXTURES (Static conditions) p* p w; 300 ^00 300 400 500 500 900 $00 700 75 9jo 900 Urn 'p-114.J ZOO 100 100 -- 100 5 TOO ZOO 30O 450 950 IOOO l *( (p,s,i t; 500 joo 400 400 600 550 1000 1000 1000 toon 1900 1900 o, bo SO 75 I OO 83 9* 90 80 70 75 5 47-5 Ignition no no no yes initial initial yes yes initial (1/4 only) yes no J. LfS/i-CtiHimou Metals, 3 (1901) 3t,7-37*t J7" K, K. I.UIMAN, r. M. CHl'IM II, E. M. KIN HI-UMAX TAUI.1C II REACTIONS OP 0.012-111. TITANIUM FOIL WITH IIF.LIUM - OX VUKN MIXTURES l'9 tfi'* *?/ trl.ldl 1i.e 1 ", r\.i /gmf wh JOO -- 300 100 yes 450 -- 45 100 yes 35 -- 35 100 no 35 *-- 35 100 yes 900 95 1830 48.5 no 900 750 1650 5-1-5 yes 900 600 1500 60.0 yes 900 400 1300 69.0 yes 900 500 1400 64.0 yes 700 300 1000 7 yes 450 50 500 90 yes S50 900 1750 48.3 no 850 700 155 55 no 6jo 280 930 7 no 600 200 Soo 75 no 300 l`0O 600 835 no tp.t.i.f.) 500 500 630 950 1000 525 750 S30 TABLE III IGNITION IN PRESENCE OF STEAM Pum 1*1*1 0. /'p.i.i.f.; to*.;.*.; f*o; Ttmpiraturc rej 15 100 ns 600 lOOO 175 300 75 515 600 875 >55 2000 700 1230 1600 97 83 74 61 5 75 60 53 100 165 200 250 2S5 iSS 241 275 Ignttton initial initial initial yes yes no no yes 'JV. - Fig, 3- Ignition of titanium poppet. near the center, resulting in a very abrupt change in diameter. Another sample was machined from i-in. round stock to resemble the poppet in the valves mentioned in J. Less-Common Metuis, 3 (1961) 367-378 M. KINI>KKMAN II,,,, -<)\ Vl.l-.N MIXTl'WhS /Cliff J 1*11 ' \Vs "l yes 1 lit) 1 yes '5 no *5 yes .o yes .o yes i.O yes yes t yes !-5 no ..0 no t no 1 no ;-5 no spun k;miion ofthanh'm 37* 1 IRSI. Report Xo. 56-S 214. Its cylindrical section was about 1 in. long, with a ho taper coming to a t/.S-in, point attached to a I /4-in. rod. (See Fig. 3: the burned sample had a longer cylindrical section.) This sample weighed 72 g. In still other runs, 1/2-in. strips of 0.012-in. titanium foil were used. All of these samples ignited readily upon breaking and were completely consumed or melted, even though in the case of the large poppet only 25% of the amount of oxygen required for complete oxidation was present. These tests indicate that, in the presence of oxygen at the necessary pressures, exposure of a very small fresh area can result in a self-sustaining reaction. Several different titanium alloys were used, such as A-55, A-iio AT, and 6 AI-4V. The results obtained with these samples did not differ within the limits of experi mental error, which are of the order of 25 p.s.i.g. The above experiments were carried out at room temperature. The effect of elevated temperature on ignition conditions was also examined. The pressure vessel STEAM 'Hfh'raturc X) 100 105 ZOO -JO I -41 -75 initial initial initial yes ves no no yes m V - 1 poppet. in diameter. Another sample was poppet in the valves mentioned in Less-Common Metals, 3 (1961) 367-37^ Fig. 4. Effect of initial temperature on ignition of titanium rods. was modified by insulating the pull-ram electrically from the body of the bomb. This was done by enlarging the guide hole of the rod connecting the pull-ram and the sample, filling it with a reinforced Epoxy resin and redrilling to size. With this insulating sleeve between the extension rod and the body of the bomb, a heavy AC current could be passed through the mounted sample. The temperature of the nar rowest portion of the sample as a function of the applied voltage was determined with an embedded platinum, platinum-10% rhodium thermocouple. In the actual runs the temperature was not measured, but was estimated from the applied voltage. This procedure resulted in an uncertainty of about i 5% in the temperature assumed. The sample was heated in the presence of oxygen until it reached the desired temperature (10 min). The current was then shut off and the sample was broken in tension. The results are tabulated in Table IV and shown in Fig. 4. The minimum oxygen pressure for spontaneous ignition upon rupture of the sample drops from 350 p.s.i.g. at room temperature to 210 p.s.i.g. at 300SC, to 150 p.a.i.g. at 500C, and to 75 p.s.i.g. at ioooC. The wide range of "ignition but no propagation" indicated in Table IV seems to contradict our previous experience, J. Less-Common Metals, 3 (1961) 367-378 K. K. MTTM.W, K. M. (HI'KllI, E. M. KIMH-KMAN TAHI-E IV TKMl'KK \Tl'HK UKHF.MIUNCK OF UiNITION TEMl'KRATUkE OF TITANIUM Tt'mpt'raluri' ro . IftHttwH *5 250 250 250 *50 *5 200 200 200 *5 150 I jo 150 150 150 150 1 Jo >5 150 150 IOO IOO IOO IOO 75 50 50 500 400 3 300 200 room 500 400 300 m+p. 900 830 800 700 650 600 550 500 500 400 95 900 700 5<X> I IOO xioo 9JO Yes Yes \ es \es No No Yes Yes No Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes No Yes Yes No No Yes No No which indicated that propagation occurred readily once the sample was ignited. This situation is, however, probably due only to a peculiarity of our experimental arrange ment : since the .sample was heated by passing a current through it, only the narrowest portion was at the predetermined temperature. The steep temperature gradient from the narrow section to the end of the rod may have resulted in effective quenching of the ignition. This view is supported by the fact that propagation of the ignited sample ditl occur more regularly when a sample with a longer conical section was used. To gain a more detailed understanding of the reaction taking place at the freshly formed titanium surface, the reaction was recorded by a Fastax camera, at a framing rate of about 6000 frames per second. The burning of the rod provided sufficient illumi nation to expose the film. Synchronization was achieved by prestressing the sample to the point where the next stroke of the hydraulic pump would cause rupture. The camera was started, and the sample was broken 0.2-0.4 sec later. This procedure allowed the camera to come up to speed and left 0.3-0.5 sec for recording the ignition. Selected frames are shown in Figs. 5 and 6. The first frame of Fig. 5 clearly shows the flash which enveloped the fresh surface an instant after the rupture. A puff of J. LtSi-CliMIHOH Mtilth. J (ll/ll J 367-378 DO 073001 CONFIDE: NT TAL \1. KINDKIOI \> 'U'J* <% \ 1 t Kl', Ol- >1 f \ Nir M lumtntn Ye** \\`> \\s w-s Xu Nii Yes Yes No Yes Yes N'es Yes' Yes Yes Yes Yes Yes Yes No Yes Yes No No . No once the sample was ignited. This arity of our experimental arrange ment through it, only the narrowest e steep temperature gradient from resulted in effective quenching of propagation of the ignited sample er conical section was used. action taking place at the freshly by a Fastax camera, at a framing f the rod provided sufficient illumihieved by prestressing the sample c pump would cause rupture. The 0. 2-0.4 sec later. This procedure j-0.5 sec for recording the ignition, first frame of Fig. 5 clearly shows stant after the rupture. A puff of1 1.rfS-Coinnnm MiUtls, 3 (t g61) 367-378 -.niNI \Mm)IIi.XIIHiX HI- IHANIIM 373 f *-7 JI.5 57.4 Fig. j. Ignition of titanium rod. U7-9 15-5 6. Ignition of titanium n*1, *32.0 /. I*fSs-Cowtou -Uf/ti/v, 3 (1901) 367~37^ J7- I-. It, I.ITHIAN, 1\ M. ITHKMI, K. M. KlXDKttMAN whin- smoke apix-ars ;iikI tin- rod In-gins to hum rather quietly, without much spark ing, melting down as time goes on (see last three frames of Fig. It). If.XITION LIMITS OK TITANIUM SHEET UNDER DYNAMIC CONDITIONS The reactor was equipped with a special head to hold down a titanium rupture disk, which was broken by a spring-loaded plunger. Either water or steam could be selected as the material initially in contact with the disk, depending upon the orientation of the reactor. A very high degree of turbulence existed as the contents of the reactor were discharged through the break in the rupture disk. The thickness of the disks varied from 4 to 12 mils. Fig. 7 shows the arrangement with the rupture disk facing downward. Fig. 7. Reactor assembly for use with rupture disk. The relationship between oxygen pressure and concentration necessary to produce autoignition of a titanium specimen under dynamic conditions of gas flow was deter mined in a series of runs in which the oxygen concentration at a given total pressure of oxygen and diluent was increased until ignition occurred. The results, given in Fig. 8 and Table V, show that ignition occurs much more readily under dynamic than under static conditions: pure oxygen at 50 p.s.i.g. pressure, streaming past a fresh surface, produced autoignition. The ignition limits are much lower throughout the whole range of oxygen concentrations. Fig. 9 shows the appearance of a reacted disk. A number of runs were made with the rupture disks facing down, so that the disk J. Ltss-Common Metals, 3 (1961) 367-378 DO 073003 CONFIDENTIAL \ M, KIN HICKMAN* -it" (uirtly, without much spark- .111,.ol ('). l>VS.\MIC CONDITION'S 10UI down a titanium rupture disk, er water or steam could be selected depending upon the orientation of .ted as the contents of the reactor u disk. The thickness of the disks ment with the rupture disk facing 1 1 >H iNT.WKtM*^ IfJNJUliN n|` TITANIUM 375 Fig. 8. Reactions of titanium with helium-oxygen mixtures. Dynamic tests. icentration necessary to produce :onditions of gas flow was deteritration at a given total pressure occurred. The results, given in tore readily under dynamic than tressure, streaming past a fresh are much lower throughout the he appearance of a reacted disk. ks facing down, so that the disk * J 1 ss-Common Metals, 3 (1961) 367-378 Fig. 9. Rupture disk after reaction. J Liss-Cowmvi) Metals, 3 (ttjfjt) 367-378 37<> 1\ K. L1TTMAN, I\ M. IIU'KIH, E. M. KISDEKMAN TAHI.li V REACTIONS OF TITANIUM WITH HELIUM -OX YG EN MIXTURES UNDER DYNAMIC CONDITIONS Foil tkifkHfw OhJ r* ip'* *,*./ pHm ot tfHtlioH () O.OO3 0.003 O.OO3 O.OO3 O.OO3 O.OO3 0.003 O.OO3 0.012 0.008 0.008 0.008 0.008 0.008 0.008 0.008 0.012 0.012 0.012 0,012 0.012 0.012 35 300 250 200 150 IOO 5 10 50 500 400 450 420 370 340 300 450 450 340 600 630 500 -- --. -- -- -- -- -- -- 500 600 550 180 230 *3 200 IO50 1050 960 OOO 1170 200 350 300 25O 200 150 IOO 5 IO 50 2000 IOOO IOOO 700 620 47 300 I5OO I^OO 1500 1500 1800 700 100 100 100 IOO IOO IOO too IOO IOO 5 40 45 60 60 7S-5 60 30 30 36 40 35 71.5 yes yes yes yes yes yes yes 00 yes yes no yes yes yes yes no no no no yes no yes was originally in contact with oxygenated water. The concentration of oxygen in water was varied by increasing the oxygen pressure above the water up to iSsop.s.i.g., which resulted in concentrations up to 7200 p.p.m. dissolved oxygen. In no case was there any indication of reaction of the oxygenated water with titanium even under streaming conditions. PROPAGATION STUDIES To start a self-sustaining reaction between titanium and oxygen, a very rapid rate of reaction has to be attained to raise the temperature of the specimen enough to continue the burning. If the initial energy is supplied from exterior sources, milder conditions should be adequate to maintain the reaction. A different reactor was built for these experiments. It consisted of a cylindrical body 4 in. in diameter and 8 in. long, with a standard head and closure. Two insulated lead-ins were provided in the head. The sample was fastened to these leads. Power for heating the sample was furnished by an a.c. welding transformer capable of delivering 175 amp at 20 V. A 1/4-in. high pressure line and a valve completed the setup. The samples consisted of 1/2-in. strips of titanium foil 0.012 in. thick and about 6 in. long, bent into a U shape and notched at one-end. For the first run the reactor was filled with oxygen at 1 atmosphere pressure and the reaction was initiated by passing a current through the notched sample until it melted (about 5 sec). The reaction continued, consuming the whole sample. Air was used in another run; under these conditions there was no propagation, the reaction J. ljss~Common Mrials, 3 (1961) 367-37* M. KINDI':l(MAN 'RES UNDEM DVN.DIU*1 I ONU1T10.VS I), rj iS>ulio too 100 100 100 100 100 100 100 100 50 AO 45 60 60 72.5 60 30 3 36 4 35 71*5 yes ves yes yes yes yes yes no yes yes no yes yes yes yes no no no no yes no yes SI'ONTAMIDI'S IGNITION OF TITANIUM 377 ceasing as soon us the sample melted and broke the circuit. The minimum oxygen concentration necessary at other pressures was determined similarly. The results are shown in Table VI and Fig. to. The curve in the figure is markedly to the left of the The concentration of oxygen in a .'thewateruptoi850p.s.i.g., dissolved oxygen. In no case was 1 water with titanium even under ES i and oxygen, a very rapid rate of I .ture of the specimen enough to lied from exterior sources, milder :ion. ;nts. It consisted of a cylindrical d head and closure. Two insulated fastened to these leads. Power for transformer capable of delivering a valve completed the setup. The 012 in. thick and about 6 in. long, t j i ! . { l ;en at i atmosphere pressure and ough the notched sample until it uming the whole sample. Air was was no propagation, the reaction (.rss-Comttion Melah, 3 (19(11) 367-378 j 1 i Fig. io. Ignition propagation in hclium-oxvgen and steam-oxygen mixtures. TABLE VI propagation of reaction of titanium with helium-oxygen mixtures 0. f'.j too 20 20 50 60 55 3 40 35 20 20 35 3 IO 15 20 45 3 13 21 He (%) _ (air) So 50 4 45 70 60 65 80 So 65 70 90 5 so 75 7 75 78 P~.. fp.i.t.*.) O 0 too 100 100 100 200 200 200 ^OO 450 450 45 800 800 800 800 800 1000 1300 /fXj/104 yes no no no yes yes no yes yes initial ves ves yes no no initial initial yes yes yes J. Ijsi'Comnwn MrInis. 3 (c<)6f I 367-378 37? 1`. K. IJTTMAN, F. M. I'lU'Ki II, K. M. KINDI'KMAN T\HI.K VII 1`KOI'M. VllllS OK KI-. M TIOS OK TtTA.VIl'M IVJTH STK AM - OX YGKN MIXTURES <>1 IttO /Yh-1 r'oi ) 5 10 20 7-5 J-o 5-5 5 2.1 20 10 O 0 95 90 So 9-* j 95 94-5 97 5 97.9 So 90 97-5 100 100 0 0 0 200 300 370 boo 700 800 800 830 1600 1900 no yes ycr yes yes yes yes yes yes yes yes no no curve delineating the spontaneous ignition limits of titanium in similar mixtures. This indicates that the reaction will continue under much less drastic conditions than were necessary for spontaneous initiation by exposure of a fresh surface. A similar set of experiments was performed with steam-oxygen mixtures. The pressure vessel was charged with about ioo ml of water, a calculated amount of oxygen was added, and heat was applied until the desired pressure was reached. The reaction was then initiated by passing current through the sample until it melted in the notched area. The results obtained with steam-oxygen mixtures are quite different from those obtained with helium-oxygen mixtures. As little as 10% oxygen in steam at one atmosphere was enough to sustain the reaction, and at higher pressures less than 5% was needed. Pure steam, however, did not sustain the reaction even at 1900 p.s.i.g. The results are tabulated in Table VII and shown graphically in Fig. 10. CONCLUSIONS Some of the limiting conditions for spontaneous ignition of massive shapes of titanium have been defined. From the results presented it appears that relatively massive shapes of titanium will spontaneously ignite if a fresh surface is produced, e.g., by rupturing a sample, in the presence of oxygen under pressure. The mechanism of this reaction is uncertain, but may involve the melting that results from initial reaction. Since the oxide is soluble in the molten metal, no protective surface coating is formed and the reaction becomes self-sustaining. The limiting conditions may be those that generate enough heat during the initial stages of this reaction to melt the surface of the metal. REFERENCES 1 J, C. Griese, T. M. Kegley and J. L. Gregg, Corrosion of titanium impeller No. i. ORSL- CF-55-9-65, Sept. 1955. * J. P. Hammond. T. M. Kegley, Jr. and G. M. Adamson, Failures of titanium alloy trim in HRP dump valve loop, ORSL-$6S-3J4, Aug. 1956. J. Less-Common Metals, 3 (1961) 367-37