"Applied Science, Faculty of"@en . "Materials Engineering, Department of"@en . "DSpace"@en . "UBCV"@en . "Takagi, Ken\u00E2\u0080\u0099ichiro"@en . "2010-05-24T02:50:54Z"@en . "1984"@en . "Master of Applied Science - MASc"@en . "University of British Columbia"@en . "The advantages of the melting of titanium and its alloys in an electron beam furnace are well understood in terms of homogenization, but it is clear that a major problem is that of evaporation of volatile elements. The purpose of the present study Is to investigate the evaporation of solutes such as aluminum from titanium alloys and the conditions under which the evaporation takes place.\r\nThe surface temperatures of the molten metal were measured by an\r\noptical pyrometer. Its accuracy was checked by measuring the\r\nevaporation rate of pure titanium over the range of experimental temperatures used.\r\nFrom the analysis of the aluminum evaporation rate from Ti-Al-V alloy, it has been established that the Langmuir equation may be applied; that is, the evaporation takes place from a well-mixed pool of uniform composition with a very small mass transfer resistance. Also, the specific evaporation constant and the activity coefficient of aluminum were calculated.\r\nAttempts were made to use the EDX equipment directly to measure the aluminum concentration changes during melting. This technique may be promising although the method and conditions were not fully defined."@en . "https://circle.library.ubc.ca/rest/handle/2429/24944?expand=metadata"@en . "OBSERVATIONS ON ELECTRON BEAM MELTING OF TITANIUM AND ITS ALLOY by KEN'ICHIRO TAKAGI M.Sc, Waseda University, Tokyo, 1977 A THESIS SUBMITTED IN PARTIAL FULFILMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF APPLIED SCIENCE in THE FACULTY OF GRADUATE STUDIES Department of Metallurgical Engineering We accept this thesis as conforming to the required standard THE UNIVERSITY OF BRITISH COLUMBIA JUNE 1984 Ken'Ichiro Takagi, 1984 In p r e s e n t i n g t h i s t h e s i s i n p a r t i a l f u l f i l m e n t o f the requirements f o r an advanced degree a t the U n i v e r s i t y o f B r i t i s h Columbia, I agree t h a t the L i b r a r y s h a l l make i t f r e e l y a v a i l a b l e f o r r e f e r e n c e and study. I f u r t h e r agree t h a t p e r m i s s i o n f o r e x t e n s i v e copying o f t h i s t h e s i s f o r s c h o l a r l y purposes may be granted by the head o f my department o r by h i s o r her r e p r e s e n t a t i v e s . I t i s understood t h a t copying or p u b l i c a t i o n of t h i s t h e s i s f o r f i n a n c i a l g a i n s h a l l not be allowed without my w r i t t e n p e r m i s s i o n . Department o f M e t a l l u r g i c a l Engineering The U n i v e r s i t y of B r i t i s h Columbia 1956 Main Mall Vancouver, Canada V6T 1Y3 D a t e \u00E2\u0080\u00A2\u00E2\u0080\u00A2\u00E2\u0080\u00A2 June 19. 1984 ABSTRACT The advantages of the melting of titanium and i t s alloys in an electron beam furnace are well understood in terms of homogenization, but i t is clear that a major problem is that of evaporation of volatile elements. The purpose of the present study Is to investigate the evaporation of solutes such as aluminum from titanium alloys and the conditions under which the evaporation takes place. The surface temperatures of the molten metal were measured by an optical pyrometer. Its accuracy was checked by measuring the evaporation rate of pure titanium over the range of experimental temperatures used. From the analysis of the aluminum evaporation rate from Ti-Al-V alloy, i t has been established that the Langmuir equation may be applied; that i s , the evaporation takes place from a well-mixed pool of uniform composition with a very small mass transfer resistance. Also, the specific evaporation constant and the activity coefficient of aluminum were calculated. Attempts were made to use the EDX equipment directly to measure the aluminum concentration changes during melting. This technique may be promising although the method and conditions were not f u l l y defined. - i i -Table of Contents Abstract 11 L i s t of Figures .. i v L i s t of Tables v L i s t of Photos v i Acknowledgements v i i L i s t of P r i n c i p a l Symbols v i i i Chapter I Introduction 1 Chapter II L i t e r a t u r e Survey 4 2-1 Ele c t r o n Beam Melting 4 2-2 Evaporation 10 2- 3 Melting of Titanium and Its Al l o y s i n Ele c t r o n Beam Furnace ..17 Chapter I I I Experimental Procedures and Results 20 3- 1 El e c t r o n Beam Furnace 20 3-2 Temperature Measurement 23 3-3 Pure Titanium Evaporation - Temperature Measurement Evaluation - 31 3-4 Determination of Molten Pool Volumes 34 3-5 EDX A p p l i c a t i o n to Quantitative Analysis of Aluminum 36 3-6 Observation of Aluminum D i s t r i b u t i o n i n Molten Pool by SEM and EPMA 40 3-7 Aluminum Evaporation from Titanium A l l o y 43 3- 8 Direct EDX A p p l i c a t i o n to Ele c t r o n Beam Furnace during Melting 48 Chapter IV Discussion 60 4- 1 Temperature Measurement 60 4-2 Aluminum D i s t r i b u t i o n i n Molten Pool 61 4-3 EDX Analysis for Aluminum Concentration 62 4-4 Di r e c t EDX A p p l i c a t i o n to Ele c t r o n Beam Furnace during Melting 63 4-5 S p e c i f i c Evaporation Constant and A c t i v i t y C o e f f i c i e n t of Aluminum 64 4-6 Implications of This Work f o r I n d u s t r i a l E l e c t r o n Beam Melting 72 Chapter V Summary 74 References 75 - i i i -L i s t of Figures 1. Block Diagram of El e c t r o n Beam Power Supply 8 2. C i r c u i t r y of Beam Generating System 22 3. Apparatus for Temperature Measurement 26 4. Approximate Area Seen by the Pyrometer 27 5. Cooling Curves of Pure Titanium 30 6. Evaporation of Pure Titanium 32 7. Energy-Dispersive Spectrum of Aluminum 39 8. Conversion f o r Aluminum Concentration 41 9. Aluminum D i s t r i b u t i o n from Ingot Surface 42 10. Loss of Aluminum from Ti-Al-V A l l o y at 1970\u00C2\u00B0C and 2050\u00C2\u00B0C 45 11. Loss of Aluminum a f t e r A/V Ratio Correction (A/V=2.44) 49 12. Apparatus for EDS Analysis of Molten Metal 51 13. EDS Analysis of N i - T i 53 14. EDS Analysis for Aluminum (no.41) 56 15. EDS Analysis for Aluminum (no.24) 57 - i v -L i s t of Tables 1. Pure Titanium Evaporation for 3 min 34 2. Aluminum Evaporation from Ti-Al-V Alloy 47 3. EDS Analysis of Ni-Ti 54 4. EDS Analysis of Ti-Al-V (Preset: 400Co.) 59 5. EDS Analysis of Ti-Al-V (Preset: 800Co.) 59 6. Aluminum Evaporation at 1970\u00C2\u00B0C 68 - v -L i s t of Photos 1. E l e c t r o n Beam Furnace 21 2. Copper C r u c i b l e and Feeding Stock 24 3. S t a i n l e s s S t e e l C r u c i b l e 29 4. Molten Volume at 5kV 36 5. Molten Volume at 6kV 36 6. Molten Volume at 7kV 37 7. Sec t ion of Ingot no.37 at 7kV 37 - v i -Ackowledgement I would l i k e to thank Prof. Alec M i t c h e l l for a l l the time he shared f o r me and e s p e c i a l l y f o r his s t y l e of guidance. Gus S i d l a ' s and Rudy Cardeno's assistance was invaluable and i s very much appreciated. The pleasant nature of the s t a f f i n the department and of fellow graduate students made my stay very enjoyable. I am very g r a t e f u l to Oregon M e t a l l u r g i c a l Corporation, Teledyne Titanium, and Reactive Metals Industries for the p r o v i s i o n of titanium a l l o y s for t h i s work. I also wish to express my gratitude to Riken Corporation f o r t h e i r f i n a n c i a l support. - v i i -L i s t of P r i n c i p a l Symbols A cm 2 surface area V cm 3 molten pool volume t sec time for evaporation k cm/sec s p e c i f i c evaporation constant k g cm/sec mass transfer constant for evaporation k^ cm/sec mass transfer constant i n l i q u i d phase \u00C2\u00B0 / o n g/cm^sec mass fl u x of a solute D cm 2/sec d i f f u s i o n c o e f f i c i e n t i n the l i q u i d t ' sec l i f e t i m e of a solute at the surface c m g/cm3 bulk concentration c s g/cm3 surface concentration a condensation c o e f f i c i e n t p\u00C2\u00B0 g/sec 2cm vapour pressure of a solute y a c t i v i t y c o e f f i c i e n t M g/mole molecular weight of a solute X mole f r a c t i o n of a solute at the surface R gcm 2/sec 2 oKmol gas constant cm 2/sec d i f f u s i o n c o e f f i c i e n t i n the gas phase x' cm e f f e c t i v e thickness of the boundary layer i n the l i q u i d phase c^ g/cm3 concentration i n the bulk of the gas - v i i i -S ' 3 c g/cm3 concentration at the surface of the gas a d i s t i l l a t i o n c o e f f i c i e n t y T \u00C2\u00B0K temperature W g/cm 2sec weight loss p mole/cm 3 a l l o y density r cm radius of the c r u c i b l e v cm/sec v e l o c i t y of l i q u i d motion - i x -I. INTRODUCTION The most important point for melting in vacuum is the removal of the normal atmosphere from the vi c i n i t y of the melt being processed. In this manner, the oxygen and nitrogen, other than that contained in the raw materials used in the charge, cannot react with the molten bath both during the working of the heat and after refining and fi n a l deoxidation reactions are carried out. The occurrence of gas-metal reactions at the bath surface, forming oxides and nitrides which result In Inclusions and slags, is minimized. In addition to the cleanliness of the melt, several advantages are expected in the case of electron beam melting: ( 1 ) various shapes of revert or scrap stocks can be used, ( 2 ) a well homogenized and superheated melt results, and ( 3 ) no metal-refractory reactions occur when using a watercooled copper crucible. At the same time, however, some disadvantages exist. For example, the alloying elements are also vapourizing at the same time as undesirable elements which form inclusions in the high vacuum. The evaporation w i l l be promoted i f the surface area of a hearth i s large, which i s common in the electron beam furnace and i f the molten metal i s well superheated. In addition, the high vacuum is required for the stable generation and operation of the electron beam. - 1 -On the other hand, electron beam melting is well recognized as a promising technique for applications In titanium melting and casting since i t cannot be processed by conventional melting methods due to i t s reactivity. However, as the elements which have high vapour pressure w i l l be easily removed from the melt as mentioned above, the control of the chemistry of the f i n a l products is a serious problem. At the present time, there is no thermochemical data available for titanium alloys at the process temperatures. Hence, the value of the activity coefficient is unknown. Moreover, there Is a liquid metal surface contacting a watercooled crucible at the edges and therefore the surface temperature is not a single value. As a result, the prediction and the control of the chemistry of the products become d i f f i c u l t . In this study, aluminum, which is one of the major alloying elements in titanium alloys, i s focused on as the target of interest and examined from the evaporation standpoint, as well as the various conditions surrounding the melting process such as the surface temperature and the molten pool depth, which are necessary to establish or analyse the evaporation phenomena. Another objective Is to investigate the possibility of using the EDX equipment to analyse the aluminum concentration changes during melting. If this technique is feasible, prompt analysis during melting can be ultimately expected and w i l l significantly help the effective - 2 -furnace operation and q u a l i t y c o n t r o l of the products. - 3 -I I . LITERATURE SURVEY 2-1. Electron Beam Melting Electron beam melting ls by no means a new process. The literature describing the early experiments in the production of X-rays contain references to the melting of the target anode by the impinging beam of electrons. In order to provide a background for understanding the process, i t is i l l u s t r a t i v e to consider the operation of the typical diode vacuum tube. This cosists of an evacuated enclosure, a filament (cathode), and a plate (anode). The filament i s heated e l e c t r i c a l l y u n t i l i t reaches a temperature at which the work function for electrons is exceeded. Electrons are then emitted from the surface with an i n i t i a l velocity which is a function of the filament temperature. An electrostatic f i e l d is maintained between the tube elements with the plate held positive relative to the filament. The emitted electrons are accelerated across this f i e l d and impinge on the plate. This represents a valid analogy with electron beam melting in i t s simplest form. The plate represents the melt stock and the cathode i s usually a filament. Electrons are accelerated through the f i e l d in sufficient numbers that the resulting plate power cannot be as easily dissipated as i t is in a diode tube, and the melt stock is heated to the fusion point 1. - 4 -There are three main fields of application: (1) EB melting/coating, (2) EB welding, and (3) EB evaporation/coating. Mention should be made of a few important specific properties of the 1 2 electron beam technology ' : - the energy can be applied directly by electron beam on the workpiece i f high vacuum is maintained - no heat transfer medium is needed. - High purity products are obtained as the process is performed in high vacuum and use Is made of a watercooled copper crucible (reactive metals). - The possible degree of concentration of the energy is such that temperatures of some thousands of \u00C2\u00B0C are attainable, which allows melting or evaporating of any type of known material. - The process does not contaminate the environment. The significant portions of the process are the i n i t i a l melting and dropping of the feed stock and the residence time in the molten pool in the ingot mold. The combination of low pressure, high temperature, and lack of contaminating materials i n contact with the melt provides necessary environment for the purification and devolatilization reactions which in the past have caused such d i f f i c u l t y in achieving successful melting by this technique. As the feed material is f i r s t melted, violent outgassing and splattering are usually encountered. From a practical standpoint the system must be capable of steady operation In the face of these conditions. The two most important requirements for successful furnace operation 1 are: (1) a high vacuum - 5 -pumping system of sufficient capacity to maintain an operating pressure In the chamber of 10 - i + mmHg at high melting rates, and (2) a high voltage power supply which includes a constant current input and an emission controller on the cathode. It is necessary to arb i t r a r i l y select a pumping system and match the melting rate to the existing capacity, since i t seems to be d i f f i c u l t to specify the pumping capacity necessary to handle a given melting rate of material, as the gas content of the feed stock i s , in practice, not known. For most furnaces the maximum size and number of pumps are incorporated that can be conveniently coupled to the melting chamber. However, i t is to be pointed out that i t is extremely Important that the backing pressure for the diffusion pumps be maintained at very low levels during gas bursts in melting. This prevents excessive back streaming of pump f l u i d from the diffusion pump, which would otherwise occur i f the backing pressure were permitted to approach the limiting value for the pumps. With regard to the circuitry of electron beam power supply, the variation in impedance of the gap between the cathode and the melt (anode) as gases and metal vapours are evolved requires moderately complicated electronic circuitry for the appropriate compensation. If the pressure is too high, i t is impossible to maintain the voltage difference between the cathode and the melt, and a low voltage discharge w i l l result. Depending on the nature of the discharge, there w i l l either be Insufficient power for melting or the transient current peak - 6 -w i l l actuate the c i r c u i t breakers. Usually the pressure pulses which cause disruption of the power are bursts of gas from the melt. Essentially steady operation is obtained by providing two means of control over the power input to the melt. F i g . l 1 indicates a block schematic of the ci r c u i t r y . The current limiting input to the power supply permits establishment of a maximum value of the current which may be drawn no matter how the impedance in the gap between the cathode and anode may vary. High current, low voltage arcs are thus prevented. The emission limiting control of the cathode serves to further stabilize operation, particularly with regard to rapid recovery after a voltage breakdown. The high reactivity of the liquid titanium limits the variety of the possible melting processes and leave among the technically accepted processes only the choice between vacuum consumable and non-consumable electrode arc melting, plasma torch and electron beam melting 3. Among them, the properties of vacuum arc melting technique, for example, are superior in some respects to the electron beam technique, such as the range of vacuum pressure during melting and necessary power, but the overall advantage of the electron beam is somewhat larger than that of vacuum arc, mainly due to the usability of consumable electrodes and scrap as starting material and the possibility to superheat the pool and the metal during pouring and the security against waterleaks and - 7 -cathode anode emission limiting circuitry rectifier isolation transformer f constant current network -o AC supply Fig. 1 Block Diagram of Electron Beam Power Supply - 8 -explosion . It w i l l be necessary to discuss the temperature and i t s distribution over the pool surface area in studying electron beam melting. A precise determination of the temperature distribution in a molten pool of metal contained in a watercooled copper crucible i s a very d i f f i c u l t problem. There is no qustion that there is considarable superheating of the surface layers of the melt at the point of impact of the electron beam. Also, electron beams used in applications such as melting cannot be described as having a homogeneous electron energy distribution across the cross section of the beam. Depending on the geometrical setup of the filament with repect to the crucible focusing and deflection currents, the beam shape and energy distribution can vary widely; in fact, there are \"hot spots\" in the beam where the filament i s imaged in certain setups and intense heating occurs at these \"hot spots\". In most cases, the beam spot is not large enough to f i l l a molten pool and, hence, the beam is electrically deflected around to various sections of a large pool 5. Other factors influencing the temperature distribution are the physical properties of the metal, such as i t s thermal conductivity and emissivity at various wavelengths, the voltage and current in the electron beam, the size of the crucible, and the depth of the molten pool 5. A l l these in turn determine the heat loss by conduction and convection to the watercooled copper crucible and by radiation to the surroundings. - 9 -2-2. Evaporation Since the operating pressure is very low in vacuum melting as compared to conventional melting, severe evaporation can result. The extent to which the volatile elements evaporate is dependent upon the operating pressure, the temperature, the nature of the furnace and the thermochemistry of the elements that are present in a melt. A number of s t u d i e s 6 - 2 0 have been done with respect to evaporation from induction melting, from which the evaporation kinetics are clearly described. These theories w i l l also be conveniently applied to the phenomena in electron beam melting. If the loss of solute atoms follows f i r s t order kinetics, then the rate of loss per unit area of free melt surface is given by the rela t i o n 6 : . dm. m. - T - - d t 1 = - k - v 1 - ' where A is the free surface area of the melt; V Is the volume of the melt; m^ i s the mass of the evaporating element,!, in the melt for the time,t; and k is the specific evaporation constant. The integrated form of equation (1) i s - 10 -(2) A plot of the logarithm of the concentration of the evaporating element versus time should thus be linear. A l l data are presented in plots of equation (2) as the gradient is clearly a convenient method of calculating the specific evaporation constant k. number of steps involving evaporation phenomena have been c l a s s i f i e d ' . The steps are considered to be: (I) transport of an atom through the melt to the neighbourhood of the free metal surface ( i i ) transport across a non-turbulent boundary layer to the free metal surface ( i l l ) desorption from the free metal surface into the gas phase (iv) transport across a stagnant boundary layer on the gas side of the free metal surface (v) transport through the gas to condensing site (vi) condensation. Steps ( i i ) and ( I i i ) may be the possible steps to be considered as the limiting steps in perfect vacuum. In step ( i i ) , Machlin's model gives the flux of atoms to the surface as: To interpret the results of the evaporative loss studies, a n = ( o 4D vl/2 / m St i Z (c - c ) g/cm sec, (3) - 11 -where c m i s the concentration in the bulk of the metal as determined by s analysis; c i s the mean concentration at the surface of the metal; D i s the diffusion coefficient in the metal; and t' is the lifetime of an element of surface, i.e. the time for i t to travel from the centre to the crucible wall in the induction st i r r i n g case. The equation can be simplified to yield: n = k^c\" 1 \" c s ) . (4) The evaporation of atoms from a surface into a perfect vacuum, step ( i i i ) , may be described by the Hertz-Knudsen-Langmuir equation: M 1 / 2 s n = a P \u00C2\u00B0 Y ( - T - R T ) x , (5) where a is the condensation coefficient; p\u00C2\u00B0 is the equilibrium partial pressure; y I s the activity coefficient; M i s the molecular weight; and g K i S the mole fraction at the surface. Again, this may be simplified, giving: n = k e c s . (6) Under low pressure conditions, assuming that the fluxes across the metal boundary layer and evaporating from the surface are equal: \u00C2\u00B0 , , m _ s N , s n = k,(c c ) = k c , , 1 e hence, c - ( k ) c . (7) 1 e Substituting eq.(7) into eq.(6), giving: o k-i k n = ( ) c , (8) 1 e thus, the rate i s a function of the rates of the two contributory processes of diffusion and evaporation. - 12 -Under high pressure conditions (above 70um for manganese in s t e e l 7 , for example), k i s no longer a constant, which suggests that the rate is limited predominantly by the resistance of the boundary layer on the gas side of the interface. Evaporation step (iv) can be expressed by: Q n - -p-Cc 8'- c G ) ' <9> s' G where c and c are the concentrations at the surface and in the bulk of Q the gas, respectively; D i s the diffusion coefficient in the gas; and x* is the effective thickness of the boundary layer. As the diffusion coefficient w i l l be inversely proportional to the pressure P, equation (9) may be simplified to: n = k ( c 8 ~ c )P . (10) g In the high pressure range, equation (10) suggests that an inverse dependence on gas pressure might be expected. The temperature of the melt surface also significantly affect the evaporation rate. For the low pressure range, Ward7 reported the variation of the evaporation rate constant with temperature for case of manganese in steel, where at low temperatures, the results are asymptotic to equation (5), suggesting that the rate of surface evaporation is the rate-controlling step In this temperature range, i.e. k g(for evaporation) \u00C2\u00AB k^(for diffusion) so that,k - k g. With an increase in temperature, k departs from the calculated line given by equation (5), indicating that the liquid phase - 13 -mass transfer starts to contribute the evaporation rate. In the steelmaking process, the evaporative loss of elements with higher vapour pressures than manganese, such as zinc and lead, w i l l be completely diffusion-controlled at steelmaking temperatures. The loss of elements with lower vapour pressures than manganese, such as chromium, copper, and aluminum, w i l l be controlled by the surface desorption process; accurately speaking, one should compare Y p \u00C2\u00B0 product for an element with that for manganese. At normal steelmaking temperatures, the rate of evaporation of such elements should be close to the prediction of the Langmuir equation. Q Olette' studied the evaporation of various elements from the standpoint that the evaporation i s controlling the rate. He has given the d i s t i l l a t i o n coefficient which predicts the possib i l i t i e s of elimination of elements. According to whether is greater or smaller than unity, the melt w i l l be impoverished or enriched in the element. a i s introduced as follows: the ratio of evaporation rates dm /dm for y v y x a dilute solution containing m of iron and m of an element Y at T (\u00C2\u00B0K) x y Is proportional to m^ /n^ : dm m y y = a \u00E2\u0080\u0094 dm y m x ' xthen, P\u00C2\u00B0 M^e 1/2 a \" Y \u00E2\u0080\u0094 - M - B ^ ) , (11) FFe 3 where Y , p\u00C2\u00A3, are respectively the activity coeffidient of Y in i t s i n f i n i t e l y dilute solution in iron, vapour pressure and atomic weight of - 14 -the element i . Thus, the evaporated proportions (in % ) , Y of constituent Y and X of iron, are related by the following expression: Y - 100 - 100 (1 - J L ) 0 ? . (12) In practice, must be higher than 2 or 3 to be responsible of any noticiable effect. The calculated value of a^ n in steel i s 960, which is contrary to experimental values in the range 100 - 300. According to Olette, this disagreement may be due to discrepancies of actual temperature measurements of the metal surface, the presence of surface active elements such as sulphur and oxygen, which decreases the evaporation rates of solutes and solvents, and the surface concentration which may be less than the measured bulk concentration. Ohno et a l 1 0 show that the order of evaporation constant k for a number of alloying elements i s as follows: Mn>Cu>Sn>S>Cr under 1600\u00C2\u00B0C and 10\" 2 ~ 10 - 3 mmHg circumstances. A l l the alloying elements follow f i r s t order kinetics (the melt held up to 15 min.) They show that the ratios c S / c m approximate to unity for Fe-Sn and Fe-S alloys, which indicates evaporation control. On the other hand, the ratio for Fe-Mn approaches zero, indicating that manganese w i l l behave under diffusion control in the liquid phase. With respect to Fe-Cu and Fe-Cr binary alloys, the ratios are 0.4 and 0.65, respectively, showing the contributory process of diffusion and evaporation. Surface active elements such as oxygen and sulphur may influence the evaporation behaviour of the elements as indicated by Olette 9 and - 15 -Hayakawa et al . Hayakawa et al show the dependence of the evaporation rate on the oxygen concentration. This study is with regard to the solvent (Fe) evaporation; therefore, i t is not necessary to account for the liquid phase mass transfer. Instead, the mechanism of evaporation i s , in general, controlled by the diffusion in the gas phase (step (iv)) under higher pressure, and by the free evaporation process at the surface of the metal (step ( i i i ) ) in a lower pressure range. In a very low oxygen content (, so that i t was large enough to pass through the emission from the metal surface and also small enough to protect a glass placed on the other end of the tube from coating. As a result of the use of this tube, no coatings were observed on the glass during melting. As mentioned previously, there is no question that there is a temperature gradient on the molten metal surface. The approximate area seen by the pyrometer is shown schematically in Fig.4 . Consequently, the pyrometer indicates the average temperature only of the metal surface. Special attention should be paid to lining up the apparatus and also to the height of the metal surface so that the pyrometer focuses exactly on the surface of the material. Temperature can be calibrated from the milivolt output from the pyrometer If the emissivity of the material of interest i s known. The - 25 -E B Furnace Fig.3 Apparatus for Temperature Measurement area metal pool Approximate Area Seen by the Pyrometer calibration is 28. (measured mV) , n_ ^ ~ , \u00E2\u0080\u0094 H = -t\u00E2\u0080\u0094-.\u00E2\u0080\u0094.' x 100 = output % of f u l l scale, 50 x (emi8sivity) v ' from which the conversion to a true temperature can be done by using the calibration curve 2 8. The emissivity of titanium at our particular furnace system was determined by using the thermal arrest on a cooling curve. It should be noted that the emissivity mentioned here means the total emissivity correction which includes any factor that deteriorates the emission coming up from the surface of the material such as the pyrometer angle and the window glass absorption. The melting point of titanium has been measured by a number of workers using a variety of tec h n i q u e s 2 9 - 3 3 . Values ranging from 1610 -23 34 1720\u00C2\u00B0C have been reported, but the recent data ' of 1669\u00C2\u00B0C has been accepted as the melting point of titanium. A watercooled copper crucible which i s normally used in the electron beam furnace cannot be used to obtain a cooling curve with a clear thermal arrest, because of the rapid cooling rate. Instead, a stainless steel block (Photo.3) without cooling water, was used as the crucible for the melt. A Pt-Ptl0%Rh thermocouple was inserted into the centre point of the block and the block temperature was monitored. Typical cooling curves are shown in Fig.5, from which the - 28 -I\u00E2\u0080\u0094I 1 c m Photo. 3 Stainless Steel Crucible - 29 -f 2000 ' 0 1 2 3 4 5 6 7 8 9 10 Time , sec Fig.5 Cooling Curves of Pure Titanium thermal arrests corresponded to 1669\u00C2\u00B0C when the total emlssivity correction e was 0.29. Block temperatures are also shown ln the figure. 3-3. Pure Titanium Evaporation - for Temperature Measurement Evaluation -In the previous section, the total emlssivity correction of titanium was determined to be 0.29 at the melting point. In this section, in order to ascertain whether this value of emlssivity can be applied over a range of temperature above the melting point, the weight losses of pure titanium due to evaporation were measured at several input power levels at which the temperatures were read through the pyrometer. Then, the measured temperatures were evaluated in such a way that the weight losses at specific temperatures were compared with the published data. Morrison 3 5 shows the relationship between the weight loss and the temperature: log W - 10.08 - 0.5 log T - 2 4 ^ 5 9 , (16) which is shown as line 1 in Fig.6. Koch et a l 2 3 obtain the experimental results shown as line 2. Line 3 is obtained from a similar equation to equation (16), but i t is based on the Langmuir equation: W = p / ^ , (17) or, - 31 -log W - log p - 0.5 log T + \u00E2\u0080\u0094 j \u00E2\u0080\u0094 log , (18) to which the vapour pressure of titanium shown by Kubaschewski 2 6 i s inserted, that i s : 23200 log p - - Z J ~ U U - 0.66 log T + 11.74 (mm), (19) resulting in, 23200 log W = 11.34 - 1.16 log T - =\u00C2\u00B12\u00C2\u00A3L . (20) In this study, the evaporative loss was measured in 3 min after a time of heating up (about 45 sec). Heating time i s the time for a material to reach a molten state over a whole surface area, during which a certain amount of evaporation takes place. Therefore, a net evaporative loss for 3 min was obtained by subtracting an i n i t i a l loss for 45 sec (=0.1811g) from a total weight loss. The data obtained are shown in Table 1 and plotted in Fig.6. In Fig.6, the vapour pressure of titanium, which is dependent on the temperature, is not accurately known. Consequently, a slight difference of the weight losses obtained theoretically and experimentally may be expected. The results from this study are almost within the range of lines 1 and 3, which indicates that the total emissivity correction of 0.29 and the accuracy of the average temperatures may be accepted over a range of temperature. Also, i t can be estimated that the average temperature of the pool surface i s about 1970\u00C2\u00B0C when the accelerating voltage is 6kV and about 2050\u00C2\u00B0C at 7kV as shown in Table 1. - 33 -Table 1 Pure Ti Evaporation for 3min. no. measured temp .,\u00C2\u00B0C accel. volt.KV total wt. loss , g corrected weight loss, g evaporation rat e , q cm~2.sec~1 surface area, cm 2 12 1810 A.5 0.8933 0.7122 4.946x10~* 8 11 1885 5.5 1.8783 1.6972 8.572x10~A 11 38 1975 6.0 2.5591 2.3 780 1.201x10 3^ 11 39 2050 7.0 5.1311 4.9500 2 500x10\"3 1 1 - 34 -3-4. Determination of Molten Pool Volumes In order to study the evaporation rate, it is of great importance to know the pool volume of the molten metal generated during melting. Since 6kV and 7kV were generally chosen as the accelerating voltages in this study, the pool volumes were checked under 5, 6, and 7kV conditions. Under each power condition, pure titanium was melted and kept for 3 min and then an iron wire was fed from the side above the titanium and mixed into it so that the vertically sectioned surface of the ingot showed two distinguishable phases, one of which would be etched differently from another phase due to the presence of iron. Photos. 4, 5, and 6 show the depths of the pools generated under conditions of 5kV, 6kV, and 7kV, respectively. It is obvious that the higher is the accelerating voltage, the deeper ls the pool. Photo.7 shows the molten pool depth that was obtained by the evaporation experiment under a condition of 7kV accelerating voltage, 4 min holding, and the material used was T1-6A1-4V. This shows a similar pattern to Photo.6. From these, the pool volumes were estimated to be about 4.5 cm3 - 35 -9 20 1 2 3 4' 5 !!! Photo. 4 Molten Volume at 5 KV Photo. 5 Molten Volume I 1 cm at 6 KV - 36 -Photo. 6 Molten Volume at 7KV at 6kV and 6.8 cm at 7kV conditions. These values are used for the evaporations calculations hereafter. 3-5. EDX Application to Quantitative Analysis of Aluminum Characteristic X-rays measured by EDX/SEM are rapidly obtained and are quite useful to identify the elements that are present in a material. It has been stated that the application of EDX to the quantitative analysis of the elements is questionable In terms of analytical accuracy. However, since the X-ray intensity of an element should be proportional to the concentration, a calibration curve for aluminum was established for the conversion of the intensity peak counts to the concentration so that the EDX analysis could be used quantitatively. The method for this analysis is that the X-ray counts are accumulated unt i l the highest peak intensity among the elements present (titanium in this study) attains 20,000 counts and then the integrated counts of aluminum are measured as schematically shown in Fig.7. That i s , in this figure, the integrated aluminum peak (Np) is obtained in such a way that the counts of 10 channels from 1.39 to 1.58 keV, which correspond to the aluminum peak, are summed, from which the background counts (Nb) are subtracted, using an estimate that the area Nb is equal to the summation of Na (5-channel integration from 1.20 to 1.28keV) and Nc (5-channel integration from 1.68 to 1.76keV). - 38 -A I o\u00E2\u0080\u0094region of interest 1.20 1.28 1.39 U 9 1.58 1.68 1.76 Energy , keV Fig. 7 Energy - Dispersive Spectrum of Aluminum - 39 -Having obtained the integrated counts Np for various aluminum concentrations, the counts are plotted versus true concentrations analysed by means of the emission spectroscopy, which is shown in Fig.8 . This curve shows a slight deviation from a straight line in the lower aluminum concentration region, where the X-ray peak intensities were not outstanding with respect to the background. This nonlinearity might cause some uncertainty of accuracy in the lower range of aluminum compositions. 3-6. Observation of Aluminum Distribution i n Molten Pool by SEM and EPMA EPMA is an analytical device with which, in principle, quantitative results can be obtained. SEM could also be used in a similar way. Therefore, both EPMA and SEM were used for the aluminum quantitative analysis of a specimen in order to ascertain i f there i s a difference between the two analytical methods. As the sample analysed was that obtained from the evaporation experiments, additional information concerning the aluminum distribution in a molten area was obtained as well. The nature of the concentration gradients was used to estimate the extent of convection. Fig.9 shows the aluminum profiles obtained by SEM and EPMA, with . the SEM line being obtained by using the calibration curve (Fig.8) . The - 4 0 -10000 c o 6000 2 3 4 5 6 Aluminum . %> Fig. 8 Conversion for Aluminum Concentration - 41 -no. 37 SEM I--X--X-V-\" EPMA x x- x-o I TJ cr J L surface 1 2 3 4 5 6 Distance . mm 7 8 9 Aluminum Distribution from Ingot Surface -42 -results of the EPMA analysis were questionable since the X-rays of the light element (Al) were collected in the presence of the heavy element ( T i ) , resulting that the absorption correction was exceeded by more than 200%. Low aluminum concentrations in the specimen might also influence the accuracy. Consequently, i t is not possible to judge the better technique as i t is not known which of the results are correct. On the other hand, the results show that the aluminum concentration is nearly constant in the lower concentration region which corresponds to the melted area. If the molten metal was stagnant, there should be a concentration gradient, which would increase toward the liquid-solid interface. Therefore, i t can be stated that this f l a t distribution of aluminum is associated with a well-stirred melt. 3-7. Aluminum Evaporation from Titanium Alloy It i s desirable to obtain the highest yields possible of the addition elements in titanium alloys. However, for the elements which have relatively high vapour pressure such as aluminum, i t is d i f f i c u l t to do so. As long as there is no means to protect aluminum from evaporation, the next best thing is to estimate how much evaporation takes place. If the evaporation rate can be predicted, i t w i l l be easier to control the solute concentration in the fi n a l products. The evaporative loss of aluminum was measured at various holding - 43 -times. Although the master alloy used contained 6% of aluminum, the i n i t i a l aluminum concentration at the starting time of the evaporation experiment was less than 6% as aluminum vapourized while the ingot was being made. Therefore, the following was used to establish the i n i t i a l concentration. An ingot was made from Ti-6%A1-4%V rod with a fixed power and a constant melting speed. The power was shut off right after the total mass of the feeding material was melted into the crucible. In this case there was no holding time for the evaporation. The aluminum concentration in this ingot was used as the i n i t i a l concentration. Evaporation experiments were conducted at two different temperatures. Time durations for the evaporation of 1, 2, 3, and 4 min were used. For each experiment the ingot was made from a new Ti-6%A1-4%V alloy as described above. 6kV and 7kV were chosen as the accelerating voltages, which were considered to correspond to about 1970\u00C2\u00B0C and 2050\u00C2\u00B0C for the holding temperatures, respectively. Fig.10 shows the results of the aluminum evaporation for up to 4 min. The aluminum concentrations in this figure were obtained by the conversion graph shown in Fig.8. The i n i t i a l aluminum concentration decreased to 2.6% for the 1970\u00C2\u00B0C experiments and 3.7% for the 2050\u00C2\u00B0C series from 6% during the time of making the ingots. Although the - 44 -melting speed was controlled to give a mass melt rate of about 8 g/min for each case, this difference occurred because the section size of the feeding rods being used for 1970\u00C2\u00B0C series was different from that for 2050\u00C2\u00B0C series. A plot of the logarithm of the concentration of the evaporating element versus time should be linear by equation (2): lo* < Im7T-> = ~ k 4\" 1 0 8 6 ( t2 \" V ' <2> i\u00C2\u00BB* That Fig.10 shows the linear relation verifies that f i r s t order kinetics are followed, from which the specific evaporation constant k can be calculated. Table 2 shows the experimental conditions and data, where a constant k was derived from equation (2) and the slopes of the lines ln Fig.10. Activity coefficient y was obtained by using the Langmuir equation for the solute evaporation: Y = ~ v k p / 2 it RMT, (21) based on the assumption that the aluminum evaporation was exclusively controlled by the desorption from the liquid surface. In a comparison of the two lines in Fig.10, i t showed that the two lines were almost parallel despite the fact that the slope at the higher temperature should be steeper than that at the lower temperature. This was caused by the A/V ratios which were different from each other. - 46 -Table 2 Al Evaporation from Ti-Al-V alloy 19 7 0 \u00C2\u00B0 C 2050\u00C2\u00B0C A/V 2.44 1.62 Al change 2.6% -0.39%. in 3 min. 3.7%-* 0.3 3% in 4 min. k (cm-sec^) 4.32x10 - 3 6.22x10~3 y 3.81x10\" 2 3.24x10 - 2 after correction of solvent(Ti) loss Al change 2.6%-* 0.34% in 3 min. 3.7%-* 0.2 5% in 4 min. k (cmsec-^ ) 4.61x10\" 3 6.91x10~3 Y 4.06x10~ 2 3.6 Ox 10\" 2 - 47 -Fig.11 shows these two lines after the correction of the ratio, being 2.44 for both cases. It is clear that the evaporation at a higher temperature takes place more. Another correction may be necessary in respect of the evaporation of solvent titanium. The changes of the aluminum composition actually result from both the aluminum and titanium evaporation, thus the real aluminum loss will be different If the solvent loss is compensated. The results from this correction are also shown in Table 2. Finally, the values of the activity coefficients result in an enthalpy of evaporation of approximately 50 kcal/mol Al. 3-8. Direct EDX Application to Electron Beam Furnace during Melting The principle of electron beam melting is the same as that of SEM; that is, in both cases, the electron beam is impinging the surface of a material. In electron beam melting, because of high input energy per unit area, the material can be melted, and at the same time, X-rays are also emitting. Therefore, in principle, it will be possible to collect X-rays emitting from the melted material for the purpose of a quantitative analysis of elements that are present. In this study, by using the electron beam furnace, the X-ray - 48 -6 5 4 0 1 2 3 4 5 Time , min Fig.11 Aluminum Content as a Function of Time, after A/V Ratio Correction (A/V= 2.44 ) analysis was done through EDX which was connected to the multi-channel analyser. As the f i r s t step, since the electron beam furnace has a relatively low accelerating voltage and a high beam current as compared to those of SEM, the objective is to establish the condition that makes i t possible to collect useful Information of X-rays and to check i f the characteristic X-ray peaks of the elements are reasonably outstanding with respect to the background. Also, when peaks appear in a X-ray spectrum, the peak intensities are examined to see i f they are proportional to the concentrations. The arrangement of EDX in the electron beam furnace is shown in Fig.12. The basic idea of this arrangement i s the same as that for the temperature measurement. However, there is some protector necessary to cover the EDX detector from the metal vapour coming up from the molten metal. For this purpose, a thin film of mylar (6urn) was set in front of the detector. A copper ring which i s supporting the mylar is loose inside the flange so that a good vacuum is obtained throughout inside the tube from the melting chamber to the detector. The accelerating voltage should be as high as possible and the beam current should be as low as possible in order to obtain a good X-ray spectrum. In this study, the electron beam furnace conditions of 8kV and 0.5A were chosen from this standpoint. Although a higher beam voltage than for previous experiments, these conditions are not optimum for X-ray generation, particularly i n comparison to SEM conditions. - 50 -unit: mm Fig.12 Apparatus for EDX Analysis of Molten Metal As the i n i t i a l results under these conditions showed that the collected X-ray intensity was too strong, lead foils were placed between the detector and the melted material so that the X-rays were reduced to a certain extent. After several trials, two 0.2mm-thick foils showed relatively clear X-ray spectra at 8kV accelerating voltage, one of which was placed right before the mylar with a 4mm-diameter hole in the centre and the other was placed on the end of the tube with a 2.5mm-diameter hole as shown in Fig.12. Experiments being done when using this apparatus and under the conditions described above are divided into two parts. One is to check if the peak intensities of elements that have relatively large energy such as titanium are sufficiently sensitive with respect to the concentration because the accelerating voltage in this study may be too low for those elements to be excited. Another is to check i f the aluminum peak counts are proportional to the concentrations while the evaporation is taking place when using Ti-Al-V alloy. Fig.13 shows the results of pure nickel and Ni-Ti alloys which were for the purpose of titanium sensitivity. Table 3 shows the data related to Fig.13. First of a l l , for 1 in Fig.13, which was the case of pure nickel, there was no peak of Ni-Ka, the Ni-La peak appeared at 0.85keV instead. From 2 to 3 and to 4, titanium peak counts increased in accordance with a piece of titanium being added Into the nickel pool. - 52 -117 _ A .0 44 t * r- * \u00E2\u0080\u00A2 \u00E2\u0080\u00A2 2 9 3 ^ 52 \u00E2\u0080\u00A2; O \u00C2\u00BB cn O c O o 416 72 \u00E2\u0080\u00A2 -1 v . . \u00E2\u0080\u00A2 H i Ni-Loc T i - K o l \u00E2\u0080\u00A2 t>. J. .. \u00E2\u0080\u00A2 M , \u00E2\u0080\u00A2\u00E2\u0080\u00A2 t \u00C2\u00BB \u00C2\u00BB\u00E2\u0080\u00A2. t.... . '\u00E2\u0080\u00A2 I tM. \u00C2\u00BB IJJJ-'- ' Fig Energy , keV .13 EDX; Analysis of Ni - Ti - 53 Table 3 EDX Analysis of Ni-Ti (1min. accumulation ) no. peak count total bkgd note N i - U Ti-K* count count* 0 117 - 131880 500 pure Ni Tt* XXX 8kV-0.7A \u00C2\u00A9 44 242 34890 100 (7)*Ti(1g ) 7kV\u00E2\u0080\u0094 0.4A \u00C2\u00A9 52 293 44750 140 (T) \u00E2\u0080\u00A2 T i ( 2 g ) 7kV- 0.4-0.5A \u00C2\u00A9 72 416 50126 150 ( 3 ) * T i ( 4 g ) 7kV-0.4-0.5A * background count near Ti peak ** accelerat ing_j^lta;ge *** beam current Experimental conditions, especially the beam current, affect the count rate. Actually, the beam current was not very stable. Therefore, totall y reliable information may not be expected from these data, but at least the indication is that for these particular conditions the X-ray intensity of titanium may be sufficiently sensitive to show the difference at various concentrations. Secondly, by using Ti-Al-V alloys, the aluminum peak counts were measured while the evaporation took place. It is desirable for the accumulation time to be short because the concentration of aluminum in the alloy i s continuously changing. It i s also desirable that the time duration of the accumulation be controlled by the counts of the highest peak among the elements present in a spectrum rather than counting for a fixed time. For previous experiments, the accumulation time was generally 1 min, but in this experiment the accumulation was stopped when the titanium peak showed 400 counts, in which case the time taken was about 40 sec. Fig.14 shows the results. Right after the material reached the molten state in the furnace and the voltage reached 7kV, the X-ray counts were accumulated unt i l the titanium peak attained 400 counts (no.l in Fig. 14). In 3 min, the X-rays were again accumulated in a same manner (no.2 in Fig.14), and no.3 was obtained after 3 min of no.2. Fig.15 shows similar experimental results, but the highest peak of titanium was accumulated up to 800 counts, in which case i t took 40 -- 55 -400 peak co.\u00E2\u0080\u0094163 ,. int. co!\u00E2\u0080\u00941 : (525 ) \u00E2\u0080\u00A2 \u00E2\u0080\u00A2 4 \u00E2\u0080\u00A2 \u00E2\u0080\u00A2 \u00E2\u0080\u00A2 400 . \u00C2\u00A3 * o o cn o .147 0 \u00E2\u0080\u00A2 \u00E2\u0080\u00A2 \u00E2\u0080\u00A2 \u00E2\u0080\u00A2 V \u00E2\u0080\u00A2 . . * % \u00E2\u0080\u00A2 \u00C2\u00AB... ... \u00E2\u0080\u00A2 (415) Kv*.\u00C2\u00BB' ,'r \u00C2\u00BB \u00E2\u0080\u00A2 \u00E2\u0080\u00A2 \u00E2\u0080\u00A2 134 \u00C2\u00A9 400 i \u00E2\u0080\u00A2 ... /\u00E2\u0080\u00A2 V(348) *\u00E2\u0080\u00A2.. .. ? \u00E2\u0080\u00A2 V/v.- v.* 4 \u00E2\u0080\u00A2 \u00E2\u0080\u00A2 \u00E2\u0080\u00A2 1 \u00C2\u00BB \u00E2\u0080\u00A2 \u00E2\u0080\u00A2 \u00E2\u0080\u00A2\u00E2\u0080\u00A2 C 3 O Al Ti-K* Ti-Krj Fig.14 Energy , keV EDX Analysis for Aluminum (no.41 ) peak co.-*-219 ;*800 I. int.co.\u00E2\u0080\u0094-.v(799) .800 Vf \.--\u00E2\u0080\u00A2 ''-V. t ,.;(629) 209 ;i8oo \u00C2\u00BB J ;(8A5) SVC ' '\u00E2\u0080\u00A2fi1!-1-\" A- 1 \u00E2\u0080\u00A2 ,800 r!: it ' (j ft d u in cn O c D O o 7- 228 :(9oo) Al-Ka 7 1* Ikf \u00E2\u0080\u00A2 \u00E2\u0080\u00A2 \u00E2\u0080\u00A2 .t i \u00E2\u0080\u00A2 \u00E2\u0080\u00A2. Ti-KotTi-KR i 4 1 ' i >. .M. I Energy , keV Fig. 15 EDX Analysis of Al ( no. 24 ) - 57 - i 45 sec to accumulate. Also, the accelerating voltage was Increased to 8kV. The time interval of each X-ray measurement was 3 min; that i s , no.2 was obtained 3 min after no.l was measured, and the f i n a l no.4 was obtained 9 min after no.l was measured. Table 4 and 5 show the related data to Fig.14 and 15, respectively. Integrated counts in the tables were obtained by the same method in section 3-5. In Table 5, neither the peak counts nor the integrated counts of aluminum show a single decrease despite the fact that the concentration should have decreased, although Table 4 shows this decrease. This may be caused by either having conditions which were not optimum for X-ray collection or the level of the aluminum concentration was too low to detect the concentration changes. Therefore, It is not possible at this moment to convert the aluminum counts obtained here to quantitative concentrations, even though previously this relation has been established for the case of SEM. - 58 -Table 4 EDX Analysis of Ti-Al-V ( preset :400 Co. ) no. Al peak count Al int. count Ti peak count total count note \u00C2\u00A9 163 525 400 17031 7kV-0.4-0.5A \u00C2\u00A9 147 41 5 400 17096 3 min. after (T^ \u00C2\u00A9 134 348 400 - 6 min. after (7) Table 5 EDX Analysis of Ti-Al-V ( preset : 800 Co. ) no. Al peak count Al int. count Ti peak count total count note \u00C2\u00A9 219 799 800 29396 8kV-0.5A \u00C2\u00A9 - 629 800 24120 3min. after(T) \u00C2\u00A9 209 84 5 800 25725 6 min. af ter(T) \u00C2\u00A9 22 8 900 800 27194 9 min. after(T) IV. DISCUSSION 4-1. Temperature Measurement The total emissivity correction is 0.29 from the cooling curve observation at the melting point. This value may also be used over the temperature range above the melting point, which has been verified by the evaporation experiment of pure titanium. It should be pointed out that the temperatures obtained by the pyrometer are the average temperatures over the area seen by the pyrometer as shown in Fig.4. The temperature gradient of the pool surface must exist; hence, the evaporation rate w i l l be Influenced by the temperature at the particular point of the surface. The surface temperature is very sensitive to the electron beam energy and the scanning mode, as well as the material to be melted and the crucible. The Influence of the beam energy on the surface temperature has been examined in the case of a 200kw furnace 3 6. The temperature difference between the centre and the edge of the surface reaches about 300\u00C2\u00B0C at 1 kwh/kg input and about 150\u00C2\u00B0C at 5 kwh/kg. Although temperature differences are expected to vary with each furnace, these values indicate that the temperature gradient i s considerable. Therefore, the evaporation rate obtained in this study should be - 60 -considered as the average evaporation rate at an average temperature. 4-2. Aluminum Distribution in Molten Pool The line analysis of aluminum in the liquid phase of the titanium alloy by EPMA (Fig.9) gives no concentration difference between the surface and the bulk of the pool, from which i t could be assumed that the metal generated by the electron beam furnace is well mixed and that the surface-to-bulk concentration ratio ( c S / c m ) is close to 1. The thickness of the boundary layer in the liquid phase which w i l l exist where the diffusion of the solutes in the liquid phase is controlling the evaporation rate is hard to predict. If the boundary layer i s of the order of magnitude of what Ward7 predicts: about 50 \xa for chromium In steel in induction melting when c S / c m is 0.95 and about S 131 100 \m for manganese when c /c is 0.22, a finer analysis of the surface concentration of aluminum may be necessary. However, as far as EPMA analysis is concerned, the aluminum concentration In the melt is constant, indicating that the c S / c m ratio is close to 1, which shows that the evaporation of aluminum is exclusively controlled by surface desorption. The extent to which the melt is stirred partially governs the thickness of the boundary layer in the liquid phase and the evaporation mechanism. If the layer is thin, caused by the melt being well mixed, - 61 -the evaporation is l i k e l y to be controlled by the surface desorption. s in If the layer is thick, indicating a relatively stagnant melt, the c /c ratio decreases and diffusion control results. To measure the extent of mixing is a d i f f i c u l t problem, but the visual observation of the melt motion in the electron beam furnace is possible. This observation indicates that the metal motion looks similar to that of induction melting, the latter being postulated 8 to be of the order of magnitude of 10 cm/sec. By taking some assumptions such as the st i r r i n g velocity, the surface concentration of aluminum (c S) and the c S / c m ratio can be calculated. The result shows the c /c ratio of approximately 0.8 (as w i l l be discussed later), suggesting that the surface desorption essentially controls the evaporation for this particular case. 4-3. EDX Analysis for Aluminum Concentration In the case of the aluminum analysis in Ti-Al-V alloy by EDX, there are no other peaks showing up near the aluminum peak in X-ray spectra. As a result, the method of using EDX for the aluminum analysis is potentially applicable. For elements such as vanadium, whose peaks are close to other peaks of the elements such as titanium, this method cannot be applied because the background counts are affected by other peaks. - 62 -In Fig.8, the calibration line is nearly straight, although i t deviates slightly in the lower aluminum concentration range. By this method, as low as 0.5% Al may be analysed, below which relatively large error w i l l be expected due to the small peak of aluminum which i s not outstanding relative to the background surrounding i t . The average max/min spread of the composition determined by this method is approximately 0.5%. 4-4. Direct EDX Application to Electron Beam Furnace during Melting Due to the significant difference of the conditions during melting as compared to when EDX was connected to SEM, the quantitative analysis of aluminum, which was the ultimate purpose, was not accomplished by this method. The factors that made the difference were mainly the accelerating voltage, the beam current, and the accumulation time. In addition, the angle at which the EDX detector was looking at the molten metal and the distance between the detector and the target may also be among the causes. However, for the particular conditions employed in this study, useful X-ray spectra were obtained which were similar to those of SEM when two lead f o i l s with holes In the middle were placed between the melt and the detector. Identification of the elements that were present was easily done, although not quantitatively. - 63 -The number of X-ray counts accumulated also may not be large enough. In general, the longer is the accumulation time, the more reliable results are expected. However, there is the problem that the aluminum composition in the melt is continuously changing due to the evaporation during melting. Therefore, the accumulation time should be as short as possible from the evaporation standpoint. The method of using EDX given here, however, is promising for the purpose of rapid analysis of the elements present during melting If problems associated with factors such as accelerating voltage and accumulation of sufficient number of X-rays can be solved. If the accelerating voltage of the electron beam is taken higher, the conditions established in this study, such as the hole sizes of the lead f o i l s , may have to be re-established. Furthermore, as the beam current directly affects the count rate of X-rays, i t should be kept constant. The results otherwise would be unreliable although i t is very d i f f i c u l t to do so while melting. 4-5. Specific Evaporation Constant and Activity Coefficient of Aluminum From the experimental results shown in Fig.10, the specific evaporation constant k can be obtained. F i r s t order equation (2) shows that the slope in the graph corresponds to -2.3k(A/V), from which k was determined as 4.32 x 10~ 3 cm/sec at 1970\u00C2\u00B0C and 6.22 x 10~ 3 cm/sec at - 64 -2050\u00C2\u00B0C. The A/V ratio does not affect the value of k but does affect the apparent slope in the graph and the total evaporative loss. Since the molten metal volume at 2050\u00C2\u00B0C is larger (estimated as 6.8 cm3) than that at 1970\u00C2\u00B0C (4.5 cm3), resulting in the A/V ratio being smaller, the aluminum loss i s replotted in Fig.11. Thus, i t is clear that k at 2050\u00C2\u00B0C Is larger than that at 1970\u00C2\u00B0C. Simultaneously with the aluminum evaporation, the solvent titanium i s evaporating; therefore, in order to obtain the true concentration change of aluminum, It is necessary to compensate for the solvent titanium evaporation. That i s , at 1970\u00C2\u00B0C for example, aluminum actually decreases from 2.6% to 0.39% in 3 min as shown in Table 2, during which time titanium also should have evaporated in the quantity that the Langmuir equation predicts to be: \u00C2\u00AB \" p Al RMT ^ \u00E2\u0080\u00A2 <22> where y may be taken as 1 and c s is about 93 wt% (4.19 g/cm3), resulting in the evaporation rate being 1.176 x 10 - 3 g/cm2sec. Therefore, for 3 min, 2.328 g of titanium must have evaporated. Then, 0.34 %A1 is obtained instead of 0.39 % after 3-min evaporation after accounting for vapourized titanium. Similarly, the evaporation rate of titanium of 2.567 x 10~3 g/cm2sec is calculated at 2050\u00C2\u00B0C from equation (22), and the aluminum composition of 0.25 % results instead of 0.33 % for 4-min evaporation. From this calculation, another k is obtained (Table 2), which - 65 -shows some increase but does not diffe r far from that before the correction, suggesting that this correction may be negligible except at very low rates of solute loss. Strocchi et a l 2 7 report the evaporative loss of aluminum from Ti-Al-V alloy, where the specific evaporation constant i s calculated to be about 3.5 x 10~3 cm/sec. They estimate the temperature as 1880 -1945\u00C2\u00B0C, the spread of which i s considerable. The k obtained in this study is slightly higher than their result, the increase being caused by the higher temperature in this study. Another interesting matter is that from the specific evaporation constant k, i t Is possible to calculate the activity coefficient Y of aluminum from equations (2) and (22) If the surface concentration i s known or i f i t can be assumed that the desorption from the surface i s controlling the evaporation rate. In Table 2, y is calculated based on s in the assumption that c /c ls equal to 1. Specific evaporation constant k is given as: k k k * WT+T- > \u00E2\u0080\u00A2 <23> 1 e where k^ i s a constant for diffusion control represented by penetration theory following Machlin's model as: k l - 2 ' TV > <24> where D i s the diffusion coefficient in the metal and t' is the lifetime of an element of surface, and k g is a constant for evaporation control - 66 -as: ke \u00E2\u0080\u00A2 p AVRMT * < 2 5 > That i s , the values of y in Table 2 is obtained at a case: k, \u00C2\u00BB k and hence, k = k . l e e Therefore, i t w i l l be necessary to evaluate i f k^ Is really greater than k g. D and t' must be known in order to obtain k^. Assuming that D i s equal to IO-1* cm2/sec and t 1 is 0.2 sec and 0.4 sec which are based on the velocities of the melt motion to be 10 cm/sec and 5 cm/sec, respectively, k^ is calculated, from which k g is obtained by using equation (23). As long as k is In the order of 10~ 3 and k^ is in the order of 10 - 2 as shown in Table 6, the value of k g remains very close to k. Or, when considering the radius of the melt surface is 1.9 cm in this study and D is 10_i* cm2/sec, the velocity of the melt motion should be more than 1.5 cm/sec in order for k^ to remain in the order of 10\" 2 cm/sec. If the average velocity in induction melting can be assumed as 10 cm/sec, the velocity of 1.5 cm/sec or more in electron beam melting is l i k e l y to occur. As a result, the assumption that the surface desorption is controlling the rate is a good approximation of the actual situation. The activity coefficient, y, of aluminum obtained from k g Is also very close to that directly obtained from k, which is logical as long as k e i s close to k. A l l the results are tabulated in Table 6. The surface concentration c is calculated as follows: - 67 -Table 6 Al Evaporation at 1970\u00C2\u00B0C V (sec ) 0.2 0.4 k (cmsec\"l) 4.61X10\" 3 kj (cm sec\"1) 2.59x10\"2 1.83x10\"2 k e (cm-sec~1) 5.61x10-3 6.16x10-3 y 4.95x10-2 5.43x10\"2 C m (gem\" 3 ) 0.117 ( 2.6wt%) C s ( g e m \" 3 ) 0.096 0.0 88 c s / c m 0.8 2 0.7 5 combining equations (8) and (23) results i n : n - k c m , (26) o where k has been obtained experimentally. The fluxes n in any stage of evaporation steps are identical, which gives: o n = k c m = k g c s . (27) Therefore, as k g i s known since the activity coefficient Y has been obtained, c 8 can be calculated. Table 6 includes the results of c 8 , from which the ratio of c 8 / c m is obtained as 0.75 or 0.82, indicating the desorption control. As a comparison, a practical example of an industrial electron beam hearth of surface dimension 150 x 45 cm, with a depth of liquid alloy of 10 cm is found to produce an aluminum loss of 0.52 % at an input level of 3.75 % and a flow rate of 500 kg/hr. Based on two average surface temperatures, 1800\u00C2\u00B0C and 1900\u00C2\u00B0C, y is calculated as 2.35 x 10\" 2 and 1.09 x 10~2, respectively. The result indicates a f a i r l y good agreement with the experimental values reported above, leading to confirmation of Langmuir evaporation conditions, within the errors imposed by the assumption of an average surface temperature. From the discussion above, the evaporation of aluminum from titanium alloys i s li k e l y to be controlled by the surface desorption. Now, let us consider the cases where the liquid phase mass transfer (diffusion) could predominate in controlling the rate. The factors considered that affect the evaporation mechanism are: (1) the vapour - 69 -pressure of a solute element, (2) time t', the lifetime of a liquid element on the surface, and (3) the temperature T. First of a l l , manganese is chosen as an example of (1), whose standard vapour pressure is higher than that of aluminum. Other variables that affect the evaporation rate constant are assumed to be constant: T = 1970\u00C2\u00B0C, t' = 0.2 sec (based on Machlin's model 8), D = 10\"** cm2/sec, and y =1. At the temperature of interest, p\u00C2\u00B0 = 444 mmHg, hence, k j - 2.59 x IO - 2 (28) k - 7.84 x 10 - 1, (29) then, k = 2.51 x 10\" 2 and k/^ = 0.97. (30) That i s , the liquid phase mass transfer offers 97 % of the total resistance and the surface concentration c S w i l l be only 3 % of c m . Thus, we may say that i t essentially controls the, rate. More accurately speaking, the larger is the value of the yP\u00C2\u00B0 product, the more the liquid phase mass transfer control i s l i k e l y to result, independent of absolute solute concentration. In a second case (2), we calculate the time t*, leading to the conditions under which the liquid phase mass transfer controls the rate: k^ \u00C2\u00AB k g. In this case, t' is the only variable and other conditions are the same as the previously described aluminum evaporation from titanium alloys. Then, 2 / -5-, \u00C2\u00AB k = 5.61 x 10~ 3 , (31) it t' e ' v ' - 70 -therefore, t' \u00C2\u00BB 4 sec. The consequent velocity of the liquid motion v, given as : v = r / t\u00C2\u00BB, (32) where r is the radius of the liquid surface (1.9 cm), then, v \u00C2\u00AB 0.5 cm/sec. (33) As a result, i t may be said that more stagnant liquid pool offers more probablity that the rate is controlled by the liquid phase mass transfer. The value of v here, however, i s very small from the standpoint of visual observation of electron beam melting, indicating that this rate-control is not l i k e l y to occur. In a third case (T is the only variable), k g is affected by T, but k^ remains to be constant. Again, considering the case k^ \u00C2\u00AB k g for the aluminum evaporation, kj= 2.59 x 10\" 2 (34) 2.59 x 10- 2 \u00C2\u00AB p j , (35) then, T \u00C2\u00BB 2530\u00C2\u00B0K. (36) Consequently, at 2530\u00C2\u00B0K (about 2260\u00C2\u00B0C), the liquid phase mass transfer and the desorption process are almost equivalent in their contribution to k, above which the former becomes essential. Finally, i t may be Interesting to comment in respect of solute concentration. It is clear that the concentration affects the total mass flux of a solute such that the evaporation rate decreases in accordance with decrease in the concentration. However, as k.. and k - 71 -are independent of the concentration, the mechanism of the evaporation control should not be influenced even when the bulk liquid becomes a dilute solution. 4-6. Implications of This Work for Industrial Electron Beam Melting From the above calculations and results we may derive some observations relevant to the industrial practice of electron beam melting in Ti-Al-V alloy and similar alloys. F i r s t , we can predict how much evaporation takes place and hence, the resulting composition changes i f the rate-control step and the factors relevant to the evaporation rate such as surface temperature of the melt and the value of YP\u00C2\u00B0 \u00C2\u00B0f a solute element are known. Second, i t is clear that extensive evaporation of aluminum is to be expected and that i t s rate is c r i t i c a l l y dependent on surface temperature (I.e. beam power and scanning mode); therefore, i t is necessary to apply a very constant level of beam power and scanning in order to obtain a reproducible evaporative loss. If this is done, i t i s possible to melt alloys with evaporating constituents even when the evaporation rate constant is as high as 10 - 3 cm/sec. Third, we should optimise surface area/volume/residence time parameters from the point of view of a balance between purification and solute evaporation. Of the parameters of the furnace design, these - 72 -parameters directly affect the evaporation rate and therefore are to be determined to minimize i t . From the evaporation standpoint, the surface area/volume ratio should be small and a high throughput rate w i l l be necessary i f reasonable alloy composition is to be achieved. It is to be noted that the throughput rates projected from such an analysis are much larger than those desirable from an ingot casting viewpoint 3 7. The electron beam hearth melting process w i l l therefore probably remain confined to the production of cast bars which are subsequently to be re-processed for structure control by remelting. Finally, use of EDX for aluminum composition analysis during melting may be applicable. If the technique can be defined, i t w i l l significantly assist the furnace operation in respect of composition control In the alloy, thus reducing the process impact of non -reproducible evaporative losses. - 73 -V. SUMMARY In conclusion, 1. The conditions under which the average temperatures of the molten metal surface are measured by using an optical pyrometer have been established in the electron beam furnace. The validity of the resulting temperatures was confirmed by measuring the evaporation rate (weight loss) of pure titanium over the range of experimental temperatures used. 2. Basic assumptions of the Langmuir equation are correct in the case of the evaporation of aluminum from the Ti-Al-V alloy; that i s , the evaporation takes place from a well-mixed pool of uniform composition with a very small mass transfer resistance. 3. The specific evaporation constant k and the activity coefficient Y of aluminum ln the Ti-Al-V alloy were determined as about 4.6 x 10 - 3 cm/sec and 4 x 10 - 2 at 1970\u00C2\u00B0C, respectively, the latter indicating that the deviations from Raoult's law are considerable. 4. A t r i a l of using the EDX equipment for directly measuring the composition on the molten metal surface indicates that this technique may be promising in terms of the analysis of the aluminum concentration changes during melting, although this has not yet been accomplished quan t i ta t ively. - 74 -REFERENCES 1. H.R.Smith, C.d'A.Hunt, C.W.Hanks, AIME 3rd Reactive Metals Conference, May (1952), pp.131-146. 2. T.H.Harrington, Canadian Metallurgical Quarterly, vol.10, no.2, (1971), pp.137-145. 3. H.Stephan, W.Dietrich, H.Aichert, Proceedings of 4th Internationl Conference on Vacuum Metallurgy, Tokyo, (1973), pp.306-309. 4. S.Schiller, H.Forster, G.Jasch, Metallurgia, Nov. (1980), pp.554-568. 5. R.F.Bunshah, M.A.Cocca, Techniques of Metals Research, John Willy & Sons. Inc.,New York, ed. by R.F.Bunshah, vol.1, part 2, (1968), pp.745-747. 6. R.G.Ward, T.D.Aurini, J.I.S.I., Sep. (1966), pp.920-923. 7. R.G.Ward, J.I.S.I., Jan. (1963), pp.11-15. 8. E.S.Machlin, Trans, of Met. Soc. of AIME, vol.218, Apr. (1960), pp.314-326. 9. M.Olette, Proceedings of 4th International Conference on Vacuum Metallurgy, Tokyo, (1973), pp.29-34. 10. R.Ohno, T.Ishida, J.I.S.I., Sep. (1968), pp.904-908. 11. S.Hayakawa, T.Choh, M.Inouye, Trans, of I.S.I.J., vol.22, no.8, no.8, (1982), pp.637-645. 12. V.D.Sehgal, A.Mitchell, J.I.S.I., no. 202, (1964), pp.216-220. 13. E.Ozberk, R.I.L.Guthrie, Proceedings of 6th International Conference on Vacuum Metallurgy, California, (1979), pp.248-267. 14. G.M.Gill, E.Ineson, G.W.Austin, J.I.S.I., vol.191, (1959), pp.172-175. 15. R.Ohno, Met. Trans., vol.7B, Dec. (1976), pp.647-653. 16. Y.Kenu,S.Congxi, Acta Metallurgica Sinica, vol.18, no.4, Aug. (1982), pp.411-417. 17. J.S.Foster, W.H.Pfeifer, Journal of Vacuum Science Technology, vol. 9, no.6, Nov. (1972), pp.1379-1384. - 75 -18. R.G.Ward, British Chemical Engineering, Jun. (1960), pp.401-403. 19. V.I.Yavoisky, B.V.Linchevsky, V.A.Grigorian, Priceedings of 4th International Conference on Vacuum Metallurgy, Tokyo, (1973), pp. 79-85. 20. V.N.Eremenko, G.M.Lukashenko, V.L.Pritula, ibid., pp.44-46. 21. T.Landig, R.McKoon, M.Young, Journal of Vacuum Science Technology, vol.14, no.3, May/Jun. (1977), pp.808-814. 22. L.G.Carpenter, W.N.Main, Proceedings of Physical Society, 64B, London, (1957), pp.57-66. 23. R.K.Koch, E.D.Calvert, C.R.Thomas, R.A.Beall, Bureau of Mines Report of Investigations, no.7271, J u l . (1969), pp.1-14. 24. W.E.Anable, R.K.Kock, E.D.Calvert, R.A.Beall, Trans, of the Vacuum Metallurgy Conference, (1966), pp.1-14. 25. L.R.Roller, Scientific Foundations of Vacuum Technique, 2nd ed., ed. by S.Dushman, J.M.Lafferty, John Willy & Sons. Inco., New York, (1962), pp.691-701. 26. O.Kubaschewski, C.B.Alcock, Metallurgical Thermochemistry, 5th ed., Pergamon Press, Oxford, (1979), p.374. 27. P.M.Strocchi, M.Socci, Titanium and Titanium Alloys, ed. by J.C.Williams, A.F.Belov, Prenum Press, London, (1976), vol.1, pp. 196-208. 28. Operations Manual of Automatic Optical Pyrometer Series 1100, IRCON Inco., I l l n o i s . 29. T.H.Schofield, A.E.Bacon, Journal of the Institute of Metals, vol. 82, no.54, (1953), pp.167-169. 30. R.A.Oriani, T.S.Jones, The Review of Scientific Instruments, vol.25, no.3, Mar.(1954), pp.248-250. 31. D.J.Maykuth, H.R.Ogden, R.I.Jaffee, Journal of Metals, Feb. (1953), pp.225-230. 32. M.Hansen, E.L.Kamen, H.D.Kessler, D.J.McPherson, Journal of Metals, Oct. (1951), pp.881-888. 33. I.Codoff, J.P.Nielsen, ibid., Feb. (1953), pp.248-252. 34. D.K.Deardorff, E.T.Hayes, ibid., May (1956), pp.509-511. 35. S.Dushman, Scientific Foundations of Vacuum Technique, (1962), p.692 - 76 -36. G.Scharf, F.Esser, D.Rumberg, Neue Hutte, vol.11, no.l, Jan. (1966), pp.19-27. 37. A.Mitchell, K.Takagi, 5th International Conference on Titanium, Munich, (1984), to be published. - 77 -"@en . "Thesis/Dissertation"@en . "10.14288/1.0078753"@en . "eng"@en . "Materials Engineering"@en . "Vancouver : University of British Columbia Library"@en . "University of British Columbia"@en . "For non-commercial purposes only, such as research, private study and education. Additional conditions apply, see Terms of Use https://open.library.ubc.ca/terms_of_use."@en . "Graduate"@en . "Observations on electron beam melting of titanium and its alloy"@en . "Text"@en . "http://hdl.handle.net/2429/24944"@en .