- Email: [email protected]
Spray rolling aluminum alloy strip
Materials Science and Engineering A 383 (2004) 96–106
Spray rolling aluminum alloy strip Kevin M. McHugh a,∗ , J.-P. Delplanque b , S.B. Johnson b , E.J. Lavernia c , Y. Zhou c , Y. Lin c a
Idaho National Engineering and Environmental Laboratory, P.O. Box 1625, Idaho Falls, ID 83415-2050, USA b Colorado School of Mines, Golden, CO 80401-1887, USA c University of California, Davis, One Shield Avenue, Davis, CA 95616, USA Received 27 January 2004
Abstract Spray rolling combines spray forming with twin-roll casting to process metal flat products. It consists of atomizing molten metal with a high velocity inert gas, cooling the resultant droplets in flight and directing the spray between mill rolls. In-flight convection heat transfer from atomized droplets teams with conductive cooling at the rolls to rapidly remove the alloy’s latent heat. Hot deformation of the semi-solid material in the rolls results in fully consolidated, rapidly solidified product. While similar in some ways to twin-roll casting, spray rolling has the advantage of being able to process alloys with broad freezing ranges at high production rates. This paper describes the process and summarizes microstructure and tensile properties of spray-rolled 2124 and 7050 aluminum alloy strips. A Lagrangian/Eulerian poly-dispersed spray flight and deposition model is described that provides some insight into the development of the spray rolling process. This spray model follows droplets during flight toward the rolls, through impact and spreading, and includes oxide film formation and breakup when relevant. © 2004 Elsevier B.V. All rights reserved. Keywords: Spray rolling; Spray forming; Aluminum strip casting
1. Introduction High strength aluminum alloys such as 2124 and 7050 are used extensively for aerospace applications. Flat products are manufactured by conventional ingot metallurgical (I/M) processing. Ingots are direct-chill (DC) cast to about 0.6 m thickness, scalped, homogenized and hot rolled to the desired thickness. Following this, the material is further processed (e.g. heat treated, cold rolled to final gauge, etc.) according to temper requirements and desired properties. I/M processing remains the most reliable, versatile production method for these alloys. However, it is energy and capital equipment intensive, reflecting the need to homogenize ingots and hot work casting flaws. Twin-roll casting, was originally proposed by Bessemer in the mid 1800s [1]. It combines solidification and hot rolling in a single operation. Liquid metal is fed into the gap between large water-cooled hollow rolls, where it solidifies to ∗
Corresponding author. E-mail addresses: [email protected] (K.M. McHugh), [email protected] (J.-P. Delplanque), [email protected] (E.J. Lavernia). 0921-5093/$ – see front matter © 2004 Elsevier B.V. All rights reserved. doi:10.1016/j.msea.2004.02.041
form strip up to about 6 mm thick. Since its use began in industry, about 50 years ago, the technology has improved steadily, particularly in the last 20 years. Today, more than 150 twin-roll casters are in use worldwide [2–4]. However, twin-roll casters operate much slower than their theoretical production-rate limit to satisfy quality requirements. And, due to production rate and quality issues, commercial sheet has been limited to alloys that have a suitably narrow freezing range [2–12]. At this time, twin-roll casting (and other continuous casting approaches such as Lauener block casting, Hazelett belt, and thin slab casting) are not used commercially to process the 2124 and 7050 alloys. A new strip/sheet casting process, termed “spray rolling” (also “spray strip casting”), is currently under development at the Idaho National Engineering and Environmental Laboratory (INEEL) in a collaborative project with the University of California-Davis, Colorado School of Mines, Alcoa, Pechiney Rolled Products, Inductotherm Corp., and Metals Technology Inc. [13–17]. The general concept of spray rolling, i.e., deposition into a roll gap to form strip, is credited to A.R.E. Singer who conducted pioneering work during the 1970s [18]. In general terms, spray rolling combines features of twin-roll casting and spray forming. A schematic of
K.M. McHugh et al. / Materials Science and Engineering A 383 (2004) 96–106
97
Fig. 1. Schematic of spray rolling approach.
the approach is shown in Fig. 1. Spray rolling consists of atomizing molten metal with a high velocity inert gas, extracting most of the metal’s latent heat in-flight via convective cooling by entrained inert gas (to about 70% solid), and depositing the atomized droplets between mill rolls. The metal is consolidated into strip/sheet while still in a semi-solid and highly formable condition. As with twin-roll casting, it is believed that some solid state compaction (hot rolling) occurs as the strip advances through the roll gap. While spray rolling shares many similarities with twin-roll strip casting, there are important differences: 1. In twin-roll casting, the metal’s superheat and latent heat are dissipated almost exclusively by conductive heat transfer to water-cooled rolls. In spray rolling, convective heat transfer from atomized droplets plays a prominent role and teams with conductive transfer at the rolls to increase production rate. 2. The metal introduced to the rolls in twin-roll casting is molten, while in spray rolling, it has a “slushy” character. Solid particles in the slush act as nucleation sites, producing an equiaxed grain structure and limiting segregation (Fig. 2). Aluminum alloys with high solute content and broad freezing ranges, such as 2124, 5083, 6111, and 7050, have been successfully spray rolled. While still in the early stages of development, spray rolling shows promise for reducing strip/sheet manufacturing costs while improving quality. The inherent rapid solidification and solid solubility extension may, in the future, provide an interesting avenue for the development of alloys tailored for the process and which show unique combinations of properties. Advantages of spray rolling over conventional I/M processing appear to include cost reduction and the elimination of energy intensive unit operations such as ingot casting, homogenization and hot rolling. When compared to twin-roll strip casting, advantages appear to be the high quality and production rate, and the ability to process a broader range of alloys. The laboratory-scale strip caster at INEEL has, to date, been used to produce strips up to 200 mm wide and 1.6–6.4 mm thick. The as-spray-rolled strip is characterized by a uniform, flat profile, a very fine-grained equiaxed microstructure and the absence of porosity (Fig. 2). To date,
Fig. 2. Cross section of as-spray-rolled strip (a) 2124 Al, 12 m grain size; (b) 7050 Al, 10 m grain size.
two non-heat-treatable alloys (3003 and 5083) and three heat-treatable alloys (2124, 6111 and 7050) have been processed. The compositions and melting ranges of these alloys are summarized in Table 1. This paper summarizes processing conditions, microstructure and material properties of the heat treatable alloys 2124 and 7050. Modeling and simulation are an important part of process development, providing insight into the mechanisms controlling process performance [16,19]. In order to develop a numerical model efficient enough to be used to investigate various scenarios, while still retaining the salient features of the process, a one-way coupled multi-scale approach has been adopted. This approach is based on the selective integration of micro-scale models for sub-processes such as droplet cooling, solidification, and oxidation in flight [20] or droplet impact and spreading on the mill rolls [19]. Johnson et al. [16] have shown how these droplet-scale models can be coupled to identify the influence of process parameters on “injector-to-roll” droplet behavior. The transition from droplet-scale to spray-scale model using a sectional approach has also been documented for pre-impact behavior [15]. Table 1 Nominal composition and melting range of alloys processed at INEEL by spray rolling to date Alloy
Composition
Melting range (◦ C)
3003 5083 2124 6111 7050
Al–1.2 Al–4.4 Al–4.4 Al–0.9 Al–6.2
643–654 574–638 502–638 587–650 524–635
Mn–0.12 Cu Mg–0.7 Mn–0.15 Cr Cu–1.5 Mg–0.6 Mn Si–0.7 Cu–0.75 Mg–0.3 Mn Zn–2.3 Cu–2.3 Mg–0.12 Zr
98
K.M. McHugh et al. / Materials Science and Engineering A 383 (2004) 96–106
2. Experimental analysis Commercial 2124-T851 and 7050-T7651 plate was remelted and spray rolled. The alloys were induction melted under a nitrogen atmosphere, heated about 100 ◦ C above the liquidus temperature, and pressure-fed into an atomizer designed and built in-house. Atomized droplets were deposited into the roll gap of a 0.2 m × 0.3 m 2-HI Fenn rolling mill (Model 4-081) operating at a roll surface speed of 4.1 m/min to produce strips measuring about 100 mm wide and 2.5 mm thick. The mill has standard tool-steel mill rolls that are not water-cooled. Currently, a coiler is not installed so multiple strips about 1.8 m long are produced by starting/stopping the spray. An inert gas atmosphere within the spray apparatus minimizes in-flight oxidation of the atomized droplets. Microstructure was evaluated using an Olympus Model PME-3 metallograph and a Philips XL-30 ESEM scanning electron microscope. Tensile testing was performed using an Instron 4505 screw-driven test machine following the ASTM E-8 procedure. Differential thermal analysis was performed with a Rheometrics Model DTA 1500 using a scan rate of 20 ◦ C/min. For STEM analysis, thin slabs of 0.5–0.75 mm thickness were cut from samples of spray-rolled and commercial 2124 and ground using silicon carbide paper until approximately 0.25 mm thick. TEM blanks, 3 mm in diameter, were punched from each thinned slab and electropolished at 25 V to perforation using a solution of ethanol–5% perchloric acid held at approximately −10 ◦ C. The electropolished samples were rinsed with ethanol and examined using a Philips EM401 scanning transmission electron microscope in the STEM mode at 120 kV. Prior to tensile testing, as-spray-rolled strip was edge trimmed. Both commercial and spray-rolled 2124 strip were heat treated to the T85 and T851 tempers, with cold rolling substituted for the normal commercial practice of stretching. The following recipes were used: • solution treated (493 ◦ C, 1 h) → water quenched → aged (190 ◦ C, 12 h); • solution treated (493 ◦ C, 1 h) → water quenched→ cold rolled 5% → aged (190 ◦ C, 12 h); • solution treated (493 ◦ C, 2h) → water quenched → cold rolled 3% → aged (190 ◦ C, 12 h); • solution treated (493 ◦ C, 2h) → water quenched → cold rolled 10% → aged (190 ◦ C, 12 h); • annealed using the standard industrial practice for the alloy. Prior to heat treatment, commercial plate was machined to the same thickness as the spray-rolled strip (about 2.5 mm). Both commercial and spray-rolled samples were heat treated side-by-side in a furnace, and tensile tested at the same time. Spray-rolled 7050 alloy was heat treated to simulate the T76 and T7651 tempers using the following recipes, with 3% cold rolling substituted for the normal commercial practice
of 3% stretching: • solution treated (471 ◦ C, 2–15 h) → water quenched → aged (125 ◦ C, 12 h + 166 ◦ C, 15 h); • solution treated (471 ◦ C, 2–15 h) → water quenched → cold rolled 3% → aged (125 ◦ C, 12 h + 166 ◦ C, 15 h); 3. Theoretical analysis To model spray rolling, the process is broken up into distinct sub-processes: atomization gas flow, in-flight droplet behavior, droplet impact behavior and oxide breakup. The state of the whole spray and deposited material is reconstructed using a sectional approach and calculations for representative droplets. 3.1. Droplet in-flight behavior The in-flight behavior of each representative droplet must be calculated. This is achieved using the model proposed by Delplanque et al. [20]. To model thermal energy transfer within a single droplet, the spherically symmetric conduction equation is employed. Heat balances are applied at the interfaces formed between the oxide, solid and liquid portions of the droplet. A convective cooling condition is applied on the exterior surface of the droplet, where the convective heat transfer coefficient is estimated using the Ranz–Marshall correlation [21]. Calculation of the convective heat transfer coefficient on the droplet surface requires knowledge of the droplet/gas relative velocity. To this end, a simplified version of the particle equation of motion is used. The droplet dynamics calculations and the convective cooling condition require that the velocity and temperature of the surrounding gas be known. These fields are evaluated by numerical solution (using a computational fluid dynamics code: CFD-ACE, CFD-RC, Huntsville, Alabama) of the Navier-Stokes and energy equation for a three-dimensional, steady state flow in which the influence of the spray is neglected (one-way coupling) [17]. An example flow field is shown in Fig. 3. These three-dimensional fields are calculated once for the particular configuration considered and the data is then used as an interpolation library. 3.2. Droplet impact behavior Droplet impingement is modeled using an integral energy-conservation approach based on that taken by Madjeski [22]. Impact conditions are defined by the outcome of the in-flight model. A more detailed description of the current impact model is provided by Johnson and Delplanque [19]. To model the spreading dynamics of a droplet impinging obliquely on a solid substrate, a three-dimensional velocity field within the splat and splat shape are prescribed. These velocity fields and shapes are then used to evaluate a global
K.M. McHugh et al. / Materials Science and Engineering A 383 (2004) 96–106
99
Fig. 3. Example of simulation results for the spray-rolling flow field (a) mesh; (b) computed temperature field, and (c) pathlines.
mechanical energy balance, which yields a differential equation describing the spreading dynamics. To determine the solid fraction in the spreading splat, it is necessary to model the thermal energy transfer between the splat and substrate. Using a lumped parameter approach, the cooling rate of the droplet may be estimated. 3.3. Oxide breakup model Oxidation may occur as a result of leaks in the processing chamber or because oxygen has been purposefully added to the atomization gas in order to capitalize on the grain boundary pinning effects of oxide dispersoids to obtain a smaller grain size. At impact, the oxide film present around each droplet will fracture and produce dispersoids. Both the oxide volume fraction and the oxide dispersoid size distribution play an important roll in the microstructural and mechanical properties of the final products. The oxide film deformation and fragmentation process is described, following the model proposed by Lin et al. [23], as a uniaxial compression. The oxide film is then assumed to fracture into disks. The number of new interfaces produced is obtained by equating the energy required for their creation to that of the elastic strain at the elastic limit. Strain is then allowed to re-accumulate and the fracturing process to repeat. 3.4. Spray reconstruction As noted above, the approach chosen is based on the evaluation of spray behavior from the integration of the behavior
of individual, representative droplets [15]. To this end, the approach used by Schmehl et al. [24] for spray combustion is implemented here to reconstruct the spray. The parameters defining droplet initial conditions include: droplet diameter, velocity magnitude as well as azimuthal and tangential injection angles. Droplets in the spray are grouped in classes based on their initial size and velocity direction and magnitude and the behavior of an average droplet in each class is taken to be representative of the behavior of all the droplets in that class. The melt flow rate that each class represent is in turn calculated based on the distribution of each parameter.
4. Results and discussion 4.1. Experimental Both alloys were spray rolled at a rate of 2230 kg/h/m strip width. No major problems were encountered processing these alloys despite their broad freezing ranges. 2124 alloy did have a tendency to stick to unlubricated rolls, particularly under high load conditions. Sticking, which appears to be analogous to soldering in die casting, is overcome by applying a thin layer of graphite or other solid lubricant/parting layer to the rolls. With 7050 alloy, care must be exercised to ensure that the melt superheat temperature is not excessive, as the combination of high equilibrium vapor pressure of zinc (about 100 Torr at 750 ◦ C) together with the large increase in melt surface area during atomization, can lead to zinc depletion. Chemical analysis of the starting material
100
K.M. McHugh et al. / Materials Science and Engineering A 383 (2004) 96–106
Table 2 Tensile properties of 2124 Al Material and heat treatmenta
Ultimate tensile strength (k.s.i.)
Yield strength (k.s.i.)
Elongation at failure (%)
Commercial 2124-T851 As spray-rolled 2124 SR 2124-ST (1 h)-A COMM 2124-ST (1 h)-A SR 2124-ST (1 h)-CR 5%-A COMM 2124-ST (1 h)-CR 5%-A SR 2124-ST (2 h)-CR 3%-A COMM 2124-ST (2 h)-CR 3%-A SR 2124-ST (2 h)-CR 10%-A COMM 2124-ST (2 h)-CR 10%-A SR2124-ANN COMM-ANN
62 46 66 63 74 70 70 68 74 70 28 27
57 38 52 55 70 68 68 66 69 68 17 12
4 10 7 6 4 3 7 2 5 2 15 19
a
SR is spray rolled; ST is solution treated (493 ◦ C); CR is cold rolled; A is artificially aged (190 ◦ C, 12 h); ANN is annealed (348 ◦ C, 2 h + 232 ◦ C, 4 h).
and as-spray-rolled strip indicate that no change in chemistry occurred for either alloy during processing. Tensile results for 2124 Al, summarized in Table 2, indicate that spray-rolled strip tensile properties compare favorably to those of commercial I/M strip with the same temper. Moreover, tensile properties of spray-rolled material appear to be less sensitive to rolling direction, because the amount of rolling is very small compared to I/M processing. To help ensure equivalence when comparing tensile properties, commercial plate was machined to the same thickness as the spray-rolled strip, and was solution heat treated, cold rolled, precipitation hardened, and tensile tested
side-by-side with spray-rolled strip. Similarly, commercial and spray-rolled samples were annealed side-by-side in a furnace. Better combinations of properties were observed for both spray-rolled and commercial 2124 strip if the materials were given small cold rolling reductions after solution heat treatment due to improved nucleation of precipitates along dislocations during aging. Small but consistent improvement in tensile properties, particularly in ductility, of spray-rolled strip over commercial strip followed inclusion of a cold rolling step in the heat treatment recipes. Spray-rolled strip also appeared to benefit from a somewhat longer solution aging treatment than did the commercial strip. Annealed
Fig. 4. Cast: (a) photomicrographs of 2124 Al; (b) commercial plate (T85); (c) as-spray rolled; (d) spray rolled and heat treated (T85). Keller’s etch.
K.M. McHugh et al. / Materials Science and Engineering A 383 (2004) 96–106
Fig. 5. STEM photomicrographs of (A) commercial 2124-T851 plate; (B) as-spray-rolled 2124; and (C) spray-rolled 2124-T851 strip.
Fig. 6. EDS element maps of (a) commercial 2124 plate, and (b) spray-rolled 2124 strip.
101
102
K.M. McHugh et al. / Materials Science and Engineering A 383 (2004) 96–106
spray-rolled 2124 had lower ductility but higher yield strength than commercial 2124, suggesting that the commercial recipe that was followed should be modified somewhat to increase ductility for the spray-rolled alloy, if desired. Fig. 4 compares the microstructure of 2124 in the cast, as-spray-rolled and heat-treated conditions. Extensive segregation and shrinkage voids characterize the cast material (Fig. 4a). Commercially, the material is homogenized at high temperature for an extended period to reduce interdendritic segregation and solutionize precipitates that form during casting. In contrast, the as-spray-rolled 2124 is characterized by a fine-grained (about 10 m) equiaxed structure with relatively small constituents that tend to be located along grain boundaries (Fig. 4c). Following solution heat treatment and aging to a T85 temper, the recrystallized commercial longitudinal microstructure appears as in Fig. 4b. Spray-rolled 2124-T85 has a similar microstructure except that the recrystallized grains do not show significant directionality due to the relatively modest amount of rolling during processing (Fig. 4d). STEM analysis was performed on commercial 2124-T851 plate, as-spray-rolled 2124 strip and spray-rolled 2124-T851 strip. Representative images are shown in Fig. 5. The commercial material and spray-rolled 2124-T851 were nearly identical. Both exhibited very large grains (also see Fig. 4) that exceeded the electron transparent area. Large (>300 nm) manganese-containing precipitates and much finer (<50 nm) round precipitates were observed in both samples. In contrast, the as-spray-rolled sample was characterized by very fine (1–10 m) grains and fewer of the large manganese-containing precipitates than was found in the heat treated samples. The as-spray-rolled 2124 strip also had a relatively high dislocation density indicating that the strip experiences some degree of cold work during spray rolling.
Energy dispersive spectroscopic (EDS) element maps of commercial 2124 plate and as-spray-rolled 2124 are shown in Fig. 6. These maps show the distribution of Al, Cu, Mg, Fe, and Mn in the primary constituent phase and surrounding matrix. K-line transition signals were accumulated for about 1 h to generate them. In these maps, light colored areas indicate the presence of that particular element, while dark areas indicate the absence of the element. The main constituent phase of the commercial 2124 plate is found to be enriched in Cu, Fe, and Mn relative to the surrounding matrix, which is enriched in Al and Mg. The primary constituent phase in as-spray-rolled 2124 is found to be enriched in Cu, depleted in Al, with Mg, Fe, and Mn uniformly distributed in the matrix and constituent phases. Spray rolling conditions that were used to process 7050 alloy were identical to those used for 2124 Al. Spray-rolled strips of 2.5 mm thickness were solution heat treated, water quenched and aged to yield the T76 temper. Additional samples were cold rolled 3% following the water quench to simulate the industrial practice of stretching 3% for the T7651 temper. DTA analysis was performed on samples of solution heat treated commercial plate and spray-rolled strip (471 ◦ C soak for 2 h), and as-spray-rolled 7050. DTA is useful for analyzing heat absorption or heat evolution that accompanies phase transformations, precipitation of new phases, resolution of phases, etc. Scan data and peak assignments are summarized in Fig. 7. The scans for both solution heat treated samples were very similar. Peaks corresponding to the exothermic precipitation of phases at low temperature during aging were observed for both solution treated samples but not the as-spray-rolled 7050 strip. In addition, a small endothermic peak, assigned to the melting of nonequilibrium eutectic was seen for the as-spray-rolled 7050.
Fig. 7. Differential thermal analysis scans on spray-rolled and commercial 7050 aluminum alloy samples.
K.M. McHugh et al. / Materials Science and Engineering A 383 (2004) 96–106
103
Table 3 Tensile properties of spray-rolled and commercial 7050 aluminum strip Material and heat treatmenta
Ultimate tensile strength (k.s.i.)
Yield strength (k.s.i.)
Elongation at failure (%)
Commercial 7050-T7651 plate As spray-rolled 7050 SR 7050-ST (2 h)-A SR 7050-ST (5 h)-A SR 7050-ST (10 h)-A SR 7050-ST (15 h)-A SR 7050-ST (2 h)-CR 3%-A SR 7050-ST (5 h)-CR 3%-A SR 7050-ST (10 h)-CR 3%-A SR 7050-ST (15 h)-CR 3%-A
74 58 79 76 78 79 72 77 75 75
65 43 72 67 71 73 64 69 66 66
12 14 10 10 10 7 8 10 10 10
a
SR is spray rolled; ST is solution treated (471 ◦ C); CR is cold rolled; A is aged (125 ◦ C, 12 h + 166 ◦ C, 15 h).
The latter observation suggested that the length of time used in normal commercial practices for solution heat treatment of sheet that has been homogenized and hot rolled may not be appropriate for the as-spray-rolled material. To evaluate this, samples were soaked at 471 ◦ C for 2, 5, 10, and 15 h, water quenched and aged (125 ◦ C, 12 h + 166 ◦ C, 15 h). Additional samples were cold rolled 3% after solution treatment and aged. Tensile test results conducted on these samples are summarized in Table 3. For comparison, tensile results for commercial 7050-T7651 plate are included. To help ensure equivalence when comparing tensile properties, commercial plate was machined to the same thickness as the spray-rolled strip, and tensile tested at the same time as spray-rolled strip. Results indicate that the overall prop-
erties of the spray rolled and heat treated 7050 are similar to those of the commercial material. 3% cold reduction of the spray-rolled 7050 resulted in a small decrease in strength compared with material that was not cold rolled. This is likely due to coarsening of η precipitates that form along dislocations during aging. Fig. 8 compares the microstructure of 7050 in the cast, as-spray-rolled and heat treated conditions. Interdendritic segregation, coarse constituents, and shrinkage voids were found in the cast material (Fig. 8a) that formed during slow cooling of the alloy. As with 2124 alloy, the as-spray-rolled 7050 is characterized by a fine-grained (about 10 m) equiaxed structure with relatively small constituents that tend to be located along grain boundaries
Fig. 8. Cast: (a) photomicrographs of 7050 Al; (b) commercial plate (T7651); (c) as-spray-rolled; (d) spray rolled and heat treated (T76).
104
K.M. McHugh et al. / Materials Science and Engineering A 383 (2004) 96–106
Fig. 9. SEM photomicrographs of (a) commercial 7050-T7651 plate; and (b) as-spray-rolled 7050. Tables summarize EDS composition analysis of phases.
(Fig. 8c). Following solution heat treatment, stretching and aging to a T7651 temper, the recrystallized microstructure of the commercial 7050 appears as in Fig. 8b, when viewed in the longitudinal direction. Spray-rolled 7050-T76 has a similar microstructure to that of the commercial material except that the recrystallized grains exhibit less directionality. Also, constituents tend to be more randomly distributed and somewhat finer than in the commercial plate (Fig. 8d). The SEM photomicrographs of Fig. 9 compare the morphology and distribution of primary constituent particles in as-spray-rolled and commercial 7050 sheets. EDS analysis of the constituents (light phase) and surrounding matrix (dark phase) is summarized in the tables accompanying the photomicrographs. Constituent phases in the commercial material tended to be aligned in the direction of hot rolling. Finer constituent particle sizes were found near the surface of spray-rolled strip than in the middle due to a higher cooling rate at the rolls, and constituents near the surface were also somewhat rounder than those in the middle and tended to follow the contour of grain boundaries. As was observed with 2124, the as-spray-rolled and commercial 7050 materials showed some variation in the composition of constituent and matrix phases. In particular, the constituents in the commercial material were particularly en-
riched in Fe and Cu while those in the as-spray-rolled 7050 were enriched in Cu, Mg, and Zn. 4.2. Modeling The integrated model described above is used to evaluate the influence of process parameters on spray in-flight and impact characteristics including oxide dispersoid size distribution. A baseline case is considered first: melt mass flow rate, 0.05 kg/s; melt superheat, 100 K and gas (90%nitrogen– 10%oxygen) flow rate, 450 slpm, with an inlet temperature of 850 K. This results in a droplet mass median diameter of 61 m. The spray angle at injection is taken to be 20◦ . The predicted radial spray distribution and solid fraction at various axial locations downstream from the injection point are shown in Fig. 10. As expected the spray spreads and solidifies as the distance from the injection point increases. The results also indicate that the solid fraction of the spray is not uniform across the spray; it is larger in the core than in the periphery. This has been shown to be the result of non-uniform droplet size distribution [15]; smaller droplets, which solidify faster, follow the gas more closely and stay relatively close to the core, while larger droplets, which solidify slower, are more likely to wander away from the core.
K.M. McHugh et al. / Materials Science and Engineering A 383 (2004) 96–106
105
Fig. 10. Spray distribution and solid fraction (a) 0.1 m; (b) 0.2 m; (c) 0.3 m; and (d) 0.4 m from injection.
Fig. 11. Comparison of the spray fractional enthalpy content as a function of the distance from injection.
Fig. 12. Comparison of the spray solid fraction as a function of the distance from injection.
106
K.M. McHugh et al. / Materials Science and Engineering A 383 (2004) 96–106
Some of the process parameters have been varied to investigate their effect on spray characteristics. Each case is defined in reference to the baseline case. Only one parameter is changed in each case: (i) the inlet temperature of the gas is decreased to 600 K; (ii) the gas flow rate is increased to 600 slpm; and (iii) the melt flow rate is increased to 0.075 kg/s. Fig. 11 shows the fractional enthalpy content of the spray as a function of the axial distance from injection for the four cases studied here. The fractional enthalpy is defined as the total enthalpy of the spray over the enthalpy of the spray at injection, measured from a reference temperature of 300 K. In the baseline case about 37% of the spray enthalpy content has been extracted at 40 cm from injection. Decreasing ambient gas temperature naturally enhances the effectiveness of heat extraction by increasing the temperature difference between the spray and the ambient gas. Increasing the melt flow rate raises the average droplet size, which reduces the area-to-volume ratio and, therefore, hinders heat transfer. Accordingly, residual enthalpy is higher at any given axial location for an increased melt flow rate than in the baseline case. The effect of increased gas flow rate is less definite but seems to induce a somewhat more efficient heat transfer because of the inherent smaller mean droplet size. Similar arguments may be made for the influence of processing parameters on the solid fraction of the spray. The solid fraction of the spray is shown in Fig. 12 as a function of the axial distance from injection for each of the four cases studied here.
5. Conclusions Spray rolling is a new strip/sheet casting technology that shows promise for processing a wide variety of aluminum alloys. By combining conductive cooling by rolls, which is inherent to twin-roll casting, with convective cooling of atomized droplets, inherent to spray forming, aluminum alloys with wide freezing ranges can be processed at rates that significantly exceed those of today’s commercial twin-roll casters. Preliminary results indicate that strip tensile properties meet or exceed those of strip processed by conventional I/M while eliminating ingot casting, homogenization and hot rolling unit operations. Future development work is required, however, prior to commercialization. A Lagrangian/Eulerian model based on a sectional representation of the spray has been developed to describe the spray-rolling process from injection to impact on the rolls, including oxidation and oxide film breakup, if appropriate. The insight provided by this integrated model is directly relevant to scale-up issues.
Acknowledgements The authors gratefully acknowledge support by the Office of Industrial Technologies, Energy Efficiency and Renewable Energy, U.S. Department of Energy under grant DE-FC07-00ID13816 and the Brown Foundation at the Colorado School of Mines and the Center for Combustion and Environmental Research at the Colorado School of Mines. References [1] H. Bessemer, US Patent No. 49,053 (25 July 1865). [2] D.G. Altenpohl, Aluminum: Technology, Applications, and Environment, sixth ed., The Aluminum Association, Washington, DC, 1999, pp. 86–94. [3] Ben Q. Li, JOM 47 (1995) 29. [4] R.V. Singh, Aluminum—Rolling (Process, Principles, & Applications) The Minerals, Metals, & Materials Society, Warrendale, PA, 2000, pp. 126–137. [5] L.H. Schwartz, Adv. Mater. Process. 147 (6) (1995) 104. [6] G.S. Cole, A.M. Sherman, Mater. Charact. 35 (1995) 23. [7] J.R. Dieffenback, A.E. Mascarin, JOM 45 (1993) 16. [8] C.B. Burger, A.K. Gupta, P.W. Jeffrey, D.J. Lloyd, Mater. Charact. 35 (1995) 23. [9] A. Guillet, F.G. Hamel, in: Proceedings of the 1991 Steelmaking Conference, 1991, p. 817. [10] Ben Q. Li, JOM 47 (1995) 13. [11] M. Yun, S. Lokyer, J.D. Hunt, Mater. Sci. Eng. A 280 (2000) 116. [12] S.A. Lockyer, M. Yun, J.D. Hunt, D.V. Edmonds, Mater. Charact. 37 (1996) 301. [13] K.M. McHugh, B.R. Wickham, Kolloquiums band Spruhkompaktieren, Universitat Bremen, Band 5 (2001) 47–62. [14] K.M. McHugh, E.J. Lavernia, Y. Zhou, Y. Lin, J.-P. Delplanque, S.B. Johnson, Spray rolling aluminum strip—process development and strip properties, in: Proceedings of the Symposium on Hot Deformation of Aluminum Alloys, 132 TMS Annual Meeting &Exhibition, San Diego, CA, 2–6 March 2003, p. 443. [15] J.-P. Delplnque, S.B. Johnson, E.J. Lavernia, Y. Zhou, Y. Lin, K.M. McHugh, in: Proceedings of the Symposium on Hot Deformation of Aluminum Alloys, 132 TMS Annual Meeting & Exhibition, San Diego, CA, 2–6 March 2003, p. 433. [16] S.B. Johnson, Y. Lin, Y. Zhou, J.-P. Delplanque, K.M. McHugh, E.J. Lavernia, in: Proceedings of the International Conference on Process Modeling in Powder Metallurgy & Particulate Materials, Newport Beach, CA, 28–29 October 2002. [17] J.-P. Delplanque, S.B. Johnson, Y. Lin, Y. Zhou, N. Yang, K.M. McHugh, E.J. Lavernia, in: Proceedings of the International Conference on Process Modeling in Powder Metallurgy & Particulate Materials, Newport Beach, CA, 28–29 October 2002. [18] A.R.E. Singer, Met. Mater. 4 (1970) 246. [19] S.B. Johnson, Modeling and Numerical Simulation of Droplet Spreading and Solidification after Impact on a Solid Substrate, Master of Science Thesis, Colorado School of Mines, 2001. [20] J.-P. Delplanque, E.J. Lavernia, R.H. Rangel, J. Heat Transfer 122 (2000) 126. [21] W.E. Ranz, W.R. Marshall, Chem. Eng. Prog. 48 (3) (1952) 141. [22] J. Madejski, Int. J. Heat Transfer 19 (1975) 1009. [23] Y. Lin, Y. Zhou, E.J. Lavernia, Metall. Mater. Trans. A, submitted. [24] R. Schmehl, G. Maier, S. Wittig, in: Proceedings of the Eighth International Conference on Liquid Atomization and Spray Systems, July 2000.
Spray rolling aluminum alloy strip Kevin M. McHugh a,∗ , J.-P. Delplanque b , S.B. Johnson b , E.J. Lavernia c , Y. Zhou c , Y. Lin c a
Idaho National Engineering and Environmental Laboratory, P.O. Box 1625, Idaho Falls, ID 83415-2050, USA b Colorado School of Mines, Golden, CO 80401-1887, USA c University of California, Davis, One Shield Avenue, Davis, CA 95616, USA Received 27 January 2004
Abstract Spray rolling combines spray forming with twin-roll casting to process metal flat products. It consists of atomizing molten metal with a high velocity inert gas, cooling the resultant droplets in flight and directing the spray between mill rolls. In-flight convection heat transfer from atomized droplets teams with conductive cooling at the rolls to rapidly remove the alloy’s latent heat. Hot deformation of the semi-solid material in the rolls results in fully consolidated, rapidly solidified product. While similar in some ways to twin-roll casting, spray rolling has the advantage of being able to process alloys with broad freezing ranges at high production rates. This paper describes the process and summarizes microstructure and tensile properties of spray-rolled 2124 and 7050 aluminum alloy strips. A Lagrangian/Eulerian poly-dispersed spray flight and deposition model is described that provides some insight into the development of the spray rolling process. This spray model follows droplets during flight toward the rolls, through impact and spreading, and includes oxide film formation and breakup when relevant. © 2004 Elsevier B.V. All rights reserved. Keywords: Spray rolling; Spray forming; Aluminum strip casting
1. Introduction High strength aluminum alloys such as 2124 and 7050 are used extensively for aerospace applications. Flat products are manufactured by conventional ingot metallurgical (I/M) processing. Ingots are direct-chill (DC) cast to about 0.6 m thickness, scalped, homogenized and hot rolled to the desired thickness. Following this, the material is further processed (e.g. heat treated, cold rolled to final gauge, etc.) according to temper requirements and desired properties. I/M processing remains the most reliable, versatile production method for these alloys. However, it is energy and capital equipment intensive, reflecting the need to homogenize ingots and hot work casting flaws. Twin-roll casting, was originally proposed by Bessemer in the mid 1800s [1]. It combines solidification and hot rolling in a single operation. Liquid metal is fed into the gap between large water-cooled hollow rolls, where it solidifies to ∗
Corresponding author. E-mail addresses: [email protected] (K.M. McHugh), [email protected] (J.-P. Delplanque), [email protected] (E.J. Lavernia). 0921-5093/$ – see front matter © 2004 Elsevier B.V. All rights reserved. doi:10.1016/j.msea.2004.02.041
form strip up to about 6 mm thick. Since its use began in industry, about 50 years ago, the technology has improved steadily, particularly in the last 20 years. Today, more than 150 twin-roll casters are in use worldwide [2–4]. However, twin-roll casters operate much slower than their theoretical production-rate limit to satisfy quality requirements. And, due to production rate and quality issues, commercial sheet has been limited to alloys that have a suitably narrow freezing range [2–12]. At this time, twin-roll casting (and other continuous casting approaches such as Lauener block casting, Hazelett belt, and thin slab casting) are not used commercially to process the 2124 and 7050 alloys. A new strip/sheet casting process, termed “spray rolling” (also “spray strip casting”), is currently under development at the Idaho National Engineering and Environmental Laboratory (INEEL) in a collaborative project with the University of California-Davis, Colorado School of Mines, Alcoa, Pechiney Rolled Products, Inductotherm Corp., and Metals Technology Inc. [13–17]. The general concept of spray rolling, i.e., deposition into a roll gap to form strip, is credited to A.R.E. Singer who conducted pioneering work during the 1970s [18]. In general terms, spray rolling combines features of twin-roll casting and spray forming. A schematic of
K.M. McHugh et al. / Materials Science and Engineering A 383 (2004) 96–106
97
Fig. 1. Schematic of spray rolling approach.
the approach is shown in Fig. 1. Spray rolling consists of atomizing molten metal with a high velocity inert gas, extracting most of the metal’s latent heat in-flight via convective cooling by entrained inert gas (to about 70% solid), and depositing the atomized droplets between mill rolls. The metal is consolidated into strip/sheet while still in a semi-solid and highly formable condition. As with twin-roll casting, it is believed that some solid state compaction (hot rolling) occurs as the strip advances through the roll gap. While spray rolling shares many similarities with twin-roll strip casting, there are important differences: 1. In twin-roll casting, the metal’s superheat and latent heat are dissipated almost exclusively by conductive heat transfer to water-cooled rolls. In spray rolling, convective heat transfer from atomized droplets plays a prominent role and teams with conductive transfer at the rolls to increase production rate. 2. The metal introduced to the rolls in twin-roll casting is molten, while in spray rolling, it has a “slushy” character. Solid particles in the slush act as nucleation sites, producing an equiaxed grain structure and limiting segregation (Fig. 2). Aluminum alloys with high solute content and broad freezing ranges, such as 2124, 5083, 6111, and 7050, have been successfully spray rolled. While still in the early stages of development, spray rolling shows promise for reducing strip/sheet manufacturing costs while improving quality. The inherent rapid solidification and solid solubility extension may, in the future, provide an interesting avenue for the development of alloys tailored for the process and which show unique combinations of properties. Advantages of spray rolling over conventional I/M processing appear to include cost reduction and the elimination of energy intensive unit operations such as ingot casting, homogenization and hot rolling. When compared to twin-roll strip casting, advantages appear to be the high quality and production rate, and the ability to process a broader range of alloys. The laboratory-scale strip caster at INEEL has, to date, been used to produce strips up to 200 mm wide and 1.6–6.4 mm thick. The as-spray-rolled strip is characterized by a uniform, flat profile, a very fine-grained equiaxed microstructure and the absence of porosity (Fig. 2). To date,
Fig. 2. Cross section of as-spray-rolled strip (a) 2124 Al, 12 m grain size; (b) 7050 Al, 10 m grain size.
two non-heat-treatable alloys (3003 and 5083) and three heat-treatable alloys (2124, 6111 and 7050) have been processed. The compositions and melting ranges of these alloys are summarized in Table 1. This paper summarizes processing conditions, microstructure and material properties of the heat treatable alloys 2124 and 7050. Modeling and simulation are an important part of process development, providing insight into the mechanisms controlling process performance [16,19]. In order to develop a numerical model efficient enough to be used to investigate various scenarios, while still retaining the salient features of the process, a one-way coupled multi-scale approach has been adopted. This approach is based on the selective integration of micro-scale models for sub-processes such as droplet cooling, solidification, and oxidation in flight [20] or droplet impact and spreading on the mill rolls [19]. Johnson et al. [16] have shown how these droplet-scale models can be coupled to identify the influence of process parameters on “injector-to-roll” droplet behavior. The transition from droplet-scale to spray-scale model using a sectional approach has also been documented for pre-impact behavior [15]. Table 1 Nominal composition and melting range of alloys processed at INEEL by spray rolling to date Alloy
Composition
Melting range (◦ C)
3003 5083 2124 6111 7050
Al–1.2 Al–4.4 Al–4.4 Al–0.9 Al–6.2
643–654 574–638 502–638 587–650 524–635
Mn–0.12 Cu Mg–0.7 Mn–0.15 Cr Cu–1.5 Mg–0.6 Mn Si–0.7 Cu–0.75 Mg–0.3 Mn Zn–2.3 Cu–2.3 Mg–0.12 Zr
98
K.M. McHugh et al. / Materials Science and Engineering A 383 (2004) 96–106
2. Experimental analysis Commercial 2124-T851 and 7050-T7651 plate was remelted and spray rolled. The alloys were induction melted under a nitrogen atmosphere, heated about 100 ◦ C above the liquidus temperature, and pressure-fed into an atomizer designed and built in-house. Atomized droplets were deposited into the roll gap of a 0.2 m × 0.3 m 2-HI Fenn rolling mill (Model 4-081) operating at a roll surface speed of 4.1 m/min to produce strips measuring about 100 mm wide and 2.5 mm thick. The mill has standard tool-steel mill rolls that are not water-cooled. Currently, a coiler is not installed so multiple strips about 1.8 m long are produced by starting/stopping the spray. An inert gas atmosphere within the spray apparatus minimizes in-flight oxidation of the atomized droplets. Microstructure was evaluated using an Olympus Model PME-3 metallograph and a Philips XL-30 ESEM scanning electron microscope. Tensile testing was performed using an Instron 4505 screw-driven test machine following the ASTM E-8 procedure. Differential thermal analysis was performed with a Rheometrics Model DTA 1500 using a scan rate of 20 ◦ C/min. For STEM analysis, thin slabs of 0.5–0.75 mm thickness were cut from samples of spray-rolled and commercial 2124 and ground using silicon carbide paper until approximately 0.25 mm thick. TEM blanks, 3 mm in diameter, were punched from each thinned slab and electropolished at 25 V to perforation using a solution of ethanol–5% perchloric acid held at approximately −10 ◦ C. The electropolished samples were rinsed with ethanol and examined using a Philips EM401 scanning transmission electron microscope in the STEM mode at 120 kV. Prior to tensile testing, as-spray-rolled strip was edge trimmed. Both commercial and spray-rolled 2124 strip were heat treated to the T85 and T851 tempers, with cold rolling substituted for the normal commercial practice of stretching. The following recipes were used: • solution treated (493 ◦ C, 1 h) → water quenched → aged (190 ◦ C, 12 h); • solution treated (493 ◦ C, 1 h) → water quenched→ cold rolled 5% → aged (190 ◦ C, 12 h); • solution treated (493 ◦ C, 2h) → water quenched → cold rolled 3% → aged (190 ◦ C, 12 h); • solution treated (493 ◦ C, 2h) → water quenched → cold rolled 10% → aged (190 ◦ C, 12 h); • annealed using the standard industrial practice for the alloy. Prior to heat treatment, commercial plate was machined to the same thickness as the spray-rolled strip (about 2.5 mm). Both commercial and spray-rolled samples were heat treated side-by-side in a furnace, and tensile tested at the same time. Spray-rolled 7050 alloy was heat treated to simulate the T76 and T7651 tempers using the following recipes, with 3% cold rolling substituted for the normal commercial practice
of 3% stretching: • solution treated (471 ◦ C, 2–15 h) → water quenched → aged (125 ◦ C, 12 h + 166 ◦ C, 15 h); • solution treated (471 ◦ C, 2–15 h) → water quenched → cold rolled 3% → aged (125 ◦ C, 12 h + 166 ◦ C, 15 h); 3. Theoretical analysis To model spray rolling, the process is broken up into distinct sub-processes: atomization gas flow, in-flight droplet behavior, droplet impact behavior and oxide breakup. The state of the whole spray and deposited material is reconstructed using a sectional approach and calculations for representative droplets. 3.1. Droplet in-flight behavior The in-flight behavior of each representative droplet must be calculated. This is achieved using the model proposed by Delplanque et al. [20]. To model thermal energy transfer within a single droplet, the spherically symmetric conduction equation is employed. Heat balances are applied at the interfaces formed between the oxide, solid and liquid portions of the droplet. A convective cooling condition is applied on the exterior surface of the droplet, where the convective heat transfer coefficient is estimated using the Ranz–Marshall correlation [21]. Calculation of the convective heat transfer coefficient on the droplet surface requires knowledge of the droplet/gas relative velocity. To this end, a simplified version of the particle equation of motion is used. The droplet dynamics calculations and the convective cooling condition require that the velocity and temperature of the surrounding gas be known. These fields are evaluated by numerical solution (using a computational fluid dynamics code: CFD-ACE, CFD-RC, Huntsville, Alabama) of the Navier-Stokes and energy equation for a three-dimensional, steady state flow in which the influence of the spray is neglected (one-way coupling) [17]. An example flow field is shown in Fig. 3. These three-dimensional fields are calculated once for the particular configuration considered and the data is then used as an interpolation library. 3.2. Droplet impact behavior Droplet impingement is modeled using an integral energy-conservation approach based on that taken by Madjeski [22]. Impact conditions are defined by the outcome of the in-flight model. A more detailed description of the current impact model is provided by Johnson and Delplanque [19]. To model the spreading dynamics of a droplet impinging obliquely on a solid substrate, a three-dimensional velocity field within the splat and splat shape are prescribed. These velocity fields and shapes are then used to evaluate a global
K.M. McHugh et al. / Materials Science and Engineering A 383 (2004) 96–106
99
Fig. 3. Example of simulation results for the spray-rolling flow field (a) mesh; (b) computed temperature field, and (c) pathlines.
mechanical energy balance, which yields a differential equation describing the spreading dynamics. To determine the solid fraction in the spreading splat, it is necessary to model the thermal energy transfer between the splat and substrate. Using a lumped parameter approach, the cooling rate of the droplet may be estimated. 3.3. Oxide breakup model Oxidation may occur as a result of leaks in the processing chamber or because oxygen has been purposefully added to the atomization gas in order to capitalize on the grain boundary pinning effects of oxide dispersoids to obtain a smaller grain size. At impact, the oxide film present around each droplet will fracture and produce dispersoids. Both the oxide volume fraction and the oxide dispersoid size distribution play an important roll in the microstructural and mechanical properties of the final products. The oxide film deformation and fragmentation process is described, following the model proposed by Lin et al. [23], as a uniaxial compression. The oxide film is then assumed to fracture into disks. The number of new interfaces produced is obtained by equating the energy required for their creation to that of the elastic strain at the elastic limit. Strain is then allowed to re-accumulate and the fracturing process to repeat. 3.4. Spray reconstruction As noted above, the approach chosen is based on the evaluation of spray behavior from the integration of the behavior
of individual, representative droplets [15]. To this end, the approach used by Schmehl et al. [24] for spray combustion is implemented here to reconstruct the spray. The parameters defining droplet initial conditions include: droplet diameter, velocity magnitude as well as azimuthal and tangential injection angles. Droplets in the spray are grouped in classes based on their initial size and velocity direction and magnitude and the behavior of an average droplet in each class is taken to be representative of the behavior of all the droplets in that class. The melt flow rate that each class represent is in turn calculated based on the distribution of each parameter.
4. Results and discussion 4.1. Experimental Both alloys were spray rolled at a rate of 2230 kg/h/m strip width. No major problems were encountered processing these alloys despite their broad freezing ranges. 2124 alloy did have a tendency to stick to unlubricated rolls, particularly under high load conditions. Sticking, which appears to be analogous to soldering in die casting, is overcome by applying a thin layer of graphite or other solid lubricant/parting layer to the rolls. With 7050 alloy, care must be exercised to ensure that the melt superheat temperature is not excessive, as the combination of high equilibrium vapor pressure of zinc (about 100 Torr at 750 ◦ C) together with the large increase in melt surface area during atomization, can lead to zinc depletion. Chemical analysis of the starting material
100
K.M. McHugh et al. / Materials Science and Engineering A 383 (2004) 96–106
Table 2 Tensile properties of 2124 Al Material and heat treatmenta
Ultimate tensile strength (k.s.i.)
Yield strength (k.s.i.)
Elongation at failure (%)
Commercial 2124-T851 As spray-rolled 2124 SR 2124-ST (1 h)-A COMM 2124-ST (1 h)-A SR 2124-ST (1 h)-CR 5%-A COMM 2124-ST (1 h)-CR 5%-A SR 2124-ST (2 h)-CR 3%-A COMM 2124-ST (2 h)-CR 3%-A SR 2124-ST (2 h)-CR 10%-A COMM 2124-ST (2 h)-CR 10%-A SR2124-ANN COMM-ANN
62 46 66 63 74 70 70 68 74 70 28 27
57 38 52 55 70 68 68 66 69 68 17 12
4 10 7 6 4 3 7 2 5 2 15 19
a
SR is spray rolled; ST is solution treated (493 ◦ C); CR is cold rolled; A is artificially aged (190 ◦ C, 12 h); ANN is annealed (348 ◦ C, 2 h + 232 ◦ C, 4 h).
and as-spray-rolled strip indicate that no change in chemistry occurred for either alloy during processing. Tensile results for 2124 Al, summarized in Table 2, indicate that spray-rolled strip tensile properties compare favorably to those of commercial I/M strip with the same temper. Moreover, tensile properties of spray-rolled material appear to be less sensitive to rolling direction, because the amount of rolling is very small compared to I/M processing. To help ensure equivalence when comparing tensile properties, commercial plate was machined to the same thickness as the spray-rolled strip, and was solution heat treated, cold rolled, precipitation hardened, and tensile tested
side-by-side with spray-rolled strip. Similarly, commercial and spray-rolled samples were annealed side-by-side in a furnace. Better combinations of properties were observed for both spray-rolled and commercial 2124 strip if the materials were given small cold rolling reductions after solution heat treatment due to improved nucleation of precipitates along dislocations during aging. Small but consistent improvement in tensile properties, particularly in ductility, of spray-rolled strip over commercial strip followed inclusion of a cold rolling step in the heat treatment recipes. Spray-rolled strip also appeared to benefit from a somewhat longer solution aging treatment than did the commercial strip. Annealed
Fig. 4. Cast: (a) photomicrographs of 2124 Al; (b) commercial plate (T85); (c) as-spray rolled; (d) spray rolled and heat treated (T85). Keller’s etch.
K.M. McHugh et al. / Materials Science and Engineering A 383 (2004) 96–106
Fig. 5. STEM photomicrographs of (A) commercial 2124-T851 plate; (B) as-spray-rolled 2124; and (C) spray-rolled 2124-T851 strip.
Fig. 6. EDS element maps of (a) commercial 2124 plate, and (b) spray-rolled 2124 strip.
101
102
K.M. McHugh et al. / Materials Science and Engineering A 383 (2004) 96–106
spray-rolled 2124 had lower ductility but higher yield strength than commercial 2124, suggesting that the commercial recipe that was followed should be modified somewhat to increase ductility for the spray-rolled alloy, if desired. Fig. 4 compares the microstructure of 2124 in the cast, as-spray-rolled and heat-treated conditions. Extensive segregation and shrinkage voids characterize the cast material (Fig. 4a). Commercially, the material is homogenized at high temperature for an extended period to reduce interdendritic segregation and solutionize precipitates that form during casting. In contrast, the as-spray-rolled 2124 is characterized by a fine-grained (about 10 m) equiaxed structure with relatively small constituents that tend to be located along grain boundaries (Fig. 4c). Following solution heat treatment and aging to a T85 temper, the recrystallized commercial longitudinal microstructure appears as in Fig. 4b. Spray-rolled 2124-T85 has a similar microstructure except that the recrystallized grains do not show significant directionality due to the relatively modest amount of rolling during processing (Fig. 4d). STEM analysis was performed on commercial 2124-T851 plate, as-spray-rolled 2124 strip and spray-rolled 2124-T851 strip. Representative images are shown in Fig. 5. The commercial material and spray-rolled 2124-T851 were nearly identical. Both exhibited very large grains (also see Fig. 4) that exceeded the electron transparent area. Large (>300 nm) manganese-containing precipitates and much finer (<50 nm) round precipitates were observed in both samples. In contrast, the as-spray-rolled sample was characterized by very fine (1–10 m) grains and fewer of the large manganese-containing precipitates than was found in the heat treated samples. The as-spray-rolled 2124 strip also had a relatively high dislocation density indicating that the strip experiences some degree of cold work during spray rolling.
Energy dispersive spectroscopic (EDS) element maps of commercial 2124 plate and as-spray-rolled 2124 are shown in Fig. 6. These maps show the distribution of Al, Cu, Mg, Fe, and Mn in the primary constituent phase and surrounding matrix. K-line transition signals were accumulated for about 1 h to generate them. In these maps, light colored areas indicate the presence of that particular element, while dark areas indicate the absence of the element. The main constituent phase of the commercial 2124 plate is found to be enriched in Cu, Fe, and Mn relative to the surrounding matrix, which is enriched in Al and Mg. The primary constituent phase in as-spray-rolled 2124 is found to be enriched in Cu, depleted in Al, with Mg, Fe, and Mn uniformly distributed in the matrix and constituent phases. Spray rolling conditions that were used to process 7050 alloy were identical to those used for 2124 Al. Spray-rolled strips of 2.5 mm thickness were solution heat treated, water quenched and aged to yield the T76 temper. Additional samples were cold rolled 3% following the water quench to simulate the industrial practice of stretching 3% for the T7651 temper. DTA analysis was performed on samples of solution heat treated commercial plate and spray-rolled strip (471 ◦ C soak for 2 h), and as-spray-rolled 7050. DTA is useful for analyzing heat absorption or heat evolution that accompanies phase transformations, precipitation of new phases, resolution of phases, etc. Scan data and peak assignments are summarized in Fig. 7. The scans for both solution heat treated samples were very similar. Peaks corresponding to the exothermic precipitation of phases at low temperature during aging were observed for both solution treated samples but not the as-spray-rolled 7050 strip. In addition, a small endothermic peak, assigned to the melting of nonequilibrium eutectic was seen for the as-spray-rolled 7050.
Fig. 7. Differential thermal analysis scans on spray-rolled and commercial 7050 aluminum alloy samples.
K.M. McHugh et al. / Materials Science and Engineering A 383 (2004) 96–106
103
Table 3 Tensile properties of spray-rolled and commercial 7050 aluminum strip Material and heat treatmenta
Ultimate tensile strength (k.s.i.)
Yield strength (k.s.i.)
Elongation at failure (%)
Commercial 7050-T7651 plate As spray-rolled 7050 SR 7050-ST (2 h)-A SR 7050-ST (5 h)-A SR 7050-ST (10 h)-A SR 7050-ST (15 h)-A SR 7050-ST (2 h)-CR 3%-A SR 7050-ST (5 h)-CR 3%-A SR 7050-ST (10 h)-CR 3%-A SR 7050-ST (15 h)-CR 3%-A
74 58 79 76 78 79 72 77 75 75
65 43 72 67 71 73 64 69 66 66
12 14 10 10 10 7 8 10 10 10
a
SR is spray rolled; ST is solution treated (471 ◦ C); CR is cold rolled; A is aged (125 ◦ C, 12 h + 166 ◦ C, 15 h).
The latter observation suggested that the length of time used in normal commercial practices for solution heat treatment of sheet that has been homogenized and hot rolled may not be appropriate for the as-spray-rolled material. To evaluate this, samples were soaked at 471 ◦ C for 2, 5, 10, and 15 h, water quenched and aged (125 ◦ C, 12 h + 166 ◦ C, 15 h). Additional samples were cold rolled 3% after solution treatment and aged. Tensile test results conducted on these samples are summarized in Table 3. For comparison, tensile results for commercial 7050-T7651 plate are included. To help ensure equivalence when comparing tensile properties, commercial plate was machined to the same thickness as the spray-rolled strip, and tensile tested at the same time as spray-rolled strip. Results indicate that the overall prop-
erties of the spray rolled and heat treated 7050 are similar to those of the commercial material. 3% cold reduction of the spray-rolled 7050 resulted in a small decrease in strength compared with material that was not cold rolled. This is likely due to coarsening of η precipitates that form along dislocations during aging. Fig. 8 compares the microstructure of 7050 in the cast, as-spray-rolled and heat treated conditions. Interdendritic segregation, coarse constituents, and shrinkage voids were found in the cast material (Fig. 8a) that formed during slow cooling of the alloy. As with 2124 alloy, the as-spray-rolled 7050 is characterized by a fine-grained (about 10 m) equiaxed structure with relatively small constituents that tend to be located along grain boundaries
Fig. 8. Cast: (a) photomicrographs of 7050 Al; (b) commercial plate (T7651); (c) as-spray-rolled; (d) spray rolled and heat treated (T76).
104
K.M. McHugh et al. / Materials Science and Engineering A 383 (2004) 96–106
Fig. 9. SEM photomicrographs of (a) commercial 7050-T7651 plate; and (b) as-spray-rolled 7050. Tables summarize EDS composition analysis of phases.
(Fig. 8c). Following solution heat treatment, stretching and aging to a T7651 temper, the recrystallized microstructure of the commercial 7050 appears as in Fig. 8b, when viewed in the longitudinal direction. Spray-rolled 7050-T76 has a similar microstructure to that of the commercial material except that the recrystallized grains exhibit less directionality. Also, constituents tend to be more randomly distributed and somewhat finer than in the commercial plate (Fig. 8d). The SEM photomicrographs of Fig. 9 compare the morphology and distribution of primary constituent particles in as-spray-rolled and commercial 7050 sheets. EDS analysis of the constituents (light phase) and surrounding matrix (dark phase) is summarized in the tables accompanying the photomicrographs. Constituent phases in the commercial material tended to be aligned in the direction of hot rolling. Finer constituent particle sizes were found near the surface of spray-rolled strip than in the middle due to a higher cooling rate at the rolls, and constituents near the surface were also somewhat rounder than those in the middle and tended to follow the contour of grain boundaries. As was observed with 2124, the as-spray-rolled and commercial 7050 materials showed some variation in the composition of constituent and matrix phases. In particular, the constituents in the commercial material were particularly en-
riched in Fe and Cu while those in the as-spray-rolled 7050 were enriched in Cu, Mg, and Zn. 4.2. Modeling The integrated model described above is used to evaluate the influence of process parameters on spray in-flight and impact characteristics including oxide dispersoid size distribution. A baseline case is considered first: melt mass flow rate, 0.05 kg/s; melt superheat, 100 K and gas (90%nitrogen– 10%oxygen) flow rate, 450 slpm, with an inlet temperature of 850 K. This results in a droplet mass median diameter of 61 m. The spray angle at injection is taken to be 20◦ . The predicted radial spray distribution and solid fraction at various axial locations downstream from the injection point are shown in Fig. 10. As expected the spray spreads and solidifies as the distance from the injection point increases. The results also indicate that the solid fraction of the spray is not uniform across the spray; it is larger in the core than in the periphery. This has been shown to be the result of non-uniform droplet size distribution [15]; smaller droplets, which solidify faster, follow the gas more closely and stay relatively close to the core, while larger droplets, which solidify slower, are more likely to wander away from the core.
K.M. McHugh et al. / Materials Science and Engineering A 383 (2004) 96–106
105
Fig. 10. Spray distribution and solid fraction (a) 0.1 m; (b) 0.2 m; (c) 0.3 m; and (d) 0.4 m from injection.
Fig. 11. Comparison of the spray fractional enthalpy content as a function of the distance from injection.
Fig. 12. Comparison of the spray solid fraction as a function of the distance from injection.
106
K.M. McHugh et al. / Materials Science and Engineering A 383 (2004) 96–106
Some of the process parameters have been varied to investigate their effect on spray characteristics. Each case is defined in reference to the baseline case. Only one parameter is changed in each case: (i) the inlet temperature of the gas is decreased to 600 K; (ii) the gas flow rate is increased to 600 slpm; and (iii) the melt flow rate is increased to 0.075 kg/s. Fig. 11 shows the fractional enthalpy content of the spray as a function of the axial distance from injection for the four cases studied here. The fractional enthalpy is defined as the total enthalpy of the spray over the enthalpy of the spray at injection, measured from a reference temperature of 300 K. In the baseline case about 37% of the spray enthalpy content has been extracted at 40 cm from injection. Decreasing ambient gas temperature naturally enhances the effectiveness of heat extraction by increasing the temperature difference between the spray and the ambient gas. Increasing the melt flow rate raises the average droplet size, which reduces the area-to-volume ratio and, therefore, hinders heat transfer. Accordingly, residual enthalpy is higher at any given axial location for an increased melt flow rate than in the baseline case. The effect of increased gas flow rate is less definite but seems to induce a somewhat more efficient heat transfer because of the inherent smaller mean droplet size. Similar arguments may be made for the influence of processing parameters on the solid fraction of the spray. The solid fraction of the spray is shown in Fig. 12 as a function of the axial distance from injection for each of the four cases studied here.
5. Conclusions Spray rolling is a new strip/sheet casting technology that shows promise for processing a wide variety of aluminum alloys. By combining conductive cooling by rolls, which is inherent to twin-roll casting, with convective cooling of atomized droplets, inherent to spray forming, aluminum alloys with wide freezing ranges can be processed at rates that significantly exceed those of today’s commercial twin-roll casters. Preliminary results indicate that strip tensile properties meet or exceed those of strip processed by conventional I/M while eliminating ingot casting, homogenization and hot rolling unit operations. Future development work is required, however, prior to commercialization. A Lagrangian/Eulerian model based on a sectional representation of the spray has been developed to describe the spray-rolling process from injection to impact on the rolls, including oxidation and oxide film breakup, if appropriate. The insight provided by this integrated model is directly relevant to scale-up issues.
Acknowledgements The authors gratefully acknowledge support by the Office of Industrial Technologies, Energy Efficiency and Renewable Energy, U.S. Department of Energy under grant DE-FC07-00ID13816 and the Brown Foundation at the Colorado School of Mines and the Center for Combustion and Environmental Research at the Colorado School of Mines. References [1] H. Bessemer, US Patent No. 49,053 (25 July 1865). [2] D.G. Altenpohl, Aluminum: Technology, Applications, and Environment, sixth ed., The Aluminum Association, Washington, DC, 1999, pp. 86–94. [3] Ben Q. Li, JOM 47 (1995) 29. [4] R.V. Singh, Aluminum—Rolling (Process, Principles, & Applications) The Minerals, Metals, & Materials Society, Warrendale, PA, 2000, pp. 126–137. [5] L.H. Schwartz, Adv. Mater. Process. 147 (6) (1995) 104. [6] G.S. Cole, A.M. Sherman, Mater. Charact. 35 (1995) 23. [7] J.R. Dieffenback, A.E. Mascarin, JOM 45 (1993) 16. [8] C.B. Burger, A.K. Gupta, P.W. Jeffrey, D.J. Lloyd, Mater. Charact. 35 (1995) 23. [9] A. Guillet, F.G. Hamel, in: Proceedings of the 1991 Steelmaking Conference, 1991, p. 817. [10] Ben Q. Li, JOM 47 (1995) 13. [11] M. Yun, S. Lokyer, J.D. Hunt, Mater. Sci. Eng. A 280 (2000) 116. [12] S.A. Lockyer, M. Yun, J.D. Hunt, D.V. Edmonds, Mater. Charact. 37 (1996) 301. [13] K.M. McHugh, B.R. Wickham, Kolloquiums band Spruhkompaktieren, Universitat Bremen, Band 5 (2001) 47–62. [14] K.M. McHugh, E.J. Lavernia, Y. Zhou, Y. Lin, J.-P. Delplanque, S.B. Johnson, Spray rolling aluminum strip—process development and strip properties, in: Proceedings of the Symposium on Hot Deformation of Aluminum Alloys, 132 TMS Annual Meeting &Exhibition, San Diego, CA, 2–6 March 2003, p. 443. [15] J.-P. Delplnque, S.B. Johnson, E.J. Lavernia, Y. Zhou, Y. Lin, K.M. McHugh, in: Proceedings of the Symposium on Hot Deformation of Aluminum Alloys, 132 TMS Annual Meeting & Exhibition, San Diego, CA, 2–6 March 2003, p. 433. [16] S.B. Johnson, Y. Lin, Y. Zhou, J.-P. Delplanque, K.M. McHugh, E.J. Lavernia, in: Proceedings of the International Conference on Process Modeling in Powder Metallurgy & Particulate Materials, Newport Beach, CA, 28–29 October 2002. [17] J.-P. Delplanque, S.B. Johnson, Y. Lin, Y. Zhou, N. Yang, K.M. McHugh, E.J. Lavernia, in: Proceedings of the International Conference on Process Modeling in Powder Metallurgy & Particulate Materials, Newport Beach, CA, 28–29 October 2002. [18] A.R.E. Singer, Met. Mater. 4 (1970) 246. [19] S.B. Johnson, Modeling and Numerical Simulation of Droplet Spreading and Solidification after Impact on a Solid Substrate, Master of Science Thesis, Colorado School of Mines, 2001. [20] J.-P. Delplanque, E.J. Lavernia, R.H. Rangel, J. Heat Transfer 122 (2000) 126. [21] W.E. Ranz, W.R. Marshall, Chem. Eng. Prog. 48 (3) (1952) 141. [22] J. Madejski, Int. J. Heat Transfer 19 (1975) 1009. [23] Y. Lin, Y. Zhou, E.J. Lavernia, Metall. Mater. Trans. A, submitted. [24] R. Schmehl, G. Maier, S. Wittig, in: Proceedings of the Eighth International Conference on Liquid Atomization and Spray Systems, July 2000.