Ceramic Mold Inserts

For many years, the diecasting process has been employed by the nonferrous foundry industry. In this process, castings are produced by forcing molten metal under high pressure into reusable permanent metal molds commonly called dies. Figure 1 shows the sequence of steps normally used in diecasting. Although zinc-base and aluminum alloys account for the bulk of die-cast products, some copper-base materials also are die-cast. However, because of the higher temperatures involved and because of the subsequent problems associated with die life, diecasting of copper alloys is not widely practiced. Within the last 10 years some attempts have been made to produce ferrous die-castings. But diecasting of ferrous materials has met with less success than diecasting of copper alloys because of the inherent problems associated with the use of metal dies at the higher temperatures required.

ceramic mold inserts diecasting

Despite these problems, diecasting has certain advantages over most casting processes:

  1. High production rates can be obtained, especially when multiple-cavity dies are used.
  2. Castings of greater dimensional accuracy and better surface finish can be produced.
  3. Better casting yield can be accomplished (that is, finished casting weight to total metal poured) for thick-wall castings—especially as compared with sand castings—with no risers required.
  4. Castings with thinner walls and greater length-to-thickness ratios can be produced because the dies are pressure-filled.
  5. For large production quantities diecasting results in the most economical process for producing complex shapes of nonuniform sections from alloys.
  6. Diecasting processes can be automated.

There are, however, limitations connected with the diecasting process:

  1. Size—Castings of 75 pounds or less of aluminum or zinc are usually produced.
  2. Porosity—Trapped gas and air in the die can lead to porosity in the final casting.
  3. Economy—Because of the high cost of the machine and dies, large quantities of castings are required for the process to be considered economical.
  4. Melting point—The present state of diecasting limits the process almost exclusively to nonferous alloys having melting points no higher than those of copper-base alloys.

If, however, die materials were available that could withstand repeated thermal shock and if new concepts in machine design were developed and tested, ferrous diecasting could compete favorably with other ferrous casting processes.

Work conducted in the late 1960’s indicated the feasibility of casting small amounts of ferrous metal into refractory metal dies. These refractory metal die materials were selected because of their ability to withstand high temperatures, their high thermal conductivity, and their resistance to failure by heat checking. Based on this research, two companies in recent years have been active in making commercial products by the ferrous diecasting process. In both instances standard commercial diecasting equipment has been employed with refractory metal dies. Although this work has shown that the ferrous diecasting process is feasible, major technical problems associated with die life, die design, and die parting have prevented the overall economic success of the process.

To aid in solving these technique problems, research on ferrous diecasting has been performed at two universities. At Pennsylvania State University, Draper studied the mechanisms of die failures when casting molten stainless steel in molds with refractory-coated substrates on conventional die steels, refractory metals, zirconium boride, silicon nitride composites, silicon carbide, and cermets as potential die materials. The work indicated the feasibility of using materials other than refractory metals to make ferrous diecasting a viable process.

Meanwhile, research at the Massachusetts Institute of Technology emphasized the development of a process to cast partially solid nondendritic alloy slurries. The production and casting of these special slurries has been termed rheocasting. The advantages claimed for the use of the semisolid charges are that thermal shocks to the die are reduced and that castings with lower porosity are produced because of less turbulent flow conditions.

Research conducted to date, as well as commercial ferrous diecasting practices, indicates that the major obstacle impeding growth and application of the process relates to the life of the dies and the other machine components that come in contact with the molten ferrous metals. Because of repeated severe thermal shock, the use of refractory metals for die components has met with limited success. The use of ceramic materials, ceramic metal composites, or ceramic coatings as die inserts in ferrous diecastings offers potential improvement in die life for ferrous diecasting. As a result, the research described herein was conducted to evaluate the potential use of ceramic materials as die inserts for ferrous diecastings.

The specific objectives of this research were—

  • To identify potential ceramic and composite ceramic materials that might have application as molds in the ferrous diecasting process.
  • To develop a computerized model for evaluating the possible performance of these materials for this application,
  • To evaluate under simulated diecasting conditions the actual performance of the more promising materials.

This report presents the results of this research.

Selection of Materials

A literature survey was conducted to characterize potential ceramic materials for use in ferrous diecasting die inserts. The survey included thermal and mechanical properties, thermal shock resistance, chemical reactions with molten metal, processes required to form die components, commercial availability, and relative costs of the ceramic materials. The majority of the mechanical and thermal data were obtained from “Engineering Properties of Selected Ceramic Materials”.

This project considered diecasting of ferrous metals only, hence only those ceramic materials with melting points above 2,700° F were evaluated. Originally 60 ceramic materials were listed, but those that were not commercially available, those that would react at high temperatures with molten metal or air, and those about which the data were incomplete were rejected. The list of ceramic materials was thus reduced to 40.

From the reduced list the selection of ceramic materials for preliminary evaluation was based primarily on the thermal-shock resistance factors R and R’, where

ceramic-mold-inserts-factors

The unit for R is ° F (° C) and for R’ is Btu/hr/ft (w/m).

The resistance factor R provides a measurement of the thermal gradient or change in temperature required to create a crack or thermal stress fracture in the ceramic material. R’, the product of Rk, provides a parameter to evaluate thermal shock and reflects the rate of heat transfer. In both cases, the higher the values of R and R’, the more resistant the ceramic material will be to thermal shock and/or thermal fatigue.

When the ratio of tensile or flexural strength to Young’s modulus is high, and when the coefficient of thermal expansion is low, the resistance to thermal shock is usually high. Inasmuch as R varies inversely with E and α, the product of these two values can also be used to determine the thermal shock or stress resistance of ceramic materials. The lower the Eα value, the higher the resistance.

The 40 materials evaluated were listed in descending order of the resistance factors R and R’ and ascending order of α and Eα. The 29 materials that appeared to meet requirements for use in ferrous die casting are listed in table 1.

ceramic-mold-inserts-physical-properties

Development of Computer Model

Concurrent with characterizing and selecting the ceramic materials for testing, a computer model was developed to simulate temperature profiles induced in the die during ferrous diecasting operations. The model’s original purpose was to measure the possible performance of a ceramic material without having to make large expenditures for research equipment and casting evaluation tests. However, further studies showed that this model could be used to measure possible performance in any diecasting operation regardless of the type of dies used or metal to be cast.

One-quarter-inch-thick ceramic die inserts were considered to be optimum. Thicker ceramic die inserts would produce a greater insulating effect for the heat transfer from the diecasting and slow the casting production. Thinner ceramic die inserts were not expected to have the mechanical strength necessary for multicycle operations, and in addition the thermal stresses would be greater because of the higher temperature gradients resulting from the thinner sections.

Thermal stresses at several points in the ceramic die inserts (nodes) were calculated at various time elements (0.1 second) by the computer using the temperature profile it had calculated from coefficient of thermal expansion, Poisson’s ratio, modulus of elasticity, and tensile or flexural strength property values of the ceramic insert. The metal die in all the computer model calculations was considered an infinite heat sink.

In the computer model the die insert was divided into 11 one-dimentional segments (nodes) as shown in figure 2. The temperature of each node was calculated at each 0.1-second interval, and the stress of each node in the ceramic was also calculated.

Figure 3 shows the simulated stress at various points in a ¼-inch boron nitride (BN) die insert resulting from injection of 1010 steel into a 1-inch die cavity at 2.6 seconds after metal injection. The 200-psi tensile stress shown at the ceramic insert-die interface was the maximum calculated at any point or time during the casting cycle. Since the flexural strength of BN is approximately 550 psi, its strength-maximum stress ratio is 550/200 = 2.75,

ceramic-mold-inserts-calculation

ceramic-mold-inserts-analysis-of-stress

which indicates that BN would be shock resistant and suitable for die insert material for that application.

Figure 4, on the other hand, shows the stress levels generated in a ¼-inch mullite insert at 0.1, 1, and 5 seconds. The maximum stress in the mullite during the diecasting cycle is at the ceramic insert-die interface 5 seconds after metal injection and is a tension of 3.2 x 10 4 psi. The mullite has a flexural strength of 2.5 x 10 4 psi, giving a strength-maximum stress ratio of 2.5 x 10 4 = 0.78, which indicates that mullite would be only marginally resistant to thermal fatigue and therefore not suitable for a ceramic insert application.

The stress at 11 points in the ceramic die insert was calculated at 0.1- second intervals for 10 seconds of simulated diecasting operation to obtain the maximum stress likely to be encountered for all 29 samples. The maximum

ceramic-mold-inserts-analysis-of-stress-versus-distance

stress, flexural strength, and strength-maximum stress ratio for each ceramic material are shown in table 2. The materials were classified as—

  1. Resistance to thermal fatigue if the strength-maximum stress ratio was greater than 1.0,
  2. Marginal resistance to thermal fatigue if the strength-maximum stress ratio was between 0.75 and 1.0.
  3. Will fail if the strength-maximum stress ratio is less than 0.75. Using this classification, 18 of the ceramic materials listed in table 2 are expected to fail, 5 would be marginal, and only 6 would be resistant to thermal shock.

ceramic-mold-inserts-computer-model

Ceramic Materials Testing

Small Scale

Ceramic materials were tested in a specially designed thermal-shock and fatigue testing apparatus. The apparatus shown in figure 5 consisted of

ceramic-mold-inserts-testing-apparatus

(1) a water-cooled specimen holder connected to (2) an air-actuated piston by (3) a rod assembly. The time each specimen was in contact with molten metal was adjusted by the air flow to the piston. The air flow was controlled by (4) a four-way electrically operated solenoid valve. The depth the specimen was immersed into the molten metal contained in (5) the furnace crucible was controlled to within one-eighth of an inch by adjustment of the rod assembly. The ceramic-metal interface temperature and the metal specimen holder temperature were continuously recorded during the tests. Molten-metal temperatures were recorded intermittently by using a dip-tip thermocouple. The specimen holder was made of H-13 die steel. (Figure 6 show a cutaway sketch of the

ceramic-mold-inserts-ceramic-test

specimen holder as well as the thermocouples for measuring temperatures and cooling water ports for controlling the die temperature.) The temperature of the specimen holder could be maintained as high as 1,100° F. This was done by controlling the water flow with a temperature-controlled, electrically operated two-way solenoid valve on the inlet water line. A mantle placed on top of the furnace crucible prevented air from contacting the molten metal. Furthermore, an inert atmosphere of argon was used in the tests.

The thermal fatigue tests were designed so that the total heat conducted into a sample would be virtually the same for each ceramic tested. Different cycle periods (the times the samples were in and out of the molten metal) were required for the various samples because of differences in thermal properties. The cycle period for each ceramic material was determined by the computer model. The cycle time was the time required for a 1-inch-thick steel casting to solidify and cool to 2,100° F. For each material tested the out-time in the cycle was twice the in-time to allow each ceramic to cool sufficiently before the next cycle. This timing cycle was designed to provide the required thermal shock.

The preliminary tests cycled each ceramic specimen 25 times into molten cast iron at 2,750° F with the holder maintained at 500° F. If, after 25 cycles, a specimen showed no visual evidence of thermal shock or fatigue, it was scheduled for a 1-000-cycle test. Because it was decided to test only those ceramic materials that were available either commercially or from research organizations, the 25-cycle tests were made on only 15 of the 29 ceramics listed in table 2.

These tests, listed in table 3, were also used to check and verify the accuracy of the computer model. It should be noted that the computer model was designed to give stress calculations for only one cycle and not multi-cycle. Accuracy was based on one-cycle stress calculations only. Figure 7 shows some of the ceramic materials subjected to 25-cycle tests.

ceramic-mold-inserts-fatigue-test

ceramic-mold-inserts-specimen

ceramic-mold-inserts-specimen-tested

The thermal shock-fatigue cycling apparatus was designed for specimens 1 inch long and either 1 inch in diameter or 1 inch square. However, a number of ceramic samples were available only in ¼-inch-thick plates. Therefore, the specimen holder was modified so that the specimen could be fastened with a high-temperature cement to a metal plug that fit into the holder. This arrangement allowed one side of the specimen to contact the molten metal while the other side interfaced with the cooler die metal. These two procedures tested the ceramics specimens for thermal shock but could not be used to evaluate the surface finish of the casting that came in contact with the ceramic. Therefore, an alternate procedure was devised for testing the smaller ceramic samples. The ceramic specimens were held in the mold shown in figure 8 with the bottom side of the ceramic in contact with a water-cooled steel plate and the top side exposed to the hot metal. The temperature of the steel plate was maintained at 500° F, while the temperature of the hot metal was 2,750° F. The cycle period for each pour was maintained at 2 minutes. The procedure was to (1) pour hot metal into mold covering ceramic specimen with ½-inch thickness of metal; (2) allow metal to cool to where it was a dull red; (3) strip metal from mold by lifting with tongs; and (4) ready mold for next pour. The 1- by 1- by ¼-inch specimens were tested for 250 pour cycles using this procedure.

The results of using this alternate test procedure are shown in tables 4 and 5. Table 4 lists those ceramic materials that failed this testing. The six ceramics that had passed the 25-cycle tests also passed the pour tests; they are listed in table 5. The roughness of the metal and ceramic surfaces was measured after each 50 metal pouring cycles using a profilometer. This instrument measures the surface roughness, calculates the rms (root mean square) of the peak to peak distance, and indicates the roughness in micro-inches. The surface roughness of the ceramic materials and metal samples were about the same. The highest value was still only about half the roughness of the surface of gray iron cast in fine molding sand.

The six ceramic samples that were thermal-shock and fatigue resistant in the 25-cycle and 250-pour tests were then subjected to the 1,000-cycle thermal- shock test. Of the six ceramics, only three passed this testing: 3N-M, PBN, and B4C:BN. The 70 pct Si3N4:30 pct SiC ceramic cracked after 800 cycles, and the ZRBSC-M cracked after 952 cycles. The BN did not crack, but it was severely eroded owing to the softness of this ceramic and possibly to the fact that BN starts to oxidize as low as 1,800° F.

ceramic-mold-inserts-alternate-testing-method

ceramic-mold-inserts-alternate-testing-method-with-surface-roughness

Large Scale

Based on the results of the small-scale tests, the three remaining ceramic materials—BN-M, PBN, and B4C:BN—were tested on a larger scale. The B4C:BN (50 pct B4C:50 pct BN) material was included for further evaluation because the earlier tests indicated thermal shock resistance even though the computer model calculations had indicated failure (see table 2).

A gravity-flow die unit was designed and constructed for large-scale evaluation of the candidate ceramic materials. The unit consisted of two H-13 steel dies machined to hold a 6- by 6- by ¾-inch ceramic block. The inserts were made so that the clearance between the steel dies and sides of the insert was 0.005 inch. This was sufficient clearance for the expansion of all inserts tested. Three flathead machine screws were used to insure that the inserts remained in the die when opened. A stand supported the dies and allowed rapid opening and closing of the dies.

Thermocouples were positioned to monitor and record the temperature of the dies, the ceramic insert, and the molten metal. Water channels were machined in the steel blocks so that the temperature of the die and the back of the insert could be controlled. Figure 9 shows the unit before it was put

ceramic-mold-inserts-gravity-flow

into operation. The steel dies without the graphite inserts are shown in figure 10, and the insert configuration is shown in figure 11. The same testing procedure was employed for each of the specimens. Molten cast iron was

ceramic-mold-inserts-steel-dies

ceramic-mold-inserts-graphite

poured at 2,750° F into the mold cavity formed by the inserts in the steel die blocks. The blocks were maintained at 500° F. Upon solidification, the dies were opened and the casting removed. Then the dies were closed and the cycle repeated. The ejection temperature of the castings was 2,000° F, and the cycle time was 2 minutes.

Initial tests were made using graphite inserts so that operational characteristics could be determined before the ceramic tests series. The cast iron could readily be poured into the mold in 10 seconds. The casting, approximately 4- by 4- by 1-inch, solidified rapidly and was easily ejected from the mold. The temperatures recorded during the runs compared quite closely with the temperatures calculated from the computer model. Figure 12 shows the casting unit immediately after ejection of a casting. The three candidate ceramics performed as follows.

ceramic-mold-inserts-casting

Boron Carbonitride

(50 pct B4C:50 pct BN)

Test 1.—The B4C:BN shattered after a few castings. The complete failure of this ceramic was not anticipated based on the earlier preliminary testing, although the computer model did indicate marginal success. Under actual conditions, however, the material was not resistant to thermal shock.

Test 2.—The B4C:BN was annealed at 1,000° F before and after machining in an attempt to relieve any stress formed during the machining. This ceramic insert did not shatter as it had during the previous test but cracked after six castings. It is therefore not considered to be thermal-shock resistant and not applicable for ferrous diecasting.

Further tests of boron carbonitride show that while the MOR (modulus of rupture) of 23,000 psi at 70° F, it dropped to only 1,100 psi at 2,550° F. The calculated internal stresses of 5,500 psi at 2,550° F indicated failure, since the stress was greater than the MOR. Figure 13 shows the B4C:BN insert after failure.

ceramic-mold-inserts-after-failure

Boron Nitride-Grade M (40 pct BN:60 pct SiO2)

Test 1.—One insert of the BN-M successfully completed 80 castings without any evidence of thermal fatigue or shock. The second, however, cracked.

Test 2.—One insert cracked in 44 castings while the other insert remained intact. Evaluation of the testing procedure indicated that the cracking of the BN-M can be attributed to mechanical shock possibly aided by thermal fatigue. The castings tended to stick in the BN-M mold even after using a mold wash. To remove the casting, a gentle tapping with a small hammer was necessary. It is possible that this tapping created a small crack that was further propagated by thermal action on the ceramic.

Test 3.—The compressive strength of the BN-M was good, but any mechanical shock usually resulted in a crack. Figure 14 shows one BN-M insert after

ceramic-mold-inserts-after-casting

75 castings. As can be seen, two of the ridges were fractured. The fracture occurred on the 75th or last casting. The casting stuck to the ceramic. When lightly tapped to eject the casting from the mold, the ridges fractured. The absence of other cracks in the insert supports the supposition that BN-M is thermal-shock resistant, but because of the apparent poor mechanical shock properties, is not acceptable for use in ferrous diecasting.

Pyrolytic Boron Nitride (PBN)

PBN is produced by vapor deposition at high temperature onto a graphite mandrel. Because it is a deposited material, a thickness of 0.035 inch is maximum. Therefore, the PBN inserts were made by depositing the material on graphite inserts. The graphite was coated with successive layers of PBN to a total thickness averaging 0.025 inch (0.6 mm).

The PBN-coated graphite inserts appeared to be thermal-shock resistant. A total of 205 castings were made in the gravity-flow casting unit. No cracking of the PBN coating or the graphite substrate was noted. Only one problem was encountered in this series of casting tests: a portion of the PBN coating delaminated. This delamination occurred mostly on the three ridges that formed the channels in the casting. Most of the delamination occurred in the first 25 to 40 castings. However, when the ceramic thickness was reduced by delamination to about 0.005 inch (0.125 mm), no further delamination was observed until the 205th casting. This further delamination resulted in halting this testing; however, it had been previously decided that 200 castings would be sufficient to indicate whether it was feasible to use ceramic for inserts in ferrous diecasting.

Several factors could cause the delamination, singularly or in combination:

  1. Thermal expansion differences between the ceramic and the substrate.
  2. Anisotropic thermal expansion characteristics of the PBN.
  3. The thickness of the original PBN coating.
  4. The strength of the bond between the PBN layers.

Figure 11 shows the PBN-coated graphite inserts prior to casting, while figure 15 shows the inserts after 205 castings. Castings made as the experiment progressed are shown in figure 16. The surface finish of the 200th casting is equal to the surface on the initial casting: both are excellent.

ceramic-mold-inserts-pbn-graphite

ceramic-mold-inserts-metal

Conclusions

Twenty-nine ceramic materials that appeared to meet requirements for use as molds in ferrous diecasting were originally evaluated, and 15 were tested. A computer model was developed for simulating the temperature and stress patterns induced in ceramic materials during ferrous diecasting operations. This model predicted the performance of different ceramic materials without large expenditures for full-scale casting experiments. It was further determined that this computer model can also be used to measure the performance in any diecasting operation regardless of the type of metal being die-cast or the die materials being used. The overall performance of the computer model was good. The predicted temperature within the die agreed very well with actual temperatures obtained during testing. The ratios of flexural strength to predicted maximum stress values were very good in predicting whether a ceramic would fail during diecasting. Large-scale test results showed that ferrous metal could be die-cast using ceramic inserts. The PBN-coated graphite inserts were suitable in producing ferrous diecastings with excellent surface finish even after 200 castings.

Although the results from 200 castings could not be used to determine the economical feasibility of using ceramics as inserts for ferrous diecasting, this study did indicate that ceramics could be used as die inserts. It is recommended that further studies be made to determine if ceramic coatings of the dies could be used in ferrous diecastings. Such work should include a study on the effect of casting thickness on wear characteristics and thermal resistance of coatings.