Click here to see this story as it appears in the November 2017 issue of Modern Casting

Lead has been commonly used to reduce microporosity in copper alloys. However, the toxic effects of lead on humans is well documented, and over the past few decades, organizations such as the World Health Organization and U.S. Environmental Protection Agency have been setting mandates aimed at eliminating the use of lead in copper alloys. Bismuth has been found to be a suitable replacement for lead, yet bismuth’s ability to decrease microporosity is less effective than that of lead.

Understanding the parameters that influence the behavior of bismuth in solution is paramount when attempting to optimize its porosity-reducing effects in copper alloys.

Primary sources of porosity in copper alloys are solidification shrinkage and dissolved gases. The formation of solidification microporosity in metals has been studied for decades. A method for calculating the amount of microporosity formed during solidification using the Niyama criterion has been proposed in the past. This method was shown to yield reasonable results for steel, aluminum, and magnesium alloys. More recently, Niyama was also found to have a correlation with the amount of hot tearing that occurs during solidification. The key parameters for microporosity prediction include details of the solidification such as cooling rate, thermal gradient and some material-specific information such as the solidification range, solid fraction curve with temperature, the density difference between the solid and liquid phases, and the viscosity of the liquid. Using this information, the propensity for an alloy to form micro-pores can be approximated, even with new and untested alloys.

The Niyama criterion, however, does not directly include the effect of dissolved gases on microporosity. Instead, dissolved gases reduce the critical pressure drop required to initiate microporosity. During copper melting, dissolved gases can originate from atmospheric exposure, combustion products from gas fired furnaces, or various other means. Since dissolved oxygen will displace hydrogen from the melt, foundries naturally rely on an oxidizing atmosphere to remove hydrogen. The melt is then deoxidized using additions of phosphorus, lithium, or boron to remove the oxygen. Therefore, the control of oxygen is typically critical to density, melt cleanliness, and properties. Measuring hydrogen in copper can be difficult, but hydrogen can greatly affect the mechanical properties as well as defect formation in copper alloys.

A series of casting experiments were conducted to develop thermal property data for a Cu-Bi copper alloy and produce a wedge-shaped casting for microstructural and numerical analysis. In addition, a series of solidification simulations were ran using the wedge shape geometry and a commercial simulation software. Unfortunately, neither simulation software nor commercial materials databases include bismuth-containing copper alloys. Therefore, the thermal properties from leaded and unleaded C83600 compositions were used as reference compositions with which to compare the cast Cu-Bi alloy. The combination of the experimental and simulation data will allow a conclusion on the accuracy of the simulation predictions.

The theory behind the formation of microporosity in alloy solidification is well developed. The combination of a long solidification range, low thermal gradient, high cooling rate, and dissolved gas leads to significant amount of microporosity during solidification. To study the formation of microporosity in a low lead copper alloy, a wedge casting with variation in cooling rate and thermal gradient was studied. Cooling curves of the material were obtained and simulation-based models were developed to evaluate the effects of composition on the alloy’s microstructure.

Experimental results exhibited a maximum microporosity of approximately 3.8%. In addition, the amount of the bismuth phase decreased slightly at the top of the casting. The amount also decreased in the traverse direction within the riser of the casting. The simulation predictions of porosity followed the experimental results.

Copper-Bismuth Alloy Analysis
Some thermal property data was measured from the copper-bismuth alloy for comparison to the C83600 alloy data. Cooling curves were acquired from a bismuth- containing alloy before and after deoxidation treatment. This would provide an indication of the effect of deoxidation on solidification range.  Two cooling curves were collected from the un-deoxidized metal in the furnace and two curves from three different ladles of metal after deoxidation treatments were performed.

One representative cooling curve from a bismuth-containing alloy and an integration of that curve were used to determine the liquidus and solidus temperatures. Table 1 shows the experimentally determined liquidus and solidus temperatures measured for bismuth-containing alloys and the database values for the leaded and lead/bismuth-free alloy compositions.

The measured solidification range for the bismuth-containing alloy was approximately 90°F (50°C) greater than the leaded alloy solidification range and approximately 81°F (45°C) smaller than the lead/bismuth free alloy. The liquidus of the bismuth containing alloys was lower than both the leaded and lead/bismuth free alloys. The solidus was below that of the leaded alloy but higher than the lead/bismuth free alloys. The deoxidation of the Cu-Bi alloy appeared to reduce the solidification range by approximately 10.8°F (6°C).

If solidification range is an indication of an alloy’s tendency to form shrinkage porosity, the copper-bismuth alloy should be somewhere between a leaded and unleaded C83600 alloy. Also, deoxidation should reduce the microporosity formation by some small amount.

Copper-Bismuth Wedge Analysis
Figure 1a shows a representative light micrograph of the sample taken from the Cu-Bi casting. The microporosity is clearly visible under the light microscope; however, the bismuth phase was a bit difficult to distinguish. Using back scatter electron emission and energy-dispersive X-ray spectroscopy, the bismuth phases became much more distinguishable (see Figure 1b).

After these microstructural analyses were completed, the data was compared to the simulation results to determine if the amount of microporosity that formed in the casting could be predicted accurately. There is a comparison of the volume fraction of microporosity found in the simulated unleaded C83600 and leaded C83600 to that found in the experimental Cu-Bi alloy. These comparisons are a function of height from the bottom of the casting. As the height and section thickness increased, the amount of porosity increased. The volume percent porosity in the leaded alloy was predicted to range from nearly 0% up to 1.2%.

The unleaded alloy was predicted to have more porosity with amounts ranging from 1% at the bottom of the casting up to 3.8% near the top.  The measured results from the Cu-Bi alloy matched very well with the leaded alloy at the bottom and center samples, but was closer to that of the unleaded alloy in the thicker section. Figures 2 and 3 show the macro-etched samples (R1, R2, R3, SW1, SW2, and SW3) that were taken from the casting. The etching clearly reveals a large amount of microporosity formation in the SW3 sample as well as some in the R1 and SW2 samples.

The location of the porosity in the experimental sample correlated with the simulation results. Visible areas of porosity were at the bottom of the riser samples.

The etching also revealed large grain structure within the casting. The samples taken from the riser portion of the casting were seen to have large columnar grains near its center which transitioned to equiaxed grains up the height of the casting.

The upper and lower row samples taken from the risers were compared to corresponding points in the simulated castings for microporosity values.

Analysis of Bismuth Phase
A crucial part of understanding the microporosity formation in the copper alloys was to analyze the bismuth distribution throughout the sample. The measurements of samples S1, S2, and S3 were done with backscatter electron SEM images while the measurements from the riser were made from light microscope images.

Therefore, the absolute values of the two measurements techniques may not be comparable.

However, it is possible to comment about the trends with respect to its location in the casting from each measurement type. The bismuth percentage appears to decrease as casting height increases.

The highest levels of bismuth were found near the bottom the casting with a volume fraction that was slightly above 1.5%; the amount of bismuth did decrease up the height of the casting to a volume fraction of approximately 0.83% at a height of 7 in. (17.78 cm) from the bottom of the casting. While the amount of bismuth phase decreased up the height of the casting, the amount of bismuth increased along the transverse direction. The amount of bismuth present at the edge of the riser was approximately 0.85% but it increased to approximately 1.5% at the very center of the riser. This suggests some convection currents might be in effect inside the melt, but additional measurements using mass spectroscopy method analysis is needed to be made to verify these results.

Discussion
The simulation predictions of porosity using the leaded alloy reasonably predicted the amount and location of the porosity in a lead-free Cu-Bi alloy.

Macro-etching the riser samples revealed large grains with low levels of porosity as predicted by the simulations. The longitudinal wedge sample had porosity predicted near the center of the slice at the top third of the casting. The experimental results validated these predictions.

Micrograph analysis on the microstructure of the alloy showed that the volume fraction of the bismuth phase appeared to increase as it approached the center of the riser portion of the casting. During solidification of copper-bismuth systems, the bismuth phase tended to segregate to the grain boundaries in the form of long films. The large columnar grains that solidify towards the center of the riser may provide a possible explanation for the increasing bismuth phase that occurred in the center of the riser. While there appeared to be an increase in bismuth phase transversely through the riser portion, there was a decrease in the volume percentage of bismuth up the height of the casting. Convection may be leading to bismuth enrichment near the bottom of the casting geometry.    

This article is based on paper 17-103 that was presented at the 2017 Metalcasting Congress in Milwaukee