This article examines the effects of reducing, neutral and oxidizing environments and their dew points on the debinding rate in powder injection molding (PIM) processes.

The thermal decomposition of an organic binder component is an operation integral to all PIM processes and most other binder- or lubricant-assisted powder forming processes. An understanding of how to manipulate this decomposition is key to those developing, controlling or improving these processes. The furnace atmosphere represents the largest and most diverse set of variables available to manipulate thermal decomposition behavior.

The combinations and applications of different powder/binder systems are also diverse. This article examines a powder/binder system of broad applicability (316L SS powder with a polyolefin binder) to map its response to manipulation of the furnace atmosphere.

The variables considered were the type of atmosphere and the dew point. Thermo-gravimetric (TGA) can be used to indicate the onset and rate of decomposition. To quantify completeness of binder removal and relative oxygen pickup, final carbon and oxygen contents were analyzed.

The majority of previous work on thermal debinding examines lubricant removal from pressed compacts. Moyer[Ref. 1] compared air and disassociated ammonia atmospheres during the delubrication of 316L compacts and concluded that less lubricant residue remained when air was used rather than dissociated ammonia. Re-nowden and Pourtalet[Ref. 2] compared nitrogen-based atmospheres with various hydrogen and water contents during the delubrication of pressed iron compacts and concluded that water had a negative influence on binder lubricant removal. Hwang and Lin[Ref. 3] compared nitrogen and dissociated ammonia atmospheres of varied dew points and reported that high dew points did not significantly influence the debinding rate, but were effective in reducing the final lubricant residue.

An inherent difficulty that should be noted with some of these studies is the use of rapid heating rates and the high green densities of pressed articles relative to injection molded articles. Rapid heating may allow purely thermal mechanisms to prevail before atmosphere-related affects can be observed. The high green density and the low permeability of pressed compacts (compared with a PIM article with an extractable phase already removed) will also tend to minimize observable binder/atmosphere interactions. This study focused on PIM concerns and uses slower heating rates and a multiphase binder system that helps to elucidate the affect of atmosphere and dew point on debinding behavior.



Fig. 1 Thermogravimetric measurement of the decomposition of poly(vinyl butyral) in films containing various oxides heated in (a) air at 5¯C/min, and (b) argon at 5¯C/min. (Masia et al., Ref. 6).

GENERAL CONCERNS IN GRAVIMETRY

There are several other concerns in thermogravimetric practice that bare consideration in certain circumstances and applications. Three are addressed here. The first is interaction of the binder with the powder surface to yield changes in the decomposition behavior of the binder. Masia, Calvert, Rhine and Bowen[Ref. 4] examined the effect of certain oxide surfaces on the decomposition behavior of poly (vinyl butyral) in both air and argon. The specific nature of a powder surface can lead to catalytic interaction with the binder to radically alter its decomposition behavior. This interaction is more apparent in air, but is also present in argon. The consequence is that although there may be a thorough understanding of the decomposition behavior of a specific powder/binder system, this may not be transferable to another powder system. Fig. 1 shows the decomposition behavior of poly (vinyl butyral) in films containing various oxides heated in air at 5 C/min (1a) and in argon at 5 C/min (1b).

Fig. 2 Thermogravimetric decomposition of polyolefin based binder heated at 2¯/min for fine shavings compared to a cube. (Woodthorpe et al., Ref. 5).

The second concern is the size of the part being analyzed. The surface to volume ratio of the part dictates the ability of the atmosphere to interact with the binder for a given time period during the debinding cycle. Additionally, thermal lag concerns may be present as the part size and heating rate increase. Woodthorpe, Edirisinghe and Evans[Ref. 5] compared the behavior of a cube versus fine shavings for several binder formulations. A typical result of this study is shown in Fig. 2. The onset and completion of decomposition occur much earlier for the fine shavings as opposed to the bulk pieces. Again, thermogravimetric data from one specific system cannot be applied directly to another very similar system.



Fig. 3 Relative weight loss over time at different isothermal hold temperatures. The dashed line indicates the onset of the isothermal condition. (Angermann et al., Ref. 6).

The third concern is that decomposition can start to occur at temperatures lower than the isothermal hold temperature, but will proceed slowly. This can be used to optimize debinding cycles once the general decomposition behavior is understood. Anger-mann and Van Der Biest[6] observed the slow decomposition of a PP/LDPE/SA binder at several isothermal hold conditions. The rate of decomposition varied greatly depending on the temperature of the isothermal hold. Fig. 3 shows the changes in decomposition behavior with respect to different isothermal hold temperatures. A set of curves similar to this can be used to predict where and how long an isothermal hold can be useful. This point is somewhat intuitive but it is mentioned because of its importance in developing and optimizing thermal debinding cycles.

EXPERIMENTAL PROCEDURE

To examine the proposed variables, injection molded parts were thermally debound in a macro thermogravimetric analyzer under varied atmosphere conditions. The macro TGA was designed and built specifically for these studies[7] and for PIM development work. It is capable of analyzing full-size parts under actual conditions. The process gas can be reducing, neutral or oxidizing at atmospheric pressure. Part weights of up to 65g can be accommodated in a cylindrical working chamber with a diameter and height of 6 cm and 5 cm, respectively. Temperature is controlled via a thermocouple adjacent to the sample and can be profiled in multiple ramps or holds for optimization purposes. Temp-erature and weight data are recorded and exported into a spreadsheet by a data acquisition system.

To minimize powder/binder interaction, 316 stainless steel powder was used because of its relatively unreactive nature. A two-phase binder system was used system incorporating a polyolefin backbone and a solids loading of 67 volume percent. The exact details of the binder system are proprietary. The extractable phase was removed prior to processing the samples. This binder approach allowed for maximum binder/atmosphere interaction. The samples were cylindrical with a diameter of 0.6 cm and length of 1.5 cm. Three samples were processed per run.

In order to examine atmosphere effects, process gas and dew point were varied. The process gases were reducing (100% H2), neutral (100% N2) and oxidizing (20% O2, 80% N2, to simulate air). The dew point was varied between -40 C and + 20 C. The low dew point was achieved using as-received bottled gas and the high dew point was achieved by running the gas through a water bubbler. These two variables yielded a set of six conditions.

When examining thermogravimetric behavior, ramp rate is also an important consideration. Rapid ramp rates greater than 10 C/min distort the results by allowing completely thermal mechanisms to overshadow the slower atmosphere related mechanisms. Extremely slow ramps (< 1 C/min) are not especially practical. To keep the study practical and sensitive to the effects of atmosphere, a 2 C/min ramp was chosen.

Additionally, carbon and oxygen contents of the thermally debound samples were analyzed. In order to differentiate between compositional changes due to atmosphere during debinding and changes causes by atmosphere interactions after debinding, the ramps were stopped within 50 C after the cessation of weight loss. Cycles using an air atmosphere were stopped at 400 C. Cycles using N2 and H2 were stopped at 500 C.

A continuous ramp cycle was chosen, rather than a profile with ramps and isothermal holds, because the intent of this work was not to optimize any particular condition, but to understand the relative behavior between conditions.



Fig. 4 Thermogravimetric behavior of 316 MIM samples in different atmospheres using a heating rate of 2¯C/min.

RESULTS AND DISCUSSION

It is understood that some of these atmosphere manipulations could be detrimental to the post-debinding process; however, the primary intent of the study was to examine binder/atmosphere interactions, not powder/atmosphere interactions.

The thermogravimetric results from the experiments above are shown in Fig. 4. The behavior for wet and dry air is almost identical, showing an early onset of decomposition, an initially sharp rate of weight loss, and completion of debinding below 375 C. The behavior for the remaining four conditions is almost identical as well. For wet and dry conditions of both H2 and N2, there is a slow onset of decomposition followed by a more rapid rate of weight loss with completion of debinding by 475 C. The offsets of the curves are associated with an inherent drift in the load cell upon completion of debinding and can be disregarded. The thermogravimetric curves presented are intended to demonstrate the debinding rate behavior, thus the comparisons of binder removal or oxygen pick up should be made from the elemental analyses.



The carbon analyses presented in Table I can be interpreted as completeness of debinding under certain atmospheric conditions. It is important to remember that these measurements were taken shortly after debinding was complete. In actual processing, these values could be further manipulated by atmospheric conditions in the temperature ranges following debinding. Because of the ready formation of carbonaceous oxides, air-containing atmospheres were the most favorable for carbon removal. Hydrogen atmospheres were the second most effective for carbon removal, this most likely due to removal of carbon by the formation of CH4. Nitrogen based atmospheres were the least effective for carbon removal. Notably, for nitrogen, the wet condition yielded a significantly lower carbon content than the dry.

Oxygen pickup was highest for the air atmospheres. Oxygen pickup was lower for all other atmospheres. Because of the relative similarity of these values it is difficult to assign significance to the differences in oxygen content for these conditions.

Ideally, a gentle slope of the debinding curve is preferable to a sharp slope. This allows not only for more rapid heating rates during debinding but relaxes the degree of temperature control necessary by widening the processing window. Additionally, in a batch system, a more gradual burnout creates less of an impulse of effluent into the debinding trap. Also, rapid decomposition can build up defect initiating pressures inside of the parts.



Fig. 5 Thermogravimetric behavior of a blended atmosphere compared with its unblended components. A heating rate of 2 ¯C/min was used.

An interesting aspect of these experiments is the extreme difference between oxidizing and non-oxidizing conditions. Both conditions have desirable and undesirable characteristics. The beginning of the oxidizing cycles and the end of the non-oxidizing cycles exhibit undesirable rapid rates of weight loss. However, the beginning of the non-oxidizing and the end of the oxidizing cycles exhibit more desirable low rates of weight loss. A more ideal curve would encompass the initial behavior of the non-oxidizing conditions and the final behavior of the oxidizing cycles. Based on this, the idea of combining these two conditions was explored by blending nitrogen and air. Because of the demonstrated extreme affect of oxidizing environments, a nitrogen /air ratio of 10:1 was selected. Fig. 5 shows the thermogravimetric response for dry air, dry nitrogen and a 10:1 blend of nitrogen and dry air.

Fig. 5 demonstrates that blending atmospheres can combine the desirable aspects of different debinding atmospheres. The resultant curve from the blended atmosphere exhibits a more gradual rate of weight loss than those of the individual constituents. This type of behavior should provide a more robust and easier to control process. The final temperature for the blended cycle was 450ÝC, 50ÝC higher than the cycles using an air atmosphere. Because of this elevated temperature, these conditions picked up more oxygen than samples run in air.



CONCLUSIONS

The main factor affecting the onset of decomposition and the decomposition rate in the examined system was the presence of an oxidizing phase in the atmosphere. The more of an oxidizing presence in an atmosphere, the earlier and more rapid were the onset and the rate of decomposition. Dew point had a negligible effect on the onset and rate of decomposition for atmosphere conditions in the examined system.

Based on residual carbon analysis, atmospheres containing air or a partial pressure of air were the most effective for debinding. An elevated dew point may assist in the debinding in a nitrogen atmosphere when compared to a dry nitrogen atmosphere. Debinding in hydrogen is less effective than oxidizing environments, but more effective than neutral environments over the examined temperature ranges.

Reducing the oxidizing potential of an environment by the introduction of a neutral species can be used to modify the thermogravimetric behavior of a system while still providing highly effective binder removal. IH



ACKNOWLEDGMENTS

The authors wish to thank Ann Taylor and Dorothy Dye of Flowmet LLC., Deland, FL, for carbon analysis, and ElectroCopper Products LTD., San Manuel, AZ, for oxygen analysis.