Accepted Manuscript

Title: Determining the oxidation behavior of metal powders

during heating through thermogravimetric and evolved gas
analysis using a coupled thermogravimetry-gas
chromatography-mass spectrometry technique

Author: Catherine A. Whitman Julian T. O’Flynn Addison J.

Rayner Stephen F. Corbin

PII: S0040-6031(16)30158-7
DOI: http://dx.doi.org/doi:10.1016/j.tca.2016.06.019
Reference: TCA 77543

To appear in: Thermochimica Acta

Received date: 21-9-2015

Revised date: 20-6-2016
Accepted date: 21-6-2016

Please cite this article as: Catherine A.Whitman, Julian T.O’Flynn, Addison J.Rayner,
Stephen F.Corbin, Determining the oxidation behavior of metal powders during
heating through thermogravimetric and evolved gas analysis using a coupled
thermogravimetry-gas chromatography-mass spectrometry technique, Thermochimica
Acta http://dx.doi.org/10.1016/j.tca.2016.06.019

This is a PDF file of an unedited manuscript that has been accepted for publication.
As a service to our customers we are providing this early version of the manuscript.
The manuscript will undergo copyediting, typesetting, and review of the resulting proof
before it is published in its final form. Please note that during the production process
errors may be discovered which could affect the content, and all legal disclaimers that
apply to the journal pertain.
Determining the oxidation behavior of metal powders during heating through thermogravimetric

and evolved gas analysis using a coupled thermogravimetry-gas chromatography-mass

spectrometry technique

Catherine A. Whitman, Julian T. O’Flynn, Addison J. Rayner, Stephen F. Corbin

Department of Civil and Resource Engineering and Institute for Research in Materials, Dalhousie
University, Halifax, Nova Scotia, Canada, B3H 4R2 Ph: (902)-494-2929, cathywhitman@dal.ca

1
HIGHLIGHTS
 Evolved gas analysis of superalloy powders using coupled TG-GCMS method
 Develop techniques to study powder injection moulding (PIM) part processing
 Determine oxidation of superalloy metal powders in industrially relevant processing conditions

2
Abstract

The oxidation behavior of a fine Ni superalloy (NiSA) powder, Inconel 625, within an alumina support

powder during heating has been determined using combined thermogravimetry (TG) and evolved gas

analysis (EGA). A gas chromatographic-mass spectrometer (GCMS) was used for EGA. The GCMS

was able to semi-quantitatively measure the rate at which air was removed from the furnace due to a He

purge, where no pre-evacuation was used. It also was able to measure the effectiveness of a three cycle

evacuation and He purge in reducing the oxygen content in the furnace and identify an important

alumina off-gassing process that exposed the NiSA powder to H2O and CO2. The parallel use of the TG

signal was able to quantitatively determine the relationship between NiSA oxidation and relative

concentrations of the oxidizing gases created by the differing furnace pre-conditions and the alumina

powder off-gassing process. It was concluded that the developed TG-GCMS method can be widely

applied to quantitatively study oxygen pick-up in metal powders under industrially relevant processing

conditions.

Keywords: TG-GCMS; evolved gas analysis; superalloy; oxidation

1. Introduction

During powder metallurgy (PM) processing, metallic alloy powders are heated to high

temperatures under controlled atmospheres to initiate debinding, sintering and densification. With an

increasing emphasis on the use of micron, submicron and nano scale sized powders, contamination of

the powders by reaction with the sintering atmosphere and support materials has become increasingly

important to consider [1]. This is particularly true for materials containing highly reactive alloying

elements such as Al, Cr, Mn and Ti.

3
 In the particular case of the PM sintering of steels, a number of studies have examined the

influence of sintering atmosphere on the surface oxide present on alloyed powders [2–8]. The emphasis

of this work has been placed on determining atmosphere conditions that would lead to a reduction of Cr

and Mn based oxide films present on the surface of the as-processed powders, in order to enhance

sintering. Some of these investigations have used post-sintering chemical and microstructural analysis to

infer the influence of sintering atmosphere. Other studies have used in-situ thermogravimetric analysis

(TG) to better understand the time-temperature mass change of the sample due to reduction/oxidation

reactions resulting from the furnace atmosphere.

The current material of interest for this study is a Ni superalloy (NiSA) powder. The sintering

behavior of Ni-based superalloys has been studied by a number of researchers over the last 25 years [9–

22]. The focus of these investigations has been to develop sintering profiles that are capable of achieving

high density components with acceptable mechanical properties. Unlike the sintering of PM steels, a

detailed study of the influence of sintering atmosphere on superalloy PM materials has not been reported

in the literature as far as the authors are aware. This aspect of processing superalloys is important given

the fact that their compositions include measureable amounts of reactive metals including Al, Cr, Mn

and Ti.

The overall objective of this research is to develop advanced thermal analysis tools and

techniques and apply them to a study of Ni based superalloy powder injection molded (PIM) part

processing. The processing steps of interest include thermal debinding, pre-sintering, cleaning and

sintering. During the thermal debinding stage, the binder system originally used to fluidize the metal

powder for injection molding is removed from the part through melting and evaporation. Alumina

powder is used to give structural support to the PIM part as it loses binder and to aid in wicking the

binder [23]. Removal of all binder phases too quickly can result in distortion and cracking of the part

while residual binder phases can contaminate and hinder sintering if present post-debind. Therefore, a

4
thorough understanding of the de-binding stage is critical to the success of the PIM process [23]. The

specific objective of the current work is to perform an in-depth analysis of the role of furnace

atmosphere and presence of alumina during the debinding process on the superalloy powder itself, i.e.

the binder was not used in these experiments to reduce the complexity of interpretation.

While TG work applied to sintering has been capable of providing detailed insight into the

process, the inability of obtaining parallel in-situ measurements of the composition of the gas

atmosphere within the furnace limits its capability. TG analysis coupled with measurements of evolved

gas using mass spectrometry (MS), Fourier-transform infrared spectroscopy (FTIR) or gas

chromatography (GC) have proven to be very effective in the study of decomposition gases in organic

and inorganic materials [24–32]. A primary goal of this research was to develop a methodology which

utilized the EGA approach, in this case using GCMS, to determine the changing furnace gas

composition as a result of reactions with a metal powder. The combined use of TG and GCMS has the

potential to simultaneously determine the extent of oxidation of a metal powder but also the reaction

mechanisms involved.

2. Experimental

2.1 Materials

The characteristics of the NiSA and alumina powders used in this study are typical of those used

in the powder injection molding (PIM) industry [23] and are listed in Table 1. The superalloy powder

described herein as NiSA was Inconel 625 with a particle size of 22 µm, obtained from Sandvik Osprey.

The alumina powder was obtained from Almatis GmbH at 0.999 mole fraction purity. Helium gas from

Air Liquide with 0.99999 mole fraction purity was used for the TG-GCMS experiments.

2.2 TG-GCMS Procedure

5
 The temperature-dependent off-gassing and oxidation behavior of the alumina and NiSA

powders were studied using a commercially available thermogravimetry - gas chromatography/mass

spectrometry instrument herein called a TG-GCMS. The ultimate goal of this work was to determine the

oxidation behavior of a fine metal powder in the context of a PIM thermal debinding operation, where

the furnace atmosphere is nominally an inert gas environment. The arrangements of interest are

illustrated in Fig. 1 as a schematic of the TG crucible and powders studied, with Fig. 1d giving an

approximation of a thermal debinding operation.

To simulate an as-moulded PIM component being debound in alumina powder, loose NiSA

powder was poured into an indentation in the alumina powder bed and then covered with more alumina

powder. This alumina and NiSA powder arrangement was contained in an open alumina crucible with a

maximum capacity of 2 g or 3.4 mL (empty crucible dimensions shown in Fig. 1a). A Netzsch STA 449

F1 coupled to an Agilent 5975E GCMS was used for TG-GCMS analysis. All TG experiments were

performed at atmospheric pressure. The standard uncertainty, u, of the temperature is the larger of 1.5 °C

or u(T) = 0.0025×T. The use of a relatively large volume crucible allowed the embedded article

arrangement of Fig. 1d to be constructed using an adequately-sized sample. The level of gas evolution

from the powders and oxidation in a nominally inert gas atmosphere was expected to be very small.

Therefore gram-sized samples were required to obtain the TG and GCMS results.

Correct interpretation of the influence of the furnace atmosphere and alumina on the oxidation of

the NiSA required a series of comparative experiments to first be performed. These included full TG-

GCMS measurements obtained on an empty crucible (Fig. 1a), alumina only in the crucible (Fig. 1b)

and NiSA only in the crucible (Fig. 1c).

In a full PIM process the green body consists of fine powder embedded in a binder mixture

consisting of several organic components. In this case one of the roles of the alumina powder is to wick

the organic binders out of the PIM part during heating. In the current study, the binder component was

6
omitted in order to fully develop the methodology and realize the full capability of the TG-GCMS

analysis. Future studies will investigate the more complex case of the presence of binder.

For EGA using TG-GCMS, the samples were heated in a dynamic atmosphere which simulated a

thermal debinding heating profile. A decrease in mass indicated evaporation or decomposition while an

increase in mass indicated that the material was ad- or absorbing or reacting with the gas in the furnace.

For example, if oxygen is present in the furnace, the NiSA could oxidize resulting in a measurable mass

change. The STA 449 has a TG resolution of 0.025 µg. The standard uncertainty in mass was

determined from four empty crucible runs and was found to be a maximum of ±0.025 mg at 848 °C. The

TG furnace was automatically evacuated and refilled with inert gas a set number of times (up to 5) in

advance of heating. The system used a rough vacuum pump to pump down to ~10-2 mbar before refilling

with the selected purge gas. The inert atmosphere (99.999% pure helium in this case to allow for optimal

GCMS analysis) was used to refill during the evacuation/refill process. Helium was used as the dynamic

atmosphere flowing into the furnace at 70 mL min-1 for 5 to 10 minutes before the start of the

experiments to allow for the TG signal to stabilize. The flow rate of He was maintained at 70 mL min -1

for the duration of the experiment.

TG-GCMS analysis was carried out for all four of the configurations of Fig. 1 while heating up

to 850 °C at 4.5 °C min-1, a rate typical of an industrial PIM process [23]. Under this heating profile

experiments were carried out with no evacuation (0×EF) and with 3 evacuation/inert gas refill cycles

(3×EF) to determine the effect of atmosphere on oxidation behavior. In the particular case of alumina

only (Fig. 1b), further evacuation cycles of 1×EF and 5×EF were carried out to examine off-gassing in

more detail.

The large alumina crucible used in these experiments can adsorb air and will gain mass in humid

conditions, so care was taken to keep the crucible as dry as possible to obtain repeatable furnace

atmosphere conditions during heating. The crucible was conditioned at the beginning of a set of

7
experiments and after each sample to prevent air and/or sample contamination. The conditioning

procedure consisted of heating the sample to 850 °C at 40 °C min-1. A correction file of the empty,

conditioned crucible in the same heating profile used for the samples (i.e. 4.5 °C min-1 to 850 °C) was

also performed. This correction took into account the temperature-dependent buoyancy effects of the

gases in the furnace and was subtracted from subsequent runs.

The GCMS was programed to sample the TG furnace atmosphere at set intervals throughout an

experiment. The sampling interval used in this investigation was 2 minutes, which allowed enough time

for the gas sample to pass into the GC column before another sample was acquired. During sampling, a

small volume of atmosphere is drawn through a heated capillary (250 °C) using a vacuum and is

transferred to the inlet of the GC. The sampled gas is mixed with the GC carrier gas (99.999% He), then

passes through the capillary column of the GC which, when under the optimal conditions, can separate

the sample into individual chemical components based on properties such as polarity and molecular

mass. A HP-5ms GC column (30 m long, 0.25 mm ID was used in these experiments. The GC

temperature was kept at 250 °C and separation of the individual chemical components did not occur in

the column. The sample further continued to the MS, where it was ionized, underwent fragmentation,

mass filtration (based on the mass to charge ratio (m/z) of the ions) and detection. The end result was a

collection of time-dependent spectra for a range of detected m/z values which can be analyzed to

identify the chemical components within the evolved gas sample. This information was used to

determine qualitatively how the concentration of these components changed during the TG experiment.

This extracted ion chromatogram (EIC) took the form of the intensity of detection by the mass

spectrometer (relative abundance or abundance) of a particular ion of m/z value as a function of time

(i.e. an abundance vs time graph for the ion of interest). The abundance is a count related to the number

of times an ion of m/z strikes the detector. Relative abundance is calculated by determining the most

abundant ion, assigning it a relative abundance of 1 and displaying all other ions as a fraction of the

8
most abundant ion. In this study, m/z = 28 from N2+ at the start of the experiment is the most abundant

ion. Consequently, the term relative abundance used in this study refers to being relative to N2+ at time =

0. The EIC’s were simplified by determining the maximum signal corresponding to each 2 minute

injection, removing the baseline arising from the carrier gas, and plotting these as a continuous line.

3. Results and Discussion

3.1 Influence of evacuation procedure on furnace atmosphere

GCMS analysis revealed that, aside from the He used as carrier and purge gas, other atmospheric

gases were present to various degrees depending on the experimental condition, namely, m/z = 18

(H2O+), m/z = 28 (N2+), m/z = 32 (O2+), and m/z = 44 (CO2+). The positive charges on the ions will be

omitted going forward for simplicity. Fig. 2 and Fig. 3 give the EIC of the four atmospheric gases

tracked for the empty crucible condition with no evacuations and three evacuation/refill cycles,

respectively. Typically the mass change of the crucible was < ±0.05 mg throughout the experiment.

The starting abundance at zero time represents the composition of the atmosphere within the

furnace after the evacuation and refill process and 5 to 10 minutes of flowing purge gas. This abundance

varied depending on the laboratory atmospheric condition on the day of the experiment for a given

evacuation process. However, the rate of atmosphere reduction due to the He purge gas (i.e. the shape of

the abundance-time curve) was very repeatable.

As expected for the no evacuation condition (Fig. 2), measurable amounts of air remain at the

beginning of the experiment since the He gas flow has had limited time to purge the furnace. O2 and N2

are the most abundant gases with comparatively smaller amounts of H2O and CO2. Note that the relative

ratio of ~78% N2 and 21% O2 confirms that this is air remaining in furnace chamber. A rapid decrease in

the abundance of all four gases occurs within the first 100 minutes of the heating profile up to 500 °C as

9
the He purge displaces the air atmosphere. However, a residual air atmosphere persists up to the peak

temperature of 850 °C.

For the 3×EF condition (Fig. 3) a small decrease in the atmosphere also occurs in the first 100

minutes as a result of the He purge. A steady state level of O2, N2, H2O and CO2 is obtained after the

100 minute mark. It is emphasized that the relative abundance measurements of Figs. 2 and 3 are both

based on the N2 abundance measurement made in each condition. For example, the relative abundance

measurements for N2 and O2 in both the 0×EF and 3×EF cases begin at 1.00 and 0.28 respectively.

However, the absolute abundance of N2 and O2 for the 3×EF case was ≈15× lower due to the evacuation

cycles. Comparing the measurements for H2O and CO2 in Figs. 2 and 3 reveal that these gases are in a

higher relative concentration in the 3×EF case, indicating they are less easily removed by the evacuation

cycles.

It is important to note that the rate of atmosphere decrease in both the 0×EF and 3×EF conditions

was repeatable over multiple runs with the empty crucible. As such, this condition was able to generate a

characteristic “background” abundance profile that could be used for comparison purposes with the runs

containing alumina, NiSA and NiSA in alumina. The 3×EF condition (Fig. 3) did give much more

consistent results for the repeats than 0×EF (Fig. 2) as should be expected since the atmosphere would

rely less on day-to-day conditions and efficiency of sampling vacuum that pulls samples into the GC

column.

3.2 Off-gassing behavior of the alumina powder support material

A series of experiments designed to investigate the off-gas behavior of the alumina powder under

different evacuation/refill cycles was also completed. In all these cases, the alumina powder was

sampled directly from the as-received bottle. The TG and GCMS results for 0×EF and 3×EF are shown

in Fig. 4 and 5, respectively. Included in the EIC plots of Fig. 4 and 5 are the atmospheric gas levels

10
determined for the empty crucible condition. Any deviations from this “background” data could then be

directly attributed to the presence of alumina powder in the crucible.

In both experiments a nearly identical, small mass loss of ~ 0.1% (1×10-3 g for ~1 g initial mass)

was experienced by the alumina powder (Fig. 4a and 5a). The mass loss was attributed to the off-gassing

of H2O (m/z = 18) and CO2 gases (m/z = 44) since these exhibit a major difference for the alumina

compared to the background (i.e. empty crucible) in Fig. 4b, 4e, 5b and 5e. For the alumina sample, the

H2O and CO2 start to evolve almost immediately, peak at about 130 °C and finish at about 320 °C. The

evolution of H2O continues gradually until about 850 °C. For CO2, a second evolution starts at 340 °C

and continues to 660 °C.

There is very little difference in the shape of relative abundance vs time for the N2 (m/z = 28)

and O2 gases (m/z = 32) for the alumina compared to the empty crucible (Fig. 4c, 4d, 5c, 5d). Again note

more of a difference in 0×EF than 3×EF as show in Fig. 2 and 3. The abundance profile and magnitude

for N2 and O2 in the presence of alumina also overlap with the background levels measured in the empty

crucible case for both the 0×EF and 3×EF conditions, indicating that the alumina has no effect on the

abundance of these two atmospheric gases.

A major influence that the evacuation cycle has on the composition of the furnace atmosphere is

clear when comparing the relative abundance measurements for H2O and CO2 in Figs. 4 and 5. During

the peak evolution of these off-gases at ≈20 minutes, the relative concentration of water and carbon

dioxide is ≈1% of the N2 level in the 0×EF case. Conversely this relative concentration increases to

approximately 30% for the 3×EF condition, the same level as the O2 abundance. This further confirms

that the evacuation cycle preferentially removes N2 and O2, while not removing H2O and CO2.

To further investigate this marked difference between the H2O/CO2 and N2/O2, additional

evacuation cycles of 5×EF, 1×EF and 1×EF with a 2 hour evacuation of the furnace were performed.

Fig. 6 presents the mass change and extracted ion chromatogram for m/z = 18, 28, 32 and 44 for alumina

11
powder for the different evacuation and fill cycles using He as purge gas. These were presented as

abundance per mass of alumina vs time instead of relative abundance vs time to highlight the influence

of alumina. The abundance of m/z = 44 and 18 do not change with evacuation method, (Fig. 6e and b),

confirming that, since the thermally activated off-gassing is similar, the evacuation procedure was

unable to remove these gases from the alumina prior to heating. The reproducibility of the H2O and CO2

abundance curves illustrate that no level of evacuation steps tested were effective in removing these

gases.

The m/z = 28 and 32 abundance curves of Fig. 6c and d do change with evacuation method.

However, the data indicates that evacuation acts rapidly and only an 3×EF or 1×EF + 2hr. procedure in

needed to bring the N2 and O2 within the furnace down to its minimum level.. Some variability in the

data is expected due to the different humidity conditions present in the room when the test was

performed and MS detector variability. Taking that into account, the largest deviation in the mass

change over the 5 data sets occurs at 180 min where the average and standard deviation is 99.908 ±

0.007% (Fig. 6a). The 0×EF trial does experience a greater mass drop at low temperature (<100 °C)

which could be attributed to loss of weakly bound water due to thermal activation. In the 1×EF to 5×EF

cases this water is likely removed by the evacuation prior to heating. After this initial mass drop, the

0×EF trial follows a mass loss profile similar to average for the remainder of the temperature cycle.

Overall, the results show that the evacuation of the furnace using a roughing pump vacuum (i.e.

≈ 10-2 bar) and inert gas refill was not able to cause a significant removal of the H2O and CO2 gases

from the alumina powder surface. This indicates that the H2O and CO2 are bound strongly enough to the

alumina powder surface that thermal activation is needed to initiate the off-gassing process.

An additional feature of all the EIC abundance profiles for alumina is a distinctive decrease in

the H2O and CO2 levels in the furnace atmosphere upon cooling from 850 °C (see for example Figs. 4

and 5 b and e). In the case of the H2O abundance and to a lesser extent with the CO2, the value actually

12
drops below the background level measured in the empty crucible case (i.e. the ambient moisture level

in the purge gas stream). This indicates that, when the thermal activation is removed, the alumina

powder re-adsorbs H2O and CO2. Since the alumina was made very dry during the thermal cycle to 850

°C, it has the capacity to re-adsorb gases to the point where it scrubs or getters both H2O and CO2 from

the ambient atmosphere. This observation is further supported by the TG signal (Fig. 4a, 5a and 6a)

where a small mass gain occurs in the alumina powder during the cooling segment at times above 180

minutes.

This surface re-adsorption of alumina was even more pronounced when the alumina powder was

re-introduced to the laboratory atmosphere after a thermal debinding cycle. More specifically it was

observed that the alumina increased in mass when re-exposed to laboratory air following a TG-GCMS

experiment. This process was investigated in more detail through measuring this increase by weighing

manually while the alumina plus crucible was placed on a bench top analytical balance. The results (see

Fig. 7) indicated that mass gain is rapid in the first 15 minutes but then stabilizes to a steady state

condition. There is some variability in the % mass reached at steady state which may be due to slightly

different humidity conditions in the room. Also note that additional mass gain of the alumina powder

would have occurred in the 20-30 seconds it took to transfer the crucible from the TG-GCMS to the

bench top balance. On average it would appear that the alumina comes close but does not regain all of

the mass loss that occurred during the original thermal exposure up to 850 °C in the STA.

A final investigation involved re-heating one of the above samples in the TG-GCMS in order to

determine the origins of the mass gain when exposed to laboratory air. Figs. 8a and b present the GCMS

data for a reheated sample in comparison to a sample tested in the as received state (i.e. directly out of

the bottle). Fig. 8a shows the TG results for the 1st heat and 2nd heat (after air exposure) for the alumina

sample. The data in Fig. 8b and c indicate that alumina regains both H2O and CO2 from the atmosphere

after an 850 °C thermal cycle, since both of these gases evolved upon re-heating. It appears that a lower

13
amount of H2O and CO2 is off-gassing during the re-heat. In particular, the second higher temperature

stage of CO2 evolution does not occur. This suggested that the CO2 evolved at higher temperatures in the

as received powder may have developed during powder production and does not reform through the 30-

60 minutes of atmosphere exposure of this test.

3.3 NiSA powder

Approximately 0.8 g of NiSA powder was placed in the TG crucible and TG-GCMS experiments

were carried out with no evacuation (0×EF) and with 3 evacuation/refill cycles (3×EF) to determine the

effect of atmosphere on oxidation behavior. Fig. 9 and 10 present the analysis which includes TG data

and EIC abundance plots for H2O, N2, O2 and CO2. In all cases the background data measured for an

empty crucible is presented for comparison. The TG data (Fig. 9a, b and 10a, b) shows that the NiSA

powder gradually gains mass in both 0×EF and 3×EF as it is heated to 850 °C, 0.66% for 0×EF (5.0 mg

for 750 mg of NiSA) and 0.06% for 3×EF (0.5 mg for 803 mg of NiSA) This mass gain in the TG

initiates at approximately 200 °C. The superalloy powder underwent the largest mass gain in the 0×EF

atmosphere condition, over 10× higher compared to the 3×EF condition.

For the GCMS analysis, the NiSA EIC profiles for N2 (m/z = 28) (Fig. 9d and 10d) are identical

to the background profile measured from the empty crucible, clearly indicating the inert nature of

nitrogen with respect to the superalloy powder. CO2 shows an increase above the background from 185

to 440 °C indicating a release of CO2 from the NiSA powder at low temperatures (Fig. 9f and 10f).

Close examination of the TG curves reveals a small downward deflection in the signal indicating some

mass loss over the same time scale as this CO2 evolution. This is most notable in the 3×EF case (Fig.

10a and b) where the mass gain starts later in minutes than in the 0×EF case so mass gain and mass loss

are not overlapping. A small decrease in the CO2 abundance in the temperature range of 650 to 850 °C

14
over the background condition in the 0×EF case (Fig. 9f) suggests that the CO2 in the furnace begins to

react with the NiSA powder at the higher temperature.

The EIC for O2 (m/z = 32) indicates a reduction in abundance of this gas below the background

ambient atmosphere level as heating proceeds for both the 0×EF and 3×EF conditions (Fig. 9e and 10e

respectively). This is an indication that the NiSA is reacting with O2 resulting in its removal from the

furnace atmosphere. This interpretation is better illustrated in Fig. 11 where the relative amount of gas

released or adsorbed from the sample vs the empty crucible experiment is more directly compared. To

allow this direct comparison, in cases where the empty crucible and NiSA curves are misaligned, i.e.

H2O from Fig. 9c and 10c, the curves are aligned to better characterize the NiSA/H2O active area. For

0×EF, the empty crucible signal was divided by 1.4 to align relative abundances at the end of the

experiment (253 min). For 3×EF, the empty crucible signal was divided by 0.9.

The total area under the relative abundance-time profile was calculated for both the empty and

“NiSA in crucible” cases over the entire time of the experiment. The difference in this area was

calculated and is illustrated by the hatched regions in Fig. 11 and is presented numerically in Table 2

with units of relative abundance×minutes (RA·min). The total abundance of O2 due to the presence of

NiSA is reduced for both the 0×EF and 3×EF experiments. By calibrating the EIC signal using the TG

mass change data for each condition (for the 3×EF condition, 1 mg = 0.79 RA·min and for the 0×EF

condition, 1 mg = 11.66 RA·min), a calculation of the mass of each gas that reacted with the NiSA

powder was performed. These were included in Fig. 11 and Table 2 as mass change in NiSA. The lower

reduction in O2 abundance for the 3×EF case corresponds to the lower mass gain of the NiSA for this

case. The drop in O2 concentration in the TG sample chamber at 150 min is about 500 ppm for 0×EF

and 25 ppm for 3×EF in the presence of NiSA. This can used as an estimated of the amount of oxygen in

the furnace at this stage assuming the NiSA reacts with all available O2. The time-temperature range

over which the reduction in O2 abundance occurs (i.e. ≈ 60 to 210 minutes) matches the same time-

15
temperature range over which the mass % increase occurs. Also, upon cooling, the mass of the NiSA

remains constant indicating no further oxidation, while the O2 abundance in the furnace rises back to the

background level measured for the empty crucible.

The EIC for H2O (m/z = 18, Fig. 11a and b for 0×EF and 3×EF respectively) also indicates a

reduction in abundance over a time-temperature range where mass gain occurs. While this reduction is

quantitatively different than the EIC for O2, it does confirm that some of the NiSA mass gain involves a

reaction with H2O as well as O2.

Chemical analysis for oxygen and carbon impurity content was measured for the NiSA powders

after heating in the TG (see Table 3). The oxygen gain indicated by the chemical analysis is 10× greater

for the sample heated in the 0×EF condition compared to 3×EF, which is consistent with the TG mass

change data. The carbon content shows little change compared to the as received powder further

confirming that the mass gain measured in the TG is due mostly to oxidation.

3.4 NiSA in Alumina powder

Approximately 0.8 g of NiSA powder was placed in 1.1 g of alumina and TG-GCMS

experiments were carried out with no evacuation (0×EF) and with 3 evacuation/refill cycles (3×EF) to

determine the combined effect of furnace atmosphere and alumina on the oxidation behavior of NiSA

powder. About half of the alumina was first placed in the TG crucible, then the NiSA was placed in the

center of the alumina powder and the remaining alumina was placed on top of the metal powder. Fig. 1d

gives a schematic of the powder in the alumina. The TG results for these experiments are shown in Fig.

12 and 13. To compare more directly the behavior of NiSA in alumina vs alone, the mass change signal

for the alumina (taken as the average of 5 runs shown in Fig. 6) was also plotted in Fig. 12a, b and 13 a,

b.

16
 At a temperature of approximately 100 °C the TG signal for the NiSA in alumina deviates in a

positive direction from the alumina only signal (see Fig. 12a, b and 13a, b). This indicates an overall

mass gain in the NiSA powder. This trend was made more evident when the alumina alone signal was

subtracted from the NiSA in alumina data. The resulting data is called NiSA = [NiSA in alumina]-

[alumina] in the legends of Fig. 12 and 13 for clarity.

As these graphs indicate, the TG-GCMS analysis shows that the NiSA powder in alumina

gradually gains mass in both 0×EF and 3×EF as it is heated to 850 °C, 0.46% for 0×EF (3.6 mg for 793

mg of NiSA) and 0.07% for 3×EF (0.55 mg for 791 mg of NiSA) with the alumina mass change data

subtracted. This mass gain in the TG initiates at approximately 100 °C which is earlier than that for

NiSA alone. The superalloy powder underwent the largest mass gain in the 0×EF atmosphere condition,

over 7× higher compared to the 3×EF condition, which is consistent with the trend exhibited in the

NiSA alone case.

The relative abundance plots of Figs. 12 and 13 for H2O (c), N2 (d), O2 (e) and CO2 (f), indicate

behavior that would be expected for a mixture containing both alumina and NiSA. N2 is unchanged from

the empty crucible condition confirming its inert nature. The O2 abundance indicates a drop below the

empty crucible baseline in the same temperature range observed in Figs. 9 and 10, indicating that

oxidation of the NiSA is qualitatively similar within the alumina. The H2O and CO2 abundance traces

are qualitatively similar to the alumina only case, demonstrating that alumina off-gassing behavior is not

significantly influenced by the presence of the NiSA. In the particular case of CO2, the small off-gassing

peak observed for the NiSA alone (centered at ~70 minutes) is superimposed on the more major off-

gassing behavior of the alumina. This is particularly evident for the 0×EF case plotted in Fig. 12 f).

Fig. 14 shows the NiSA alone data on the same graph as the NiSA=[NiSA in alumina]-[alumina]

data to allow improved interpretation of the influence of alumina over NiSA mass-gain behavior. Under

the 0×EF condition, NiSA powder underwent less overall mass gain when packed in alumina (see Fig.

17
14a), indicating that the rate and extent of oxidation is reduced in the superalloy powder under this

condition. An additional difference is the earlier onset of mass gain when packed in alumina, which is

more easily seen in the expanded time scale plot of Fig. 14d.

This earlier onset of mass gain is also prevalent in the 3×EF condition (Fig. 14b), where the

NiSA undergoes nearly half of its mass gain within ≈ 90 minutes (i.e. before the NiSA alone sample

begins to oxidize). Interestingly, despite this early onset of mass gain, and in contrast to the 0×EF case,

the NiSA powder packed in alumina undergoes a similar overall mass gain during the thermal profile

compared to the NiSA alone in the 3×EF condition.

Fig. 14c and d compare the full data sets of NiSA alone and in alumina for both 0×EF and 3×EF

conditions. The early onset of mass gain is clearly indicated to be a consequence of the NiSA in alumina

and occurs to a similar magnitude up to approximately 320 °C regardless of the furnace atmosphere

condition. Beyond 320 °C, mass gain behavior becomes very dependent on the furnace atmosphere

condition. Table 4 gives mass change as a percent of the initial mass of powder for alumina, NiSA alone

and NiSA in alumina at 320 °C (where initial mass gain levels off) and at the end of the experiment (at

253 minutes). For NiSA alone, at 320 °C the mass change is similar for the two furnace conditions,

0.013% and -0.003% for 0×EF and 3×EF respectively. The same is the case for NiSA in alumina,

0.031% and 0.030% for 0×EF and 3×EF respectively. At the end of the experiments (253 min), the mass

changes percentages vary more, 0.66% vs 0.064% for NiSA alone, and 0.46% vs 0.068% for NiSA in

alumina.

The origins of this initial mass gain can be better understood from Fig. 15 which presents the TG

data for the NiSA in alumina with EIC abundance plots for H2O and CO2 superimposed for both the

0×EF and 3×EF. For the 3×EF case, the TG signal exhibits three distinct regions where the slope of the

mass gain changes (most evident in the 3×EF case). The first stage corresponds to the time frame over

which the majority of alumina H2O and CO2 off-gas occurs. The early mass gain at low temperature

18
when NiSA is surrounded by alumina (<60 min or 320 °C in Figure 15) would be due to the adsorption

of the H2O and CO2 that is evolving from alumina onto the NiSA. The second stage, with a lower slope,

occurs where lower levels of evolved gases are measured. The third stage occurs at higher temperatures

and is expected to be due to thermally activated oxidation.

To help fully understand the higher temperature mass gain behavior of the NiSA in alumina, Fig.

16 directly compares the H2O, O2 and CO2 abundance for the alumina and NiSA in alumina for both the

0×EF and 3×EF cases. This is given as abundance per mass of alumina vs time for m/z = 18 and 44 on

the same y-axis scale since the signals do not vary with evacuation process, as shown in Figure 6.

However, for the m/z = 32 data, the plot is abundance per mass of alumina vs time but since there is a

greater variation in O2 with respect to evacuation, the y-axis scale is different for 0×EF and 3×EF cases.

The empty crucible EIC data was subtracted from the NiSA in alumina EIC data to give a signal of the

sample as if it was not in the crucible.

Above 320 °C (60 minutes) the EIC data for O2 at 0×EF (Fig. 16c) indicates a depletion in

oxygen below the background compared to alumina alone, similar to that observed for the NiSA only

case. The depletion for O2 at 3×EF (Fig. 16d) is less than 0×EF. The EIC for H2O at 3×EF (Fig. 16b)

shows a depletion from 100 min to 200 min when NiSA is in alumina indicating that the superalloy is

reacting with water lost by alumina powder. CO2 at 3×EF (Fig. 16f) also show a reduction in signal after

100 minutes when NiSA is present, most obvious is the reduction of the large peak at 120 minutes.

Overall the results of Fig. 16 indicate that thermally activated oxidation of NiSA in alumina occurs

predominantly through reaction with O2 in the 0×EF case whereas oxidation in the 3×EF case occurs

through reaction with O2, H2O and CO2.

5. Summary and Conclusions

19
 The above investigation has demonstrated that a combined TG-GCMS thermal analysis

technique is capable of quantitatively describing the oxidation behavior of metal powders under furnace

pre-conditioning environments commonly used in industry. The inclusion of the GCMS capabilities has

also allowed a semi-quantitative capability to determine the relative concentrations of oxidants (i.e. O2,

H2O and CO2) within the furnace as a function of pre-conditioning. The additional novelty of using a

large sample crucible format, has allowed the metal powder oxidation process to be analyzed under the

important industrial condition of an alumina support material. This has led to the identification of an off-

gassing behaviour in the alumina powder and its role in metal powder oxidation, under practical furnace

pre-conditioning states. A more detailed summary of the main results are described below.

Figure 17 summarizes the important aspects of the gas atmosphere as determined by GCMS

analysis for the case of alumina only. When no evacuation is performed prior to a He gas purge, the

oxygen concentration reduces over the thermal profile but remains at a level at least 10 times higher than

the concentration of H2O and CO2. This is the case even when alumina powder, which off-gases H2O

and CO2 during heating, is present in the furnace (Fig. 17a). When evacuation and He purge cycles are

performed prior to heating, a significant reduction in the oxygen concentration occurs which remains

approximately constant over the heating process. However, the H2O and CO2 present on the alumina

powder surface, which must be removed by thermal activation, is not influenced by the evacuation

cycle. Therefore the overall concentrations of H2O and CO2 within the nominally pure He atmosphere

(i.e. their partial pressures) are similar for the 0×EF and 3×EF conditions while the concentration or

partial pressure of oxygen is greatly reduced for the 3×EF. Consequently, the 0×EF and 3×EF conditions

create different relative concentrations or partial pressures of O2, H2O and CO2. Most notably the 3×EF

condition creates a relatively higher partial pressure H2O and CO2 with respect to the oxygen

concentration, as indicated in Fig. 17b.

20
 The potential influence of the changing relative partial pressure of O2, H2O and CO2 on the

oxidation of the NiSA powder can be elucidated by considering the following three reactions, assuming

that the Cr within the NiSA is the primary element undergoing oxidation;

Cr + O2 ⇌ Cr2O3 (1)

Cr + 2H2O ⇌ Cr2O3 + 2H2 (2)

Cr + CO2 ⇌ Cr2O3 + CO (3)

At 800 °C the Gibbs energy, Go, of these reactions is approximately -560, -185, -90 kJ respectively,

with a similar Go ranking at lower temperatures. Therefore it is more thermodynamically feasible for

O2 to oxidize Cr than H2O or CO2. In the case of 0×EF, where the overall abundance of O2 is high, the

introduction of the less oxidizing H2O and CO2 off-gases leads to less NiSA oxidation compared to the

NiSA case alone. In the 3×EF case, where the O2 is lower, the relative concentrations of H2O and CO2

increase. Under this condition, while the overall oxidation extent is lower than in the 0×EF case, the

H2O and CO2 contribute more to oxidation. This results in a mass gain and extent of oxidation of NiSA

in alumina that is similar to that of NiSA on its own.

Acknowledgement

The authors would like to thank the Natural Science and Engineering Research Council of Canada

(NSERC) and Pratt and Whitney Canada for their financial support of this research. They would also

like to thank Alain Bouthillier, Thomas Georges and Orlando Scalzo of Pratt and Whitney Canada for

their guidance, fruitful discussions and support of this work. Thanks to Dan Cluff of Dalhousie

University for his advice and assistance with data analysis.

21
References

[1] Baril E, Lefebvre LP, Thomas Y. Powder Metall 2011;54:183.

[2] Hryha E, Nyborg L. J Therm Anal Calorim 2014;118:825.

[3] Chasoglou D, Hryha E, Nyborg L. Mater Chem Phys 2013;138:405.

[4] Hryha E, Dudrova E, Nyborg L. Metall Mater Trans A 2010;41:2880.

[5] Hryha E, Dudrova E, Nyborg L. J Mater Process Technol 2012;212:977.

[6] Hryha E, Nyborg L. Metall Mater Trans A 2013;45:1736.

[7] Bergman O. Powder Metall 2007;50:243.

[8] Ortiz P, Castro F. Powder Metall 2004;47:291.

[9] Özgün Ö, Gülsoy HÖ, Yılmaz R, Fındık F. J Alloys Compd 2013;576:140.

[10] Chan JL, Alcock JR, Stephenson DJ. J Mater Sci 1998;33:5131.

[11] Tang CF, Pan F, Qu XH, Duan BH, He XB. J Alloys Compd 2009;474:201.

[12] Hens KF, Grohowski JA, German RM, Valenica JJ, T.McCabe. Processing of Superalloys by
Powder Injection Moulding, in:. Met. Powder Rep., vol. 49. 1994.

[13] Bose A, Valencia JJ, Spirko J, Schmees R. Adv Powder Metall Part Mater Vol 4 Met Ceram
Inject Molding, Proc 1997 Int Conf Exhib Powder Metall Part Mater 1997:18.

[14] Valencia .J., Spirko J, Schmees R. Superalloys 718 , 625 , 706 Var Deriv TMS 1997:753.

[15] Jeandin M, Koutny JL, Bienvenu Y. Powder Metall 1983;26:17.

[16] Gessinger GH. Powder Metallurgy of Superalloys. London: Butterworths; 1984.

[17] Westerman EJ. Trans Metall Soc AIME 1962;224:159.

[18] German RM. Int J Powder Metall 1990;26:23.

[19] Lal A, Iacocca RG, German RM. Metall Mater Trans A 1999;30:2201.

[20] Lal A, Iacocca RG, German RM. J Mater Sci 2000;35:4507.

[21] Simchi A. Metall Mater Trans A 2006;37:2549.

22
[22] German RM. Metall Mater Trans A 1997;28:1553.

[23] German RM, Bose A. Injection Molding of Metals and Ceramics. Princton: Metal Powder
Industries Federation; 1997.

[24] Rufino B, Coulet M-V, Bouchet R, Isnard O, Denoyel R. Acta Mater 2010;58:4224.

[25] Arii T, Senda T, Fujii N. Thermochim Acta 1995;267:209.

[26] Madarász J, Nagygyörgy V, Stathatos E, Pokol G. J Therm Anal Calorim 2013;113:1055.

[27] Gembinski C, Charola AE, Price B, Mcghie AR. J Therm Anal Calorim 2000;59:601.

[28] Xie W, Pan W-P. J Therm Anal Calorim 2001;65:669.

[29] Hoffmann S, Schmidt M, Scharsach S, Kniep R. Thermochim Acta 2012;527:204.

[30] Eigenmann F, Maciejewski M, Baiker A. Thermochim Acta 2006;440:81.

[31] Maciejewski M, Baiker A. Thermochim Acta 1997;295:95.

[32] Emmerich W-D, Post E. J Therm Anal 1997;49:1007.

23
FIGURES:

a) empty crucible b) alumina in crucible c) NiSA in crucible d) NiSA in alumina

in crucible
Ø 19 mm

22 mm

alumina ≈ 1.1 g NiSA ≈ 0.8 g NiSA ≈ 0.8 g

alumina ≈ 1.1 g
Fig. 1. Schematic of TG crucible, mass of powder used and configuration of powder(s) in crucible.

24
Fig. 2. EICs for four empty crucible trials showing time-dependent signal for the atmospheric gases with
no evacuation (0×EF) for m/z=18 (H2O), m/z=28 (N2), m/z=32 (O2) and m/z=44 (CO2). Temperature
profile is shown as broken line.

25
 1.0

TG temp
0.9 800

0.8
Relative Abundance

0.7
600
m/z=28

Temperature / °C
0.6

0.5
400
0.4

0.3

200
0.2 m/z=32

0.1 m/z=18
m/z=44
0.0 0
0 50 100 150 200 250
Time / minute
Fig. 3. EICs for four empty crucible trials showing time-dependent signal for three evacuation and fill
cycles (3×EF) for m/z=18 (H2O), m/z=28 (N2), m/z=32 (O2) and m/z=44 (CO2). Temperature profile is
shown as broken line.

26
 a)

b)

c)

d)

e)

Fig. 4. Mass change, a), and EIC for m/z = b) 18, c) 28, d) 32, e) 40, f) 44 for alumina powder (solid
line) and empty crucible (dotted line) after 0 evacuation and fill cycles. Temperature profile is shown as
broken line in a).
27
 a)

b)

c)

d) m/z=32
Relative Abundance

0.3

0.2

e) 0.1

Fig. 5. Mass change, a), and EIC for m/z = b) 18, c) 28, d) 32, e) 40, f) 44 for alumina powder (solid
line) and empty crucible (dotted line) after 3 evacuation and fill cycles. Temperature profile is shown as
broken line in a).
28
 0xEF 1xEF 1xEF 2hr
3xEF 5xEF Average
empty crucible 3xEF TG temperature
a) Time / minute
0 100 200
100.02 1000
Mass change / %

100.00
800

Temperature / °C
99.98
99.96 600

99.94 400
99.92
200
99.90
99.88 0
b) 40
m/z=18
(mg of alumina)
Abundance /

30

20

10

0
c) 10,000
m/z=28
Abundance / (mg of

1,000
alumina)

100

10
d) 1,000
m/z=32
Abundance / (mg of
alumina)

100

10
e) 50
m/z=44
40
Abundance/(mg of

30
alumina)

20

10

0
0 100 200
Time / minute
Fig. 6. Mass change, a), and EIC for m/z = b) 18, c) 28, d) 32 and e) 44 of alumina powder for various
different evacuation and fill cycles. The graphs for m/z = 28 and 32 are plotted on a log scale. The error
bars are the calculated standard deviation. The empty crucible for 3×EF is shown for comparison. The
mass of the 3×EF alumina sample was used to calculate Abundance/(mg of alumina) for the empty
crucible.
29
 100.06

100.05
Mass change / %

100.04

100.03

100.02

100.01

100
0 10 20 30 40
Time / minute
Fig. 7. Mass gain for alumina sample in air after a heat to 850 °C in STA.

30
 alumina out of bottle empty crucible
heat after in air TG temperature

a) Time / minute
0 100 200
100.02 1000

100.00 800
Mass change / %

Temperature / °C
99.98
600
99.96
400
99.94

99.92 200

99.90 0
b)
0.30
Relative Abundance

m/z=18

0.20

0.10

0.00
c) 0.30
Relative Abundance

m/z=44
0.20

0.10

0.00
0 100 200
Time / minute
Fig. 8. TG results for alumina powder re-heated following atmosphere exposure after an initial 850 °C
thermal profile are in a). EIC measured for alumina powder re-heated following atmosphere exposure
after an initial 850 °C thermal profile for b) H2O and c) CO2.

31
 NiSA, 0xEF Time / minute
0 100 200
a) 100.80
100.03 600
1000

b)
400
100.60 800
+0.66%
200
Mass change / %

Temperature / °C
100.40 99.98 0 600
0 50 100

100.20 400

100.00 200

99.80 0
c)
Relative Abundance

0.020
m/z=18

0.010

d) 0.000
Relative Abundance

1.00

m/z=28

0.50

0.00
e)
0.30
Relative Abundance

m/z=32

0.20

0.10

0.00
f) 0.006
Relative Abundance

m/z=44
0.004

0.002

0.000
0 100 200
Time / minute
Fig. 9. Mass change a) and b), and EIC for m/z = c) 18, d) 28, e) 32 and f) 44 of NiSA powder (solid
line) and empty crucible (dotted line) for 0×EF. Temperature is plotted in a) and b) as a broken line. The
error bars are the standard deviation calculated at 50 min.

32
 NiSA, 3xEF Time / minute
0 100 200
a) 100.10 1000
100.03 600
b)
400
800
Mass change / %

Temperature / °C
200
100.05
99.98 0 +0.06% 600
0 50 100

400
100.00

200

99.95 0
c)
Relative Abundance

0.10
m/z=18

0.05

d) 0.00
1.00
Relative Abundance

m/z=28

0.50

e) 0.00
0.30
m/z=32
Relative Abundance

0.20

0.10

0.00
f) 0.04
m/z=44
Relative Abundance

0.02

0.00
0 100 200
Time / minute
Fig. 10. Mass change, a) and b), and EIC for m/z = c) 18, d) 28, e) 32 and f) 44 of NiSA powder (solid
line) and empty crucible (dotted line) for 3×EF. Temperature is plotted in a) and b) as a broken line. The
error bars are the standard deviation calculated at 50, 130 and 230 min.

33
 NiSA, 0xEF Time / minute NiSA, 3xEF Time / minute
0 100 200 0 100 200
a) 0.015 1000 b) 0.15 1000
m/z=18, 0xEF m/z=18, 3xEF
Area difference = 0.07 RA·min Area = 2.3 RA·min
Relative Abundance

mass gain due to H2O = 0.09 mg 800 mass gain due to H2O = 0.20 mg 800

Relative Abundance
Temperature / °C

Temperature / °C
0.010 0.10
600 600

empty crucible/1.4 400 400

0.005 0.05 empty crucible/0.9

200 200

c) 0.000 0 d) 0.00 0

Relative Abundance
0.30 0.30
Relative Abundance

m/z=32, 0xEF m/z=32, 3xEF

Area difference = 3.7 RA·min Area diffference = 4.6 RA·min
0.20 mass gain due to O2 = 4.71 mg mass gain due to O2 = 0.39 mg
0.20

0.10 0.10

0.00 0.00
e) 0.006 f) 0.06
m/z=44, 0xEF Relative Abundance m/z=44, 3xEF
Relative Abundance

Area difference = 0.06 RA·min Area difference = 0.6 RA·min

0.004 mass loss due to CO2 = 0.08 mg mass loss due to CO2 = 0.05 mg
0.04

0.002 0.02

0.000 0.00
0 100 200 0 100 200
Time / minute Time / minute
Fig. 11. Areas of difference in m/z = a) and b) 18, c) and d) 32, and e) and f) 44 EIC data for NiSA only
(solid line) vs empty crucible (dotted line) at the 0×EF and 3×EF condition. The difference between the
empty crucible and NiSA EIC and TG data is used to calculate mass change in NiSA for each gas.
Abundance adjustment for crucible for m/z = 18 noted in graphs and temperature shown as broken line.

34
 NiSA=[NiSA in alumina]-[alumina] NiSA in alumina
average alumina TG temperature
Time / minute
a) 0 100 200
100.80 1000
100.10 600
b)
100.60 400 800

Temperature / °C
Mass change / %

100.00
200
100.40 600
99.90 0
0 50 100
100.20 400
+0.46%

100.00 200

99.80 0
c) 0.015
Relative Abundance

NiSA+alumina
m/z=18 empty crucible
0.010

0.005

0.000
d) 1.00
Relative Abundance

m/z=28

0.50

0.00
e) 0.30
Relative Abundance

m/z=32
0.20

0.10

0.00
f) 0.008
Relative Abundance

m/z=44
0.006

0.004

0.002

0.000
0 100 200
Time / minute
Fig. 12. Mass change, a) and b), and EIC for m/z = c) 18, d) 28, e) 32 and f) 44 of NiSA powder in
alumina powder and empty crucible for 0×EF using He as purge gas. The error bars are the standard
deviation calculated at 50 min.

35
 NiSA=[NiSA in alumina]-[alumina] NiSA in alumina
average alumina TG temperature
Time / minute
a) 0 100 200
100.20 1000
100.10 600
b)
400
100.00 800
200

Temperature / °C
Mass change / %

100.10
99.90 0 600
0 50 100

+0.07%
400
100.00

200

c) 99.90 0
0.20
Relative Abundance

NiSA+alumina
m/z=18
empty crucible

0.10

0.00
d)
1.00
Relative Abundance

m/z=28
0.80

0.60

e) 0.40
0.30
Relative Abundance

m/z=32
0.20

0.10

0.00
f)
Relative Abundance

0.20
m/z=44

0.10

0.00
0 100 200
Time / minute
Fig. 13. Mass change, a) and b), and EIC for m/z = c) 18, d) 28, e) 32 and f) 44 of NiSA powder in
alumina powder for 3×EF using He as purge gas. The error bars are the standard deviation calculated at
50 min.

36
 NiSA alone 0xEF NiSA=[NiSA in alumina]-[alumina]0xEF
NiSA alone 3xEF NiSA=[NiSA in alumina]-[alumina]3xEF
a) NiSA alone NiSA=[NiSA in alumina]-[alumina] TG temperature c) TG temperature
100.80 1000 100.80 1000

100.60 800 100.60 800

=0.20%

Mass change / %
0xEF +0.66%

Temperature / °C
Mass change / %

Temperature / °C
100.40 600 100.40 600

+0.46%
100.20 400 100.20 400
=0.01%
100.00 200
100.00 200

99.80 0
99.80 0
0 100 200
0 100 200 Time / minute
Time / minute d) NiSA alone0xEF NiSA=[NiSA in alumina]-[alumina]0xEF
b) NiSA alone NiSA=[NiSA in alumina]-[alumina] TG temperature NiSA alone3xEF NiSA=[NiSA in alumina]-[alumina]3xEF
TG temperature

100.10 1000 100.10

3xEF 400
+0.07%

Mass change / %
800
Mass change / %

similar alumina effect until ~320 °C

Temperature / °C
Temperature / °C

100.05 100.05
600

200
400
100.00 100.00
+0.06%
200

99.95 0 99.95 0
0 100 200 0 50 100
Time / minute
Time / minute
Fig. 14. Mass change showing effect alumina has on NiSA a) NiSA alone and in alumina for 0×EF, b)
NiSA alone and in alumina for 3×EF, c) all data sets compared, d) all data sets compared during early
stages of heating. The error bars are the standard deviation calculated at 50 min.

37
 a) 15 100.10 b) 15 100.10
m/z=18 m/z=18
m/z=44 3xEF m/z=44
0xEF 100.08 TG 100.08
TG
Abundance / mg of alumina

Abundance / mg of alumina
10 10
100.06 100.06

Mass Change / %
Mass Change / %
5 100.04 5 100.04

100.02 100.02
0 0
100.00 100.00

-5 99.98 -5 99.98
0 100 200 0 100 200
Time / minute Time / minute
Fig. 15. TG data for the NiSA in alumina with EIC abundance plots for m/z = 18 (H2O) and m/z = 44
(CO2) superimposed for a) 0×EF and b) 3×EF. The data for empty crucible was subtracted from the data
for NiSA in alumina to give an approximate measure of the sample as if it was not in crucible.

38
 Time / minute Time / minute
0 100 200 0 100 200
a)
15 1000 b) 15 1000
Abundance / mg of alumina

Abundance / mg of alumina
m/z=18, 0xEF m/z=18, 3xEF
800 800

Temperature / °C
10 10

Temperature / °C
600 600
5 5
400 400

0 0
200 200

-5 0 d) -5 0
c) m/z=32, 3xEF

Abundance / mg of alumina
Abundance / mg of alumina

20
2
m/z=32, 0xEF
0
0
-20

-2
-40

-60 -4
e) f)
Abundance / mg of alumina

15 15
Abundance / mg of alumina

m/z=44, 0xEF m/z=44, 3xEF

10 10

5 5

0 0

-5 -5
0 100 200 0 100 200
Time / minute Time / minute
Fig. 16. EIC abundance for NiSA in alumina (dotted line) and alumina alone (solid line) for 0×EF for a)
water (m/z = 18), c) oxygen (m/z = 32) and e) carbon dioxide (m/z = 44) and 3×EF b), d) and f). Note
that results are reported as Abundance/(mg of alumina). The empty crucible signal was subtracted from
sample signal for each of these runs to remove background (i.e. empty crucible signal). Temperature is
shown as broken line.

39
 a) 1000 b) 1000
3xEF H2 O
H2O
0xEF CO2
CO2
0.30 0.30
O2
O2
H2O
H2 O 800 800
TG temp
CO2
CO2

Relative Abundance

Temperature / °C
Relative Abundance

Temperature / °C
O2
O2
TG temp 600 600
0.20 0.20

400 400

0.10 0.10
200 200

0.00 0 0.00 0
0 100 200 0 100 200
Time / minute Time / minute
Fig. 17. Oxygen, water and carbon dioxide EIC for alumina under a) no evacuation (0×EF) and b) three
evacuation/refill procedures (3×EF).

40
TABLES:
Table 1
List of samples.
Chemical Name Source Mole Fraction Purity
Helium Air Liquide 0.99999
Alumina powder Almatis GmbH 0.999
Inconel 625 Sandvik Osprey Conforms to AMS 5599
Chemical Composition Alloy Inconel 625 (wt%)
Ni Cr Fe Mo Nb Ti Al C
Bal. 22.0 ≤5 9.0 3.5 ≤0.4 ≤0.4 ≤0.1

41
Table 2
Area of peak or valley shown in Fig. 11 and NiSA mass change corresponding to peak calculated from
EIC data. For the 3×EF condition, 1 mg = 0.79 RA·min and for the 0×EF condition, 1 mg = 11.66
RA·min.
Sample m/z=18, H2O m/z=32, O2 m/z=44, CO2
valley = 0.07 RA·min valley = 3.7 RA·min peak = 0.06 RA·min
NiSA 0×EF
mass gain = 0.09 mg mass gain = 4.71 mg mass loss = 0.08 mg
valley = 2.3 RA·min valley = 4.6 RA·min peak = 0.6 RA·min
NiSA 3×EF
mass gain = 0.20 mg mass gain = 0.39 mg mass loss = 0.05 mg

42
Table 3
Chemical analysis of NiSA powder for C and O, mass gain from the TG experiment and calculated mass
gain due to increase in O + C content based on initial mass used in experiment.
Calculated mass
O/ C/ Initial mass gain due to O + C
Sample Mass gain in TG
ppm ppm in TG / mg increase from
chem. analysis
NiSA stock 1020 120 NA NA NA
NiSA 0×EF 8300 140 750.5 5.0 mg [0.66%] 5.5 [0.73%]
NiSA 3×EF 1770 130 802.7 0.5 mg [0.06%] 0.6 [0.07%]
NiSA=[NiSA+alumina]-
15000* 110 793.1 3.6 mg [0.46%] 10.5 [1.32%]
[alumina] 0×EF
NiSA=[NiSA+alumina]-
2300* 150 791.1 0.6 mg [0.07%] 1.0 [0.12%]
[alumina] 3×EF
*NiSA contamination by alumina support powder invalidated oxygen analysis

43
Table 4
Mass change as percent of initial mass for 0×EF and 3×EF condition at 320 °C (where mass gain due to
alumina levels off) and at the end of the experiment (at 253 minutes).
0×EF 3×EF 0×EF 3×EF
Mass change % point point at end at end
at 320 °C at 320 °C (253 min) (253 min)
NiSA alone 0.013 -0.003 0.663 0.064
Alumina (average) -0.072 -0.072 -0.088 -0.088
NiSA+Alumina -0.041 -0.042 0.370 -0.020
NiSA=[NiSA+alumina]-[alumina] 0.031 0.030 0.458 0.068

44