Analysis of the reduction of magnesium oxide and the corrosion of
calcium nitride by hydrogen as potential parts of a proposed process
for solar thermochemical synthesis of ammonia
Brian Peterson
NSF REU Program, Kansas State University Center for Sustainable Energy
Department of Chemical Engineering, Missouri University of Science and Technology
Executive Summary
Reducing magnesium oxide to either magnesium metal or magnesium nitride is a possible
step in a solar thermochemical ammonia production process. It appears from preliminary data
that the addition of chromium(III) oxide is more effective than iron(III) oxide at increasing the
reduction of magnesium oxide. Also, more magnesium oxide is reduced with a greater presence
of the reducing agent carbon. For some yet undetermined reason, x-ray diffraction analysis of
solids composition has proved ineffective for this experiment.
The composition of the solids remaining from reacting calcium nitride with hydrogen was
to be determined. The remaining solids after several reactions show a variety of compounds with
varying ratios of calcium, nitrogen, and hydrogen. This could be an indication of a complex
pathway to liberated nitrogen as ammonia with several intermediate steps. The composition of
the remaining solids varies dramatically with temperature. At higher temperatures volatility of
certain material can also change the solids’ composition, such as the calcium hydride. Calcium
hydride apparently was able to vaporize and leave the system somewhere between 700 and 1000
°C. Preliminary results also indicate that ammonia production may level off after about 700 °C.
That is why that temperature was chosen for the kinetics analysis. Also, it was found that
varying the hydrogen flow rate has no clear influence on the reaction rate. This indicates a
diffusion limited reaction.
Introduction
Nitrogen fertilizers are critical to agricultural. Crops feed us directly and feed the
livestock which we then consume. Additionally, agricultural products are in increasing demand
for various forms of bio-energy. The fields in which these crops are grown must be fed a steady
diet of digestible nitrogen in the form of ammonia or various ammonium salts in order to avoid
depleting the soil nutrition.
Presently ammonia is produced by the Haber-Bosch Process, which has been the standard
for nearly 100 years. Without this process, roughly one third of the world's population could not
be sustained. Through chemistry, food is made from the nitrogen in the air as it is "fixed" in the
form of ammonia. Despite the necessary benefits, there is room for improvement. Haber-Bosch
is not a very green process. World-wide the Haber-Bosch Process accounts for almost 2% of the
world's energy consumption. It requires one ton of natural gas for every ton of ammonia
produced and operates at extremely high temperatures and pressures.
Kansas State Chemical Engineering PhD student Ronny Michalsky, under the direction
of his mentor professor Dr. Peter Pfromm, has been exploring possible alternatives. The potential
process would utilize a metal nitride which would react with a hydrogen source (water or
hydrogen gas) to form ammonia and a remaining solid. The solid, usually a metal oxide, must be
somehow then reduced to its original metal so it can then be nitridated back to the metal nitride,
and the cycle can begin again. Reducing this metal oxide can prove difficult. It requires high
temperatures and the presence of a reducing agent such as carbon. So theoretically this process
would only require inputs of water, nitrogen (from air), and carbon for outputs of the desired
ammonia and carbon monoxide which could be utilized in other chemical syntheses. The process
would operate at ambient pressure, and high temperatures required by certain legs of the cycle
are produced by magnified sunlight (solar energy).
The search for the ideal reactants and the ideal reacting conditions is critical to
developing this alternative process. First, a reaction involving the optimization of the reduction
of magnesium oxide was studied. This was done by reacting the magnesium oxide with nitrogen
in the presence of carbon as a reducing agent.
Figure 1: Basic overall reaction cycle of magnesium oxide. The reaction with the smaller dashed line around it is the one
being studied
Additionally an alternative metal oxide will be added in hopes of reducing the reaction
temperature. Either iron(III) oxide or chromium(III) oxide was added during experiments. The
ultimate goal of this reaction is to form magnesium nitride and to maximize its yield by varying
the stoichiometric ratios of the carbon and alternative metal oxide. Data on the kinetics of the
reaction were gathered, but the results after analysis proved wildly inaccurate. It is important to
note that this reaction was heated using solar energy from a solar concentrator described later.
The other reaction studied involves reacting calcium nitride and hydrogen. Originally the
metal nitrides were readily reacted with water, forming ammonia and a metal oxide. The metal
in the metal oxides have proved very difficult to reduce and recycle back to the easily reacted
metal nitride. The fact that hydrogen contains no oxygen molecule like water does should
circumvent the production of a metal oxide.
Figure 2: The overall reaction cycle using calcium nitride and hydrogen. The reaction with the smaller dashed line around it
is the one being studied.
How much nitrogen will be released (in the form of ammonia rather than gaseous nitrogen) from
the solid state calcium nitride into the gas phase is unknown. What is also uncertain is the nature
of the “solids” left behind. Additionally, the kinetics of the reaction at 700 °C were studied.
Materials and Methods
Reduction of magnesium oxide
Samples of solid reactant were prepared by mixing magnesium oxide (99.95% pure, -325
mesh, NOAH Technologies Corporation) with various stoichiometric ratios of carbon (graphite
powder, crystalline, -300 mesh, 99%, Alfa Aesar) and either iron(III) oxide (anhydrous, 99.95%
pure, -325 mesh, typ. 0.3 micron, NOAH Technologies Corporation) or chromium(III) oxide
(anhydrous, 99% pure, -325 mesh, NOAH Technologies Corporation). The powder was then
thoroughly crushed and mixed to ensure uniformity. The various mixtures (eight total) were
stored in a refrigerator until needed. Once needed, enough of each mixture was weighed so that
there were 50 mg of Magnesium Oxide present in the reaction. The mixture was then placed into
a tubular flow through reactor (27 mm inner diameter, ID, 30 mm outer diameter, OD, 200 mm
length, fused quartz, Technical Glass Products). The nitrogen for the reaction was provided at a
Table 1: molar composition of mixtures
Mixture
1
Molar Ratio
1:1:4 MgO:Fe2O3:C
2
1:1:8 MgO:Fe2O3:C
rate of 2 liters/minute, measure by a gas flow meter (Omega
Engineering).
The appropriate temperature was attained by mounting
a Fresnel lens (0.93 x 1.24 m, 0.7 mm thickness, 1.03 m focal
4
1:1:8 MgO:Cr2O3:C
length, UV filtering acrylic, geometrical concentration ratio ~
5
2:1:5 MgO:Fe2O3:C
2440 [49], Mitsubishi TV) onto a custom aluminum scaffold
6
2:1:10 MgO:Fe2O3:C
(Skarda Equipment Company). The scaffold had to be
7
2:1:5 MgO:Cr2O3:C
manually adjusted approximately every three minutes to
8
2:1:10 MgO:Cr2O3:C
maintain alignment with the sun. This device is referred to as
the Solar Concentrator. Temperature was checked before the experiment using a thermocouple
(High Temperature Flexible Ceramic Fiber-Insulated Probe, Type K; IR-Pro Infrared
Thermometer, 1371 °C maximum temperature (Tmax), both ThermoWorks). To determine the
temperatures above the thermocouple’s limits and the maximum temperature, silicon, iron, and
chromium powders were melted (Tmax≈1600 °C at a solar power flux of Psolar≈0.85 ± 0.03
kW/m2). This process also allowed the Solar Concentrator to be calibrated for temperature at
various distances from the focal point for ease of temperature adjustment. Additionally, the solar
power flux before entering the lens was monitored using a solar meter (SP1065, 300-1100 nm
wavelength detection range, EDTM Glass, Window & Film Test Equipment). The first set of 8
experiments was performed at 1200±100 °C. The temperature varies so much because of the
constantly moving focal point.
3
1:1:4 MgO:Cr2O3:C
The end of the tubular reactor was plugged with quartz wool to capture any vaporized
magnesium that may have been formed by reduction. The solid samples were exposed to the
flowing nitrogen and the specified temperature for 30 minutes. The samples of quartz wool and
the reacted solid were collected for analysis. Each of those two samples were then divided and
part was dissolved in 10 mL of 0.0107 molar hydrochloric acid, the other part was inerted with
argon and stored. Thus each experiment yielded four samples to analyze: two liquid and two
solid.
For kinetics data, reactions were run using the indoor reactor and furnace (explained in
the corrosion of calcium nitride section) to insure a stable temperature at 1200 °C. Mixtures 2, 4,
and a reference analogous to them without iron or chromium oxides were reacted 3 different
times each at time intervals of 30, 60, and 120 minutes. Thus, 9 total reactions were run for
kinetics data. Only the remaining solid samples were collected.
Corrosion of calcium nitride by hydrogen to form ammonia
The reactant, calcium nitride, was prepared by using the indoor reactor setup, rather than
the solar concentrator. Approximately 4 grams calcium metal (99%, granular, Acros Organics)
was placed in a quartz boat which was then placed in a tubular reactor (60 mm ID, 1 m length,
quartz, model HTF55347C, Lindberg/Blue). Nitrogen was allowed to flow over the solids for 4
hours at 2 L/min. The reaction was heated to a temperature of 750 °C by an electric furnace and
a temperature controller (model CC58434C, Lindberg/Blue). The product was then crushed to a
fine, uniform powder in the presence of argon gas to prevent unwanted corrosion. The sample
was then inerted with argon and stored under refrigeration until needed for experiments. It is
important to note that both water and oxygen will cause unwanted corrosion of the calcium
nitride. To prevent this, reactants or products of these reactions must always be inerted by argon
or nitrogen gas with the exception of weighing samples, moving from one inert vessel to the
next, or during solid state analysis (after which the sample is discarded anyway). When
transporting exposed calcium nitride over any significant distances, a trash bag full of argon was
used as a temporary vessel.
The first set of experiments with calcium nitride revolved around varying temperature.
The same indoor reactor setup as described in the previous paragraph was used with one
modification. A flask filled 25 mL of 0.0107 molar hydrochloric acid was placed in an ice bath,
and the effluent gas from the reaction was allowed to bubble through the liquid. This setup
allows the ammonia formed by the reaction to be captured in the very stable form of the
ammonium ion. While the reaction was being prepared, the tubular reactor was allowed to purge
(~5 min.) by nitrogen at 2 L/min. at 200 °C. Roughly 0.5 g calcium nitride was placed in a
quartz boat and then into the reactor. The reactor temperature was elevated to desired levels
while the inert nitrogen was still flowing. Once the reaction temperature was reached, reactive
hydrogen was turned on at a flow rate of 0.5 L/min. For 1 hour hydrogen gas was allowed to
flow over the solid reactant. Experiments at 4 different temperatures (300, 500, 700, 1000 °C)
were performed.
After analyzing the results of the previous 4 reactions, it was determined that the reaction
at 700 °C was the best case for analyzing reaction kinetics. This is because the 700 and 1000 °C
reactions both produced similar amounts of ammonia (see Results and Discussion), but the lower
temperature is favored because of the lower energy requirements. The experimental setup to
analyze the reaction kinetics at 700 °C was very similar the previous 4 experiments. The
reaction was run for two hours while 4 mL samples from a flask initially filled with 50 mL of
0.0107 molar hydrochloric acid were taken at 1, 5, 10, 30, 60, and 120 minutes into the reaction.
This procedure was repeated for 4 different flow rates of hydrogen gas (0.2, 0.5, 1.0, and 2.0
L/min.).
Chemical analysis
The solid products of the reactions were analyzed using x-ray diffraction (XRD). The
particular machine utilized was the Miniflex II diffractometer (Cu-target X-ray tube, 30 kV / 15
mA output, diffracted beam monochromator, Rigaku). The machine was set to scan through
specific angles of detection (5-80 °2θ range, 1 °2θ/min scan speed, 0.02 data points/°2θ,
continuous mode). The results of this process were analyzed using PDXL Software (Version
1.6.0.0). The solid state analysis allowed the approximate weight percent composition of solid
products to be determined.
The concentration of the ammonium ion within liquid samples was identified using an
ammonium Ion Selective Electrode (Denver Instruments) which was attached to a pH/ISE
Controller (model 270, Denver Instruments). This device measures the flux of ammonium ions
across a membrane by reporting a voltage. On one side of the membrane is 2 mL of 0.1 molar
ammonium chloride at a pH of ~13 which was attained using 0.2 mL of 10 molar sodium
hydroxide. The ISE had to be calibrated by measuring the voltage outputs of known samples of
ammonium ions. Values were taken at 0.01, 0.1, 1, 10, 100, and 1000 ppm ammonium ion. A
logarithmic regression of the results was then used to determine ammonium ion concentration as
a function of voltage.
Results and Discussion
Reduction of magnesium oxide
The goal of this reaction was to reduce the magnesium oxide. Ideally magnesium nitride
will be the compound found. Magnesium nitride is desired because it can be directly reacted
with water to ultimately produce ammonia. It is also possible that magnesium metal itself was
formed. In this case, the magnesium metal would likely vaporize at the reactor temperature and
subsequently recondense onto the quartz wool which was packed into the end of the tubular
reactor. In the analysis of the quartz wool, finding magnesium oxide is most likely because of
the reactivity of magnesium in the presence of oxygen. After XRD analysis of the quartz wool,
it became apparent that if there was any magnesium or magnesium oxide present, the silicon
dioxide of the quartz wool proved to abundant and masked any other compounds present.
The results of the XRD analysis for mass composition of the remaining solids of the 8
mixtures at 1200±100 °C are shown in Table 2.
Table 2: XRD composition results for the mixtures from Table 1
Compound
Name
Formula
Carbon
C
Chromium
Cr
Chromium Carbide
Cr7 C3
Chromium Nitride
Cr2 N
Eskolaite, syn
Cr2 O3
Iron Nitride
Fe8 N
Iron Oxide
Fe0.925 O
Iron(III) Oxide
Fe2 O3
Magnesium Nitride Mg3 N2
Periclase, syn
Mg O
Tongbaite
Cr3 C2
Wustite, high
Fe0.909 O
Wustite, syn
Fe0.911 O
Mixture
1
17.18%
2
55.00%
3
19.40%
7.00%
20.26%
25.81%
10.86%
17.00%
13.14%
0.97%
35.36%
1.37%
13.00%
13.80%
2.12%
25.42%
4
23.97%
5
33.56%
6
44.88%
7.89%
21.06%
23.84%
10.57%
12.67%
10.31%
0.44%
41.94%
3.37%
15.95%
0.15%
35.65%
7
22.02%
8
41.69%
3.36%
40.07%
33.30%
0.43%
31.02%
3.10%
22.48%
13.75%
The results from Table 2 demonstrate the relative amounts of magnesium nitride formed,
but it is also apparent from the large variety of compounds detected that there are fairly large
inaccuracies in the XRD analysis. For instance, the detection of iron nitrides are most likely due
to the presence of nitrogen gas entrained in iron. This uncertainty makes it difficult to judge the
accuracy of the results.
Due to a malfunctioning ISE, results from all the soluted samples could not be found.
For analysis of the kinetics experiments, a simple mass balance was done on magnesium
atoms. If they were retained in the sample as magnesium oxide, the magnesium was not
reduced. If magnesium metal were produced it would very likely vaporize and exit the system.
Thus the weight percent of magnesium lost should indicate reduction of that metal as a function
of reaction time.
Table 3: Weight percent of magnesium lost in the reaction at various reaction times
% Mg Lost
Molar Ratio
MgO:C:FeO3:Cr2O3 30 min. 60 min.
1:8:0:0
49.21%
50.55%
1:8:1:0
-60.80% -74.01%
1:8:0:1
7.09%
-24.82%
120 min.
-43.35%
-154.78%
44.06%
It is apparent that there are wild errors involved. All highlighted sections of Table 3
indicate a negative magnesium loss, and thus magnesium was created. This cannot be true. This
data is useless until more effective analytical techniques can be found.
0.17%
24.83%
Corrosion of calcium nitride by hydrogen to form ammonia
The two aspects of the analysis for this reaction at 4 different temperatures were to
determine the amount of ammonia formed and the composition of the solids left after the
duration of the reaction. The results of the XRD analysis for mass composition of the remaining
solids are in Table 4.
Table 4: Composition of the solids following reactions at various temperatures
Compound
Reaction Temperature (°C)
Name
Molar Mass (g/mol) Formula
300
500
Calcium Nitride
148.24
Ca3N2 56.51% 57.01%
Calcium Oxide
36.08
CaO
4.86% 8.88%
Calcium Hydroxide
74.08
Ca(OH)2 18.61% 11.70%
Calcium Nitride
94.16
Ca2N 20.02% 0.00%
Calcium Hydride
42.08
CaH2
0.00% 14.73%
Calcium Hydride Nitride
95.16
Ca2HN 0.00% 7.68%
700
0.00%
15.75%
19.17%
0.00%
11.57%
53.54%
700
0.00%
16.59%
20.20%
0.00%
21.51%
41.71%
It is interesting to note that calcium hydride appears at higher temperatures and then
disappears suddenly in the 1000 °C reaction. This is due to vaporization of that product.
Due to ISE failure, all liquid samples remain unanalyzed.
Further analysis can determine the percent of nitrogen liberated to the gas phase. This is
a very close indicator of the amount of ammonia formed.
Figure 3: Mole percent of nitrogen atoms released to the gas phase as a function of temperature. This should closely mimic
the production of ammonia.
70%
Percent N Liberated to Gas Phase
60%
50%
40%
30%
20%
0
200
400
600
Temperature (°C)
800
1000
1200
1000
0.00%
16.88%
20.39%
0.00%
0.00%
62.73%
The next step in analyzing this reaction was to study the effects of different hydrogen
flow rates. A plot of the mole percent of nitrogen liberation which correlates to ammonia
production as a function of hydrogen flow rate can be seen in Figure 4.
Figure 4: Mole percent of nitrogen atoms released to the gas phase as a function of hydrogen flow rate. This should closely
mimic the production of ammonia.
100.00%
Percent N Liberated to the Gas Phase
80.00%
60.00%
40.00%
20.00%
0
0.5
1
1.5
2
2.5
Hydrogen Flow Rate (L/min.)
No clear trend appears with flow rate change.
Liquid samples taken at various time intervals during the reaction could not be analyzed
due to ISE failure.
Additionally, minimum energy requirements to separate the ammonia from the unreacted
hydrogen were performed.
Figure 5: Minimum energy requirements for separation of ammonia from hydrogen at various hydrogen flow rates.
Calculated using wmin=-RT[xNH3ln(xnh3)+xH2ln(xH2)]
19
18
17
16
15
14
13
Miniumum Energy of Separation
(kBTU/ton ammonia)
12
11
10
0
0.5
1
1.5
Hydrogen Flow Rate (L/min.)
2
2.5
When increasing hydrogen flow, there appears to be no effect on ammonia production,
but it does effect the energy of separation.
Conclusions and Outlook
Reduction of magnesium oxide
The first part of analyzing this reaction is studying the effects of adding iron(III) oxide
and chromium(III) oxide and various amounts of carbon to the reaction. Table 2 indicates that
chromium oxide has the greater influence on the reduction of magnesium oxide. Data also
suggests that adding more of the reducing agent carbon increases the reduction. Further analysis
of the effectiveness of these additional reactants should be pursued, perhaps without using the
Solar Concentrator which has high temperature fluctuations. Instead, an indoor reactor and
furnace may create more reproducible results.
Attempts at analyzing the kinetics of this reaction proved difficult. Results from x-ray
diffraction produced impossible data. Further kinetics analysis should be performed once more
reliable analysis can be achieved.
Corrosion of calcium nitride by hydrogen to form ammonia
The remaining solids after several reactions show a variety of compounds with varying
ratios of calcium, nitrogen, and hydrogen. This could be an indication of a complex pathway to
liberated nitrogen as ammonia with several intermediate steps. The composition of the
remaining solids varies dramatically with temperature. At higher temperatures volatility of
certain material can also change the solids’ composition, such as the calcium hydride. Calcium
hydride apparently was able to vaporize and leave the system somewhere between 700 and 1000
°C. Preliminary results also indicate that ammonia production may level off after about 700 °C.
That is why that temperature was chosen for the kinetics analysis.
A key piece of data is that hydrogen flow rate has little or no effect ammonia production.
This is an indicator of nature of the reaction. Diffusion appears to be the limiting portion of the
reaction. This means that it is very difficult to speed the reaction up by methods that are not
energy intensive. Also the calculations show that an increase in hydrogen flow rate adds extra
separation energy requirements.
Overall, some data shows promising results.
promising. And some data is just plain bad.
Some shows results that are not that