2. The carbon formed needs to deposit onto the HCFeMn slag in appreciable amounts
3. The carbon needs to stick to the surface of the HCFeMn and not easily fall off, to be able to withstand transport etc.
For optimisation and scale-up of the process, it is also useful to know the effect of different parameters on the deposition rate. The following subsections presents the results with these issues in mind.
Methane decomposition
The off-gas measurements are presented in Table 2, averaged across the 2-hour holding period during which the samples were exposed to process gas. Since the total gas-flow and -composition into the furnace was known, and disregarding higher order hydrocarbons like C2H2 (which according to thermodynamic equilibrium should be present at levels orders of magnitude below 1%), the hydrogen- carbon- and oxygen mass balances give three equations with three unknowns:
$${{\phi }}_{0}\bullet \left({\frac{1}{4}\left[{\text{C}\text{H}}_{4}\right]}_{0}+\frac{1}{2}{\left[{\text{H}}_{2}\right]}_{0}\right)={{\phi }}_{1}\bullet \left(\frac{1}{4} {\left[{\text{C}\text{H}}_{4}\right]}_{1}+\frac{1}{2}{\left[{\text{H}}_{2}\right]}_{1}+\frac{1}{2}{\left[{\text{H}}_{2}\text{O}\right]}_{1}\right)$$
$${{\phi }}_{0}\bullet {\left[\frac{1}{2}{\text{C}\text{O}}_{2}\right]}_{0}={{\phi }}_{1}\bullet \left( {\left[\text{C}\text{O}\right]}_{1}+{\left[{\text{H}}_{2}\text{O}\right]}_{1}+\frac{1}{2}{\left[\text{C}{\text{O}}_{2}\right]}_{1}\right)$$
$${{\phi }}_{0}\bullet \left({\left[{\text{C}\text{H}}_{4}\right]}_{0}+{\left[{\text{C}\text{O}}_{2}\right]}_{0}\right)={{\phi }}_{1}\bullet \left( {\left[{\text{C}\text{H}}_{4}\right]}_{1}+{\left[{\text{C}\text{O}}_{2}\right]}_{1}+{\left[\text{C}\text{O}\right]}_{1}\right)+{\left\{{\text{C}}_{\left(\text{s}\right)}\right\}}_{1}$$
Here a subscript "0" denotes incoming, a subscript "1" denotes outgoing, [X] denotes the concentration (moles/l) of species X, \({\phi }\) denotes total gas-flow (L/min), and {C(s)} denotes the net rate of solid carbon formation (mol/min). Solving these equations gives the unknowns \({{\phi }}_{1}\), [H2O]1 and {C(s)}1, and since argon is inert, the outgoing argon-concentration is simply the known incoming concentration scaled by the change in total gas flow. Multiplying{C(s)}1 by the holding time gives the net solid carbon formed during each experiment. It should be stressed that the net solid carbon formed does not equate to the amount of carbon deposited on the HCFeMn slag. Some of the formed carbon can for example deposit on the crucible wall or be carried off as fine dust. These calculated values are also included in Table 2.
The (dried) slag contains 36.6% MnO. That means that for conversion of all MnO in a 2000g slag sample to Mn after the reaction MnO + C = Mn + CO, 123.8g of carbon is required. In the following discussion, "Cstoich" is introduced to mean: the amount of carbon deposited relative to the amount required for stoichiometric conversion of all MnO (in other words, relative to 123.8g in the current work). Cstocih is used not only when referring to carbon specifically deposited on HCFeMn slag, but also for example when describing the total amount of carbon formed in the system.
Table 2
Measured (µ-GC) and calculated (from mass balance) composition of off-gas and amount of solid carbon formed for all experiments. Cstoich means percentage of carbon necessary for self-reduction of all MnO in slag. Note that carbon formed does not equate to carbon deposited.
Exp.
|
Incoming gas
|
T
|
Flow-rate
|
Outgoing Gas Composition
|
Net Solid C Formed
|
Measured averages
|
Calculated
|
Calculated
|
CO2
|
CH4
|
Ar
|
[°C]
|
[L/min]
|
CO
|
CO2
|
H2
|
CH4
|
H2O
|
Ar
|
[g]
|
Cstoich
|
1
|
27%
|
63%
|
10%
|
790
|
2.35
|
3%
|
28%
|
0%
|
66%
|
-2%
|
6%
|
-3.8
|
-3%
|
2
|
0%
|
90%
|
10%
|
790
|
2.35
|
0%
|
0%
|
1%
|
100%
|
0%
|
7%
|
0.3
|
0%
|
3*
|
0%
|
90%
|
10%
|
790
|
4.7
|
0%
|
0%
|
0%
|
0%
|
0
|
10%
|
0
|
0%
|
4
|
27%
|
63%
|
10%
|
1000
|
2.35
|
27%
|
5%
|
22%
|
36%
|
5%
|
9%
|
4.5
|
4%
|
5
|
27%
|
63%
|
10%
|
1000
|
2.35
|
30%
|
3%
|
24%
|
36%
|
7%
|
8%
|
7.1
|
6%
|
6
|
0%
|
90%
|
10%
|
1000
|
2.35
|
0%
|
0%
|
38%
|
58%
|
0%
|
9%
|
30.5
|
25%
|
7
|
0%
|
90%
|
10%
|
1000
|
2.35
|
1%
|
1%
|
34%
|
52%
|
0%
|
9%
|
31
|
25%
|
8
|
27%
|
63%
|
10%
|
1000
|
4.7
|
18%
|
8%
|
12%
|
37%
|
3%
|
8%
|
-0.8
|
-1%
|
9
|
27%
|
63%
|
10%
|
1000
|
4.7
|
25%
|
7%
|
18%
|
41%
|
8%
|
8%
|
15.6
|
13%
|
10
|
0%
|
90%
|
10%
|
1000
|
4.7
|
1%
|
0%
|
27%
|
68%
|
0%
|
9%
|
40.9
|
33%
|
11
|
27%
|
63%
|
10%
|
1100
|
2.35
|
31%
|
0%
|
43%
|
18%
|
3%
|
11%
|
18.3
|
15%
|
12
|
27%
|
63%
|
10%
|
1100
|
2.35
|
28%
|
0%
|
40%
|
14%
|
1%
|
3%
|
16.9
|
14%
|
13
|
0%
|
90%
|
10%
|
1100
|
2.35
|
1%
|
0%
|
82%
|
29%
|
0%
|
8%
|
73.6
|
59%
|
14
|
0%
|
90%
|
10%
|
1100
|
2.35
|
1%
|
0%
|
80%
|
29%
|
0%
|
8%
|
71.8
|
58%
|
15
|
0%
|
90%
|
10%
|
1100
|
2.35
|
1%
|
0%
|
67%
|
42%
|
0%
|
8%
|
55.8
|
45%
|
*) In Experiment 3, an error occurred which gave no signal to the off-gas measurements. |
In the experiments at 790°C, no hydrogen was detected in the off-gas, so no carbon decomposition took place. This means that the presence of CO2 was not sufficient to drive deposition of carbon at lower temperatures. In all other experiments hydrogen was detected in the off-gas, indicating that cracking did in fact take place. Figure 3 below shows the hydrogen levels averaged across experiments with identical experimental parameters. For data points with the same colour, the only different experimental parameter was temperature. For two sets of parameters (90% CH4, 2.35l/min; 63%CH4-27%CO2, 2.35l/min) there is significant hydrogen evolution at two different temperatures (1000°C and 1100°C) with all other parameters the same. In both cases the amount of hydrogen is much higher at 1100°C than at 1000°C. The amount of data is too limited to quantify the effect, but this indicates that temperature is a very important parameter, and that methane decomposition increases with temperature.
For two sets of parameters (63%CH4-27%CO2, 1000°C; 90%CH4, 1000°C) all parameters are the same except the total gas flow. In both cases the amount of hydrogen, and thus the degree of methane decomposition, is higher in the case of a lower total gas flow.
In experiments with CO2 in the process gas, this was converted into CO to a larger or smaller extent. At 790°C, only 3% of CO was found in the off gas. Since no hydrogen was detected and no solid carbon produced, the Boudouard reaction cannot be the source of this CO. A direct reaction between CO2 and methane that does not produce hydrogen is a possible explanation, like for example the following reaction, which is thermodynamically favourable at 790°C:
¼ CH4 + ¾ CO2 ←→ ½ H2O + CO
At 1000°C, there was between 18–30% of CO, and the CO2 level had dropped below 10%. At 1100°C, basically all CO2 had been converted.
The experiments with CO2 also had less hydrogen in the off-gas than when no CO2 was present. It is not possible, within the uncertainties, to determine to what extent this is because of methane dilution or if reactions with CO2 plays a role. On the one hand, the presence of CO in the off-gas indicates that CO2 reacts with something, which could be the produced H2. On the other hand, it is theoretically possible that all CO is produced from reactions between CO2 and deposited C, which should not influence the H2-levels.
While CO2 in the process gas giving less hydrogen in the off-gas is a drawback in isolation, there could also be an upside to having CO in the off-gas. Syngas (H2-CO-mixture) production from biogas without CO2-emissions requires the upgrading (CO2-removal) of biogas to biomethane, followed by storage/utilisation of the CO2 and steam reforming of the methane into syngas. It is possible that a competitive option would be to use non-upgraded biogas as a solid carbon source as described here, avoiding the need for CO2-removal and steam reforming.
The observations that temperature is a very important parameter, and that methane decomposition increases with temperature is in line with experiences from similar work [11, 12], as is the result that the degree of methane decomposition is higher in the case of a lower total gas flow [13].
Carbon deposition
The decomposition of methane is only the first step; the formed carbon needs to deposit onto the HCFeMn slag. Several techniques were used to verify the deposition of carbon and determine the amount deposited. The chemical analyses turned out not to be very useful for quantification since the slag contained inhomogeneously distributed carbon (likely in metallic droplets). Visual inspection of samples by eye was sufficient to identify that some samples turned completely black, suggesting carbon deposition. EPMA images showed carbon layers (verified by X-ray diffraction) on several of the samples. Example images are shown in Fig. 4.
The weight change measurements are seen as the most trustworthy estimates of amounts of deposited carbon. Figure 5 shows the results with uncertainties for all experiments, where repeat experiments have been grouped and averaged. The best results are with 90% CH4 and 2.35 l/min total gas flow at 1100°C, where an average of 46g of carbon has been deposited, corresponding to 37 ± 4% of Cstoich. In Fig. 5, the results are plotted versus temperature. The first observation that can be made is that there is no significant carbon deposition at 790°C, an observation that matches the visual appearance of the material, as well as the absence of hydrogen in the off-gas for these experiments (Fig. 3).
For experiments with low gas flow (2.35l/min), there was significant weight change at both 1000°C and 1100°C, and the weight change at 1100°C was significantly higher than at 1000°C. This was true for experiments with and without CO2 in the process gas. Increased temperature thus leads to increased carbon deposition rate.
At 1000°C, the average results with high gas flow (4.7 l/min) lie below the average results of experiments with low gas flow (2.35 l/min) for otherwise identical parameters. Since the uncertainty intervals of the low- and high gas flow experiments overlap for each set of parameters, it can not be concluded that this is a statistically significant trend. While the gas measurements showed less methane decomposition at higher gas flow, this alone should not give an expectation of less carbon deposited, since it is possible that a reduction in decomposition rate is outweighed by an increase in total amount of methane passing through the furnace. The picture remains unclear.
The final process parameter that can be discussed in terms of its influence on the carbon deposition is the process gas composition. At low gas flows (2.35 l/min) there was deposition at 1000°C and 1100°C both with and without CO2 in the process gas. This demonstrates that carbon deposition from CO2-containing gas is possible. At 1000°C and 2.35 l/min, the average deposited values are higher when there is no CO2 in the process gas, but since the uncertainty intervals overlap the result is not significant. At 1100°C however, the result is much more pronounced and clearly significant. CO2 is expected to have two potentially negative impacts on carbon deposition in this temperature range: The first is the simple dilution of methane, meaning that less carbon is available to deposit. The second is the potential consumption of already deposited carbon through the Boudouard reaction. The indicated negative impact of CO2 also at 1000°C is thus expected to be valid. However, at present it is not possible to say whether this difference is due simply to the dilution of methane, or whether chemical reactions involving CO2 also plays a role.
Abrasion strength of carbon layer
Figure 6 below shows the fraction (Wt.%) of material below 500 µm before and after the abrasion test for all experiments in the first series of experiments. In all cases, treatment in the drum increases the amount of fines. It can be noted that the fines-generation in the experiments at 790°C is comparable to that of the other samples. If the carbon layer had disintegrated easily and created a lot of fines, it would be expected that there was a big difference between samples from 790°C and the other experiments, since the samples at 790°C had no methane decomposition and no carbon deposited.
The generated fines-fraction was analysed for its carbon content. Where pre-drum analyses were available, an average of 99% of C remained in the > 500µm fraction and 96% in the > 5mm fraction. As it was mentioned earlier in the section, the Leco analysis is not the best method of comparing carbon levels in this work. As worst-case scenarios, it can be assumed that the generated fines are 100% carbon. In this scenario, for samples with significant carbon deposition, 81% of the deposited carbon material would remain in the > 500µm fraction. When also considering the 500µm-5mm fraction, the weight of material < 5mm is greater than the weight of the deposited carbon. However, this worst-case scenario is not realistic. As the comparison of fines-generation in carbon-free materials shows, the fines fractions are far from 100% carbon.
These results point towards good adhesion between the carbon and the slag. There is nothing in these results that indicate that the material would not survive normal transport to/around an industrial plant.