- Email: [email protected]
Experimental investigations on the combustion of lithium particles in CO2 and CO2-N2 mixtures
Fuel 199 (2017) 28–37
Contents lists available at ScienceDirect
Fuel journal homepage: www.elsevier.com/locate/fuel
Full Length Article
Experimental investigations on the combustion of lithium particles in CO2 and CO2-N2 mixtures Martin Schiemann a,⇑, Peter Fischer a, Jeffrey M. Bergthorson c, Guenter Schmid b, Dan Taroata b a
Department of Energy Plant Technology, Ruhr-Universität Bochum, Germany Siemens AG, Corporate Technology, Erlangen, Germany c McGill University, Montreal, Canada b
h i g h l i g h t s
g r a p h i c a l a b s t r a c t
Lithium particles were burned in CO2-
containing atmospheres with nitrogen dilution. Optical and sampling methods were used to characterize combustion time scales. In nitrogen-diluted CO2 atmospheres, the formation of Li2CO3 is retarded. Combustion temperature decreases with CO2 partial pressure.
a r t i c l e
i n f o
Article history: Received 1 November 2016 Received in revised form 13 February 2017 Accepted 22 February 2017
Keywords: Lithium combustion Particles Burnout times Metal combustion Energy storage
a b s t r a c t Energy cycles based on lithium combustion have been proposed recently and provide an option to combine the benefits of high-density energy storage with carbon-free production of heat, power and chemicals. Knowledge on efficient combustion processes for lithium is limited. In the current work, combustion experiments for single lithium particles (dp = 10–250 mm) burning in CO2 and three different CO2-N2 mixtures are carried out using a laminar flow drop tube reactor. In order to determine combustion times, temperatures and burnout constituents, the burning particles are investigated using three different measurement techniques: a reflex camera, a high speed two-color pyrometer and a sampling probe to collect burnout samples. The results confirm the existence of two combustion stages, which have been observed before: gas-phase combustion and surface reaction. As detailed investigations show, the duration of the gas-phase combustion is approximately ten times shorter compared to the surface reaction. In addition, the burnout times of both combustion stages increase clearly for lower CO2 partial pressures. The reduction of the CO2 partial pressure leads to lower combustion temperatures as well as the production of less Li2CO3 and more Li2O, while no nitrogen compounds are formed. Ó 2017 Elsevier Ltd. All rights reserved.
1. Introduction Recently, the combustion of various metals has been investigated to develop a chemical energy carrier [1]. Renewable energy ⇑ Corresponding author. E-mail address: [email protected] (M. Schiemann). http://dx.doi.org/10.1016/j.fuel.2017.02.079 0016-2361/Ó 2017 Elsevier Ltd. All rights reserved.
can be used to charge the energy carrier (i.e. splitting of metal salts). To discharge the stored energy, the metal is burned in pulverized or sprayed form to increase the surface-to-volume ratio and, therefore, the reaction rate. The heat produced during combustion can be used in internal combustion engines [2,3] or is converted to electricity using a water-steam cycle or Stirling engine [1]. In the past, aluminum and iron have been investigated regard-
M. Schiemann et al. / Fuel 199 (2017) 28–37 Table 1 Exothermic reactions of Li with exhaust gas components calculated with data from [26]. Reaction #
Chemical reaction
Reaction enthalpy [kJ/molLi] (298.15 K, 0.1 MPa)
1 2 3 4 5 6 7
6 Li + N2 ? 2 Li3N Li3N + 3 H2O ? 3 LiOH + NH3 2 Li + 2 CO2 ? Li2CO3 + CO 2 Li + 1 CO2 ? Li2O + CO 4 Li + 1 CO2 ? 2 Li2O + C 2 Li + 2 H2O ? 2 LiOH + H2 4 Li + O2 ? 2 Li2O
54 145 270 157 201 202 299
ing their potentials as chemical energy storage materials [4–11]. Another promising metal for a chemical energy carrier is lithium, as it reacts exothermally with various gases, including O2, H2O, N2 and CO2 [12–16]. Thus, lithium can be burned with the main components of fossil power plant exhaust gas, which enables the possibility to utilize the flue gas constituents. In addition, the reaction of lithium with N2 [17] or CO2 [18,19] produces valuable byproducts such as Li3N (reaction with N2,) or CO (reaction with CO2), which can be further converted to ammonia or synthetic hydrocarbon fuels. A good overview on the combustion of lithium is given in the reviews of Jeppson et al. [20], Rhein [21] and Schiemann et al. [22], who also sum up the combustion of lithium particles or droplets focusing on processes for energy utilization. The major chemical reactions for the different combustion processes are summarized in Table 1. Considering lithium as an energy carrier needs process schemes which go beyond the combustion process. The idea to use the metal as energy carrier and typical metal purification schemes lead to the idea of metal recycling. Focusing on lithium, it will be necessary to recycle the combustion products (Table 1: Li3N (R1), Li2CO3 (R3), Li2O (R4)) to reuse the lithium contained. The lithium carbonate price of 6400 US$ per metric ton
29
(2015, [23]) and the economic value of the chemical products (CO, (R3), NH3 generated from Li3N (R1, [17])) lead to an estimate for the allowed lithium loss per process cycle in the range of one percent [22]. However, there are even cheaper sources of lithium carbonate with prices below 1000 $/tLi2CO3, such that the estimated loss rate can be higher by the corresponding factor. When NH3 production is considered, the value of approx. 500 €/tNH3 increases economic feasibility to a level which reduces the recycling efficiency demand drastically. The recycling step utilizing renewable energy is possible from lithium salts [24], which seems feasible for stationary plants. Lithium-based internal combustion engines would need a very efficient lithium capture, which might be more realizable when hydrogen production from lithium (Table 1, (R6)) is considered [25], but has not been investigated in depth yet. In the past three years, the combustion of lithium particles in CO2 has been investigated both experimentally and numerically [18,19]. For that purpose, a drop tube reactor was used to analyze the combustion behavior of single particles using a two-color pyrometer for the determination of combustion temperatures as well as particle sampling to analyze the burnout composition. It has been shown that, in pure CO2, the combustion occurs in two different stages. After ignition, the particles start to burn in the gas phase with the flame detached from the particle surface. The gas-phase combustion (GPC) ends after a very short period of time (several ms) due to the accumulation of reaction products on the particle surface, which form a product layer. The layer hinders the diffusion of evaporated lithium from the particle surface to the detached flame and thus terminates the GPC and starts the second combustion stage. During the second stage, the reaction occurs at the particle surface. For the surface reaction (SR) in pure CO2, reaction kinetics were derived from experimental data using a two-color pyrometer [18]. In addition, time-dependent particle sampling was conducted to determine the product constituents, which showed a nearly complete conversion to Li2CO3.
Fig. 1. Schematic drawing of the drop tube reactor (DTR) including gas temperature profiles along the reactor tube axis and a example of the particle streak for the combustion in 100% CO2.
30
M. Schiemann et al. / Fuel 199 (2017) 28–37
Table 2 Gas flows (STP) of the investigated atmospheres. Atmosphere CO2/N2 (molar)
V_CO2 [l/min]
V_N2 [l/min]
1/0 0.5/0.5 0.25/0.75 0.1/0.9
60.0 37.3 21.3 7.8
– 37.3 63.8 70.3
The reaction of lithium particles with N2 has been investigated in the same drop tube reactor conducting particle sampling at various particle residence times [17]. The results showed that the main product is Li3N and that the temperature is likely limited by the decomposition temperature of Li3N, which is relatively low (1234 K). Only surface reaction occurs since the combustion temperature is lower than the boiling point of lithium (1620 K). In the current work, further single particle experiments are carried out to determine the combustion behavior in CO2-N2 mixtures. Compared to the results published in [17,18], where the combustion of lithium particles in pure N2 and pure CO2 was analyzed, three different CO2-N2 mixtures are investigated concerning burnout constituents and combustion temperatures. Additionally, the data gathered from pure CO2 experiments are included to show the impact of different dilution grades more clearly. In addition, experiments using a reflex camera are carried out to determine burnout times for the combustion in pure CO2 and in the CO2-N2 mixtures. To ensure comparable results, the particle size distribution is similar to the former experiments in [18]. The results are important for improving our understanding of the fundamental combustion properties of lithium with flue gases.
2. Experimental setup The drop tube reactor (DTR) is presented in Fig. 1 and will be explained only briefly in the following section as it already has been discussed in detail in [18]. The gas flow, consisting of pure CO2 and three different CO2-N2 mixtures (flow rate and mixing ratio in Table 2) is heated up to approx. 830 K with an electrical gas preheater. As CO2 and N2 differ in their heat capacity, density and thermal conductivity, the total gas flow is modified for each atmosphere to provide similar gas temperature profiles in the reactor. The resulting gas flows are presented in Table 2. The hot gas is guided through a ceramic honeycomb, to establish a laminar flow in the reactor. The reactor itself is enveloped by a quartz glass tube (d = 65 mm) to conserve the laminar flow and to provide optical access to the burning particles. The lithium particles (approx. 1 g/ h) are injected from the top into the reactor. The very low particle mass flow rate is guaranteed by a small star feeder. The cylindrical cell size is reduced to 0.5 mm diameter and 1 mm height, such that each cell contains only a small number of particles. As the particles are transported to the reactor inlet through a thin tube of approx. 0.5 m length, the small bunch of particles escaping from each cell is dispersed to let the particles burn isolated from each other. The particles are conveyed using a cold Argon carrier flow (approx. 0.4% (0.25 L/min) of the total gas flow) to prevent the particles from heating or reacting before entering the quartz glass tube. Additionally, the injector, which ends directly at the end of the ceramic honeycomb, is water cooled. For information on the particle production (base material: lithium rod with purity 99.9%; <1500 ppm metal traces, [27]) and handling process as well as the particle size distribution (10–250 mm), please refer to [18]. Fig. 1 also shows gas temperature profiles along the reactor tube axis for each gas mixture investigated. The temperature profiles were measured without Li particles using thermocouples (radiation correction has been carried out as described in [11]).
Fig. 2. Measured particle velocities for the investigated atmospheres.
After entering the reactor, the particles are heated up (heating rate of approx. 104 K/s, as calculated in [19]), melt and ignite. As the particles move further downstream in the tube while burning, they form a glowing particle streak as also depicted in Fig. 1 for the atmosphere in 100% CO2. The combustion is monitored using three different measurement techniques: a reflex camera (Canon EOS 50D), a high speed two-color pyrometer and a sampling probe. The sampling probe is used to extract burnout samples at various residence times for subsequent analysis using X-ray diffraction and CHN-analysis. As quenching of the reacting particles is necessary at the probe inlet, an Ar co-flow is applied in the sampling probe to dilute the reaction atmosphere which is aspirated together with the particles. Additionally, the probe wall is watercooled to reduce temperature as far as possible. As the quartz glass tube provides optical access to the reactor, the pyrometer system can be used to measure particle and flame diameters as well as particle temperatures during combustion. For the measurements presented, it is very crucial that particles burn well separated from each other to avoid affection of the inter-particle gas phase, which in turn would reduce the concentration of available gaseous reaction partners and reduce the burning rate, which increases burning time. The image analysis for temperature (particle) and diameter (particle/flame) measurements was therefore restricted to images which only contain one single combustion event in the field of view (5 * 4 mm). Most of the exposures fulfilled this requirement, which indicates that the criterion of well-separated particles is met. The statistical probability of a line-of-sight overlap is reduced to a minimum, justifying the assumption that the diameter data are not biased by overlapping particles. Both measurement methods have been used before during the experiments in pure CO2. For further information concerning those systems please refer to [18]. The 2D-pyrometry camera system TOSCA (Temperature measurement with electro Optical high Speed CAmeras) is utilized to determine the particle velocities inside the reactor. For this purpose, double-exposure shots of the particles were recorded using very short exposure times (40 ms) and defined exposure timeouts (500 or 800 ms) between the two shots. As the particles move further downstream in the reactor during the exposure timeout, two shots of the same particle are recorded and saved in the same 2dimensional image. As the TOSCA-system has been size calibrated before (resolution 10 mm/px), the measured distances between the two shots can be converted into particle velocities using the size calibration and the defined exposure timeouts. At the same time, the reflex camera is used to record the whole particle trajectory covered during combustion. As the reflex camera has also been size calibrated, the recorded combustion
M. Schiemann et al. / Fuel 199 (2017) 28–37
trajectories can be converted from pixels to millimeters, resulting in the distance covered from ignition to extinction. Integrating the particle velocities for the recorded combustion distances leads to the combustion times [28,29]. 3. Experimental results The particle velocities are measured for each atmosphere at the top and the bottom of the visible particle streak (see Fig. 1) as well as at two more measurement points in between. As the beginning and the end of the streak differ for each atmosphere, different measurement points are utilized depending on the investigated atmosphere. The resulting particle velocities are presented in Fig. 2. To ensure statistical significance, at least 100 particles were recorded for each measurement point. The depicted error bars denote the resulting standard deviation. The lowest particle velocities occur in 100% CO2 (1.05–1.15 m/s), as the total gas volume flow is much lower than for the other atmospheres (see Table 2). The clearly deviating measurement distances from the particle inlet reflect the differences in combustion behavior. Altogether, the measured particle velocities correspond well to the theoretical gas velocities in the reactor tube. This leads to the conclusion that the relative velocity between the particle and the gas is low. Slip free movement of particles of similar size and density have been reported before for other drop tube reactor experiments from Cope et al. and Mitchell [30,31]. In addition, the multiple exposure measurements of the transition from GPC to SR (see Fig. 6) show very round enveloping flames during GPC, which is another indication of no significant relative velocity between particle and gas. In the next step, the burning particles are recorded with the reflex camera. Typical examples of the resulting photographs for each atmosphere are depicted in Fig. 3. The first two photographs in 100% and 50% CO2 show the two combustion mechanisms: the GPC appears as a very bright and white flame, while the SR results in a red-orange glowing of the particle. Here, the GPC flame is much broader than the glowing particle, confirming that the GPC flame is detached from the particle surface, as reported before in [18,27]. For the combustion in 100% CO2, the start and the end of the GPC, as well as the end of SR, are also marked in Fig. 3. In addition, the comparison of the combustion in 100% and 50% CO2 shows that
31
both the GPC and the SR last longer in 50% CO2, and that the two combustion stages are clearly separated from each other for both atmospheres. This explicit separation between the two combustion stages is not observed for the combustion in 25% CO2. Here, the flame diameter of the GPC decreases continuously, resulting in a smooth transition from GPC to SR. For most of the recorded particles, a second, much shorter, gas-phase combustion was detected, which appears as a burst after the first transition to a surface reaction. After this second GPC event, the remainder of the particle burns through a second SR. Compared to the combustion in 50% CO2, the duration of the first GPC increases and the flame diameter decreases. For the combustion in 10% CO2, no GPC is observed at all: After ignition, the particle starts to burn directly in the surface reaction regime. Additionally, the duration of the SR significantly increases compared to the other three atmospheres. This phenomenon, that metal gas-phase combustion does not occur for low oxidizer partial pressures, has been reported before by Macek and Semple for the combustion of Beryllium in O2 [32]. These various experimental observations are all consistent with the fact that the adiabatic flame temperature decreases for lower oxidizer (CO2) partial pressures. At some point, the adiabatic flame temperature falls below the boiling temperature of the metal, inhibiting gas-phase combustion (see Glassman’s criterion for metal vapor phase combustion [33]). Most of these qualitative observations can also be confirmed quantitatively: For this purpose, all images of the burning particles shot with the reflex camera are evaluated automatically using an image interpretation macro based on the open source library OpenCV. The macro identifies the distance covered during GPC and SR by searching the starting and ending points of both combustion stages for each recorded particle. For that purpose, each image is scanned for white and red pixels above a defined threshold. To determine the threshold a defined number of random pictures was evaluated manually. Next, the same pictures were interpreted automatically using different thresholds. An optimization routine was established to compare the manually and the automatically derived results iteratively and to improve the threshold value. It is further assumed that the ending point of the GPC equals the starting point of the SR. However, the GPC flame is detached from the particle surface, which leads to an error while calculating the
Fig. 3. Typical examples for reflex photographs of single combusting particles in the investigated atmospheres.
32
M. Schiemann et al. / Fuel 199 (2017) 28–37
measurement uncertainty. To ensure a clear arrangement, the error bars are only plotted for every second measurement point. The total measurement uncertainty consists of three main effects: Both the determination of the starting and the ending point of the GPC is subject to an uncertainty of ±1 px, as several pixels might have similar color values close to the defined threshold. The same error occurs for the determination of the width of the GPC. As the distance covered during GPC has to be corrected using the flame diameter, the total error of the distance covered during GPC sums up to ±4 px. In addition, the error of the measurements of the particle velocity (up to 9%, see Fig. 2) and the uncertainty of the size calibration (±2%) have to be considered.
Fig. 4. Schematical drawing of the adjustment of the distance covered by the particle during GPC.
distance covered by the particle center during GPC. To correct the covered distance, the flame diameter has to be taken into account, as depicted in Fig. 4. The flame diameter can be estimated from the width of the recorded GPC and has to be subtracted from the recorded length of the GPC, as the particle center differs both from the flame edge at the beginning and at the end of the GPC by half of the flame diameter. The determined distances covered during both combustion stages are converted to combustion times using the size calibration and the measured particle velocities. The resulting durations of the GPC (left) and the SR (right) are depicted in Fig. 5 for each atmosphere. As no GPC occurs for the combustion in 10% CO2, only the measured data for the other three atmospheres is presented on the left side of Fig. 5. Error bars are included to depict the total
Gas-phase combustion shows a clear trend from fast reaction (5 ms in 100% CO2) to longer time scales (20 ms in 25% CO2). According with the increase in average duration the frequency distribution gets significantly widened. The analysis of the durations of the SR shows similar results compared to the GPC, except for the order-of-magnitude longer timescales. The durations of the SR also increase for lower CO2 partial pressures, whereas the difference between 100% and 50% CO2 is small. In fact, for the combustion in 100% CO2, the peak of the relative frequency is at 40 ms and for 50% CO2 the peak is at 70 ms, but the average duration of the SR shows very similar results for both atmospheres around 80 ms. For 25% CO2 and 10% CO2 the measured durations increase more clearly for decreasing CO2 partial pressures. In the next step, the pyrometry system was used to point out the transition from GPC to SR using multiple exposure measurements. As the particle moves further downstream during the exposure timeout of 500 ms, six shots of the same particle are generated in each image recorded. Typical examples for the transition from GPC to SR for the atmospheres with 100%, 50% and 25% CO2 are depicted in Fig. 6. For the combustion in 100% the first two shots show a flame diameter of approx. 400 mm while the last four shots
Fig. 5. Results of the duration of the gas-phase combustion (GPC) and surface reaction (SR) for each investigated atmosphere (top) and average durations depending on the molar fraction of CO2 (bottom).
M. Schiemann et al. / Fuel 199 (2017) 28–37
33
Fig. 6. Examples for multiple exposure shots with the pyrometry system (only cam 2) of the transition from gas phase to surface reaction for the combustion in 100%, 50% and 25% CO2.
only have a diameter of approx. 82 mm. Thus, during the first two shots, the particle burns in the gas phase with a flame standing off from the particle surface while, during the last four shots, the reaction takes place at the particle surface. The same effect can be observed for the combustion in 50% CO2. Here, the particle burns in the gas phase (first three shots) with a flame diameter of approx. 238 mm. After the GPC is finished, a SR is also visible. For the combustion in 25% CO2, the combustion behavior is slightly different as no abrupt ending of the GPC is visible. Here, the transition from GPC is smooth as the flame diameter decreases slowly from 128 to 66 mm. The combustion in 10% CO2 is not depicted in Fig. 6 as there is only SR and no transition from GPC to SR. Regarding the ratio of the flame diameter (during GPC) compared to the particle diameter (during SR) df/dp, the combustion in 100% CO2 leads to a df/dp of approx. 4.5, while in 50% CO2 the ratio is approx. 3.3 and in 25% CO2 approx. 1.9. Thus; the ratio df/ dp decreases for lower CO2 partial pressures, as shown in Fig. 7. The uncertainty in the particle or flame diameters is approximately ±20%, leading to an uncertainty in the ratios in Fig. 7 of approximately 30%. In addition to measuring the flame and particle diameters, the pyrometry system has been also used to measure combustion temperatures. Here, only the temperature during SR can be measured, as the flame during GPC is not a gray body radiator [34]. Thus, temperature determination using two-color pyrometry is only reasonable for the SR [18]. As the pyrometry system supplies twodimensional images of the particles, a pair of information containing particle temperature Tp as well as diameter dp is known for each particle recorded. At least 300 particles burning in the surface reaction regime are recorded for each atmosphere. The uncertainty of the pyrometry measurements has previously been calculated as approx. ±48 K for the temperature measurements [18]. As the position of each particle streak depends on the atmosphere, the camera system was adjusted to record burning particles in the middle of each streak. The results are depicted in Fig. 8, in which particles that have a statistical deviation of less than 2-r from both the median temperature and the median diameter are highlighted bold. These highlighted particles form a twodimensional statistical density ellipse for each atmosphere. For further information, the resulting median temperatures and standard deviations are depicted in Table 3. Here, the trend of decreasing temperatures for lower CO2 partial pressures is visible. This observation is supported statistically by comparing the four ellipses from Fig. 8 in one plot as depicted in Fig. 9: The blue ellipse
Fig. 7. Relation of flame diameter during gas-phase combustion (GPC) to particle diameter (df /dp) for the combustion in 100%, 50% and 25% CO2.
(100% CO2) has the highest temperatures while the other ellipses cover lower temperatures for decreasing CO2 partial pressures. In addition, the ellipses for the diluted atmospheres (50%, 25% and 10% CO2) show a clear trend concerning the particle diameter: For smaller particles, the measured surface temperatures decrease. For the combustion in 100% CO2, this effect was not observed. Here, the temperatures show no dependency on the particle diameter and vary around a constant median temperature of 1643 K. To investigate the burnout composition, the sampling probe is inserted bottom up in the reactor (see Fig. 1). For each atmosphere, sampling was conducted at six or more sampling points at the beginning, in the middle, and at the end of each particle streak. Thus, the quantity and the positions of the measurement points depend on the investigated atmosphere. To ensure comparable results, the measurement points are converted to particle residence times, which were determined by integrating the particle velocity measured with the pyrometry system. The extracted mix of solid particles covered with aerosol was analyzed using X-ray diffraction to determine the constituents of the extracted reaction products. The qualitative analysis identified only Li2CO3, Li2O, Li, and LiOH. No carbides or elemental carbon (neither graphite nor amorphous carbon) were observed. In addition, no products containing nitrogen such as Li3N were detected,
34
M. Schiemann et al. / Fuel 199 (2017) 28–37
Fig. 8. Particle statistics of the events burning with surface reaction (SR) for the combustion in 100% CO2, 50% CO2, 25% CO2 and 10% CO2; filled data points have a twodimensional probability density inside of a 2-r interval.
although the reaction of Li and pure N2 has been observed before in the same reactor [17]. Thus, in the presence of CO2, the reaction of lithium with CO2 to Li2O and Li2CO3 is preferred and nitrogen acts only as an inert gas. The formation of LiOH does not occur during the combustion process due to the absence of H2O in the reactor. The particle extraction from the filter and the XRD analysis could not be carried out under inert conditions and, because elemental lithium and Li2O are both hygroscopic, the samples absorb water from the ambient air forming enough LiOH to be detected at Xray diffraction analysis. The quantitative analysis using XRD is difficult since quantitative data maps are not available for Li and LiOH. Only the amount of Li2O and Li2CO3 can be derived from X-ray diffraction analysis. To compute the quantity of Li and LiOH, each probe was also examined with a CHN-analyzer (Leco TruSpec Micro) to gain information on the fraction of carbon and hydrogen. Combining the results of the CHN analysis and the quantitative X-ray diffraction as well as the overall mass balance leads to the quantitative fractions of Li, LiOH, Li2O and Li2CO3. For further information on the calculation routine please refer to [17,18]. The resulting mass fractions from the measurements in the CO2-N2 mixtures are compared to the results from the previous measurements in pure CO2 [18] in Fig. 10. Note that the Y-axis is separated for each atmosphere. As Fig. 10 already contains a lot of parameters and information, error bars from a Gaussian error propagation are not depicted to ensure clear arrangement. The biggest errors (9–15%) occur for the elemental lithium as the lithium fraction is calculated from the overall mass balance and, therefore, the errors of all three other components are summed up in the
Table 3 Median temperatures and standard deviations for the surface reaction. xCO2
100%
50%
25%
10%
Tm [K]
1643 78
1582 45
1539 51
1451 78
r [K]
Fig. 9. Measured particle temperatures during surface reaction for various atmospheres (2-r interval distributions).
error of elemental lithium. The errors of the other fractions are smaller (3–12%). The resulting uncertainties are based on the assumptions that the XRD data includes an uncertainty of approx. 25% and the error of the CHN analysis is approx. 15%, which has been discussed to be a generously calculated upper limit [18].
M. Schiemann et al. / Fuel 199 (2017) 28–37
35
Fig. 10. Burnout constituents at various particle residence times and CO2-N2 mixtures.
The analysis of the burnout constituents shows a decreasing fraction of elemental lithium with increasing residence time for all three atmospheres, while the amount of Li2O and/or Li2CO3 increases. The fraction of LiOH is below 20%, and was detected only due to the absorption of water during the removal of the sample from the probe. Comparing the four atmospheres with different CO2 partial pressures leads to the following observations: The conversion rate of lithium to Li2CO3 is fastest for the atmosphere containing pure CO2 (approx. 180 ms from 0% to 90% Li2CO3). At the end of the particle streak, the major part of the lithium is converted to Li2CO3. The reaction rate of the other atmospheres is much lower and a complete conversion to Li2CO3 was not detected in any of the cases with reduced CO2 concentrations. With decreasing CO2 partial pressure, the amount of Li2CO3 decreases while the fraction of Li2O grows significantly. Thus, the reaction from Li to Li2O seems to be much faster for low CO2 partial pressures than the reaction from Li2O to Li2CO3. At the end of the streak in the 10% CO2 atmosphere, the fraction of elemental lithium remains constant and is not converted to Li2O or Li2CO3. This could be caused by the broad particle diameter distribution, as bigger particles ignite further downstream and therefore are not, or are only partly, reacted. In addition, some particles may not ignite at all since the gas atmosphere is getting colder further downstream. 4. Discussion Comparing the measured burnout times to the other experimental results, several issues have to be addressed. For the com-
bustion in 10% CO2, the fraction of elemental lithium is not completely converted (see Fig. 10), such that it is unclear if the measured burnout times (see Fig. 5) reflect the total combustion process of a particle burning in this atmosphere. As the reaction rate depends on the CO2 partial pressure at least linearly (see. Eqs. (1) and (2)), the reaction rate in 10% CO2 is already low. It is assumed that, during combustion, a product layer is built up on the particle surface, such that the reaction rate decreases towards the end of the surface reaction as the layer gets thicker. Thus, at the end of the combustion in 10% CO2, the reaction rate, and resulting particle temperature, might be too low to be recognized with the reflex camera during the burnout time measurements. This means that the particle might require longer burn times than the presented burnout times suggest. Another interesting effect is the combustion behavior in 25% CO2, where the particles start to burn in the GPC. After the GPC is finished, due to the built-up of the surface product layer, the GPC stops and the SR starts. After a short period of time, a second GPC occurs, which occurs much shorter than the first one. After this second GPC is finished, the particle burns out with a second SR. The reason for the second GPC is not clear; however, it is known that other metal particles (e.g. aluminum, beryllium) form an oxide cap/product layer which can agglomerate on one side of the particle or peel off completely [32,35–38]. These effects might also be possible for lithium. Nevertheless, the measured burning times confirmed and quantified the known combustion mechanisms of GPC and SR for the combustion in pure CO2 and the CO2-N2 mixtures. The analysis showed that gas-phase combustion occurs only for a very short period of time (average values: 5–20 ms), while surface reaction durations are up to ten times higher (average values: 80–210 ms).
36
M. Schiemann et al. / Fuel 199 (2017) 28–37
During the pyrometry experiments, two different phenomena were observed for the temperatures of the SR. The first observation is that the measured temperatures decrease for lower CO2 concentrations. This is a direct result of the strong dependency of the diffusion limited reaction rate n_ 00Li;diff on the CO2 partial pressure pCO2 ;1 . According to Fick’s law (see Eq. (1)) the diffusion rate depends on the particle diameter dp, the universal gas constant R and the film temperature in the particle gas boundary T. D denotes the diffusion constant and pCO2 ;s the CO2 partial pressure at the particle surface. For further information on the mathematical modeling of the surface reaction, please refer to [18,19].
n_ 00Li;diff ¼ 2
D pCO2 ;s Þ ðp dp RT CO2 ;1
ð1Þ
A reduction of the CO2 concentration in the atmosphere leads directly to a decrease of diffusive flow of CO2 towards the particle surface and thus reduces the reaction rate of the surface reaction. This causes a lower heat release rate and, therefore, lower particle temperatures for lower CO2 concentrations. The second observation is that the particle temperatures decrease with decreasing particle sizes for the CO2/N2 mixtures. To understand this effect, the particle energy balance (see Eq. (2)) has to be investigated, taking several combustion phenomena into account:
dp qp cp;p dT p dmp Nu k r ðT p T 1 Þ þ ep rðT 4p T 4w Þ þ Dh ¼ dp dt 6 dt
ð2Þ
On the one hand, the change of the reaction products from Li2CO3 to Li2O for lower CO2 concentrations has to be considered, which implies a two-step reaction from Li to Li2O and from Li2O to Li2CO3. This has a direct effect on the particle energy balance, as it influences the reaction enthalpy significantly. As the quantitative analysis of the product composition is only available for the whole particle size distribution used in the experiments, the calculation of the individual reaction rates for each reaction and each particle size is difficult. Additionally, it has to be taken into account that a part of the particle is reacting during GPC. Thus, a part of the lithium is already consumed when the surface reaction starts. As it is unknown how much Li is consumed during GPC, the particle energy balance cannot be solved without a detailed and complex analysis, involving all data from all the different measurement methods. This analysis is not part of the current work and has to be conducted in future work. In spite of these complexities, it is likely that smaller particles are undergoing more effective heat exchange with the surrounding gas leading to their lower observed temperatures. 5. Conclusions In the current work, the combustion of single lithium particles with pure CO2 and three different CO2-N2 mixtures was investigated experimentally using two-color pyrometry, particle sampling and a reflex camera to measure burnout times. The experiments confirmed previous findings concerning two different combustion phenomena for the combustion of lithium in CO2 [18]. The particles burn in the gas phase first, likely with detached vapor-phase diffusion flames, followed by a reaction that collapses to the particle surface. The flame diameters of the GPC decrease for lower CO2 partial pressures until no GPC was detected for the 10% CO2 atmosphere. Simultaneously, the durations of the GPC increase for decreasing CO2 partial pressures. The combustion continues with a reaction at the particle surface, whose duration also increases with decreasing CO2 partial pressures. Adding N2 to the atmosphere reduces the temperatures of the particles burning in the SR mode, indicating lower reaction rates.
As sampling probe results indicated, N2 does not take part significantly as a reactant and acts only as an inert gas, slowing down the reaction of Li with CO2. In addition, a decreasing CO2 partial pressure shifts the fraction of reaction products from Li2CO3 to Li2O indicating a dominating reaction to Li2O for lower CO2 partial pressures. The combustion times, in both the GPC and SR modes, show a strong dependence on the CO2 concentration. The discussion of the measurements of the burnout times showed that the transition from GPC to SR should be investigated in more detail in future work. While the SR has been described by the authors in previous work up to a certain degree, modeling approaches for GPC of lithium in CO2 and/or CO2-N2 mixtures are not available in the literature. A new model for lithium combustion should include the heat and species transfers in the particle-flame boundary layer, as well as an approach to model the transient build-up of a product layer on the particle surface that limits the diffusion of fresh lithium vapor outward and CO2 inward. Modelling of the multi-stage reaction during the SR would also be needed to fully explain the decreasing particle temperatures observed for smaller particle sizes in the low-CO2 environments, although more effective heat exchange likely reduces the burning temperatures of smaller particles. Acknowledgements This work was supported by the Federal Ministry of Education and Research (BMBF) under Project Number 03EK3007D and by the Ruhr University Research School PLUS, funded by Germany’s Excellence Initiative [DFG GSC 98/3]. References [1] Bergthorson JM, Goroshin S, Soo MJ, Julien P, Palecka J, Frost DL, et al. Direct combustion of recyclable metal fuels for zero-carbon heat and power. Appl Energy 2015;160:368–82. http://dx.doi.org/10.1016/j.apenergy.2015.09.037. [2] Beach DB, Rondinone AJ, Sumpter BG, Labinov SD, Richards RK. Solid-state combustion of metallic nanoparticles: new possibilities for an alternative energy carrier. J Energy Resour Technol 2007;129:29. http://dx.doi.org/ 10.1115/1.2424961. [3] Mandilas C, Karagiannakis G, Konstandopoulos AG, Beatrice C, Lazzaro M, Di Blasio G, et al. Study of basic oxidation and combustion characteristics of aluminum nanoparticles under enginelike conditions. Energy Fuels 2014;28:3430–41. http://dx.doi.org/10.1021/ef5001369. [4] Julien P, Soo M, Goroshin S, Frost DL, Bergthorson JM, Glumac N, et al. Combustion of aluminum suspensions in hydrocarbon flame products. J Propuls Power 2014;30:1047–54. http://dx.doi.org/10.2514/1.B35061. [5] Soo M, Julien P, Goroshin S, Bergthorson JM, Frost DL. Stabilized flames in hybrid aluminum-methane-air mixtures. Proc Combust Inst 2013;34:2213–20. http://dx.doi.org/10.1016/j.proci.2012.05.044. [6] Julien P, Whiteley S, Goroshin S, Soo MJ, Frost DL, Bergthorson JM. Flame structure and particle-combustion regimes in premixed methane-iron-air suspensions. Proc Combust Inst 2014;35:2431–8. http://dx.doi.org/10.1016/j. proci.2014.05.003. [7] Elitzur S, Rosenband V, Gany A. Electric energy storage using aluminum and water for hydrogen production on-demand. Int J Appl Sci Technol 2015;5:112–21. [8] Shkolnikov EI, Zhuk AZ, Vlaskin MS. Aluminum as energy carrier: feasibility analysis and current technologies overview. Renew Sustain Energy Rev 2011;15:4611–23. http://dx.doi.org/10.1016/j.rser.2011.07.091. [9] Elitzur S, Rosenband V, Gany A. Urine and aluminum as a source for hydrogen and clean energy. Int J Hydrogen Energy 2016:3–7. http://dx.doi.org/10.1016/j. ijhydene.2016.05.259. [10] Vlaskin MS, Shkolnikov EI, Bersh AV, Zhuk AZ, Lisicyn AV, Sorokovikov AI, et al. An experimental aluminum-fueled power plant. J Power Sources 2011;196:8828–35. http://dx.doi.org/10.1016/j.jpowsour.2011.06.013. [11] Gany A, Elitzur S, Rosenband V. Compact electric energy storage for marine vehicles using on-board hydrogen production. J Shipp Ocean Eng 2015;5:151–8. http://dx.doi.org/10.17265/2159-5879/2015.04.001. [12] Markowitz MM. Alkali metal-water reactions. J Chem Educ 1963;40:633–6. [13] Markowitz MM, Boryta DA. Lithium metal-gas reactions. J Chem Eng Data 1962;7:586–91. http://dx.doi.org/10.1021/je60015a047. [14] McFarlane E, Tompkins F. Nitridation of lithium. Trans Faraday Soc 1962:997–1007. [15] Gardner M, Altermatt R. Kinetics of the reaction of hydrogen and nitrogen with molten lithium. In: Bach RO, editor. Curr. Appl. Sci. Med. Technol.. New York: Wiley-Interscience; 1985. p. 195–206.
M. Schiemann et al. / Fuel 199 (2017) 28–37 [16] Barnett D, Gil T, Kazimi M. Lithium-mixed gas reactions. Fusion Technol 1989;15:967. [17] Schiemann M, Fischer P, Bergthorson J, Schmid G, Taroata D. Combustion of lithium particles in N2 — reaction rates. Combust Sci Technol 2017;187:169–86. http://dx.doi.org/10.1080/00102202.2016.1196200. [18] Fischer P, Schiemann M, Scherer V, Maas P, Schmid G, Taroata D. Experimental characterization of the combustion of single lithium particles with CO2. Fuel 2015;153:90–101. http://dx.doi.org/10.1016/j.fuel.2015.02.098. [19] Fischer P, Schiemann M, Scherer V, Maas P, Schmid G, Taroata D. A numerical model of the combustion of single lithium particles with CO2. Fuel 2015;160:87–99. http://dx.doi.org/10.1016/j.fuel.2015.07.033. [20] Jeppson DW, Ballif JL, Chou BE, Yuan WW. Lithium literature review: lithium’s properties and interactions. Richland, WA: Hanford Engineering Development Laboratory; 1978. [21] Rhein RA. Lithium combustion : a review. China Lake, CA: 1990. [22] Schiemann M, Bergthorson J, Fischer P, Scherer V, Taroata D, Schmid G. A review on lithium combustion. Appl Energy 2016;162:948–65. http://dx.doi. org/10.1016/j.apenergy.2015.10.172. [23] Jewell S, Kimball SM. Mineral commodity summaries. Reston, Virigina: 2016. [24] Eckert H, Schiemann M, Schmid G, Taroata D, Scherer V, Fischer PH, et al. Conversion of metal carbonate to metal chloride. PCT/EP2015/052,621 (2015). [25] Bergthorson JM, Yavor Y, Palecka J, Georges W, Soo M, Vickery J, et al. Metalwater combustion for clean propulsion and power generation. Appl Energy 2017;186:13–27. http://dx.doi.org/10.1016/j.apenergy.2016.10.033. [26] Chase M, Force J. NIST-JANAF thermochemical tables. J Phys Chem Ref Data 1998. [27] Schiemann M, Fischer P, Scherer V, Schmid G, Taroata D. Combustion of lithium particles: optical measurement methodology and initial results. Chem Eng Technol 2014;37:1600–5. http://dx.doi.org/10.1002/ceat.201400011. [28] Schiemann M, Geier M, Shaddix CR, Vorobiev N, Scherer V. Determination of char combustion kinetics parameters: comparison of point detector and
[29]
[30]
[31]
[32]
[33] [34]
[35] [36] [37]
[38]
37
imaging-based particle-sizing pyrometry. Rev Sci Instrum 2014;85:75114. http://dx.doi.org/10.1063/1.4890438. Schiemann M, Vorobiev N, Scherer V. Stereoscopic pyrometer for char combustion characterization. Appl Opt 2015;54:1097–108. http://dx.doi.org/ 10.1364/AO.54.001097. Cope R, Monson C, Germane G, Hecker W. Improved diameter, velocity, and temperature measurements for char particles in drop-tube reactors. Energy Fuels 1994:925–31. http://dx.doi.org/10.1021/ef00046a016. Mitchell RE. Experimentally determined overall burning rates of coal chars. Combust Sci Technol 1987;53:165–86. http://dx.doi.org/10.1080/ 00102208708947025. Macek A, McKenzie Semple J. Experimental burning rates and combustion mechanisms of single beryllium particles. Symp Combust 1969;12:71–81. http://dx.doi.org/10.1016/S0082-0784(69)80393-0. Glassmann I, Yetter R. Combustion. 4th ed. Amsterdam: Elsevier Inc; 2008. Brockhinke A, Koppmann J, Brockhinke R, Kellermann R, Eckert H, Taroata D, et al. Spectroscopic characterization of lithium combustion. MRS Proc. 2014. 1644:mrsf13-1644-dd08-09. doi: 10.1557/opl.2014.316. Drew C, Gordon A, Knipe R, Kraeutle K, Prentice J, Price E. Metal particle combustion progress report. China Lake, CA: Naval Weapons Center; 1968. Prentice J. Aluminum droplet combustion: rates and mechanisms in wet and dry oxidizers. China Lake, CA: Naval Weapons Center; 1974. Razdobreev AA, Skorik AI, Frolov YV. Ignition and combustion mechanism in aluminum particles. Combust Explos Shock Waves 1977;12:177–82. http://dx. doi.org/10.1007/BF00744882. Zenin A, Kusnezov G, Kolesnikov V. Physics of aluminum particle combustion at convection. In: Zenin A, editor. 38th Aerosp. Sci. Meet. Exhib.. Reston, VA: American Institute of Aeronautics and Astronautics; 2000. doi: 10.2514/ 6.2000-849.
Contents lists available at ScienceDirect
Fuel journal homepage: www.elsevier.com/locate/fuel
Full Length Article
Experimental investigations on the combustion of lithium particles in CO2 and CO2-N2 mixtures Martin Schiemann a,⇑, Peter Fischer a, Jeffrey M. Bergthorson c, Guenter Schmid b, Dan Taroata b a
Department of Energy Plant Technology, Ruhr-Universität Bochum, Germany Siemens AG, Corporate Technology, Erlangen, Germany c McGill University, Montreal, Canada b
h i g h l i g h t s
g r a p h i c a l a b s t r a c t
Lithium particles were burned in CO2-
containing atmospheres with nitrogen dilution. Optical and sampling methods were used to characterize combustion time scales. In nitrogen-diluted CO2 atmospheres, the formation of Li2CO3 is retarded. Combustion temperature decreases with CO2 partial pressure.
a r t i c l e
i n f o
Article history: Received 1 November 2016 Received in revised form 13 February 2017 Accepted 22 February 2017
Keywords: Lithium combustion Particles Burnout times Metal combustion Energy storage
a b s t r a c t Energy cycles based on lithium combustion have been proposed recently and provide an option to combine the benefits of high-density energy storage with carbon-free production of heat, power and chemicals. Knowledge on efficient combustion processes for lithium is limited. In the current work, combustion experiments for single lithium particles (dp = 10–250 mm) burning in CO2 and three different CO2-N2 mixtures are carried out using a laminar flow drop tube reactor. In order to determine combustion times, temperatures and burnout constituents, the burning particles are investigated using three different measurement techniques: a reflex camera, a high speed two-color pyrometer and a sampling probe to collect burnout samples. The results confirm the existence of two combustion stages, which have been observed before: gas-phase combustion and surface reaction. As detailed investigations show, the duration of the gas-phase combustion is approximately ten times shorter compared to the surface reaction. In addition, the burnout times of both combustion stages increase clearly for lower CO2 partial pressures. The reduction of the CO2 partial pressure leads to lower combustion temperatures as well as the production of less Li2CO3 and more Li2O, while no nitrogen compounds are formed. Ó 2017 Elsevier Ltd. All rights reserved.
1. Introduction Recently, the combustion of various metals has been investigated to develop a chemical energy carrier [1]. Renewable energy ⇑ Corresponding author. E-mail address: [email protected] (M. Schiemann). http://dx.doi.org/10.1016/j.fuel.2017.02.079 0016-2361/Ó 2017 Elsevier Ltd. All rights reserved.
can be used to charge the energy carrier (i.e. splitting of metal salts). To discharge the stored energy, the metal is burned in pulverized or sprayed form to increase the surface-to-volume ratio and, therefore, the reaction rate. The heat produced during combustion can be used in internal combustion engines [2,3] or is converted to electricity using a water-steam cycle or Stirling engine [1]. In the past, aluminum and iron have been investigated regard-
M. Schiemann et al. / Fuel 199 (2017) 28–37 Table 1 Exothermic reactions of Li with exhaust gas components calculated with data from [26]. Reaction #
Chemical reaction
Reaction enthalpy [kJ/molLi] (298.15 K, 0.1 MPa)
1 2 3 4 5 6 7
6 Li + N2 ? 2 Li3N Li3N + 3 H2O ? 3 LiOH + NH3 2 Li + 2 CO2 ? Li2CO3 + CO 2 Li + 1 CO2 ? Li2O + CO 4 Li + 1 CO2 ? 2 Li2O + C 2 Li + 2 H2O ? 2 LiOH + H2 4 Li + O2 ? 2 Li2O
54 145 270 157 201 202 299
ing their potentials as chemical energy storage materials [4–11]. Another promising metal for a chemical energy carrier is lithium, as it reacts exothermally with various gases, including O2, H2O, N2 and CO2 [12–16]. Thus, lithium can be burned with the main components of fossil power plant exhaust gas, which enables the possibility to utilize the flue gas constituents. In addition, the reaction of lithium with N2 [17] or CO2 [18,19] produces valuable byproducts such as Li3N (reaction with N2,) or CO (reaction with CO2), which can be further converted to ammonia or synthetic hydrocarbon fuels. A good overview on the combustion of lithium is given in the reviews of Jeppson et al. [20], Rhein [21] and Schiemann et al. [22], who also sum up the combustion of lithium particles or droplets focusing on processes for energy utilization. The major chemical reactions for the different combustion processes are summarized in Table 1. Considering lithium as an energy carrier needs process schemes which go beyond the combustion process. The idea to use the metal as energy carrier and typical metal purification schemes lead to the idea of metal recycling. Focusing on lithium, it will be necessary to recycle the combustion products (Table 1: Li3N (R1), Li2CO3 (R3), Li2O (R4)) to reuse the lithium contained. The lithium carbonate price of 6400 US$ per metric ton
29
(2015, [23]) and the economic value of the chemical products (CO, (R3), NH3 generated from Li3N (R1, [17])) lead to an estimate for the allowed lithium loss per process cycle in the range of one percent [22]. However, there are even cheaper sources of lithium carbonate with prices below 1000 $/tLi2CO3, such that the estimated loss rate can be higher by the corresponding factor. When NH3 production is considered, the value of approx. 500 €/tNH3 increases economic feasibility to a level which reduces the recycling efficiency demand drastically. The recycling step utilizing renewable energy is possible from lithium salts [24], which seems feasible for stationary plants. Lithium-based internal combustion engines would need a very efficient lithium capture, which might be more realizable when hydrogen production from lithium (Table 1, (R6)) is considered [25], but has not been investigated in depth yet. In the past three years, the combustion of lithium particles in CO2 has been investigated both experimentally and numerically [18,19]. For that purpose, a drop tube reactor was used to analyze the combustion behavior of single particles using a two-color pyrometer for the determination of combustion temperatures as well as particle sampling to analyze the burnout composition. It has been shown that, in pure CO2, the combustion occurs in two different stages. After ignition, the particles start to burn in the gas phase with the flame detached from the particle surface. The gas-phase combustion (GPC) ends after a very short period of time (several ms) due to the accumulation of reaction products on the particle surface, which form a product layer. The layer hinders the diffusion of evaporated lithium from the particle surface to the detached flame and thus terminates the GPC and starts the second combustion stage. During the second stage, the reaction occurs at the particle surface. For the surface reaction (SR) in pure CO2, reaction kinetics were derived from experimental data using a two-color pyrometer [18]. In addition, time-dependent particle sampling was conducted to determine the product constituents, which showed a nearly complete conversion to Li2CO3.
Fig. 1. Schematic drawing of the drop tube reactor (DTR) including gas temperature profiles along the reactor tube axis and a example of the particle streak for the combustion in 100% CO2.
30
M. Schiemann et al. / Fuel 199 (2017) 28–37
Table 2 Gas flows (STP) of the investigated atmospheres. Atmosphere CO2/N2 (molar)
V_CO2 [l/min]
V_N2 [l/min]
1/0 0.5/0.5 0.25/0.75 0.1/0.9
60.0 37.3 21.3 7.8
– 37.3 63.8 70.3
The reaction of lithium particles with N2 has been investigated in the same drop tube reactor conducting particle sampling at various particle residence times [17]. The results showed that the main product is Li3N and that the temperature is likely limited by the decomposition temperature of Li3N, which is relatively low (1234 K). Only surface reaction occurs since the combustion temperature is lower than the boiling point of lithium (1620 K). In the current work, further single particle experiments are carried out to determine the combustion behavior in CO2-N2 mixtures. Compared to the results published in [17,18], where the combustion of lithium particles in pure N2 and pure CO2 was analyzed, three different CO2-N2 mixtures are investigated concerning burnout constituents and combustion temperatures. Additionally, the data gathered from pure CO2 experiments are included to show the impact of different dilution grades more clearly. In addition, experiments using a reflex camera are carried out to determine burnout times for the combustion in pure CO2 and in the CO2-N2 mixtures. To ensure comparable results, the particle size distribution is similar to the former experiments in [18]. The results are important for improving our understanding of the fundamental combustion properties of lithium with flue gases.
2. Experimental setup The drop tube reactor (DTR) is presented in Fig. 1 and will be explained only briefly in the following section as it already has been discussed in detail in [18]. The gas flow, consisting of pure CO2 and three different CO2-N2 mixtures (flow rate and mixing ratio in Table 2) is heated up to approx. 830 K with an electrical gas preheater. As CO2 and N2 differ in their heat capacity, density and thermal conductivity, the total gas flow is modified for each atmosphere to provide similar gas temperature profiles in the reactor. The resulting gas flows are presented in Table 2. The hot gas is guided through a ceramic honeycomb, to establish a laminar flow in the reactor. The reactor itself is enveloped by a quartz glass tube (d = 65 mm) to conserve the laminar flow and to provide optical access to the burning particles. The lithium particles (approx. 1 g/ h) are injected from the top into the reactor. The very low particle mass flow rate is guaranteed by a small star feeder. The cylindrical cell size is reduced to 0.5 mm diameter and 1 mm height, such that each cell contains only a small number of particles. As the particles are transported to the reactor inlet through a thin tube of approx. 0.5 m length, the small bunch of particles escaping from each cell is dispersed to let the particles burn isolated from each other. The particles are conveyed using a cold Argon carrier flow (approx. 0.4% (0.25 L/min) of the total gas flow) to prevent the particles from heating or reacting before entering the quartz glass tube. Additionally, the injector, which ends directly at the end of the ceramic honeycomb, is water cooled. For information on the particle production (base material: lithium rod with purity 99.9%; <1500 ppm metal traces, [27]) and handling process as well as the particle size distribution (10–250 mm), please refer to [18]. Fig. 1 also shows gas temperature profiles along the reactor tube axis for each gas mixture investigated. The temperature profiles were measured without Li particles using thermocouples (radiation correction has been carried out as described in [11]).
Fig. 2. Measured particle velocities for the investigated atmospheres.
After entering the reactor, the particles are heated up (heating rate of approx. 104 K/s, as calculated in [19]), melt and ignite. As the particles move further downstream in the tube while burning, they form a glowing particle streak as also depicted in Fig. 1 for the atmosphere in 100% CO2. The combustion is monitored using three different measurement techniques: a reflex camera (Canon EOS 50D), a high speed two-color pyrometer and a sampling probe. The sampling probe is used to extract burnout samples at various residence times for subsequent analysis using X-ray diffraction and CHN-analysis. As quenching of the reacting particles is necessary at the probe inlet, an Ar co-flow is applied in the sampling probe to dilute the reaction atmosphere which is aspirated together with the particles. Additionally, the probe wall is watercooled to reduce temperature as far as possible. As the quartz glass tube provides optical access to the reactor, the pyrometer system can be used to measure particle and flame diameters as well as particle temperatures during combustion. For the measurements presented, it is very crucial that particles burn well separated from each other to avoid affection of the inter-particle gas phase, which in turn would reduce the concentration of available gaseous reaction partners and reduce the burning rate, which increases burning time. The image analysis for temperature (particle) and diameter (particle/flame) measurements was therefore restricted to images which only contain one single combustion event in the field of view (5 * 4 mm). Most of the exposures fulfilled this requirement, which indicates that the criterion of well-separated particles is met. The statistical probability of a line-of-sight overlap is reduced to a minimum, justifying the assumption that the diameter data are not biased by overlapping particles. Both measurement methods have been used before during the experiments in pure CO2. For further information concerning those systems please refer to [18]. The 2D-pyrometry camera system TOSCA (Temperature measurement with electro Optical high Speed CAmeras) is utilized to determine the particle velocities inside the reactor. For this purpose, double-exposure shots of the particles were recorded using very short exposure times (40 ms) and defined exposure timeouts (500 or 800 ms) between the two shots. As the particles move further downstream in the reactor during the exposure timeout, two shots of the same particle are recorded and saved in the same 2dimensional image. As the TOSCA-system has been size calibrated before (resolution 10 mm/px), the measured distances between the two shots can be converted into particle velocities using the size calibration and the defined exposure timeouts. At the same time, the reflex camera is used to record the whole particle trajectory covered during combustion. As the reflex camera has also been size calibrated, the recorded combustion
M. Schiemann et al. / Fuel 199 (2017) 28–37
trajectories can be converted from pixels to millimeters, resulting in the distance covered from ignition to extinction. Integrating the particle velocities for the recorded combustion distances leads to the combustion times [28,29]. 3. Experimental results The particle velocities are measured for each atmosphere at the top and the bottom of the visible particle streak (see Fig. 1) as well as at two more measurement points in between. As the beginning and the end of the streak differ for each atmosphere, different measurement points are utilized depending on the investigated atmosphere. The resulting particle velocities are presented in Fig. 2. To ensure statistical significance, at least 100 particles were recorded for each measurement point. The depicted error bars denote the resulting standard deviation. The lowest particle velocities occur in 100% CO2 (1.05–1.15 m/s), as the total gas volume flow is much lower than for the other atmospheres (see Table 2). The clearly deviating measurement distances from the particle inlet reflect the differences in combustion behavior. Altogether, the measured particle velocities correspond well to the theoretical gas velocities in the reactor tube. This leads to the conclusion that the relative velocity between the particle and the gas is low. Slip free movement of particles of similar size and density have been reported before for other drop tube reactor experiments from Cope et al. and Mitchell [30,31]. In addition, the multiple exposure measurements of the transition from GPC to SR (see Fig. 6) show very round enveloping flames during GPC, which is another indication of no significant relative velocity between particle and gas. In the next step, the burning particles are recorded with the reflex camera. Typical examples of the resulting photographs for each atmosphere are depicted in Fig. 3. The first two photographs in 100% and 50% CO2 show the two combustion mechanisms: the GPC appears as a very bright and white flame, while the SR results in a red-orange glowing of the particle. Here, the GPC flame is much broader than the glowing particle, confirming that the GPC flame is detached from the particle surface, as reported before in [18,27]. For the combustion in 100% CO2, the start and the end of the GPC, as well as the end of SR, are also marked in Fig. 3. In addition, the comparison of the combustion in 100% and 50% CO2 shows that
31
both the GPC and the SR last longer in 50% CO2, and that the two combustion stages are clearly separated from each other for both atmospheres. This explicit separation between the two combustion stages is not observed for the combustion in 25% CO2. Here, the flame diameter of the GPC decreases continuously, resulting in a smooth transition from GPC to SR. For most of the recorded particles, a second, much shorter, gas-phase combustion was detected, which appears as a burst after the first transition to a surface reaction. After this second GPC event, the remainder of the particle burns through a second SR. Compared to the combustion in 50% CO2, the duration of the first GPC increases and the flame diameter decreases. For the combustion in 10% CO2, no GPC is observed at all: After ignition, the particle starts to burn directly in the surface reaction regime. Additionally, the duration of the SR significantly increases compared to the other three atmospheres. This phenomenon, that metal gas-phase combustion does not occur for low oxidizer partial pressures, has been reported before by Macek and Semple for the combustion of Beryllium in O2 [32]. These various experimental observations are all consistent with the fact that the adiabatic flame temperature decreases for lower oxidizer (CO2) partial pressures. At some point, the adiabatic flame temperature falls below the boiling temperature of the metal, inhibiting gas-phase combustion (see Glassman’s criterion for metal vapor phase combustion [33]). Most of these qualitative observations can also be confirmed quantitatively: For this purpose, all images of the burning particles shot with the reflex camera are evaluated automatically using an image interpretation macro based on the open source library OpenCV. The macro identifies the distance covered during GPC and SR by searching the starting and ending points of both combustion stages for each recorded particle. For that purpose, each image is scanned for white and red pixels above a defined threshold. To determine the threshold a defined number of random pictures was evaluated manually. Next, the same pictures were interpreted automatically using different thresholds. An optimization routine was established to compare the manually and the automatically derived results iteratively and to improve the threshold value. It is further assumed that the ending point of the GPC equals the starting point of the SR. However, the GPC flame is detached from the particle surface, which leads to an error while calculating the
Fig. 3. Typical examples for reflex photographs of single combusting particles in the investigated atmospheres.
32
M. Schiemann et al. / Fuel 199 (2017) 28–37
measurement uncertainty. To ensure a clear arrangement, the error bars are only plotted for every second measurement point. The total measurement uncertainty consists of three main effects: Both the determination of the starting and the ending point of the GPC is subject to an uncertainty of ±1 px, as several pixels might have similar color values close to the defined threshold. The same error occurs for the determination of the width of the GPC. As the distance covered during GPC has to be corrected using the flame diameter, the total error of the distance covered during GPC sums up to ±4 px. In addition, the error of the measurements of the particle velocity (up to 9%, see Fig. 2) and the uncertainty of the size calibration (±2%) have to be considered.
Fig. 4. Schematical drawing of the adjustment of the distance covered by the particle during GPC.
distance covered by the particle center during GPC. To correct the covered distance, the flame diameter has to be taken into account, as depicted in Fig. 4. The flame diameter can be estimated from the width of the recorded GPC and has to be subtracted from the recorded length of the GPC, as the particle center differs both from the flame edge at the beginning and at the end of the GPC by half of the flame diameter. The determined distances covered during both combustion stages are converted to combustion times using the size calibration and the measured particle velocities. The resulting durations of the GPC (left) and the SR (right) are depicted in Fig. 5 for each atmosphere. As no GPC occurs for the combustion in 10% CO2, only the measured data for the other three atmospheres is presented on the left side of Fig. 5. Error bars are included to depict the total
Gas-phase combustion shows a clear trend from fast reaction (5 ms in 100% CO2) to longer time scales (20 ms in 25% CO2). According with the increase in average duration the frequency distribution gets significantly widened. The analysis of the durations of the SR shows similar results compared to the GPC, except for the order-of-magnitude longer timescales. The durations of the SR also increase for lower CO2 partial pressures, whereas the difference between 100% and 50% CO2 is small. In fact, for the combustion in 100% CO2, the peak of the relative frequency is at 40 ms and for 50% CO2 the peak is at 70 ms, but the average duration of the SR shows very similar results for both atmospheres around 80 ms. For 25% CO2 and 10% CO2 the measured durations increase more clearly for decreasing CO2 partial pressures. In the next step, the pyrometry system was used to point out the transition from GPC to SR using multiple exposure measurements. As the particle moves further downstream during the exposure timeout of 500 ms, six shots of the same particle are generated in each image recorded. Typical examples for the transition from GPC to SR for the atmospheres with 100%, 50% and 25% CO2 are depicted in Fig. 6. For the combustion in 100% the first two shots show a flame diameter of approx. 400 mm while the last four shots
Fig. 5. Results of the duration of the gas-phase combustion (GPC) and surface reaction (SR) for each investigated atmosphere (top) and average durations depending on the molar fraction of CO2 (bottom).
M. Schiemann et al. / Fuel 199 (2017) 28–37
33
Fig. 6. Examples for multiple exposure shots with the pyrometry system (only cam 2) of the transition from gas phase to surface reaction for the combustion in 100%, 50% and 25% CO2.
only have a diameter of approx. 82 mm. Thus, during the first two shots, the particle burns in the gas phase with a flame standing off from the particle surface while, during the last four shots, the reaction takes place at the particle surface. The same effect can be observed for the combustion in 50% CO2. Here, the particle burns in the gas phase (first three shots) with a flame diameter of approx. 238 mm. After the GPC is finished, a SR is also visible. For the combustion in 25% CO2, the combustion behavior is slightly different as no abrupt ending of the GPC is visible. Here, the transition from GPC is smooth as the flame diameter decreases slowly from 128 to 66 mm. The combustion in 10% CO2 is not depicted in Fig. 6 as there is only SR and no transition from GPC to SR. Regarding the ratio of the flame diameter (during GPC) compared to the particle diameter (during SR) df/dp, the combustion in 100% CO2 leads to a df/dp of approx. 4.5, while in 50% CO2 the ratio is approx. 3.3 and in 25% CO2 approx. 1.9. Thus; the ratio df/ dp decreases for lower CO2 partial pressures, as shown in Fig. 7. The uncertainty in the particle or flame diameters is approximately ±20%, leading to an uncertainty in the ratios in Fig. 7 of approximately 30%. In addition to measuring the flame and particle diameters, the pyrometry system has been also used to measure combustion temperatures. Here, only the temperature during SR can be measured, as the flame during GPC is not a gray body radiator [34]. Thus, temperature determination using two-color pyrometry is only reasonable for the SR [18]. As the pyrometry system supplies twodimensional images of the particles, a pair of information containing particle temperature Tp as well as diameter dp is known for each particle recorded. At least 300 particles burning in the surface reaction regime are recorded for each atmosphere. The uncertainty of the pyrometry measurements has previously been calculated as approx. ±48 K for the temperature measurements [18]. As the position of each particle streak depends on the atmosphere, the camera system was adjusted to record burning particles in the middle of each streak. The results are depicted in Fig. 8, in which particles that have a statistical deviation of less than 2-r from both the median temperature and the median diameter are highlighted bold. These highlighted particles form a twodimensional statistical density ellipse for each atmosphere. For further information, the resulting median temperatures and standard deviations are depicted in Table 3. Here, the trend of decreasing temperatures for lower CO2 partial pressures is visible. This observation is supported statistically by comparing the four ellipses from Fig. 8 in one plot as depicted in Fig. 9: The blue ellipse
Fig. 7. Relation of flame diameter during gas-phase combustion (GPC) to particle diameter (df /dp) for the combustion in 100%, 50% and 25% CO2.
(100% CO2) has the highest temperatures while the other ellipses cover lower temperatures for decreasing CO2 partial pressures. In addition, the ellipses for the diluted atmospheres (50%, 25% and 10% CO2) show a clear trend concerning the particle diameter: For smaller particles, the measured surface temperatures decrease. For the combustion in 100% CO2, this effect was not observed. Here, the temperatures show no dependency on the particle diameter and vary around a constant median temperature of 1643 K. To investigate the burnout composition, the sampling probe is inserted bottom up in the reactor (see Fig. 1). For each atmosphere, sampling was conducted at six or more sampling points at the beginning, in the middle, and at the end of each particle streak. Thus, the quantity and the positions of the measurement points depend on the investigated atmosphere. To ensure comparable results, the measurement points are converted to particle residence times, which were determined by integrating the particle velocity measured with the pyrometry system. The extracted mix of solid particles covered with aerosol was analyzed using X-ray diffraction to determine the constituents of the extracted reaction products. The qualitative analysis identified only Li2CO3, Li2O, Li, and LiOH. No carbides or elemental carbon (neither graphite nor amorphous carbon) were observed. In addition, no products containing nitrogen such as Li3N were detected,
34
M. Schiemann et al. / Fuel 199 (2017) 28–37
Fig. 8. Particle statistics of the events burning with surface reaction (SR) for the combustion in 100% CO2, 50% CO2, 25% CO2 and 10% CO2; filled data points have a twodimensional probability density inside of a 2-r interval.
although the reaction of Li and pure N2 has been observed before in the same reactor [17]. Thus, in the presence of CO2, the reaction of lithium with CO2 to Li2O and Li2CO3 is preferred and nitrogen acts only as an inert gas. The formation of LiOH does not occur during the combustion process due to the absence of H2O in the reactor. The particle extraction from the filter and the XRD analysis could not be carried out under inert conditions and, because elemental lithium and Li2O are both hygroscopic, the samples absorb water from the ambient air forming enough LiOH to be detected at Xray diffraction analysis. The quantitative analysis using XRD is difficult since quantitative data maps are not available for Li and LiOH. Only the amount of Li2O and Li2CO3 can be derived from X-ray diffraction analysis. To compute the quantity of Li and LiOH, each probe was also examined with a CHN-analyzer (Leco TruSpec Micro) to gain information on the fraction of carbon and hydrogen. Combining the results of the CHN analysis and the quantitative X-ray diffraction as well as the overall mass balance leads to the quantitative fractions of Li, LiOH, Li2O and Li2CO3. For further information on the calculation routine please refer to [17,18]. The resulting mass fractions from the measurements in the CO2-N2 mixtures are compared to the results from the previous measurements in pure CO2 [18] in Fig. 10. Note that the Y-axis is separated for each atmosphere. As Fig. 10 already contains a lot of parameters and information, error bars from a Gaussian error propagation are not depicted to ensure clear arrangement. The biggest errors (9–15%) occur for the elemental lithium as the lithium fraction is calculated from the overall mass balance and, therefore, the errors of all three other components are summed up in the
Table 3 Median temperatures and standard deviations for the surface reaction. xCO2
100%
50%
25%
10%
Tm [K]
1643 78
1582 45
1539 51
1451 78
r [K]
Fig. 9. Measured particle temperatures during surface reaction for various atmospheres (2-r interval distributions).
error of elemental lithium. The errors of the other fractions are smaller (3–12%). The resulting uncertainties are based on the assumptions that the XRD data includes an uncertainty of approx. 25% and the error of the CHN analysis is approx. 15%, which has been discussed to be a generously calculated upper limit [18].
M. Schiemann et al. / Fuel 199 (2017) 28–37
35
Fig. 10. Burnout constituents at various particle residence times and CO2-N2 mixtures.
The analysis of the burnout constituents shows a decreasing fraction of elemental lithium with increasing residence time for all three atmospheres, while the amount of Li2O and/or Li2CO3 increases. The fraction of LiOH is below 20%, and was detected only due to the absorption of water during the removal of the sample from the probe. Comparing the four atmospheres with different CO2 partial pressures leads to the following observations: The conversion rate of lithium to Li2CO3 is fastest for the atmosphere containing pure CO2 (approx. 180 ms from 0% to 90% Li2CO3). At the end of the particle streak, the major part of the lithium is converted to Li2CO3. The reaction rate of the other atmospheres is much lower and a complete conversion to Li2CO3 was not detected in any of the cases with reduced CO2 concentrations. With decreasing CO2 partial pressure, the amount of Li2CO3 decreases while the fraction of Li2O grows significantly. Thus, the reaction from Li to Li2O seems to be much faster for low CO2 partial pressures than the reaction from Li2O to Li2CO3. At the end of the streak in the 10% CO2 atmosphere, the fraction of elemental lithium remains constant and is not converted to Li2O or Li2CO3. This could be caused by the broad particle diameter distribution, as bigger particles ignite further downstream and therefore are not, or are only partly, reacted. In addition, some particles may not ignite at all since the gas atmosphere is getting colder further downstream. 4. Discussion Comparing the measured burnout times to the other experimental results, several issues have to be addressed. For the com-
bustion in 10% CO2, the fraction of elemental lithium is not completely converted (see Fig. 10), such that it is unclear if the measured burnout times (see Fig. 5) reflect the total combustion process of a particle burning in this atmosphere. As the reaction rate depends on the CO2 partial pressure at least linearly (see. Eqs. (1) and (2)), the reaction rate in 10% CO2 is already low. It is assumed that, during combustion, a product layer is built up on the particle surface, such that the reaction rate decreases towards the end of the surface reaction as the layer gets thicker. Thus, at the end of the combustion in 10% CO2, the reaction rate, and resulting particle temperature, might be too low to be recognized with the reflex camera during the burnout time measurements. This means that the particle might require longer burn times than the presented burnout times suggest. Another interesting effect is the combustion behavior in 25% CO2, where the particles start to burn in the GPC. After the GPC is finished, due to the built-up of the surface product layer, the GPC stops and the SR starts. After a short period of time, a second GPC occurs, which occurs much shorter than the first one. After this second GPC is finished, the particle burns out with a second SR. The reason for the second GPC is not clear; however, it is known that other metal particles (e.g. aluminum, beryllium) form an oxide cap/product layer which can agglomerate on one side of the particle or peel off completely [32,35–38]. These effects might also be possible for lithium. Nevertheless, the measured burning times confirmed and quantified the known combustion mechanisms of GPC and SR for the combustion in pure CO2 and the CO2-N2 mixtures. The analysis showed that gas-phase combustion occurs only for a very short period of time (average values: 5–20 ms), while surface reaction durations are up to ten times higher (average values: 80–210 ms).
36
M. Schiemann et al. / Fuel 199 (2017) 28–37
During the pyrometry experiments, two different phenomena were observed for the temperatures of the SR. The first observation is that the measured temperatures decrease for lower CO2 concentrations. This is a direct result of the strong dependency of the diffusion limited reaction rate n_ 00Li;diff on the CO2 partial pressure pCO2 ;1 . According to Fick’s law (see Eq. (1)) the diffusion rate depends on the particle diameter dp, the universal gas constant R and the film temperature in the particle gas boundary T. D denotes the diffusion constant and pCO2 ;s the CO2 partial pressure at the particle surface. For further information on the mathematical modeling of the surface reaction, please refer to [18,19].
n_ 00Li;diff ¼ 2
D pCO2 ;s Þ ðp dp RT CO2 ;1
ð1Þ
A reduction of the CO2 concentration in the atmosphere leads directly to a decrease of diffusive flow of CO2 towards the particle surface and thus reduces the reaction rate of the surface reaction. This causes a lower heat release rate and, therefore, lower particle temperatures for lower CO2 concentrations. The second observation is that the particle temperatures decrease with decreasing particle sizes for the CO2/N2 mixtures. To understand this effect, the particle energy balance (see Eq. (2)) has to be investigated, taking several combustion phenomena into account:
dp qp cp;p dT p dmp Nu k r ðT p T 1 Þ þ ep rðT 4p T 4w Þ þ Dh ¼ dp dt 6 dt
ð2Þ
On the one hand, the change of the reaction products from Li2CO3 to Li2O for lower CO2 concentrations has to be considered, which implies a two-step reaction from Li to Li2O and from Li2O to Li2CO3. This has a direct effect on the particle energy balance, as it influences the reaction enthalpy significantly. As the quantitative analysis of the product composition is only available for the whole particle size distribution used in the experiments, the calculation of the individual reaction rates for each reaction and each particle size is difficult. Additionally, it has to be taken into account that a part of the particle is reacting during GPC. Thus, a part of the lithium is already consumed when the surface reaction starts. As it is unknown how much Li is consumed during GPC, the particle energy balance cannot be solved without a detailed and complex analysis, involving all data from all the different measurement methods. This analysis is not part of the current work and has to be conducted in future work. In spite of these complexities, it is likely that smaller particles are undergoing more effective heat exchange with the surrounding gas leading to their lower observed temperatures. 5. Conclusions In the current work, the combustion of single lithium particles with pure CO2 and three different CO2-N2 mixtures was investigated experimentally using two-color pyrometry, particle sampling and a reflex camera to measure burnout times. The experiments confirmed previous findings concerning two different combustion phenomena for the combustion of lithium in CO2 [18]. The particles burn in the gas phase first, likely with detached vapor-phase diffusion flames, followed by a reaction that collapses to the particle surface. The flame diameters of the GPC decrease for lower CO2 partial pressures until no GPC was detected for the 10% CO2 atmosphere. Simultaneously, the durations of the GPC increase for decreasing CO2 partial pressures. The combustion continues with a reaction at the particle surface, whose duration also increases with decreasing CO2 partial pressures. Adding N2 to the atmosphere reduces the temperatures of the particles burning in the SR mode, indicating lower reaction rates.
As sampling probe results indicated, N2 does not take part significantly as a reactant and acts only as an inert gas, slowing down the reaction of Li with CO2. In addition, a decreasing CO2 partial pressure shifts the fraction of reaction products from Li2CO3 to Li2O indicating a dominating reaction to Li2O for lower CO2 partial pressures. The combustion times, in both the GPC and SR modes, show a strong dependence on the CO2 concentration. The discussion of the measurements of the burnout times showed that the transition from GPC to SR should be investigated in more detail in future work. While the SR has been described by the authors in previous work up to a certain degree, modeling approaches for GPC of lithium in CO2 and/or CO2-N2 mixtures are not available in the literature. A new model for lithium combustion should include the heat and species transfers in the particle-flame boundary layer, as well as an approach to model the transient build-up of a product layer on the particle surface that limits the diffusion of fresh lithium vapor outward and CO2 inward. Modelling of the multi-stage reaction during the SR would also be needed to fully explain the decreasing particle temperatures observed for smaller particle sizes in the low-CO2 environments, although more effective heat exchange likely reduces the burning temperatures of smaller particles. Acknowledgements This work was supported by the Federal Ministry of Education and Research (BMBF) under Project Number 03EK3007D and by the Ruhr University Research School PLUS, funded by Germany’s Excellence Initiative [DFG GSC 98/3]. References [1] Bergthorson JM, Goroshin S, Soo MJ, Julien P, Palecka J, Frost DL, et al. Direct combustion of recyclable metal fuels for zero-carbon heat and power. Appl Energy 2015;160:368–82. http://dx.doi.org/10.1016/j.apenergy.2015.09.037. [2] Beach DB, Rondinone AJ, Sumpter BG, Labinov SD, Richards RK. Solid-state combustion of metallic nanoparticles: new possibilities for an alternative energy carrier. J Energy Resour Technol 2007;129:29. http://dx.doi.org/ 10.1115/1.2424961. [3] Mandilas C, Karagiannakis G, Konstandopoulos AG, Beatrice C, Lazzaro M, Di Blasio G, et al. Study of basic oxidation and combustion characteristics of aluminum nanoparticles under enginelike conditions. Energy Fuels 2014;28:3430–41. http://dx.doi.org/10.1021/ef5001369. [4] Julien P, Soo M, Goroshin S, Frost DL, Bergthorson JM, Glumac N, et al. Combustion of aluminum suspensions in hydrocarbon flame products. J Propuls Power 2014;30:1047–54. http://dx.doi.org/10.2514/1.B35061. [5] Soo M, Julien P, Goroshin S, Bergthorson JM, Frost DL. Stabilized flames in hybrid aluminum-methane-air mixtures. Proc Combust Inst 2013;34:2213–20. http://dx.doi.org/10.1016/j.proci.2012.05.044. [6] Julien P, Whiteley S, Goroshin S, Soo MJ, Frost DL, Bergthorson JM. Flame structure and particle-combustion regimes in premixed methane-iron-air suspensions. Proc Combust Inst 2014;35:2431–8. http://dx.doi.org/10.1016/j. proci.2014.05.003. [7] Elitzur S, Rosenband V, Gany A. Electric energy storage using aluminum and water for hydrogen production on-demand. Int J Appl Sci Technol 2015;5:112–21. [8] Shkolnikov EI, Zhuk AZ, Vlaskin MS. Aluminum as energy carrier: feasibility analysis and current technologies overview. Renew Sustain Energy Rev 2011;15:4611–23. http://dx.doi.org/10.1016/j.rser.2011.07.091. [9] Elitzur S, Rosenband V, Gany A. Urine and aluminum as a source for hydrogen and clean energy. Int J Hydrogen Energy 2016:3–7. http://dx.doi.org/10.1016/j. ijhydene.2016.05.259. [10] Vlaskin MS, Shkolnikov EI, Bersh AV, Zhuk AZ, Lisicyn AV, Sorokovikov AI, et al. An experimental aluminum-fueled power plant. J Power Sources 2011;196:8828–35. http://dx.doi.org/10.1016/j.jpowsour.2011.06.013. [11] Gany A, Elitzur S, Rosenband V. Compact electric energy storage for marine vehicles using on-board hydrogen production. J Shipp Ocean Eng 2015;5:151–8. http://dx.doi.org/10.17265/2159-5879/2015.04.001. [12] Markowitz MM. Alkali metal-water reactions. J Chem Educ 1963;40:633–6. [13] Markowitz MM, Boryta DA. Lithium metal-gas reactions. J Chem Eng Data 1962;7:586–91. http://dx.doi.org/10.1021/je60015a047. [14] McFarlane E, Tompkins F. Nitridation of lithium. Trans Faraday Soc 1962:997–1007. [15] Gardner M, Altermatt R. Kinetics of the reaction of hydrogen and nitrogen with molten lithium. In: Bach RO, editor. Curr. Appl. Sci. Med. Technol.. New York: Wiley-Interscience; 1985. p. 195–206.
M. Schiemann et al. / Fuel 199 (2017) 28–37 [16] Barnett D, Gil T, Kazimi M. Lithium-mixed gas reactions. Fusion Technol 1989;15:967. [17] Schiemann M, Fischer P, Bergthorson J, Schmid G, Taroata D. Combustion of lithium particles in N2 — reaction rates. Combust Sci Technol 2017;187:169–86. http://dx.doi.org/10.1080/00102202.2016.1196200. [18] Fischer P, Schiemann M, Scherer V, Maas P, Schmid G, Taroata D. Experimental characterization of the combustion of single lithium particles with CO2. Fuel 2015;153:90–101. http://dx.doi.org/10.1016/j.fuel.2015.02.098. [19] Fischer P, Schiemann M, Scherer V, Maas P, Schmid G, Taroata D. A numerical model of the combustion of single lithium particles with CO2. Fuel 2015;160:87–99. http://dx.doi.org/10.1016/j.fuel.2015.07.033. [20] Jeppson DW, Ballif JL, Chou BE, Yuan WW. Lithium literature review: lithium’s properties and interactions. Richland, WA: Hanford Engineering Development Laboratory; 1978. [21] Rhein RA. Lithium combustion : a review. China Lake, CA: 1990. [22] Schiemann M, Bergthorson J, Fischer P, Scherer V, Taroata D, Schmid G. A review on lithium combustion. Appl Energy 2016;162:948–65. http://dx.doi. org/10.1016/j.apenergy.2015.10.172. [23] Jewell S, Kimball SM. Mineral commodity summaries. Reston, Virigina: 2016. [24] Eckert H, Schiemann M, Schmid G, Taroata D, Scherer V, Fischer PH, et al. Conversion of metal carbonate to metal chloride. PCT/EP2015/052,621 (2015). [25] Bergthorson JM, Yavor Y, Palecka J, Georges W, Soo M, Vickery J, et al. Metalwater combustion for clean propulsion and power generation. Appl Energy 2017;186:13–27. http://dx.doi.org/10.1016/j.apenergy.2016.10.033. [26] Chase M, Force J. NIST-JANAF thermochemical tables. J Phys Chem Ref Data 1998. [27] Schiemann M, Fischer P, Scherer V, Schmid G, Taroata D. Combustion of lithium particles: optical measurement methodology and initial results. Chem Eng Technol 2014;37:1600–5. http://dx.doi.org/10.1002/ceat.201400011. [28] Schiemann M, Geier M, Shaddix CR, Vorobiev N, Scherer V. Determination of char combustion kinetics parameters: comparison of point detector and
[29]
[30]
[31]
[32]
[33] [34]
[35] [36] [37]
[38]
37
imaging-based particle-sizing pyrometry. Rev Sci Instrum 2014;85:75114. http://dx.doi.org/10.1063/1.4890438. Schiemann M, Vorobiev N, Scherer V. Stereoscopic pyrometer for char combustion characterization. Appl Opt 2015;54:1097–108. http://dx.doi.org/ 10.1364/AO.54.001097. Cope R, Monson C, Germane G, Hecker W. Improved diameter, velocity, and temperature measurements for char particles in drop-tube reactors. Energy Fuels 1994:925–31. http://dx.doi.org/10.1021/ef00046a016. Mitchell RE. Experimentally determined overall burning rates of coal chars. Combust Sci Technol 1987;53:165–86. http://dx.doi.org/10.1080/ 00102208708947025. Macek A, McKenzie Semple J. Experimental burning rates and combustion mechanisms of single beryllium particles. Symp Combust 1969;12:71–81. http://dx.doi.org/10.1016/S0082-0784(69)80393-0. Glassmann I, Yetter R. Combustion. 4th ed. Amsterdam: Elsevier Inc; 2008. Brockhinke A, Koppmann J, Brockhinke R, Kellermann R, Eckert H, Taroata D, et al. Spectroscopic characterization of lithium combustion. MRS Proc. 2014. 1644:mrsf13-1644-dd08-09. doi: 10.1557/opl.2014.316. Drew C, Gordon A, Knipe R, Kraeutle K, Prentice J, Price E. Metal particle combustion progress report. China Lake, CA: Naval Weapons Center; 1968. Prentice J. Aluminum droplet combustion: rates and mechanisms in wet and dry oxidizers. China Lake, CA: Naval Weapons Center; 1974. Razdobreev AA, Skorik AI, Frolov YV. Ignition and combustion mechanism in aluminum particles. Combust Explos Shock Waves 1977;12:177–82. http://dx. doi.org/10.1007/BF00744882. Zenin A, Kusnezov G, Kolesnikov V. Physics of aluminum particle combustion at convection. In: Zenin A, editor. 38th Aerosp. Sci. Meet. Exhib.. Reston, VA: American Institute of Aeronautics and Astronautics; 2000. doi: 10.2514/ 6.2000-849.