Sonnenbergstrasse 16, CH-8645 Rapperswil-Jona, Switzerland

Supplementary Data for

Investigations to convert CO2, NaCl and H2O into Na2CO3 and HCl by thermal solar energy with high solar efficiency

Martin Forster*

Sonnenbergstrasse 16, CH-8645 Rapperswil-Jona, Switzerland

1. Experimental

Analytical grade MgCl2•6H2O (coarse) and 1N HCl (Sigma-Aldrich), indicator bromothymol blue (Omikron) and analytical grade 10N NaOH (Hänseler) were used as received. Coarse MgCl2•6H2O was used to simulate real applications. Weighing was done with a scale (Mettler-Toledo) calibrated to +/-0.0002 g.

For the experiments the same equipment, slightly modified, was used as in Lit. [S1]. The reaction chamber for the thermo-chemical experiments of Eq. (S1) was a quartz tube 2 of 34 mm inner diameter and 1000 mm length, see Fig. S1. The quartz tube could be heated electrically with oven 1 (Heraeus) and a heating coil 11. Both ends of the quartz tube were equipped with flanges 3 (entrance) and 3' (exit) of stainless steel with inner Teflon lining. The flanges could be heated to T > 100 °C in order to avoid condensation. A stainless steel tube 4 with small holes 5 at the end extended into the hot zone of the oven and delivered water vapor by pumping liquid distilled water 6 into the stainless steel tube by an automatic syringe. The small holes 5 of the stainless steel tube 4 were directed up- and backwards delivering a smooth flow of water vapor 8 above the sample holder 7 towards the exit flange. For experiments which produced highly concentrated HClaq the stainless steel tube was exchanged for a quartz tube. The flow of water could be regulated from 6 to 120 mLh-1, yielding a velocity of water vapor above the sample holder from 1 to 29 cms-1 depending on a temperature between 300°C to 500°C and on the flow rate. At the entrance flange 3 air 12 could be pumped into the quartz tube 2 by a peristaltic pump.

As sample holder a quartz boat 7 with an outer diameter Da = 26 mm and inner dimensions DixL = 23x197 mm, and being cut off in the upper third was used yielding a volume of 71 cm3. The area of the cross-sections of the quartz boat and of the free space above the quartz boat were 53% and 47%, respectively. The quartz boat contained 44 g of coarse MgCl2•6H2O and was placed in the quartz tube either in the center of oven 1 at position a or at position b, see Fig. 1. At the exit flange 3' two cold traps 9 in series condensed the vapors from the thermochemical reaction and, in an adjoined washing flask 10 with distilled water, eventually gaseous HCl was collected. Temperatures at the point of the sample = middle of oven 1 = point a, within the heating coil 11 = point b and at both flanges were measured with stainless steel thermocouples of type K. At the point of the sample the thermocouple was surrounded by a ceramic tube of Degussit. Temperature data were collected digitally. For the sample holder placed at point a the temperature variation along the relatively long sample holder was +/- 8.5 °C at T = 544 °C. Similar temperature variations can occur in real solar troughs [S2]. Water vapor was produced in the quartz tube 2 at temperatures T > 150°C and the reaction temperature could be stabilized to +/- 3°C. The condensate (HCl) of a thermo-chemical reaction according to Eq. (S1) was titrated with 10N NaOH to pH = 7 (bromothymol blue) and the MgO formed was weighed. From these data the yield of the reaction was calculated.

MgCl2•6H2O(s) + xH2O(l) -> MgO(s) + 2HCl•((5+x)/2)H2O(l) (S1)

Fig. S1: Experimental setup for the reactions of MgCl2•6H2O with H2O.

Two types of experiments were performed.

Experiment Type I:

At position a in Fig. S1 the temperature was set to T = Treaction + ca. 250°C. The sample was kept at position b for 20 min with T = 170 - 180°C to melt MgCl2•6H2O and the system was purged for 15 min with water vapor of 60 ml /h H2O. The sample was then shifted from position b to a and the temperature fell to Treaction. The sample was kept at a for 45 min at Treaction +/- 3°C and varying amounts of water vapor were added. The sample was then shifted from a to b and the system purged for 15 min with water vapor of 60 ml/h H2O. The sample was removed, then weighed and the HCl produced was titrated. From these data the percentage yield hHCl of reaction (S1) could be calculated; see main paper.

Experiment Type II:

In these experiments MgCl2•6H2O was first dried in a drying stage and then the reaction stage of Eq. (S1) followed. At position a in Fig. S1 the temperature was set to T = 225 °C. The sample was kept at position b for 20 min with T = 170 - 180°C to melt MgCl2•6H2O and the system was purged for 15 min with water vapor of 60 ml/h H2O and 57 ml/min air. The sample was then shifted from position b to a and the temperature fell to T = 205°C ... 215°C. The sample was kept at this temperature at a for 90 min with 57 ml/min air to remove crystal water from MgCl2•6H2O and yielded MgCl2•xH2O together with small amounts of Mg(OH)Cl and HCl [S3]. In one experiment the drying time was increased to 200 min. During this time and with this flow of air the relation between the molar amounts of air and MgCl2•6H2O was 1:1, calculated for air = 80% N2 + 20 % O2. The sample was then shifted from position a to b and the system purged for 15 min with water vapor of 60 ml /h H2O and 57 ml/min air. The sample was removed, then weighed and the HCl produced was titrated.

Now at position a in Fig. 4 the temperature was set to T = Treaction + ca. 150°C. The sample of MgCl2•xH2O was kept at position b for 20 min with T = 170 - 180°C in order to heat up MgCl2•xH2O. The system was purged for 15 min with water vapor of 60 ml/h H2O and 57 ml/min air and the sample was shifted from position b to a, thereby lowering the temperature to Treaction. The sample was kept at a for 45 min at Treaction +/- 3°C with 57 ml/min air and varying amounts of water vapor were added. The sample was then shifted from a to b and the system purged for 15 min with water vapor of 60 ml/h H2O and 57 ml/min air. The sample was removed, then weighed and the HCl produced was titrated.

From the data taken after the drying and after the reaction stage the yield of the HCl-formation of these two stages and the residual amount of H2O in MgCl2•xH2O could be calculated. For the calculation of the overall percentage yield hHCl of reaction (S1) the amount of HCl from the drying step and from reaction (S1) have been added.

In experiments of type II rather small amounts of water were used and therefore air was needed as a second carrier gas besides water vapor to remove the HCl formed from the reaction site. Since at T > 400°C HCl might react very slowly with O2 according to the Deacon reaction (S2) to Cl2 and H2O

2HCl + 0.5O2 -> H2O + Cl2, (S2)

for experiments of type II the washing flask 10 contained some NaOH. By this measure Cl2 from Eq. S2 was consumed by NaOH according to Eq. (S3)

Cl2 + 2NaOH -> NaOCl + NaCl + H2O (S3)

and NaOCl then decomposed according to Eq. (S4)

NaOCl -> NaCl + 0.5O2. (S4)

Subsequently the decomposed HCl was compensated by reactions (S3) - (S4) and, taking into account the NaOH originally contained in the washing flask 10 during the titration, the correct amount of HCl formed in reaction (S1) was obtained. However, similar experiments performed with and without NaOH in the washing flask 10 showed no difference in the yield of HCl beyond the experimental error of +/- 2%. Obviously reaction (S2) occurred to only a very small degree.

2. Use of very diluted solutions in the MgCl2/MgO modified ammonia soda process

Lit. [S4] discusses the use of magnesite MgCO3 instead of limestone CaCO3 in the conventional Solvay ammonia soda process and proclaims that in reaction (S5) the evaporation of (y-5)H2O(l) would consume 20 GJ per t of Na2CO3 produced.

2NH4Cl (in y/2 H2O)(l) + MgO(s) + yH2O(l) -> 2NH3(g) + MgCl2•6H2O(s) + (y-5)H2O(g) (S5)

Therefore, according to [S4], reaction (S5) would make such a magnesite modified Solvay ammonia soda process uneconomic.

Assuming no heat recovery for reaction (S5), these 20 GJ per t of Na2CO3 correspond to y = 50 in Eq. (S5) and mean that a highly diluted solution would be used for Eq. (S5). With heat recovery of 75% in Eq. (S5), y would become even y >= 200, corresponding to a very highly diluted solution. To evaporate such an enormous amount of water would also need an enormous amount of energy, making the process uneconomic indeed.

Therefore, as calculated in chapter 2 and shown experimentally in chapter 4of the main paper, by keeping the concentrations high the MgCl2/MgO modified ammonia soda process becomes both thermodynamically and economically favorable as was outlined already in Lit. [S1].

3. Estimated influence of the chemical reactor on the solar efficiency

For an actual application the suggested system of two reaction chambers illuminated by two solar troughs described in the paper could look as in Fig. S2. Boats containing MgCl2•6H2O(s) are shifted through the two reaction chambers 1 and 2

Fig. S2: Suggestion for a combination of reaction chamber I at T = 215°C and II at T = 525°C irradiated by two solar troughs to perform reaction (9'); dark small rectangles = boats filled; white small rectangles = boats emptied; pointed line = gaseous curtain; HE heat exchanger

where reaction (S6) occurs:

MgCl2•6H2O(s) + 1.8H2O(l) -> MgO(s) + 2HCl•3.4H2O(l) (S6)

with an actual solar efficiency h0 (10) = 21.5% (for the definition of h0 (10) see the main paper). This efficiency considers only the thermodynamics of the chemical reactions.

The influence of the reaction chambers and of the boats on h0 (10) can be estimated as follows:

1. The reaction chambers keep the indicated temperatures. Only the boats change their temperatures while moving through the two reaction chambers.

2. Assume that boats made of quartz with a thickness of wall = 3 mm and inner dimensions of LxBxH 1000x200xY+20 mm, where Y = 20, 50 or 100 mm = filling height for MgCl2•6H2O(s). Furthermore, assume that 75 % of the sensible heat of the boats can be recovered.

3. If such boats are then included in the thermodynamic calculations the real solar efficiency h0 (10) for Eq. (S6) becomes 21.18%, 21.34% and 21.39 %, respectively for the three types of boats. If the sensible heat of the boats cannot be recovered the corresponding numbers read 20.3%, 20.9% and 21.1%.

This small dependency of h0 (10) on the chemical reactor stems from the fact that the changes of enthalpy of the chemical compounds during the reaction are far much larger than the changes of enthalpy of the boat material. For a boat with LxBxH 1000x200xY+20 mm with Y = 20 mm, the enthalpies of reaction (S6) for a boat filled up to 20 mm height are:

DrH 298-800-298 (chemicals, 75 % recuperation) = 91'238 kJ, DrH 298-800-298 (boat, 75% recuperation) = 442 kJ

4. Relative sizes of the solar troughs in Fig. S2

For reaction (S6) the necessary enthalpies at the two temperatures T = 215 °C and T = 525 °C amount to 187 kJ and 125 kJ per formula conversion, respectively. The boats should be kept at these two temperatures for 90 - 200 min. and 45 min., respectively. These times can be adjusted by the length of the two chambers I and II. The solar troughs in Fig. S2 now have to develop the correct temperatures T = 215 °C and T = 525 °C and deliver the correct enthalpies to the boats at every point along the way of the boats through the two chambers. Therefore, to fulfill these tasks the sizes and concentration ratios of the two solar troughs will have to be optimized for a process plant according to the average solar radiation available at the geographical position of this plant.

5. Literature

[S1] M. Forster, J. Clean. Prod., 23(2013)195-208.

[S2] S.D. Odeh, G.L. Morrison, M. Behnia, Solar Energy, 62(1998)395-406.

[S3] Q. Huang, G. Lu, J. Wang, J. Yu, J. Anal. Appl. Pyrolysis, 91(2011)159-164.

[S4] G. Steinhauser, J. Clean. Prod.,16(2008)833-841.