Concept development to extract sodium sulfate from an aqueous solution

(1)

Concept development to extract sodium sulfate from an aqueous solution

André Selander

Sustainable Energy Engineering, master's level 2021

Luleå University of Technology

Department of Engineering Sciences and Mathematics

(2)

Acknowledgments

This thesis was done at the energy laboratory at Luleå University of Technology. Access to the laboratory was provided thanks to Prof. Kentaro Umeki from Luleå University of Technology which was my examinator and supervisor for this thesis.

I want to thank Therese Nylander, which was my supervisor provided by SCA Energy, which was a huge support for this thesis. I also want to thank Anders Edling Hultgren to provide this thesis. The thought of doing my master’s thesis for SCA Energy has existed since the summer of 2019 when I started working for SCA Energy. SCA has helped me develop as an engineer and I hope the results of this thesis show my gratitude for this opportunity. A huge thanks to Dr. Christian Kugge from SCA R&D for all the support through the years, and especially through this thesis, with a final thanks to Kostiantyn Kulyk for the help understanding the FTIR analysis.

(3)

I

Abstract

Now when the interest is increasing to reach a sustainable infrastructure, one possibility SCA is experimenting with is the possibility to produce renewable hydrocarbons from black liquor which can be extracted from a Kraft process. However, when extracting the black liquor, a lot of sodium-based compounds are removed from the recovery process and when hydrocarbons are produced in SCA’s biorefinery, these compounds are caught in an aqueous solution. The aqueous solution is received at 50°C, and the sodium-based compounds are mainly sodium sulfate and sodium carbonate, where the solution do also contain organic compounds and a solvent that is used in the biorefinery.

This thesis focused on building a concept to extract sodium sulfate from the aqueous solution.

The thesis did also include if any additional preparatory work needs to be done to the solution before extracting sodium sulfate. Finally, a flow chart that maps the energy needed for the process was created.

The method that was used was crystallisation by cooling the solution. By cooling the solution, sodium sulfates solubility decreases which will result in that sodium sulfate falls out of the solution as crystals. It was determined that the solvent that the solution contains should be extracted if the solvents boiling temperature is below 100°C. Further, by cooling the solution under stirring to 15°C with a residence time of 3 hours, unwanted compounds can be

extracted. By later cooling the solution under stirring to 5°C with a residence time of 1 hour, it gave sodium sulfate decahydrate (Na2SO4·10H2O) with small amounts of organic

compounds. By removing the water, the dry product reached a purity of 94wt% sodium sulfate with a yield of 12% (mass of dry product/mass of aqueous solution). This result reached the specific objectives that were set at the start of this thesis, which was to reach a purity of 90wt% sodium sulfate with a yield of 5%.

The energy intensity for evaporating the solvent is expected to be high. It highly depends on which solvent is used. However, this process can use the lowest quality of steam that is

(4)

II available from the pulp mill. It is expected that the cooling will require high amounts of

cooling water and a high investment cost for the heat exchanger. Yet, this is a vital part of the process to reduce the need for coolers which is powered by electricity.

(5)

III

Sammanfattning

Nu när intresset ökar, för att nå en hållbar infrastruktur, så experimenterat SCA med möjligheten att producera förnybara kolväten från svartlut som kan extraheras från en

sulfatprocess. Vid extrahering av svartluten tas dock mycket natriumbaserade föreningar bort från återvinningsprocessen och när kolväten produceras i SCA:s bioraffinaderi fastnar dessa föreningar i en vattenlösning. Den lösningen tas emot vid 50°C och de natriumbaserade föreningarna är huvudsakligen natriumsulfat och natriumkarbonat, där lösningen också innehåller organiska föreningar och ett lösningsmedel som används i bioraffinaderiet.

Denna avhandling fokuserade på att bygga ett koncept för att extrahera natriumsulfat från vattenlösningen. Avhandlingen omfattade också om ytterligare förberedande arbete måste göras av lösningen innan man extraherar natriumsulfat. Slutligen skapades ett flödesschema som kartlägger den energi som behövs för processen.

Metoden som bestämde sig för att användas var kristallisering genom kylning av lösningen.

Genom att kyla lösningen minskar lösligheten av natriumsulfater vilket leder till att natriumsulfat faller ut ur lösningen som kristaller. Det bestämdes att lösningsmedlet som lösningen innehåller skulle extraheras om lösningsmedlets koktemperatur är under 100°C.

Vidare, genom att kyla lösningen under omrörning till 15°C med en uppehållstid på 3 timmar, kan oönskade ämnen extraheras. Genom att senare kyla lösningen under omrörning till 5°C med en uppehållstid på 1 timme gav natriumsulfatdekahydrat (Na2SO4·10H2O) med små mängder organiska föreningar. Genom att avlägsna vattnet nådde den torra produkten en renhet av 94 vikt% natriumsulfat med ett utbyte av 12% (massa torr produkt/massa

vattenlösning). Detta resultat nådde de specifika mål som sattes i början av denna avhandling, vilket var att nå en renhet av 90 vikt% natriumsulfat med ett utbyte på 5%.

Energiintensiteten för att förånga lösningsmedlet förväntas vara hög. Det beror mycket på vilket lösningsmedel som används. Denna process kan dock använda den lägsta ångkvaliteten som finns tillgänglig från massafabriken. Det förväntas att kylningen kommer att kräva stora

(6)

IV mängder kylvatten och höga investeringskostnader för värmeväxlaren. Ändå är detta en viktig del av processen för att minska behovet av kylare som drivs av elektricitet.

(7)

V

1. Introduction

1.1 Background

SCA is a Swedish company that mainly has operations within the forest management, wood processing and the pulp and paper sector. It also operates within biofuels, energy and

logistics. In the pulp and paper sector, SCA has three main facilities, Östrand, Munksund and Obbola, all of which are Kraft pulping processes where today Östrand has a yearly production of 1000 000 tonnes of pulp and Munksund and Obbola has a yearly production of 410 000 and 450 000 tonnes of packaging paper [1]. Östrand also produces electricity, district heating, tall oil and turpentine [1], where the district heating is managed by Adven [2].

1.1.1 Wood structure

Wood consists of mainly three components, cellulose, hemicellulose and lignin. Cellulose is the most important chemical for pulp production. Cellulose consists of linear unbranched chains of anhydrous glucose. The length of these polymers is usually several thousand monosaccharide. Figure 1 illustrates the structure of cellulose. [3]

Figure 1 The structure of cellulose, which only contains of carbon, oxygen and hydrogen.

Cellulose usually occurs with hemicelluloses, which is a polysaccharide with a more irregular structure than cellulose. The two most regular structures of hemicellulose in wood are

glucomannans and xylan, where xylan is a more stable polysaccharide. [3]

Finally, the last main component of wood is lignin, which contains aromatic polymers. These polymers have a very complex structure and therefore have no well-defined primary structure.

Lignin works as the glue of wood, where because of the ability to form multiple covalent bonds with polysaccharides, polysaccharides can be cross-linked together making the wood stiff and more water-resistant. Figure 2 illustrates the structure of wood where cellulose, glucomannan, xylan and lignin are included. [3]

(10)

2

Figure 2 The structure of wood. Note how the lignin is bonded and connecting the cellulose with the glucomannan and xylan.

As Figure 2 shows, it is in fact the lignin polymers that bond the hemicelluloses with each other or with cellulose. Hardwood usually contains around 40-55% of cellulose, 24-40% of hemicellulose and 18-25% of lignin, while softwood contains around 45-50% of cellulose, 25-35% of hemicellulose and 25-35% of lignin. [4]

1.1.2 Kraft pulping process

The most common process to produce pulp by chemical pulping is the Kraft pulping process.

By heating wood chips in an aqueous solution containing sodium hydroxide (NaOH) and sodium sulfide (Na2S) at a temperature around 140-170°C, the wood chips are starting to dissolve because the lignin is degraded (delignification). This aqueous solution is called white liquor. This mixture of white liquor and dissolved wood chips are afterwards washed to separate the cellulose from the mixture, which is used for the production of pulp [3]. The remaining mixture is called black liquor and mainly consists of water, organic compounds from the lignin and inorganic compounds. The majority of the inorganic compounds are sodium-based, e.g. sodium carbonate (Na2CO3), sodium sulfate (Na2SO4), sodium hydroxide, sodium sulfite (Na2SO3) and sodium oxide (Na2O) [5]. When the cellulose is separated, the black liquor has a dry solid content of 10-18 wt% and is in this stage called more specifically,

(11)

3 weak black liquor. The weak black liquor is an important substance for the reproduction of white liquor. Firstly, the weak black liquor needs to be more concentrated, so the water in the weak black liquor is evaporated in multiple steps to a concentration around 60-80 wt% and at this point, it is referred to as thick black liquor. Thick black liquor is later used in the recovery boiler, where the organic compounds are combusted. The sodium-based compounds react with carbon or carbon dioxide according to equations 1 to 4 [3].

𝑁𝑎₂𝑆𝑂₄+ 2𝐶 → 𝑁𝑎₂𝑆 + 2𝐶𝑂₂ (1)

2𝑁𝑎𝑂𝐻 + 𝐶𝑂₂ → 𝑁𝑎₂𝐶𝑂₃+ 𝐻₂𝑂 (2)

𝑁𝑎₂𝑂 + 𝐶𝑂₂ → 𝑁𝑎₂𝐶𝑂₃ (3)

2𝑁𝑎₂𝑆𝑂₄+ 2𝐶 + 𝐻₂𝑂 → 𝑁𝑎₂𝑆₂𝑂₃+ 2𝐶𝑂₂+ 2𝑁𝑎𝑂𝐻 (4) The aqueous solution that is recovered from the recovery boiler is referred to as green liquor and the main components are sodium sulfite and sodium carbonate. The reason why it is called green liquor is due to the colour of the solution that occurs because of the small amount of iron sulfides that is present in the solution. The final step of the recovery process is to convert sodium carbonate to sodium hydroxide. This is done by letting the sodium carbonate react with calcium oxide in the recausticizer, see equation 5. [3]

𝑁𝑎₂𝐶𝑂₃+ 𝐶𝑎𝑂 + 𝐻₂𝑂 → 2𝑁𝑎𝑂𝐻 + 𝐶𝑎𝐶𝑂₃ (5)

The calcium carbonate can be extracted from the solution and by heating it to high

temperature in the rotary kiln, the calcium carbonate can be converted into calcium oxide and carbon dioxide. When the calcium carbonate has been removed from the solution, the main components are sodium hydroxide and sodium sulfite, and white liquor has been reproduced.

Figure 3 shows an overview of the recovery system. [3]

(12)

4

Figure 3 The kraft pulping system. a) the Wood handling, b) the Digester and Washer, c) the additional pulp process, d) the Evaporator, e) the Recovery boiler and f) the Recausticizer. [3]

An important parameter for the white liquor is the sulfidity, which is a measure of the sulfide ion concentration. Higher sulfide ion concentration will increase the delignification of the wood chips and therefore the temperature can be lowered, which will result in higher pulp yield. [3]

1.1.3 Biorefinery

The SCA-concern have since 2013 developed a process to produce hydrocarbons from their pulp production. By extracting black liquor from the Kraft recovery system when the dry solids are between 25-45 wt% and treat the black liquor in numerous steps, lignin oil can be produced. By additional treatment of the lignin oil, renewable hydrocarbons can finally be produced in the range between petrol to diesel [6]. This production line is referred to as Line 2, where Line 1 will use solid biomass to produce renewable hydrocarbons. These two lines are planned to be built next to Östrands pulp mill and together produce 300 000 tonnes of renewable hydrocarbons. Figure 4 illustrates an example of what the facility could look like with both the pulp mill and the biorefinery. [7]

(13)

5

Figure 4 The existing pulp mill together with the possible biorefinery. (Scheiwiller Svensson Arkitektkontor) [7]

One of the prime treatments of the black liquor is the addition of solvent. The solvent is added to the black liquor in order to extract the lignin oil. This solvent has to be polar or aromatic in order to decrease the lignin oils viscosity. According to SCA’s patent (EP3350289 B1), many different solvents can be used, e.g. ethyl acetate, methyl isobutyl ketone, methyl-

tetrahydrofuran, toluene, benzene, benzyl alcohol, etc. [8]

From the production of renewable hydrocarbons, an aqueous solution is obtained at 50 °C, which contains mainly sodium sulfate, sodium carbonate, a mixture of organic compounds and the solvent that has been added at an earlier stage. SCA is interested in the possible usage of this aqueous solution and is determined to find a concept that can either be recycled back to the biorefinery, the Kraft process or extracted as a finished product. [6]

(14)

6 1.1.4 Previous work

Previously, experiments were performed by SCA R&D Centre where the compounds in the aqueous solution were investigated. Two aqueous solutions were produced in the laboratory using black liquor, one from Obbolas black liquor and one from Östrands black liquor. By evaporating the solvent and water at 110°C, solid crystals were formed, see Figure 5. [6]

Figure 5 The extracted crystals after the evaporation of the solvent.

It was calculated that the yield (kg dry solid/kg aqueous solution) was 17%. These crystals were ground and calcinated at 625°C to remove organic compounds or compounds, which evaporate between 110 to 625°C. The crystals had a weight loss of 25 wt% after calcination.

An Inductively coupled plasma atomic emission spectroscopy (ICP-AES) analysis was made to determine the concentration of different elements. The ICP-AES analysis indicated that the aqueous solution contained a substantial amount of sodium and sulfur, and also a considerable amount of aluminium and potassium, for both the samples. However, proving that the crystals contained sulfur does not necessarily mean they occur as sulfate and are bound to sodium forming sodium sulfate. Therefore, a Fourier transform infrared spectroscopy (FTIR) analysis was made on sodium sulfate and sodium carbonate, see Figure 6. [6]

(15)

7

Figure 6 The FTIR analysis of the crystals, compared to pure sodium carbonate and sodium sulfate.

As the FTIR analysis showed, there was a higher concentration of both sodium sulfate and sodium carbonate when comparing to pure substance as a reference. The absorbance reached around 0.15 to 0.18 for the sodium carbonate spike and 0.2 for both sodium sulfate spikes. By assuming that all sulfur is bound as sulfate and forms sodium sulfate and the excess of sodium is sodium carbonate, the molar ratio between sodium sulfate and sodium carbonate

(Na2SO4:Na2CO3) ratio was 67:33. [6]

An additional experiment was performed where crystallisation occurred at 0°C, overnight and without any stirring. The crystals were thereafter filtrated and dried at 105°C. When sodium sulfate is crystallised with the presence of water, sodium sulfate and sodium carbonate occur as decahydrate, i.e. bonded with water [9] [10]. It was calculated that the yield was around 8%. An FTIR analysis was performed in the crystals, see Figure 7. [6]

Figure 7 The FTIR analysis of the crystals extracted at 0°C, compared to pure sodium sulfate.

(16)

8 Note that the absorbance of the Obbola salts reached 0.35 compared to 0.2 for the previous experiments, which indicates a higher concentration of sodium sulfate. Additional purification was performed by washing the crystals with Tetrahydrofurna (THF) which removed organic compounds. The purified crystals were analysed again with the FTIR spectroscopy, see Figure 8. [6]

Figure 8 The crystals which were extracted at 0°C and washed with THF. Compare the figure with Figure 7.

As shown in the figure, both the sodium sulfate spikes increased up to 0.42, which indicates an even higher concentration of sodium sulfate.

Another ICP-AES analysis was made to determine the concentration of different elements.

The ICP-AES showed that the crystals had a concentration of 288.4 and 285.0 g/kg sodium for the Obbola and Östrand samples respectively, which is a slight increase compared to the evaporation method. The concentration of sulfur showed a significant increase with a concentration of 190.4 and 196.4 g/kg. This also indicates that the concentration of sodium sulfate has increased, where the Na2SO4:Na2CO3 ratio is approximately 90:10. [6]

(17)

9

1.2 Objective and scope

The objective of this thesis is to create a concept to crystallise sodium sulfate, where the purity, yield and energy demand will be measured to determine the optimal condition for crystallisation.

The scope of this project is to investigate how to perform crystallisation of the aqueous

solution that is received from the biorefinery when producing hydrocarbons from black liquor.

This will also include a small economical evaluation if the solvent, that contains in the solution, should be recovered. The process variables that were investigated are temperature, residence time and stirring. The purity of the crystals was determined by element analysis.

1.3 Specific objectives

The following question will be answered:

• Is it possible to extract sodium sulfate from the aqueous solution?

• Is the purity of sodium sulfate high enough that additional purification will not be needed?

• Should SCA recover the solvent that is dissolved in the aqueous solution?

• Is the energy demand of the crystallisation process low enough to be profitable?

The specific objectives of this project are as following:

• To reach a yield of at least 5% (50 kg dry solid/tonne aqueous solution).

• To reach a purity of 90wt% sodium sulfate.

(18)

10

2. Theory

An aqueous solution is a water-based solution that contains solid solutes that have been fully dissolved. If the aqueous solution contains only one solid solute, then the solution is

homogeneous and if the solution contains multiple solids, the solution is heterogeneous. [11]

2.1 Solubility and Gibbs free energy

At a given temperature there is a maximum amount of solids the water can dissolve and when this amount is reached, the solution has reached saturation, which is described in equation 6,

𝑆 = 𝐶

𝐶^∗ (6)

where C is the concentration of the solute and C* is the critical concentration of solute. The solutes critical concentration or solubility are usually presented as weight percentage (wt%) or grams of solids by 100 grams solute [11]. The solubility of a solute is often temperature- dependent, where often the solubility increases with increasing temperature. Figure 9 presents four usual common ways how solubility can change with temperature.

Figure 9 The solubility of different compounds depending on the temperature. Note that in the figure, sodium sulfate and sodium carbonate share the same solubility curve. In reality, the solubility is similar but not exactly the same.

As can be seen in Figure 9, the solubility of potassium chloride increases linearly with the temperature, benzoic acid is exponential, sodium chloride is constant and sodium sulfate is exponential at lower temperature and becomes nearly constant for higher temperature. [12]

(19)

11 The solubility of a material in a homogeneous solution can be predicted with the equilibrium constant. To predict the solubility of sodium sulfate, the equilibrium equation is used, see equation 7. [11]

𝑁𝑎₂𝑆𝑂₄ ↔ 2 𝑁𝑎⁺ + 𝑆𝑂₄⁻ (7)

By knowing the change of enthalpy ΔH and change of entropy ΔS, the change of Gibbs free energy can be determined by equation 8,

∆𝐺^° = ∆𝐻^°− 𝑇∆𝑆^° (8)

where T is the temperature in Kelvin. [11]

The change of enthalpy and entropy are calculated with equation 9 and 10,

∆𝐻^°= 2 ∙ ∆𝐻^°_{𝑓,𝑁𝑎+}+ ∆𝐻^°_{𝑓,𝑆𝑂4−}− ∆𝐻^°

𝑓,𝑁𝑎2𝑆𝑂4 (9)

∆𝑆^°= 2 ∙ ∆𝑆^°_{𝑓,𝑁𝑎+}+ ∆𝑆^°_{𝑓,𝑆𝑂4−}− ∆𝑆^°

𝑓,𝑁𝑎2𝑆𝑂4 (10)

where the data is usually presented at standard condition, making ΔG=ΔG° where ΔG° is the change of Gibbs free energy at standard conditions. [11]

The equilibrium constant can be calculated by equation 11,

𝐾_𝑠𝑝 = exp (−∆𝐺°

𝑅𝑇) (11)

where R=8.3145 J/(mol∙K) and is the ideal gas constant. [11]

The equilibrium constant can also be expressed by equation 12,

𝐾_𝑠𝑝 =(𝐶_𝑁𝑎+)²(𝛾_𝑁𝑎+)²𝐶_𝑆𝑂4−𝛾_𝑆𝑂4−

𝐶_{𝑁𝑎2𝑆𝑜4}𝛾_{𝑁𝑎2𝑆𝑜4} (12)

where C is the molar concentration and γ is the activity coefficient. [11]

If the sodium sulfate is in its stable crystal form and at atmospheric pressure, the activity (C∙γ) will be one. If the assumption is made the activity coefficient is one and the molar

concentration is the same between the sodium and sulfate ion, the critical concentration can be calculated according to equation 13,

𝐶^∗ = (𝐾_𝑠𝑝)

1

3 (13)

(20)

12 and the critical molar concentration is the solubility in moles. [11]

The equilibrium constant for a reaction can also be used to predict the reaction between several substances. If crystallisation occurs at a given temperature for an aqueous solution containing sodium, sulfate and carbonate, the amount of sodium sulfate compared to the amount of sodium carbonate can be predicted by calculation of the equilibrium constant.

Equation 14 presents the reaction. [11]

𝑁𝑎₂𝐶𝑂₃+ 𝑆𝑂₄⁻ ↔ 𝑁𝑎₂𝑆𝑂₄+ 𝐶𝑂₃⁻ (14) By calculating the change of enthalpy and entropy for the reaction according to equation 15 and 16,

∆𝐻^°= ∆𝐻^°_{𝑓,𝑁𝑎2𝑆𝑂4}+ ∆𝐻^°_{𝑓,𝐶𝑂3−}− ∆𝐻^°_{𝑓,𝑁𝑎2𝐶𝑂3} − ∆𝐻^°_{𝑓,𝑆𝑂4−} (15)

∆𝑆^°= ∆𝑆^°_{𝑓,𝑁𝑎2𝑆𝑂4}+ ∆𝑆^°_{𝑓,𝐶𝑂3−}− ∆𝑆^°_{𝑓,𝑁𝑎2𝐶𝑂3} − ∆𝑆^°_{𝑓,𝑆𝑂4−} (16) the change of Gibbs free energy and the equilibrium constant can be calculated with equation 8 and 11. If the concentration between sulfate and carbonate is equal and the quotient of the activity coefficients are one, the equilibrium constant represent the number of moles of sodium sulfate that will be produced for each mole of sodium carbonate. The weight percentage of sodium sulfate can further be calculated by equation 17,

𝑤_{𝑁𝑎2𝑆𝑂4} = 𝐾_𝑠𝑝𝑀_{𝑁𝑎2𝑆𝑂4}

𝐾_𝑠𝑝𝑀_{𝑁𝑎2𝑆𝑂4}+ 𝑀_{𝑁𝑎2𝑆𝑂4}∙ 100 (17)

where MNa2SO4 = 142.04 g/mol and MNa2CO3 = 105.9888 g/mol. [11]

2.2 Crystallisation

A crystal consists of the atoms that are arranged in a three-dimensional repeating periodic structure. The creation of crystals are called crystallisation which is a metastable process where the driving force is the level of supersaturation. There are four common ways to create supersaturation. The first is by temperature change. Most compounds solubility decreases with decreasing temperature, which then creates supersaturation. The other is by evaporation of the solvent. This increases the concentration of the solids which creates supersaturation.

The third is by chemical reaction. By adding another solid which creates a reaction forming a product with a lower solubility, supersaturation will increase. Finally, the last common

(21)

13 method is by changing the solvent composition. If a second solvent is added to create a mixed solvent system that decreases the solubility of the solids, the degree of supersaturation will increase. [11]

2.2.1 Supersaturation

If a solution has reached supersaturation, it means that the concentration of a substance is higher than the solution can dissolve. From equation 6, it means that the solubility is higher than one (S>1). However, because crystallisation is a metastable process, this does not necessarily mean that crystallisation will occur, and a higher level of supersaturation may be needed. At one point, the concentration of a substance can be so high that crystallisation start occurring spontaneously, and this point is the supersaturation limit. Crystallisation always occurs in the metastable zone, which is the area between the saturation limit and the supersaturation limit. Figure 10 illustrates the metastable zone.

Figure 10 The metastable zone is the zone between the saturation line and supersaturation limit. Point A is outside the saturation line, so no crystallisation will occur, while for point B crystallisation will occur very slowly and point C crystallisation will occur quickly.

(22)

14 When the solution is cooled from point A to B, it may take several hours for crystals to start appearing. If additional cooling is performed to point C, crystals will start appearing after a few minutes. [11]

2.2.2 Nucleation

Crystallisation can be divided into two steps. The first step is nucleation, which is the birth of a crystal, where the substance is changing the phase from liquid to solid. The second step is crystal growth where already existing crystals will start increasing in size. [13]

Nucleation is divided up into primary and secondary nucleation. Primary nucleation occurs in the absence of crystalline surfaces, either as homogeneous nucleation or heterogeneous nucleation. Homogeneous nucleation occurs when the solution is homogeneous, which means that one substance is present in the solvent, while heterogeneous contains multiple substances in the solvent. It is very rare to perform crystallisation on a homogeneous solution on an industrial scale. The presence of a foreign substance generally reduces the energy required for nucleation. This means that for a heterogeneous solution, primary nucleation occurs at a lower degree of supersaturation. The rate of primary nucleation is described in equation 18 for spherical nuclei,

𝐵₀ = 𝐴 exp [ −16𝜋𝜎³𝜐²

3𝑘³𝑇³[ln(𝑆)]²] (18)

where σ is the surface tension [dynes/cm] (1 dynes = 10^-5 N), ν is the molecular volume [cm³/g], T is the temperature [K], S is the supersaturation ratio [-], k is Boltzmann’s constant [J/K] and A is the preexponential factor with a theoretical value of 10³⁰ nuclei/(cm³∙s). As can be seen, by increasing the supersaturation ratio or temperature, the nucleation rate will

increase, while an increasing surface tension will decrease the nucleation rate. [13]

Secondary nucleation occurs when crystals are already present in the solution. When crystals interact with each other, this promotes the creation of more crystals, which by then lower the supersaturation that is needed for additional nucleation. This phenomenon can be seen as a chain reaction, where crystals create more crystals. Secondary nucleation includes several mechanisms for nucleation where the most common one is initial/dust breeding,

dendritic/needle breeding or contact nucleation. [13]

(23)

15 With initial/dust breeding, tiny crystallites are formed on the crystal surface during the growth of seed crystals. The crystallites then act as nucleation sites. This crystallisation method makes the nucleation rate independent of the degree of supersaturation or the stirring, which makes this method suitable for batch crystallisation. [13]

Dendritic/needle breeding occurs when the solution has reached a high degree of

supersaturation, close to the supersaturation limit. Nucleation at this point results in needle- like crystals called “fines” which causes filtration problems. If even higher supersaturation is reached, polycrystalline breeding can occur. This is when the crystals have a structure made of several crystallites, like a solar panel. Having crystals with polycrystalline structure can cause even higher filtration problems, but this mechanism is not considered very likely on an industrial scale. Needle breeding can easily be avoided with stirring or by slower cooling of the solution. [13]

Finally, contact nucleation is the most common mechanism for secondary nucleation. By stirring the solution, crystals can interact with other crystals, the stirrer or the crystalliser’s wall and this interaction enforce nucleation. [13]

Because secondary nucleation is such a complex phenomenon, there is no general theory to predict the nucleation rate. Instead, there are correlations based on the power law, where the equation depends on the experimental conditions. Equation 19 predicts the nucleation rate when crystallisation is performed with stirring,

𝐵 = 𝑘^′_𝑁𝑊^𝑖𝑀^𝑗_𝑇Δ𝐶^𝑛 (19)

where Wis the agitation rate [RPM] or impeller tip speed [m/s], MT is the suspension density (mass of crystal/Volume of solution) [kg/m³], ΔC is the difference between the concentration and the critical concentration and k’n is a constant where the unit depends on the unit of the agitation rate. If no agitation occurs, equation 20 is otherwise used. [13]

𝐵 = 𝑘^′′_𝑁𝑀^𝑗_𝑇Δ𝐶^𝑛 (20)

For specific cases, it is possible to create a model to predict the nucleation rate more accurate.

Equation 21 is a model to predict contact nucleation,

(24)

16 𝐵^∘_𝑒 = 𝐾_𝑒𝑛^∘𝐺 [𝑇𝐼𝑃𝑆²

𝑇𝑂 ] 𝐿⁵_𝐷 (21)

where TIPS is the tip speed of pump or impeller[m/min], TO is the active volume/circulation rate[min], Ke is an experimental determined constant, n is the population density

[no/(mm·Vaq)], G is the Growth rate [mm/h] and LD is the dominant particle size [mm]. As can be seen in equation 21, the dominant particle size has a fifth-order influence on the nucleation rate and the stirring have a second-order influence. [12]

2.2.3 Crystal growth

When nucleation has occurred, the crystals are at the smallest size possible. The next step of crystallisation is to let the crystals react with more solute molecules, thereby letting the crystals grow. Generally, molecules tend to bond at locations where the molecules have the highest surface contact because this is the most energetically favourable location. This is the reason why crystal growth usually results in layer-by-layer crystals because it is easier for molecules to bond with an already existing layer than to create a new one. A common way to define the crystal growth rate is the linear growth rate of a crystal face, expressed in cm/s.

One of the most common models of describing the linear growth rate is the mononuclear model which assumes that when a surface nucleus is formed, it spreads across the surface at an infinite velocity, making the linear growth rate proportional to the area of the face and the step height according to equation 22,

𝐺 = 𝐶₁ℎ𝐴[ln(𝑆)]^0.5𝑒𝑥𝑝 [ −𝐶₂

𝑇²ln (𝑆)] (22)

where h is the step height between layers [mm], S is the supersaturation ratio [-], A is the surface area of the crystals [mm²] and C1 [mm^-2·h^-1] and C2 [K²] is empirical constants

determined from experiments. As can be seen, the temperature and supersaturation ratio has a significant influence on the growth rate where higher temperature and supersaturation will result in a higher linear crystal growth rate. [14]

Another model is the birth and spread model. This model assumes that nuclei can form at any location, which includes incomplete layers. Therefore, the unit of the crystal growth rate for this model is the mass divided by the surface area and time, see equation 23,

(25)

17 𝑅 = 𝐶₃(𝑆 − 1)²³[ln(𝑆)]¹⁶𝑒𝑥𝑝 [− 𝐶₄

𝑇²ln(𝑆)] (23)

where C3 [kg/(m²s)] and C4 [K²]is empirical constants. Similar to the mononuclear model, the growth rate for the birth and spread model is highly dependent on the supersaturation and the temperature. [14]

2.2.4 Residence and Induction time

The residence time is an important parameter for crystallisation and especially for a batch crystalliser, it is defined from the point cooling of the aqueous solution is initiated to the point where the product is extracted. Another term is the induction time which is the time between the creation of supersaturation to the formation of the solid phase. It is defined according to equation 24,

𝑡_𝑖𝑛𝑑 = 𝑡_𝑡𝑟 + 𝑡_𝑛+ 𝑡_𝑔 (24)

where ttr is the transient time, tn is the nucleation time and tg is the time that is required for the critical nucleus to grow to a detectable size. For an aqueous solution that is moderately supersaturated, the transient time is usually negligible. In general, the induction time can be expressed as in equation 25,

𝑡_𝑖𝑛𝑑 = 𝐾𝑆^−𝑛 (25)

where n is the kinetic order of nucleation and K is a constant [s]. There are several parameters that control the constant K and the kinetic order of nucleation n. Increasing the stirring of the solution will decrease the induction time to a certain degree and beyond that, the induction time will remain unchanged. [13]

(26)

18 Figure 11 illustrates how the supersaturation level change over time.

Figure 11 There are three states for crystallisation, the nucleation state, the transient state, and quasi steady state.

As Figure 11 shows, when cooling starts and the solution has reached saturation, it takes an additional amount of supersaturation for nucleation to start, which is the critical

supersaturation. From this point, nucleation and crystal growth starts until the solution has reached the critical supersaturation point again. At this point, the solution is in the transient state and the nucleation rate and crystal growth rate is highly decreased. Finally, the solution is in the quasi-steady state where the nucleation and crystal growth rate is close to zero. As can be seen, the solution will take a very long time to reach a supersaturation of one. This means that a higher residence time will always result in higher yield, but can result in lower purity for a heterogeneous solution if the impurity starts to crystallise at a late period. [13]

(27)

19

2.3 Thermodynamics

2.3.1 Cooling, heating and evaporation

Increment of a substance temperature is described with the substance change of enthalpy according to equation 26,

𝑄_ℎ = 𝑚∆ℎ (26)

where m is the mass of the substance [kg] and Δh is the change of enthalpy [kJ/kg]. [15]

However, for an increment of temperature at constant pressure and without and phase

changes, the heat that is required to change a substance temperature is calculated according to equation 27,

𝑄_ℎ = 𝑚𝑐_𝑝 ∆𝑇 (27)

where cp is the substances specific heat capacity [kJ/(kgK)], which is dependent on the nature of the material. [15]

The phase change of a substance (fusion or vaporisation) occurs at a defined temperature which is dependent on the pressure, and the heat that is required to vaporise a substance is described in equation 28,

𝑄_𝑒 = 𝑚𝐿_𝑣 (28)

where Lv is the latent heat of vaporisation [kJ/kg] and is usually given at the vaporisation temperature at atmospheric pressure. [15]

2.3.2 Heat exchangers and coolers

The heat that is transfer though a heat exchanger is defined according to equation 29,

𝑄̇ = 𝑚̇_ℎ𝑐_𝑝,ℎ(𝑇_ℎ,𝑖− 𝑇_ℎ,𝑜) = 𝑚̇_𝑐𝑐_𝑝,𝑐(𝑇_𝑐,𝑜− 𝑇_𝑐,𝑖) (29) where ṁh is the mass flow of the heating stream [kg/s], ṁc is the mass flow for the cooling stream, cp,h is the heating streams specific heat capacity [kJ/kgK], cp,c is the cooling streams specific heat capacity, Th,i and Th,o is the heating streams in and out temperature and Tc,i and Tc,o is the cooling streams in and out temperature. [16]

(28)

20 The heat transfer across a heat exchanger can also be expressed according to equation 30,

𝑄̇ = 𝑈𝐴∆𝑇_𝑙𝑚 (30)

where U is the overall heat transfer coefficient for the heat exchanger [W/(m²K)], A is the area of the heat exchanger [m²] and ΔTlm is expressed according to equation 31,

∆𝑇_𝑙𝑚 =∆𝑇₂− ∆𝑇₁ 𝑙𝑛 (∆𝑇₂

∆𝑇₁)

(31)

where for counter flow heat exchangers, ΔT2 = Th,o – Tc,i and ΔT1 = Th,i – Tc,o. [16]

For additional cooling using an industrial compressor cooler, the electrical power input is expressed according to equation 32,

𝑃 = 𝑄̇_𝑐

𝐶𝑂𝑃 (32)

where 𝑄̇_𝑐 is the heat that is removed [W] and COP is the cooling factor [-]. [17]

2.3.3 Heat of crystallisation

A crystallisation process is usually an exothermal process, which means that the solvent will increase in temperature when crystallisation occurs. The heat of crystallisation is important to investigate if the heat of crystallisation for a specific substance is significant or if it can be neglected. A substance heat of crystallisation is usually presented as the reverse of the heat of solution. In reality, the difference between the heat of crystallisation and heat of solution is so small that it is considered negligible. The heat of crystallisation is dependent on the solution the substance is crystallised in, most commonly water, and on temperature. It is very common to neglect the temperatures impact and therefore the heat that is released during crystallisation is presented in equation 33,

𝑄_𝐶𝑟𝑦 = 𝑛∆𝐻_𝐶𝑟𝑦= −𝑛∆𝐻_𝑆𝑜𝑙 (33)

where n is the amount of substance and ΔHcry is the heat of crystallisation [kJ/kmol] and ΔHSol

is the heat of solution. [11]

(29)

21

3. Methods

3.1 Evaporation of solvent

Firstly, because the solvent that is used when extracting lignin oil will dissolve in the water, the concept of evaporating the solvent and reusing it has to be investigated. The presence of an additional liquid in the solution may affect the crystallisation process. The simplest way to remove the solvent from the aqueous solution is by evaporation at the solvents boiling

temperature. For this concept to work, the boiling temperature has to be lower for the solvent compared to water. In Table 1 all the solvents that SCA mention in their patent are presented with the boiling temperature at atmospheric pressure [8] [18].

Table 1 A list of possible solvent that is mention in SCA’s patent and are possible solvens for the process with the boiling point.

Solvent Boiling temperature [°C]

Anisole 155.5

Benzene 80.1

Benzyl alcohol 205.3

Ethyl Acetate 77

Methyl isobutyl ketone (MIBK) 115.8 2-Methyl-tetrahydrofuran (MeTHF) 78

M-xylene 139.3

O-xylene 144.4

Phenylethyl alcohol 218.2

3-Phenyl-1-propanol 235

P-cymene 177.1

P-xylene 138.35

Toluene 110.6

As can been seen, only three of the solvents has a boiling temperature below 100°C, which is Benzene, Ethyl acetate and MeTHF. Because these three solvents can easily be separated from the solution by evaporation, a simplified cost analysis was carried out to evaluate if it is economically beneficial to recover these solvents. To be able to recover other solvents that have a boiling temperature above 100°C, another method has to be used and no further evaluation was done regarding the possible technology because it was considered to be

(30)

22 outside the scope of this thesis. Necessary data for these three solvents were received and is presented in Table 2.

Table 2 Data collection for the three solvents that has a boiling temperature under 100°C.

Data Benzene Ethyl acetate MeTHF

Molar mass 78.12 [19] 88.11 [19] 86.13 [20]

Heat capacity (kJ/kmol∙K) 135.7 [19] 169.6 [19] 156.9 [20]

Latent heat of vaporisation (kJ/kmol)

33.83 [20] 31.9 [21] 34 [21]

Solubility in water (wt%) 1.79 [20] 8.0 [20] 15.59 [20]

Cost of solvent (€/kg) 0.88 (2012) [22] 0.93 (2012) [22] 5 (2015) [23]

To calculate the amount of heat which is needed to recover the solvent, the heat that is needed to increase the aqueous solution to the solvent boiling temperature, to extract 1 kg of solvent, is calculated by remaking equation 27, see equation 34,

𝑄_ℎ = ( 𝑠

100 𝑐^𝑝𝑠(𝑇_𝑏𝑠− 𝑇₁) +(100 − 𝑠)

𝑠 𝑐_𝑝𝑤(𝑇_𝑏𝑠− 𝑇₁)) ( 𝑠

100)

(34) where s is the solubility of the solvent, cps is the solvents heat capacity, cpw is the specific heat capacity for water which holds the value of 4.18 kJ/kg at 25°C [24], Tbs is the solvents boiling temperature and T1 is the solvents initial temperature, which as previously mentioned is 50°C. Note that equation 34 does not take into consideration the mass and specific heat capacity of organic and inorganic compounds that do occur in the solution. This method assumes that the solution only contains water and the specific solvent. To simplify the

calculations, it is assumed that the waters specific heat capacity is constant instead of varying dependent on the temperature. Furthermore, to compare the cost of recovering the solvent from the aqueous solution, and to repurchase the solvent, the cost of heat will be set for the sake of simplicity at 0.433 Sek/kWh (0.12 Sek/MJ), which is the cost of district heating that the Östrand factory provides [25].

(31)

23

3.2 Solubility and Gibbs free energy

To determine the optimal condition to perform the crystallisation, the solubility was compared between sodium sulfate and sodium carbonate, see Figure 12 [26] [20].

Figure 12 The solubility of sodium sulfate and sodium carbonate depending on the temperature.

Because the solubility of sodium sulfate and sodium carbonate is similar between 0 to 30°C, further evaluation is needed to determine which temperature is beneficial to perform the crystallisation. However, because both sodium sulfate and sodium carbonate have an

exponential change of solubility depending on the temperature, this makes them very suitable for crystallisation by cooling, where hopefully no evaporation of water or additional

antisolvent will be needed because these methods are more expensive to operate.

Because the change of solubility could only determine that lower temperature will a give higher yield, the change of Gibbs free energy is used. By knowing the change of Gibbs free energy, the equilibrium constant can be used which describe the molar ratio between sodium sulfate and sodium carbonate, assuming these are the only reactions that occur, see equation 14. The enthalpy and entropy for each compound are presented in Table 3.

0 5 10 15 20 25 30 35 40 45

0 5 10 15 20 25 30 35

g/100g H2O

Temp °C

Solubility

Na2CO3 Na2SO4

(32)

24

Table 3 The enthalpy and entorpy at standard state for different compounds. [27]

Compound ΔH°f [kJ/mol] ΔS°f [J/(mol∙K)] State

Na2CO3 -1130.68 134.98 Crystal

SO4- -909.27 20.1 Aqueous

Na2SO4 -1387.08 149.58 Crystal

CO3- -677.14 -56.9 Aqueous

Na⁺ -240.12 59 Aqueous

By using equation 15 and 16, the change of enthalpy and entropy can be calculated.

Furthermore, by using equation 8, the change of Gibbs free energy can be calculated where with equation 11, the equilibrium constant can be determined. Finally, with equation 17, the weight percentage of sodium sulfate can be calculated.

3.3 Execution of experiments

3.3.1 Main experiments

The aqueous solution was produced during the month of March at the SCA R&D Centre from black liquor received from Östrand as a 4.2 litre batch. Before the first experiment was

performed, it was noted that a thick layer of crystals had occurred at the bottom of the container where the aqueous solution was stored. This means that crystallisation can occur even at room temperature. However, this process is believed to take a long time. Further notes and conclusion about this phenomenon will be presented in the results.

To dissolve the crystals, the container was placed in a water bath of 50°C and occasionally mixed until the crystals have been dissolved. This heating of the aqueous solution was carried out for all the experiments. The temperature was chosen due to the fact that the aqueous solution is received from the biorefinery at 50°C.

After the solution had been heated and stirred in the container, 200 ml of the solution was poured into a beaker. The beaker was placed into a Lauda Proline Edition X cooling bath, where the temperature was regulated. A stirrer rod was contacted to a VELP LS overhead stirrer and placed into the aqueous solution as low as possible without scraping the bottom of the beaker. A thermometer was used to monitor the temperature of the aqueous solution and

(33)

25 when the aqueous solution had reached the desired temperature, the beaker was kept in the bath until it had reached the desired residence time. Figure 13 illustrated when the beaker with the solution is placed in the cooling bath. At the point the picture was taken,

crystallisation had already occurred.

Figure 13 Crystallisation of the solution when placed in the cooling bath.

During the experiments, the surroundings did heat up the solution and therefore the

temperature of the bath had to be lower than the desired temperature of the solution. Table 4 shows the temperature of the bath based on the desired temperature of the solution.

Table 4 The temperatures that were set on the bath, depending och the temperatures of the solution.

Temperature of the solution [°C] Temperature of the bath [°C]

0 -3

5 3.5

10 9

15 14

(34)

26 When the solution had reached the desired residence time, the beaker was removed from the cooling bath and the solution was vacuum filtrated. Figure 14 shows the crystals that were caught on the filter.

Figure 14 The crystals that was caught on the filter, when the solution was filtrated after crystallisation.

The crystals and the filter were thereafter placed in a dryer at 105°C until the sample was completely dried. The dried crystals were weighed together with the filter, and by removing the weight of the filter, the yield between the amounts of dry crystals compared to the amount of aqueous solution can be calculated. The dried crystals were thereafter removed from the filter, and the dried crystals are illustrated in Figure 15.

(35)

27

Figure 15 The crystals after they had been dried at 105°C.

Finally, the crystals were thereafter ground and placed in a sample vial.

In total, ten experiments were performed. Two parameters were primary tested, the temperature and the residence time. The initial plan was also to investigate the impact increased stirring would have. However, because the overhead stirrer that was provided did not specify the exact rotational speed, the focus was instead set on a concept of pre-crystallise organic compounds before performing the main crystallisation at a lower temperature.

Therefore, the rotational speed was constant (estimated between 200-300 RPM) for all experiments, except for one experiment where crystallisation without stirring was carried out to clarify if stirring is needed. Table 5 shows all the experiments that were performed, where 200 ml of aqueous solution was used for all experiments.

(36)

28

Table 5 All the experiments that were performed for this thesis. For experiment 1 to 3, two filtrations occurred for the same temperature, where the first filtration had the first residence time, and the second filtration had the second residence time.

The same goes for experiment 5, 7 and 9, but the first filtration occurred at the first mentioned temperature and the second filtration at the second mentioned temperature.

Experiment Temperature [°C] Residence time [h]

1 5 1+7

2 5 2+6

3 5 0.5+7.5

4 0 1

5 15→5 3+1

6* 5 1

7 10→5 3+1

8 -4 17

9 10→0 3+1

10 5 1

* This experiment was performed without stirring.

In the three first experiment that occurred, the objective was to investigate the impact of the residence time. The temperature of 5°C was chosen based on the fact that a lower temperature will be very costly on an industrial scale and a higher temperature may give a poor yield. By decreasing the temperature from 50 to 5°C and filtrate the crystals after the specific residence time, the yield can be measured. By then taking the mother liquor (the remaining solution) and letting it crystallise at the same temperature so the experiments total residence time reaches 8 hours, the yield from the first filtration can be compared to the total amount of crystals that can be extracted, assuming that steady-state has been reached after 8 hours [28].

For Experiment 1, 2 and 3 the first filtration was performed after 1, 2 and 0.5 hours respectively and therefore the second filtration was performed after 7, 6 and 7.5 hours

respectively. For experiment 5, 7 and 9, the residence time for the pre-crystallisation step was 3 hours while 1 hour for the main crystallisation. The Büchner funnel that was used between experiments 1 to 4 broke. Therefore, experiments 5 to 9 had a lager Büchner funnel and therefore a lager filter. To compensate for this, it was avoided to use the whole filter and instead only the central part of the filter was used.

(37)

29 3.3.2 Additional experiments and notes

When the first filtration was performed for experiment 9, only a small amount of dark samples was caught on the filter. However, after the solution had been filtrated, a noticeable amount of crystals occurred on the bottom of the vacuum flask. These crystals were quickly collected by filtrating the solution again with the same filter, see Figure 16.

Figure 16 When the solvent was filtrated again. Note that some crystals were caught on the vacuum flask.

The picture was taken at the moment where the vacuum pump has not been started yet and therefore there is a high amount of water with the crystals. Some crystals were caught on the vacuum flask and were impossible to recover.

To investigate if the crystals that are received directly from the filtration is bonded with water as decahydrate, a sample was used from experiment 10, and the sample was placed in a water bath at 40°C. This is because sodium sulfate decahydrate (Na2SO4∙10H2O) and sodium

(38)

30 carbonate decahydrate (Na2CO3∙10H2O) is stable below 32.4°C and 35.4°C respectably [9]

[10].

Finally, two samples of experiment 10 were placed in a dryer at 110°C and the weight loss was monitored to further evaluate if the sodium sulfate and sodium carbonate are bonded as decahydrate.

3.4 Assumptions for analysis

Results from the ICP-AES will give the samples mass concentration of sulfur and sodium. By assuming that all the sulfur is bonded as sulfate, the molar concentration can be calculated for both sulfate and sodium, where the assumption is made that all sulfate is bonded with sodium forming sodium sulfate. If there is an excess of sodium (2∙nNa>nSO4), this excess will be assumed to be bonded with other ions, primary carbonate. This assumption will be verified by performing a DP12 analysis of sulfate and carbonate on sample 1F1 and 5F2.

When knowing the mole concentration of sodium sulfate and sodium carbonate, these values can be calculated back to the mass concentration by using the molar weight and thereby knowing the purity of sodium sulfate. A Thermogravimetric analysis (TGA) using nitrogen will be used to investigate the amount of organic compounds and see if this can be used to clarify the actual concentration of sodium sulfate. Because sodium sulfate is stable up to 880°C and afterwards it starts to decompose [29], the amount of inorganic compounds is assumed to be the weight percentage that is left at 880°C while the amount of organic compounds is assumed to be the weight losses up to 880°C.

3.5 Heat and mass balance.

A flow chart for each specific solvent was created using heat and mass balance calculations.

Östrand have several steam flows that could provide heat for this process. One main steam flow is at a total pressure of 4.31 bar and 150°C and another is at 12.51 bar at 200°C (M.

Hjärpe, personal communication, 20^th of April 2021). For the heat balance when evaporating the solvents, the change of enthalpy is used for the steam that is heating the solution, see equation 27 and all steam data is provided from steam tables [30]. The 4.31 bar, 150°C steam is used for the evaporation of the solvent, because the higher temperature gradient is

considered not necessary. The phase of the steam is assumed to fully condensate after the heat exchanger, and the evaporators overall heat transfer coefficient is assumed to be between

(39)

31 1500 W/(m²K) and 4000 W/(m²K) [16]. The specific heat capacity for the solution without any solvent is assumed to be the same as water even though water that is saturated with sodium carbonate and sodium sulfate has a specific heat capacity of 3.56 kJ/(kgK) and 3.49 kJ/(kgK) respectively [31]. Östrand does have a stream of subcooled cooling water at 20°C that could be used for the process. The heat exchanger uses this water to cool the solution and the overall heat transfer coefficient are assumed to be between 850 W/(m²K) and 1700

W/(m²K) [16]. For the additional cooler that is used to decrease the temperature of the aqueous solution below 25°C, the electrical power input is calculated by equation 32 and the COP value is 7.3 [32]. The heat of crystallisation was assumed to be negligible due to the fact that the heat of crystallisation for sodium sulfate is 2.1 kJ/kmol or 0.015 kJ/kg [33].

(40)

32

4. Results and discussions

4.1 Evaporation of solvent

With equation 28 and 34 and data from Table 1 and 2, the heat that is required to heat the solution to the boiling point and evaporate the solvent is presented in Table 6 with the total heat QTot, which is the sum of Qh and Qe.

Table 6 The energy nedded to heat the solution and evaorate the solvent. This results is based on that the solvent only contains the solvent and water.

Solvent Benzene Ethyl acetate MeTHF

Qh [kJ/kg solvent] 6944 1357 685

Qe [kJ/kg solvent] 433 362 395

QTot [kJ/kg solvent] 7377 1719 1079

As can be seen, more heat is needed to increase the temperature of the solutions to the boiling point compared to evaporate the solvent. This is due to the low latent heat of vaporisation the solvents have and the high specific heat capacity water has compared to the solvents. Qh is also highly dependent on the solubility of the solvents, and because benzene has low solubility in water, recovery of benzene will be very energy demanding.

With the QTot and the cost of heat, which is presented in the Method, the total cost of recovering the solvent compared to repurchase it is being presented in Figure 17.

(41)

33

Figure 17 The estimated total cost of recover the solvents compared to buying the same amount.

As Figure 17 shows, the cost to repurchase the solvent is higher than the cost of recovery process. Where the cost of recovering the solvent is 9%, 2% and 0.2% of repurchasing it for benzene, ethyl acetate and MeTHF respectively.

To further analyse the cost of recovering the solvent, Figure 18 presents the cost without repurchase of the solvents.

887

9704

207

10255

130

53576

0 10000 20000 30000 40000 50000

Benzene recover Benzene buy Ethyl acetate recoverEthyl acetate buy MeTHF recover MeTHF buy Sek/tonnes

(42)

34

Figure 18 The estimated cost of recovering the solvents.

As Figure 18 shows, benzene is the most expensive solvent to recover while MeTHF is the cheapest to recover. This is due to the high solubility of MeTHF, which decreases the solutions specific heat capacity.

4.2 Equilibrium composition

For the reaction between sodium carbonate and sodium sulfate according to equation 14, the change of enthalpy was calculated to -24.27 kJ/mol while the change of entropy was

calculated to -62.4 J/(mol∙K). The change of the equilibrium constant depending on the temperature is presenting in Figure 19.

887

207

130

0 100 200 300 400 500 600 700 800 900 1000

Benzene recover Ethyl acetate recover MeTHF recover Sek/tonnes

(43)

35

Figure 19 The change of the equilibrium constant changes with changing temperature.

As Figure 19 shows, the equilibrium constant does increase exponential with decreasing temperature. This then shows that the molar ratio between sodium sulfate and sodium

carbonate greatly benefits sodium sulfate at low temperature. Figure 20 shows how the weight percentage of sodium sulfate change dependent on the temperature.

Figure 20 The change of purity with changing temperature, assuming only sodium sulfate and sodium carbonate is crystallised.

Figure 20 indicates that the weight percentage increase logarithmic with decreasing

temperature. Also note the fact that the aqueous solution does contain more sulfate ions than carbonate ions, which will benefit the creation of sodium sulfate. What can be concluded is that lower crystallisation temperature will give both higher yield based on the solubility and higher mass concentration of sodium sulfate compared to sodium carbonate.

0 5 10 15 20 25 30

270 280 290 300 310 320

Ksp

Temp [K]

Gibbs free energy

89 90 91 92 93 94 95 96 97 98

270 280 290 300 310 320

wt%

Temp [K]

Change of weight percentage Na

₂

SO

₄

(44)

36

4.3 Mass exchange and yield

To analysis the mass that is extracted for each experiment, Figure 21 shows a bar chart of the amount of solid that was extracted from experiment 1 to 9. Experiment 3 was performed two times due to the fact that no solids were received from the first experiment.

Figure 21 A compilation of the amount of solids that was extracted for experiment 1 to 9, when crystallising 200 ml aqueous solution.

Experiment 1 and 6 are both experiments that were performed at 5°C for 1 hour. The

difference was that experiment 6 was performed without any stirring. The fact that not even 1 gram of solid was received from experiment 6 indicates the benefit of performing

crystallisation with stirring to create contact nucleation. Also, experiment 8 was performed at -4°C for 17 hours and for some reason the amount of solids that was extracted was very low.

These experiments are seen as a failure and no further evaluation will be done on this experiment.

In Figure 22, experiment 1, 2, 3 and 3a are presented in terms of the yield. The residence time is presented for each bar where the first number is the residence time for the first filtration and the second number for the second filtration.

0 5 10 15 20 25 30 35

1 2 3 3a 4 5 6 7 8 9

g solid

Sample nr

2nd filtration 1st filtration

Concept development to extract sodium sulfate from an aqueous solution