Separation of water-rich n-methylmorpholine from a mixture containing n-methylmorpholine, water, methanol and high boilers

Abstract

A process for separating a mixture containing at least methanol, N-methylmorpholine (NMM), water and high boilers by distillation in a first and a second distillation column, comprising the steps: (i) feeding to the first distillation column the mixture containing at least methanol, NMM, water and high boilers, wherein the weight ratio water / (NMM + water) is lower than in an azeotropic NMM / water-mixture at the operating pressure of the second distillation column, (ii) separating off methanol as overhead stream of the first distillation column, wherein a mixture containing NMM and water is obtained as bottom withdrawal stream of the first distillation column, wherein the weight ratio water / (NMM + water) is lower than in an azeotropic NMM / water-mixture at the operating pressure of the second distillation column, (iii) adding water to the bottom withdrawal stream of the first distillation column in such an amount that the weight ratio water / (NMM + water) is still lower than but closer to the ratio water / (NMM + water) in an azeotropic NMM / water-mixture at the operating pressure of the second distillation column, or the weight ratio water / (NMM + water) corresponds to or is higher than the weight ratio water / (NMM + water) in said azeotropic NMM / water-mixture, and feeding the resulting mixture to the second distillation column, (iv) separating off a NMM / water-mixture as overhead stream of the second distillation column, (v) separating off a mixture containing water and high boilers as bottom withdrawal stream of the second distillation column.

Description

[0001] Separation of water-rich N-methylmorpholine from a mixture containing N-methylmorpholine, water, methanol and high boilers

[0002] Description

[0003] The present invention relates to an economic process for the separation of water-rich N-methylmorpholine by distillation from a mixture containing at least methanol, N-methylmorpholine (NMM), water and high boilers. In particular, the present invention relates to such a process, wherein a NMM / water-mixture having a composition close to the azeotropic composition is separated, which can be fed as an educt stream to the subsequent oxidation of NMM to N-methylmorpholine oxide (NMMO).

[0004] N-methylmorpholine (4-methylmorpholine; 4-methyltetrahydro-1 ,4-oxazine; NMM) is for instance used as a catalyst in polyurethane production. N-methylmorpholine is also the starting material for the production of N-methylmorpholine oxide, that is an important solvent for cellulose.

[0005] NMM can be prepared by the reductive amination of morpholine (MO) with formaldehyde in the presence of hydrogen.

[0006] The preferred process for preparing NMM comprises continuously feeding morpholine, a formaldehyde solution and hydrogen into a reactor equipped with a heterogenous hydrogenation catalyst and subjecting the morpholine and formaldehyde to reductive amination to obtain NMM, wherein the formaldehyde solution is a methanolic formaldehyde solution.

[0007] A key feature of this process is that the continuously fed formaldehyde solution is a methanolic formaldehyde solution. The term “methanolic formaldehyde solution” means that formaldehyde is employed in a solution that comprises methanol and preferably also water.

[0008] N-methylmorpholine oxide (N-methylmorpholine N-oxide; NMMO) is for instance used as solvent for cellulose. Aqueous solutions of N-methylmorpholine N-oxide are technically used as a solubilizer in Lyocell fiber production, wherein pulp is dissolved and spun from a solution in N- methylmorpholine N-oxide and water.

[0009] NMMO can be obtained by oxidation of NMM. The oxidation step comprises reacting NMM with a less than stoichiometric amount of aqueous hydrogen peroxide in an aqueous medium in the presence of carbon dioxide as promotor.

[0010] In a preferred embodiment, the oxidation step comprises reacting NMM with a less than stoichiometric amount of aqueous hydrogen peroxide in an aqueous medium while imposing on the aqueous medium a vapor space having a carbon dioxide partial pressure p (CO2) of less than 0.75 bar absolute, preferably less than 0.20 bar absolute. NMM forms an azeotrope with water. At about 1 ,5 bar, the NMM / water-azeotrope contains 72 wt.-% of NMM and 28 wt.-% of water. At about 1 .0 bar, the NMM / water-azeotrope contains 74 wt.-% of NMM and 26 wt.-% of water. At about 10 bar, the NMM / water-azeotrope contains 57 wt.-% of NMM and 43 wt.-% of water.

[0011] For the oxidation of NMM to NMMO, a certain amount of water in the NMM containing educt mixture is advantageous. In the presence of water, the oxidation reaction starts quicker. Furthermore, the heat produced by the exothermic oxidation reaction is buffered by the heat capacity of water. However, too much water in the NMM containing educt mixture would deteriorate the space-time-yield of the oxidation reaction.

[0012] It is an object of the present invention to provide an economic process for separating a mixture containing at least methanol, NMM, water and high boilers by distillation into essentially pure methanol, a NMM / water-mixture, and a high boiler containing stream. In particular, the present invention relates to such a process, wherein a NMM / water-mixture having a composition close to the azeotropic composition is obtained, which can be fed as an educt stream to the subsequent oxidation of NMM to NMMO.

[0013] The problem is solved by a process for separating a mixture containing at least methanol, N- methylmorpholine (NMM), water and high boilers by distillation in a first and a second distillation column, comprising the steps:

[0014] (i) feeding to the first distillation column the mixture containing at least methanol, NMM, water and high boilers, wherein the weight ratio water / (NMM + water) is lower than in an azeotropic NMM / water-mixture at the operating pressure of the second distillation column,

[0015] (ii) separating off methanol as overhead stream of the first distillation column, wherein a mixture containing NMM and water is obtained as bottom withdrawal stream of the first distillation column, wherein the weight ratio water / (NMM + water) is lower than in an azeotropic NMM / water-mixture at the operating pressure of the second distillation column,

[0016] (iii) adding water to the bottom withdrawal stream of the first distillation column in such an amount that the weight ratio water / (NMM + water) is still lower than but closer to the ratio water / (NMM + water) in an azeotropic NMM / water-mixture at the operating pressure of the second distillation column, or the weight ratio water / (NMM + water) corresponds to or is higher than the weight ratio water / (NMM + water) in said azeotropic NMM / watermixture, and feeding the resulting mixture to the second distillation column,

[0017] (iv) separating off a NMM / water-mixture as overhead stream of the second distillation column,

[0018] (v) separating off a mixture containing water and high boilers as bottom withdrawal stream of the second distillation column. A common engineer would expect that addition of water before the distillation in the second distillation column would significantly increase the energy demand in the second distillation column, since additional water has to be evaporated during rectification. However, it has been surprisingly found that adding water before the distillation in the second distillation column leads to a less energy consuming overall process than distillation of a NMM I water-mixture containing significantly less water than the NMM / water-azeotrope, and adding water only after separating off a (non-azeotropic) NMM / water-mixture in the second distillation column.

[0019] In step (i) the mixture containing at least methanol, NMM, water and high boilers, wherein the weight ratio water / (NMM + water) is lower than in an azeotropic NMM / water-mixture at the operating pressure of the second distillation column, is fed to the first distillation column. A typical product mixture of the reductive amination of morpholine with formaldehyde using a methanolic formaldehyde solution contains from 50 to 80 wt.-%, preferably from 66 to 70 wt.-% of NMM, from 8 to 24 wt.-%, preferably from 14 to 18 wt.-% of water, from 6 to 22 wt.-%, preferably from 12 to 16 wt.-% of methanol, from 0.01 to 5.0 wt.-%, preferably from 0.2 to 1 .0 wt.-% of high boilers.

[0020] High boilers may include morpholine (MO), tetramethylendiamine (TMEDA), 2,2-bis(dimethyl- aminoethyl)eter (BDMAEE), 2,4-dimethylmorpholine (DM MO), and further high boling substances.

[0021] In step (ii), methanol is separated off as an overhead stream in the first distillation column. Preferably, the water content of the methanol obtained as an overhead stream in the first distillation column is 0.02 wt.-% (200 wt.-ppm) or lower.

[0022] A mixture containing NMM and water is obtained as bottom withdrawal stream in the first distillation column. Preferably, the methanol content of the bottom withdrawal stream of the first distillation column is 0.1 wt.-% (1000 wt.-ppm) or lower.

[0023] In general, the first distillation column has from 17 to 44 theoretical stages, and the reflux ratio is correspondingly from 2.0 mole / mole (44 stages) to 10 mole / mole (17 stages). The feed stage can be at stage 7 if the first distillation column has 17 theoretical stages, or it can be at stage 25 if the first distillation column has 44 theoretical stages. For example, the first distillation column can have 30 theoretical stages, with the feed stage being stage 15 (the sump being stage 1), and the reflux ratio being 2.7 mole / mole.

[0024] The first distillation column is in general operated at a pressure of from 1.0 to 10 bar, preferably from 1 .0 to 2 bar.

[0025] Preferably, the first distillation column is operated at a pressure as close to atmospheric pressure as possible, but high enough to make up for pressure losses, since higher operating pressures require higher temperatures, which is energetically disadvantageous. Operating pres- sures higher than 10 bar are unfavourable since the sump temperature may rise up to and above 200 °C, which may cause colorization of NMM.

[0026] The weight ratio water I (NMM + water) is essentially as it is in the mixture fed to the first distillation column and is significantly lower than in an azeotropic NMM / water-mixture at the operating pressure of the second distillation column. With the values given above for the product mixture, this ratio is from 12 to 26 wt.-% of water, based on the total amount of water and NMM.

[0027] In step (iii), water is added to the bottom withdrawal stream of the first distillation column in such an amount that the weight ratio water / (NMM + water) is still lower than but closer to the ratio water / (NMM + water) in an azeotropic NMM / water-mixture at the operating pressure of the second distillation column, or the weight ratio water / (N M M + water) corresponds to or is higher than the weight ratio water / (NMM + water) in said azeotropic NMM / water-mixture.

[0028] The weight ratio water / (NMM + water) can be lower than in the original (product) mixture but closer to the weight ratio water / (NMM + water) in an azeotropic NMM / water-mixture at the operating pressure of the second distillation column. This case is also called “left of azeotrope” or “LAZ”. Preferably, the weight ratio water / (NMM + water) is as close as possible to the respective ratio of the azeotrope. Preferably, it is less than 2 wt.-% of water lower than the respective ratio of the azeotrope.

[0029] The weight ratio water / (NMM + water) can corresponds to or can be higher than the weight ratio water / (NMM + water) in an azeotropic NMM / water-mixture at the operating pressure of the second distillation column. The latter case is also called “right of azeotrope” or “RAZ”. Preferably, the weight ratio water / (NMM + water) is as close as possible to the respective ratio of the azeotrope. Preferably, it is up to 5 wt.-% of water higher, preferably 1 to 3 wt.-% of water higher than the weight ratio water / (NMM + water) in said azeotropic NMM / water-mixture.

[0030] The “right of azeotrope” case is preferred to the “left of azeotrope” case. It has been found that “right of azeotrope” (RAZ) is always more economic than “left of azeotrope” (LAZ). Furthermore, the NMM loss can only be minimized in the “right of azeotrope” case.

[0031] The addition of water in step (iii) is carried out in a mixing device. Examples for suitable mixing devices are a vessel with pump around and residence time, a static mixer, a pipe where in pipemixing occurs due to turbulent flow.

[0032] The resulting mixture is fed to the second distillation column.

[0033] In step (iv), a NMM / water-mixture is separated off as overhead stream in the second distillation column. Preferably, the content of high boilers obtained as an overhead stream of the second distillation column is 0.1 wt.-% (1000 wt.-ppm) or lower. More preferably, the morpholine content of the NMM / water-mixture obtained as an overhead stream in the second distillation column is 0.1 wt.-% (1000 wt.-ppm) or lower. In step (v), a mixture containing water and high boilers is separated off as bottom withdrawal stream of the second distillation column. Preferably, the NMM loss, defined as the NMM mass in the top withdrawal stream of the first distillation column plus the bottom withdrawal stream of the second distillation column divided by the NMM mass fed to the first distillation column, is 0.002 kg / kg or less. The NMM loss can be essentially zero in the “right of azeotrope” case.

[0034] In general, the second distillation column has from 5 to 20 theoretical stages, and the reflux ratio is correspondingly from 0.02 mole / mole (20 stages) to 0.5 mole / mole (5 stages). The feed stage can be at stage 4 if the second distillation column has 5 theoretical stages, or it can be at stage 18 if the first distillation column has 20 theoretical stages. For example, the second distillation column can have 10 theoretical stages, with the feed stage being stage 8 (the sump being stage 1), and the reflux ratio being 0.03 mole / mole.

[0035] In general, investment costs are correlated with rising numbers of theoretical stages, and operating costs, including energy consumption of evaporators (evaporator duties), are correlated with rising reflux ratios.

[0036] The second distillation column is in general operated at a pressure of from 1 .0 to 10 bar, preferably from 1 .0 to 2 bar.

[0037] Preferably, the first distillation column is operated at a pressure close to atmospheric pressure, but high enough to make up for pressure losses, since higher operating pressures require higher temperatures, which is energetically disadvantageous. Operating pressures higher than 10 bar are unfavourable since the sump temperature may rise up to and above 200 °C, which may cause colorization of NMM.

[0038] The problem is also solved by an alternative process for separating a mixture containing at least methanol, N-methylmorpholine (NMM), water and high boilers by distillation in a single distillation column having a side withdrawal, comprising the steps:

[0039] (i) providing the mixture containing at least methanol, N-methylmorpholine (NMM), water and high boilers, wherein the weight ratio water / (NMM + water) is lower than in an azeotropic NMM I water-mixture at the operating pressure of the distillation column,

[0040] (ii) adding water to the mixture in such an amount that the weight ratio water / (NMM + water) is still lower than but closer to the weight ratio water / (NMM + water) in an azeotropic NMM / water-mixture at the operating pressure of the distillation column, or the weight ratio water / (NMM + water) corresponds to or is higher than weight ratio water I (NMM + water) in said azeotropic NMM / water-mixture, and feeding the resulting mixture to the distillation column,

[0041] (iii) separating off methanol as overhead stream of the distillation column,

[0042] (iv) separating off a NMM / water-mixture as side withdrawal stream of the distillation column, (v) separating off a mixture containing high boilers as bottom withdrawal stream of the distillation column.

[0043] A common engineer would expect that addition of water before the distillation in a single distillation column would significantly increase the energy demand in the distillation column, since additional water has to be evaporated during rectification. However, it has been surprisingly found that adding water before the distillation in a single distillation column leads to a less energy consuming overall process than distillation of a NMM I water-mixture containing significantly less water than the NMM / water-azeotrope, and adding water only after separating off a (non- azeotropic) NMM / water-mixture in the single distillation column.

[0044] In step (i), a mixture containing at least methanol, NMM, water and high boilers, wherein the weight ratio water / (NMM + water) is lower than in an azeotropic NMM / water-mixture at the operating pressure of the single distillation column, is provided. The composition of the mixture provided is that of a typical product mixture of the reductive amination of morpholine with formaldehyde using a methanolic formaldehyde solution, as indicated above.

[0045] In step (ii), water is added to the mixture in such an amount that the weight ratio water I (NMM + water) is still lower than but closer to the weight ratio water / (NMM + water) in an azeotropic NMM / water-mixture at the operating pressure of the distillation column, or the weight ratio water / (NMM + water) corresponds to or is higher than the weight ratio water / (NMM + water) in said azeotropic NMM / water-mixture.

[0046] The weight ratio water / (NMM + water) can be lower than in the original (product) mixture but closer to the weight ratio water / (NMM + water) in an azeotropic NMM / water-mixture at the operating pressure of the single distillation column. This case is also called “left of azeotrope” or “LAZ”. Preferably, the weight ratio water / (NMM + water) is as close as possible to the respective weight ratio of the azeotrope. Preferably, it is less than 2 wt.-% of water lower than the respective ratio of the azeotrope.

[0047] The weight ratio water / (NMM + water) can correspond to or can be higher than the weight ratio water / (NMM + water) in an azeotropic NMM / water-mixture at the operating pressure of the single distillation column. The latter case is also called “right of azeotrope” or “RAZ”. Preferably, the weight ratio water / (NMM + water) is as close as possible to the respective ratio of the azeotrope. Preferably, it is up to 5 wt.-% of water higher, preferably 1 to 3 wt.-% of water higher than the weight ratio water / (NMM + water) in said azeotropic NMM / water-mixture.

[0048] The “right of azeotrope” case is preferred to the “left of azeotrope” case. It has been found that “right of azeotrope” (RAZ) is always more economic than “left of azeotrope” (LAZ) also in case of a single column having a side withdrawal. The addition of water in step (iii) can be carried out in a mixing device as described above for the two column-setup. The addition of water can also occur in a feed tank to which the mixture containing methanol, NMM, water and high boilers is provided.

[0049] The resulting mixture is fed to the single distillation column having a side withdrawal.

[0050] In step (iii), methanol is separated off as overhead stream of the distillation column. Preferably, the water content of the methanol obtained as an overhead stream is 0.02 wt.-% (200 wt.-ppm) or lower.

[0051] In step (iv) a NMM I water-mixture is separated off as side withdrawal stream of the distillation column. Preferably, the content of high boilers obtained as an overhead stream of the second distillation column is 0.1 wt.-% (1000 wt.-ppm) or lower. More preferably, the morpholine content of the NMM / water-mixture obtained as an overhead stream of the second distillation column is 0.1 wt.-% (1000 wt.-ppm) or lower.

[0052] In step (v), a mixture containing high boilers is separated off as bottom withdrawal stream of the distillation column. Preferably, the NMM loss, defined as the NMM mass in the top withdrawal stream of the first distillation column plus the bottom withdrawal stream of the second distillation column divided by the NMM mass fed to the first distillation column, is 0.0020 kg / kg or less. The NMM loss can be essentially zero in the “right of azeotrope” case.

[0053] In general, the single distillation column has from 24 to 55 theoretical stages, and the reflux ratio is correspondingly from 2.6 mole / mole (55 stages) to 9.1 mole / mole (24 stages). The feed stage can be at stage 13 if the distillation column has 24 theoretical stages, or it can be at stage 37 if the distillation column has 55 theoretical stages. The side withdrawal can be at stage 5 if the distillation column has 24 theoretical stages, or it can be at stage 9 if the distillation column has 55 theoretical stages. For example, the distillation column can have 40 theoretical stages, with the feed stage being stage 25 (the sump being stage 1), the side withdrawal being at stage 7, and the reflux ratio being 3.1 mole / mole.

[0054] The distillation column is in general operated at a pressure of from 1.0 to 10 bar, preferably from 1 .0 to 2 bar.

[0055] Preferably, the first distillation column is operated at a pressure as close as possible to atmospheric pressure, but high enough to make up for pressure loss, since higher operating pressures require higher temperatures, which is energetically disadvantageous. Operating pressures higher than 10 bar are unfavourable since the sump temperature may rise up to and above 200 °C, which may cause colorization of NMM.

[0056] The single distillation column process turns out to be more economic in terms of investment costs and operating costs than the two column-process. In a preferred embodiment of the process of the invention, the mixture containing at least methanol, N-methylmorpholine (NMM), water and high boilers provided in step (i) is manufactured by reductive amination of morpholine with formaldehyde using a methanolic formaldehyde solution.

[0057] Preferably, the mixture containing at least methanol, N-methylmorpholine (NMM), water and high boilers provided in step (i) has a composition as follows: from 50 to 80 wt.-%, preferably from 66 to 70 wt.-% of NMM, from 8 to 24 wt.-%, preferably from 14 to 18 wt.-%of water, from 6 to 22 wt.-%, preferably from 12 to 16 wt.-% of methanol, from 0.01 to 5.0 wt.-%, preferably from 0.2 to 1 .0 wt.-% of high boilers.

[0058] The invention further related to a process for the preparation of N-methylmorpholine oxide (NMMO), comprising the steps of:

[0059] (a) separating off an NMM / water-mixture in step (iv) of the process described above;

[0060] (b) optionally adding water to said NMM / water-mixture; and

[0061] (c) subjecting the NMM / water-mixture obtained in step (a) or (b) to an oxidation reaction to obtain NMMO.

[0062] Oxidation step (c) comprises reacting N-methylmorpholine with a less than stoichiometric amount of aqueous hydrogen peroxide in an aqueous medium in the presence of carbon dioxide as promotor.

[0063] In a preferred embodiment, oxidation step (c) comprises reacting N-methylmorpholine with a less than stoichiometric amount of aqueous hydrogen peroxide in an aqueous medium while imposing on the aqueous medium a vapor space having a carbon dioxide partial pressure p (CO2) of less than 0.75 bar absolute, preferably less than 0.20 bar absolute. Although carbon dioxide is an effective promoter for the conversion of a tertiary amine into its amine oxide, applicants have found that amounts of free carbon dioxide in excess of a promoting amount can lead to undesired discoloration.

[0064] In an embodiment, the initial concentration of tertiary amine in the aqueous medium, that is the concentration of tertiary amine prior to the addition of hydrogen peroxide, is in the range of from 40 to 95 vol.-%, preferably 60 to 85 vol.-%. Any aqueous hydrogen peroxide can be used. In view of practical considerations, the concentration of hydrogen peroxide in the aqueous hydrogen peroxide that is added to the aqueous medium is in the range of from 10 to 71 wt.-%, preferably from 29 to 51 wt.-%.

[0065] The invention is further illustrated by the following examples. Examples

[0066] All examples are calculated in an equation-oriented inhouse flowsheet simulator with the same working principles of commonly known commercial software packages, i.e. by AspenTech or AVEVA.

[0067] Table 1 summarizes all components with short name, CAS number and molar mass. During the example section theoretical stage amounts are determined and these stages are always counted from the bottom of the column.

[0068] Table 1 : Table of all components considered in the simulation with short name, CAS number and molar mass.

[0069] ‘Substitute component for all residual and not specified high boilers in the mixture

[0070] Simulation results for the two-column setup

[0071] For the two-column setup, three cases are described in this example section. All cases comprise one MEOH separation (see figures 1 and 2: column 1), one NMM / H2O separation (see figures 1 and 2: column 2), and a mixing step (see figure 1 and 2: “Mix”). All these cases share the same feed from the synthesis section denoted as stream “1” in the figures below.

[0072] Figure 1 : „Water Before" setup with two columns. Synthesis: Reaction section. Column 1 : MEOH-separation. Mix: Mixing device to add water. Column 2: NMM / H2O-separation.

[0073] Figure 2: „Water After" setup with two columns. Synthesis: Reaction section. Column 1 : MEOH separation. Column 2: NMM / H2O separation. Mix: Mixing device to add water. While the NMM / H2O-separation is different for the cases, the MEOH-separation is always the same. The preferred setup comprises a column with 30 theoretical stages and the feed stage on stage 15. Moreover, the head pressure is set to 1.50 bara, and the pressure drop over the column is assumed to be 4 mbara per theoretical stage. In addition to this, the pressure loss over the evaporator is assumed to be 162 mbara, and the pressure loss over the condenser is assumed to be 150 mbara. The condenser is set to 53 °C outlet temperature given by the cooling water conditions at the production site though condensation temperature is roughly at 69 °C. A total condensation is not possible because of the residual inerts from synthesis such as hydrogen and carbon monoxide. The MEOH column is further characterized by requiring 1000 wt- ppm MEOH in the sump draw stream (stream “3”) and 200 wt-ppm H2O in the distillate draw stream (stream “2”). Therefore, the reflux ratio is calculated to be 2.69 mole / mole and the sump draw stream is calculated to be 6447 kg / h. The evaporator requires 1610 kW and the condenser 1267 kW.

[0074] Table 2 summarizes the composition of the streams of the MEOH separation for the two-column setup.

[0075] Table 2: Stream table for „Water Before" and “Water After” setup with two columns. Stream numbering according to Figure 1 and 2.

[0076] The mixing device (see figures 1 and 2: “Mix”) is set previously to the NMM / H2O-separation for the “Water Before” cases. For comparison purposes, there is also a mixing device added after the NMM / H2O-separation in the “Water After” cases to ensure that the same NMM / H2O- concentrations are compared in the two cases. This would technically not be necessary if the resulting NMM / H2O-concentrations are sufficient. This does not change the energy demand of column 2 and thus does not affect the results in this example. Nevertheless, it ensures the same product concentrations for the “Water Before” and “Water After” cases in stream “6”. The added water stream (stream “4”) is calculated such that the H2O concentration in the product stream “6” is 29 wt-% H2O / (NMM+H2O).

[0077] Now, the second distillation column for NMM / H2O-separation differs in stages and energy demand depending on the respective case but the following assumptions are the same for all cases. The head pressure is set to 1 .50 bara and the pressure drop over the column is assumed to be 4 mbara per theoretical stage. In addition to this, the pressure loss over the evaporator is assumed to be 162 mbara, and the pressure loss over the condenser is assumed to be 150 mbara. Following this, the condenser is set as total condenser because no inerts are present and the condensation temperature is significantly above 53 °C with about 102 °C. Thus, cooling water or air may be used as cooling medium. The NMM / H2O column is further characterized by requiring 1000 wt-ppm MO in the product draw (stream “6”) and only 0.002 kg / kg NMM loss throughout the separation sequence. The NMM loss is calculated as the sum of the NMM mass flow in stream “2” and stream “7” divided by the NMM mass flow in stream “1 ”. Therefore, the reflux ratio and the sump draw are calculated variables.

[0078] For the “Water Before” case according to the invention there are two cases considered, the “LAZ” and “RAZ” case. For the “Water After” case only the RAZ case is presented here for comparison.

[0079] Figure 3: Stage (N) versus energy demand (Q) optimization curve for „Water Before" and “Water After” setup with two columns for “RAZ” case. “Water Before” (line): Water added before column 2, see figure 1 . “Water Before NoNMMLoss” (narrow dashed line): Water added before column 2, see figure 1 , and no NMM loss in the sump of column 2 allowed. “Water After” (wide dashed line): Water added after column 2, see figure 2. “RAZ”: Added as much water to the feed of column 2 that the water concentration is narrow to but slightly on the water rich side of the azeotropic concentration.

[0080] Figure 4: Stage (N) versus reflux ratio (v) optimization curve for „Water Before" and “Water After” setup with two columns for “RAZ” case. “Water Before” (line): Water added before column 2, see figure 1 . “Water Before NoNMMLoss” (narrow dashed line): Water added before column 2, see figure 1 , and no NMM loss in the sump of column 2 allowed. “Water After” (wide dashed line): Water added after column 2, see figure 2. “RAZ”: Added as much water to the feed of column 2 that the water concentration is narrow to but slightly on the water rich side of the azeotropic concentration.

[0081] Beginning with the latter, figures 3 and 4 emphasize the design difference between “Water Before - RAZ” and “Water After - RAZ”. They show the energetically optimum designs at given theoretical stage numbers and specifications. In other words, at a given theoretical stage number, the minimum reflux ratio and favorable feed stage and thus minimum energy demand is calculated while maintaining the specifications of 1000 wt-ppm MO in the product stream and 0.002 kg / kg NMM loss throughout rectification. Following this approach for different stage numbers results in a theoretical stage number (N) versus energy (Q) or reflux ratio (v) plot presenting the best designs possible at different stage numbers. This optimization approach is commonly used in the design of rectification columns. For instance, at 30 theoretical stages the minimum energy demand of the “Water Before” case is 1932 kW while in the “Water After” case it is significantly increased with 3465 kW.

[0082] Summarizing, figures 3 and 4 show that at any given stage number, the energy demand of adding water before NMM / H2O-rectification is lower than the energy demand of adding water after NMM / H2O-rectification considering same stage amount. (Conclusion 1).

[0083] In addition to this, the specification of 0.002 kg / kg NMM loss is derived from a plant trial and taken for comparison reasons in both cases. Another advantage of adding water before the NMM / H2O-rectification is that it is the only energetically reasonable way to reach essentially no NMM loss during rectification. Decreasing the NMM loss for the “Water After” case to zero will drive the energy demand to the extreme. For instance, at 30 stages the “Water After” case’s energy demand would increase from 3465 kW to 10660 kW to decrease the NMM loss from 0.002 kg / kg to 0.0003 kg / kg NMM loss. While the “Water Before” case’s energy demand would marginally increase from 1932 kW to 1940 kW to decrease NMM loss from 0.002 kg / kg to only 0.0003 kg / kg considering the same theoretical stage number of 30. Moreover, the NMM loss for the “Water Before” setup may even be decreased to 10’8kg / kg, that means essentially no NMM loss in the NMM / H2O-column. As shown in figure 3, decreasing the NMM loss to 10’8kg / kg (“NoNMMLoss”) may not increase the energy demand significantly above 25 theoretical stages but below 25 theoretical stages the decreasing NMM loss demands more energy.

[0084] Summarizing, only the “Water Before” setup may decrease the NMM loss during rectification to essentially zero within reasonable energy demand limits. (Conclusion 2)

[0085] Finally, there may be another way of adding water instead of a mixing device in the feed. It might also be possible to add water directly into the column using a second feed stage. This way the added water would be used as stripping medium within the column directly. This may also lead to a column setup with no NMM loss. However, as compared to the “Water Before - RAZ - NoNMMLoss” setup, this “direct stripping” setup would be energetically unfavorable. For instance, at the same stage amount of 30 stages, the “direct stripping” setup would require 2369 kW while the “Water Before - RAZ - NoNMMLoss” requires 1937 kW.

[0086] Summarizing, mixing water into the feed instead of directly feeding the water into the column with a second feed stage is energetically favorable. (Conclusion 3). Any column design along the curve of figure 3 and 4 is possible. For “Water After - RAZ” a preferable design might be 37 stages, feed on stage 29 and a reflux ratio of 0.92 mole / mole representing 2422 kW evaporator and 2474 kW condenser duty. Table 3 summarizes the streams for the “Water After - RAZ” case excluding the MEOH separation in column 1 . The stream numbers are given in figure 2.

[0087] Table 3: Stream table for “Water After - RAZ” setup with two columns. Stream numbering according to figure 2.

[0088] Though any column design along the curve of figure 3 and 4 is possible for “Water Before - RAZ”, for “Water Before - RAZ” a preferable design might be 10 theoretical stages, feed on stage 8 and a reflux ratio of 0.028 mole / mole representing 1933 kW evaporator and 1923 condenser duty. Table 4 summarizes the streams for the “Water Before - RAZ” case excluding the MEOH separation in column 1. The stream numbers are given in figure 1.

[0089] Table 4: Stream table for “Water Before - RAZ” setup with two columns. Stream numbering according to figure 1 .

[0090] As discussed above, there is also the possibility to decrease the NMM loss to zero NMM loss.

[0091] Though any column design along the curve of figure 3 and 4 is possible for “Water Before - RAZ - NoNMMLoss”, for this setup a preferable design might be 22 stages, feed on stage 20 and a reflux ratio of 0.029 mole / mole representing 1939 kW evaporator and 1929 condenser duty. Table 5 summarizes the streams for the “Water Before - RAZ - NoNMMLoss” case excluding the MEOH separation in column 1. The stream numbers are given in figure 1. Table 5: Stream table for “Water Before - RAZ - NoNMMLoss” setup with two columns. Stream numbering according to figure 1 .

[0092]

[0093] Knowing that the addition of water immediately before the rectification of NMM / H2O is advantageous to adding water after the rectification, it is still necessary to estimate the amount of water that should be added. A common engineer would add as low amount of water as possible to decrease additional energy costs during rectification. Following this, the “Water Before” case is investigated for different amounts of added water, beginning with the case of no added water, followed by the case of water addition to approach the azeotropic point (“LAZ” and “RAZ” case), and finally the case of significantly adding water to reach 40 wt% H2O / (NMM+H2O) in the product. Thus, there are four cases presented. Figure 5 showing another stage number (N) versus energy demand (Q) curve as described in the previous paragraph. Accordingly, the same specifications are set, with maximum 1000 wt-ppm MO in the product stream and 0.002 kg / kg NMM loss throughout rectification. It can be seen that from adding no water (“NoH2OFeed”) over adding water close to the NMM rich side of the azeotropic concentration (“LAZ”) up to adding water slightly above the azeotropic concentration on the H2O rich side of the azeotrope (“RAZ”), the energetically optimum design improves. For instance, at the same stage number of 31 stages the “NoH2OFeed” case requires 3605 kW and the “LAZ” case requires 2166 kW, while the “RAZ” case requires only 1932 kW. If significantly more water is added, i. e. 40 wt% H2O / (NMM+H2O) (“40wt%_H2O”), the theoretical stage number does not change but the energy demand significantly increases. For instance, at the same theoretical stage number of 7 theoretical stages, the “40wt%_H2O” setup requires 3252 kW while the “RAZ” setup requires only 1952 kW.

[0094] Figure 5: Theoretical stage number (N) versus energy demand (Q) optimization curve for „Water Before" setup with two columns (see figure 1). “NoH2OFeed” (wide dashed line): No additional water is added to the feed of column 2. “LAZ” (narrow dashed line): Added as much water to the feed of column 2 that the water concentration is narrow to but slightly on the NMM rich side of the azeotropic concentration. “RAZ” line): Added as much water to the feed of column 2 that the water concentration is narrow to but slightly on the water rich side of the azeotropic concentration. “40wt%_H2O” (dotted and dashed line): Added as much water to the feed of column 2 that the water concentration is significantly higher than the azeotropic concentration with 40 wt% H2O / (NMM+H2O).

[0095] Summarizing, figure 5 shows that at any given number of theoretical stages, the energy demand is lower by adding water in such an amount that the water concentration is slightly on the water rich side of the azeotropic concentration (“RAZ”). (Conclusion 4) Simulation results for the one-column setup

[0096] After discussing the two-column setup, the one-column with side draw setup is also discussed in this example section. For the one-column setup all four conclusions (1) to (4) as for the two- column setup are valid:

[0097] (1 ) At any given stage amount, the energy demand of adding water before N M M / H2O rectification is lower than the energy demand of adding water after NMM / H2O rectification considering same stage amount.

[0098] (2) Only the “Water Before” setup may decrease the NMM loss during rectification to essentially zero within reasonable energy demand limits.

[0099] (3) Mixing the additional water into the feed stream instead of directly feeding the water into the column via a second feed stage is energetically favorable at any design constellation.

[0100] (4) At any given stage amount, the energy demand is lower by adding water in such an amount that the water concentration is slightly on the water rich side of the azeotropic concentration (“RAZ”).

[0101] Figure 6 shows the “Water Before” setup for a one-column design with a side draw.

[0102] Figure 6: „Water Before" setup with one side draw column. Synthesis: Reaction section. Mix: Mixing device to add water. Column 1 : Side draw column.

[0103] The one-column case is specified as the two-column case for comparison reasons. Firstly, the head pressure is set to 1 .50 bara and the pressure drop over the column is assumed to be 4 mbara per theoretical stage. Secondly, the pressure loss over the evaporator is assumed to be 162 mbara, and the pressure loss over the condenser is assumed to be 150 mbara. Thirdly, the top condenser is set to 53 °C outlet temperature given by the cooling water conditions at the production site though condensation temperature is about 69 °C. A total condensation is not possible because of the residual inerts from synthesis such as hydrogen and carbon monoxide. Fourthly, the side condenser is set as total condenser because no inerts are present and the condensation temperature is significantly above 53 °C with about 107 °C. Thus, cooling water or air may be used as cooling medium. Fifthly, the MEOH product stream (stream “2”) is specified with 200 wt-ppm H2O and thus the reflux ratio is calculated. Sixthly, the NMM product stream (stream “6”) is specified with 1000 wt-ppm MEOH and therefore the side draw is calculated. Seventhly, the NMM product stream (stream “6”) is specified with 1000 wt-ppm MO and hence the sump draw is calculated. Eighthly, the NMM product stream (stream “6”) is specified with 29 wt% H2O / (NMM+H2O) and thus the added mass flow of water (stream “4”) is calculated. Ninthly, a maximum of 0.002 kg / kg NMM loss during rectification is still specified but may not be fixed in simulation because there are no degrees of freedom left in the equation system.

[0104] Nonetheless, the optimum design may still be expressed in a theoretical stage number (N) versus operating costs (C) curve. This approach is analogue to a theoretical stage number (N) ver- sus energy demand (Q) optimization because the energy demand is directly linked to operating costs for the heating means such as steam. Now, the operating costs (€ / h) are the sum of the steam consumption (kg / h) times steam costs (€ / kg) and the NMM product loss (kg / h) times the product price (€ / kg). This leads to a design representing the minimal operating costs at a given theoretical stage number while maintaining the specifications stated above. In other words, the best design is always driven towards the lowest NMM loss while maintaining reasonable energy demand within the product specifications. All optimization cases resulting in a NMM loss greater than 0.002 kg / kg are discarded because they do not meet the minimum specification of a maximum NMM loss of 0.002 kg / kg. All other cases with less NMM loss than 0.002 kg / kg are regarded as valid optimization results.

[0105] Figures 7 and 8 show the result of this optimization for a one-column setup with side draw (“Water Before - OneCol - RAZ”: line). It displays the energetically optimum design at given theoretical stage numbers resulting in a theoretical stage number (N) versus energy demand (Q) or reflux ratio (v) curve. In comparison to that, the two-column setup is also shown either with the previously described 30 stage MEOH separation (“Water Before - TwoCol - RAZ - 30stMEOH”: wide dashed line) or with a 44 stage MEOH separation (“Water Before - TwoCol - RAZ - 44stMEOH”: narrow dashed line) or with a 20 stage MEOH separation (“Water Before - TwoCol - RAZ - 20stMEOH”: dotted and dashed line). The total amount of theoretical stage numbers for the two-column setup is therefore the sum of the NMM / H2O column’s theoretical stage number that can vary as shown in figures 3 and 4 and the MEOH column’s stage amount being either 20, 30 or 44 theoretical stages. Analogue to this, the energy demand for the two-column setup is derived by summing up the energy demand of the MEOH separation being 2557 kW for 20 stages, 1610 kW for 30 stages and 1434 kW for 44 stages and the energy demand of the NMM / H2O separation that can vary as shown in figure 3 and 4.

[0106] Figure 7: Stage (N) versus energy demand (Q) optimization curve for „Water Before" setup with one or two column setup for “RAZ” case an different MEOH separation stages.

[0107] “TwoCol_20stMEOH” (dotted and dashed line): Two-column setup as in figure 1 with a 20 stages MEOH separation. “TwoCol_30stMEOH” (wide dashed line): Two-column setup as in figure 1 with a 30 stages MEOH separation. “TwoCol_44stMEOH” (narrow dashed line): Two-column setup as in figure 1 with a 44 stages MEOH separation. “OneCol” line): One-column setup with a side draw as in figure 6.

[0108] Figure 8: Stage (N) versus reflux ratio (v) optimization curve for „Water Before" setup with one- column setup for “RAZ” case. “OneCol” (line): One-column setup with a side draw as in figure 6.

[0109] If the two-column setup is set to 20 or 30 theoretical stages for the MEOH separation, the one- column setup is at any stage amount energetically advantageous over the two-column setup. For instance at 50 theoretical stages, the one-column setup requires 3477 kW while the two- column setup requires 3560 kW with a 30 theoretical stage MEOH separation or 4498 kW with a 20 theoretical stage MEOH separation. Only if the theoretical stage number of the MEOH separation is significantly increased to 44 theoretical stages, then the two-column setup may achieve energetical favorable results. For instance at 73 theoretical stages, the one-column setup requires 3445 kW while the two-column setup requires 3370 kW with a 44 stage MEOH separation.

[0110] Anyhow, the one-column setup may be achieved with less overall theoretical stages at similar energy demand. The presented energetically minimum two-column setup with 44 theoretical stages for MEOH separation will firstly require increased investment costs due to the high MEOH column, secondly due to the second equipment and thirdly may not achieve the separation in less than overall 50-60 stages. In contrast to that, the one-column design may achieve separation at reasonable energy demand even between 30-50 stages. For instance, the energy demand at 50 stages is 3477 kW for the one-column setup, while at roughly the same energy demand of 3442 kW the two-column setup with 44 stages MEOH separation requires about 62 stages. In other words, 24 % more theoretical stages and thus investment costs are required to achieve 1 % less energy demand and thus energy costs. In addition to this, the two-column setup requires two columns with full periphery in contrast to a one-column design meaning the investment costs difference will probably higher than 24 % in reality.

[0111] Summarizing, the one-column design is in most cases economically more favorable in terms of investment costs and operating costs than the two-column process.

[0112] Any column design along the curve of figure 7 and 8 may be possible for the one-column setup. A preferable design may be 40 stages, feed on stage 25, gaseous side draw on stage 7 and a reflux ratio of 3.09 mole / mole representing 3611 kW evaporator and 1405 kW condenser duty. Table 6 summarizes the streams for the “Water Before - RAZ” case with the one-column setup. The stream numbers are given in figure 6.

[0113] Table 6: Stream table for “Water Before - RAZ” setup with one-column. Stream numbering according to figure 6.

Claims

Patent claims1 . A process for separating a mixture containing at least methanol, N-methylmorpholine (NMM), water and high boilers by distillation in a first and a second distillation column, comprising the steps:(i) feeding to the first distillation column the mixture containing at least methanol, NMM, water and high boilers, wherein the weight ratio water / (NMM + water) is lower than in an azeotropic NMM / water-mixture at the operating pressure of the second distillation column,(ii) separating off methanol as overhead stream of the first distillation column, wherein a mixture containing NMM and water is obtained as bottom withdrawal stream of the first distillation column, wherein the weight ratio water / (NMM + water) is lower than in an azeotropic NMM / water-mixture at the operating pressure of the second distillation column,(iii) adding water to the bottom withdrawal stream of the first distillation column in such an amount that the weight ratio water / (NMM + water) is still lower than but closer to the ratio water / (NMM + water) in an azeotropic NMM / water-mixture at the operating pressure of the second distillation column, or the weight ratio water I (NMM + water) corresponds to or is higher than the weight ratio water / (NMM + water) in said azeotropic NMM / water-mixture, and feeding the resulting mixture to the second distillation column,(iv) separating off a NMM / water-mixture as overhead stream of the second distillation column,(v) separating off a mixture containing water and high boilers as bottom withdrawal stream of the second distillation column.

2. The process according to claim 1 , wherein the weight ratio water / (N M M + water) of the mixture obtained in step (iii) and fed to the second distillation column is higher than the weight ratio water / (NMM + water) in said azeotropic NMM I water-mixture.

3. The process according to claim 2, wherein the weight ratio water / (NMM + water) of the mixture obtained in step (iii) and fed to the second distillation column is up to 5 wt.-% of water higher, preferably 1 to 3 wt.-% of water higher than the weight ratio water I (NMM + water) in said azeotropic NMM / water-mixture.

4. The process according to any one of claims 1 to 3, wherein the first column has from 17 to 44 theoretical stages.

5. The process according to claim 4, wherein the reflux ratio in the first column is from 2 to10 mole / mole.

6. The process according to any one of claims 1 to 5, wherein the second column has from 5 to 20 theoretical stages.

7. The process according to claim 6, wherein the reflux ratio in the second column is from 0.02 to 0.5 mole / mole.

8. A process for separating a mixture containing at least methanol, N-methylmorpholine (NMM), water and high boilers by distillation in a single distillation column comprising the steps:(i) providing the mixture containing at least methanol, N-methylmorpholine (NMM), water and high boilers, wherein the weight ratio water / (NMM + water) is lower than in an azeotropic NMM / water-mixture at the operating pressure of the distillation column,(ii) adding water to the mixture in such an amount that the weight ratio water I (NMM + water) is still lower than but closer to the weight ratio water / (NMM + water) in an azeotropic NMM / water-mixture at the operating pressure of the distillation column, or the weight ratio water / (NMM + water) corresponds to or is higher than weight ratio water / (NMM + water) in said azeotropic NMM / water-mixture, and feeding the resulting mixture to the distillation column,(iii) separating off methanol as overhead stream of the distillation column,(iv) separating off a NMM / water-mixture as side withdrawal stream of the distillation column,(v) separating off a mixture containing high boilers as bottom withdrawal stream of the distillation column.

9. The process according to claim 8, wherein the weight ratio water / (NMM + water) of the mixture obtained in step (ii) and fed to the distillation column is higher than the weight ratio water / (NMM + water) in said azeotropic NMM / water-mixture.

10. The process according to claim 9, wherein the weight ratio water / (NMM + water) of the mixture obtained in step (iii) and fed to the second distillation column is up to 5 wt.-% of water higher, preferably 1 to 3 wt.-% of water higher than the weight ratio water I (NMM + water) in said azeotropic NMM / water-mixture.11 . The process according to any one of claims 8 to 10, wherein the column has from 24 to 55 theoretical stages.

12. The process according to claim 11 , wherein the reflux ratio in the column is from 2 to 10 mole / mole.

13. The process according to any one of claims 1 to 12, wherein the mixture containing at least methanol, N-methylmorpholine (NMM), water and high boilers provided in step (i) is manufactured by reductive amination of morpholine with formaldehyde using a methanolic formaldehyde solution14. The process according to any one of claims 1 to 13, wherein the mixture containing at least methanol, N-methylmorpholine (NMM), water and high boilers provided in step (i) has a composition as follows: from 50 to 80 wt.-%, preferably from 66 to 70 wt.-% of NMM, from 8 to 24 wt.-%, preferably from 14 to 18 wt.-% of water, from 6 to 22 wt.-%, preferably from 12 to 16 wt.-% of methanol, from 0.01 to 5.0 wt.-%, preferably from 0.2 to 1 .0 wt.-% of high boilers.

15. A process for the preparation of N-methylmorpholine oxide (NMMO), comprising the steps of:(a) separating off an NMM I water-mixture in step (iv) of the process according to any one of claims 1 to 14;(b) optionally adding water to said NMM / water-mixture; and(c) subjecting the NMM I water-mixture obtained in step (a) or (b) to an oxidation reaction to obtain NMMO.

Images