Purifying aqueous solutions

Abstract

Provided is a process for purifying an aqueous solution, wherein the aqueous solution comprises one or more sugars and additionally comprises furfural, hydroxymethylfurfural, or a mixture thereof; wherein the process comprises passing the aqueous solution through a collection of resin particles; and wherein the resin particles comprise covalently bound acid functional groups.

Claims

1. A process for purifying an aqueous solution, wherein the aqueous solution comprises one or more sugars, two or more organic carboxylic acids and additionally comprises furfural, hydroxymethylfurfural, or a mixture thereof, wherein the process comprises passing the aqueous solution through a bed of resin particles at a temperature from 40° C. to 80° C., followed by passing an aqueous solution containing one or more inorganic acids through said bed to elute the one or more sugars, two or more organic carboxylic acids, furfural, and hydroxymethylfurfural from the bed, wherein the furfural and hydroxymethyfurfural elute from the bed much later than the one or more sugars and two or more organic carboxylic acids thereby separating furfural and hydroxymethylfurfural from said one or more sugars and said two or more organic carboxylic acids without binding furfural and hydroxymethylfurfural to the resin particles, wherein the resin particles comprise 90 wt % or more polymerized units of vinyl aromatic monomer and comprise covalently bound sulfonic acid functional groups and wherein 50 mole % or more of the acid functional groups covalently bound to the resin particles are in hydrogen form.

2. The process of claim 1, wherein the resin particles have volume-average diameter of 400 μm or less.

3. The process of claim 1, wherein the resin particles have uniformity coefficient of 1.2 or less.

4. The process of claim 1, wherein the aqueous solution comprises four or more organic carboxylic acids and additionally comprises one or more inorganic salts, one or more alcohols, or a mixture thereof.

Description

EXAMPLE 1: PULSE TEST

(1) A preliminary aqueous solution was prepared by dissolving exactly one of the compounds below in a solvent made by dissolving 2 grams of sulfuric acid per liter of water. Each solute was present at 20% by weight or else by the maximum solubility in the solvent, whichever was lower.

(2) The resin was a gel resin having polymerized units of styrene and divinylbenzene and having sulfonic acid groups. The amount of polymerized units of divinylbenzene, by weight based on the weight of the resin, was between 3 and 6%. D50 was 310 μm, and UC was 1.2. Total capacity was between 1.3 and 1.75.

(3) A packed bed of resin particles (91 cm long by 2.7 cm diameter) was prepared in a chromatography column. The amount of preliminary aqueous solution loaded onto the top of the column was 0.05 BV (26.1 mL). The eluent was a solution of 2 g/L of sulfuric acid in water. Eluent flowed through the column at 3.0 BV per hour. The process was conducted at 53° C. Elution fractions were collected at the exit of 8 mL each (0.015 BV)

(4) Each fraction was analyzed using a Reichert™ AR200 refractometer. Compound standards of known concentration were used to convert refractometer signal into concentration in g/L. For each compound, this procedure generated a curve of concentration versus BV.

(5) The entire procedure was repeated for each of the compounds listed below.

(6) The mean (μ.sub.1) and variance (μ.sub.2) for each such curve is reported below. Mean and variance were calculated as follows. Retention “times” were measured in BV. Equation 1 below was used to calculate the retention time (t.sub.i) of a given fraction i:
t.sub.i=t.sub.delay+t.sub.elution,i+t.sub.deadvolume  (Eq. 1)
Where t.sub.delay is the delay before starting the fraction collector, t.sub.elution,i is the time after starting is the fraction collector where the midpoint of a given fraction i comes out, and t.sub.deadvolume is added delay caused by system dead volume as discussed above. t.sub.delay was measured directly by the pulse test operator, while t.sub.elution,i and t.sub.deadvolume were calculated using Equations 2 and 3 respectively below:

(7) $\begin{array}{cc}{t}_{\mathrm{elution},i}=\left(i-0.5\right)*t_{\mathrm{fraction}}& \left(\mathrm{Eq}.2\right)\\ t_{\mathrm{deadvolume}}=\frac{1}{F}*\left(V_{\mathrm{piping}}+\frac{V_{\mathrm{Loop}}}{2}\right)& \left(\mathrm{Eq}.3\right)\end{array}$
Where in Eq. 2 i is the fraction number and t.sub.fraction is the collection time of each fraction programmed into the fraction collection. In Eq. 3 F is the system flow rate, V.sub.piping is the estimated volume of piping (outside of the column itself) that the sugar pulse must travel to get from the sample loop to the fraction collector, and V.sub.Loop is the sample loop volume. The sample loop is a small piece of tubing which holds the preliminary aqueous solution prior to loading on the column.

(8) C.sub.x,i is the normalized concentration of a component x (e.g. glucose, fructose, or maltose) in fraction i. The concentrations of each component were then normalized to the pulse feed concentration. Thus, the calculation for C.sub.x,i—the normalized concentration of component x in a pulse test fraction—is:

(9) $\begin{array}{cc}{C}_{x,i}=\frac{A_{x,i}}{A_{x,\mathrm{Pulse}}}& \left(\mathrm{Eq}.4\right)\end{array}$
where A.sub.x,i is the peak area (undiluted) of component x in fraction i, and A.sub.x,Pulse is the peak area (undiluted) of component x in the pulse feed.

(10) With t.sub.i and C.sub.x,i determined, the first peak moment, μ.sub.1, (the arithmetic mean of the peak), was calculated for each component x using Eq. 5 below:

(11) $\begin{array}{cc}{\mu }_{1,x}=\frac{\underset{n}{.Math.}\left[C_{x,i}*t_i*t_{\mathrm{fraction}}\right]}{\underset{n}{.Math.}\left[C_{x,i}*t_{\mathrm{fraction}}\right]}& \left(\mathrm{Eq}.5\right)\end{array}$
where t.sub.fraction on is again the time between fractions and n is the total number of fractions analyzed.

(12) The value of μ.sub.2 for component x, the peak variance, is calculated using Eq. 6 below:

(13) $\begin{array}{cc}{\mu }_{2,x}=\frac{\underset{n}{.Math.}\left[C_{x,i}*t_i^2*t_{\mathrm{fraction}}\right]}{\underset{n}{.Math.}\left[C_{x,i}*t_{\mathrm{fraction}}\right]}-{\mu }_{1,x}^2& \left(\mathrm{Eq}.6\right)\end{array}$
In the results below, standard deviation (sd) is reported, where sd is the square root of μ.sub.2.

(14) Results were as follows, in units of BV:

(15) TABLE-US-00001 compound mean sd Furfural 1.677 0.281 Hydroxymethyl- 1.724 0.106 furfural Acetic Acid 0.705 0.079 Ascorbic Acid 0.573 0.098 Citric Acid 0.544 0.109 Gluconic Acid 0.663 0.133 Maleic Acid 0.604 0.108 Malic Acid 0.645 0.141 Propionic Acid 0.771 0.117 Sodium Sulfate 0.472 0.052 Succinic Acid 0.668 0.102 Glucose 0.522 0.085 Glycerol 0.715 0.088 Glycolic Acid 0.746 0.09 Itaconic Acid 0.698 0.12 Lactic Acid 0.77 0.128 Levulinic Acid 0.754 0.156 Tartaric Acid 0.61 0.103 2-Ketoglutaric Acid 0.572 0.107 Ethanol 0.794 0.097 Saccharic Acid 0.539 0.125
The furfural and the hydroxymethylfurfural both elute much later that all the other compounds, as shown by the mean values of elution time. Also, the furfural and hydroxymethylfurfural peaks do not significantly overlap with any of the peaks from the other compounds, as shown by the standard deviation (sd) values. This result demonstrates that the furfurals could readily be separated from any mixture of the other compounds using the method employed.

(16) It is contemplated that the pulse test is an accurate predictor of the efficacy of an SMB method. That is, it is expected that an SMB method conducted using materials similar to Example 1 would yield results similar to those of Example 1.

COMPARATIVE EXAMPLE 2C

(17) Pulse tests using the procedure in Example 1 were conducted. The resin was a crosslinked acrylic gel resin with tertiary amine functional groups, total capacity of 1.6 eq/L or greater, harmonic mean particle size of 400 to 500 μm, and uniformity coefficient of 1.3 or less. Each of the following compounds showed an elution curve (i.e., a curve of concentration in the exit fraction versus time) that overlapped substantially with the elution curve of either furfural or hydroxymethylfurfural or both: sodium sulfate, acetic acid, lactic acid, gluconic acid, glucose, glycerol, ethanol, levulinic acid, and propionic acid. The procedure was not useful for separating furfurals from aqueous solutions containing any of these compounds.

COMPARATIVE EXAMPLE 3C

(18) Pulse tests using the procedure in Example 1 were conducted. The resin had quaternary ammonium functional groups, volume-average particle diameter of 300 μm, and uniformity coefficient of less than 1.2.

(19) Each of the following compounds showed an elution curve (i.e., a curve of concentration in the exit fraction versus time) that overlapped substantially with the elution curve of either furfural or hydroxymethylfurfural or both: citric acid, tartaric acid, propionic acid, succinic acid, acetic acid, and lactic acid. The procedure was not useful for separating furfurals from aqueous solutions containing any of these compounds.

COMPARATIVE EXAMPLE 4C

(20) Pulse tests using the procedure in Example 1 were conducted. The resin was an adsorbent resin with no acid functional groups and no amine functional groups. Furfural and hydroxymethylfurfural bound to the resin and did not elute with the eluent described above. Consequently it was judged that the adsorbent resin is unsuitable for the present method, because, even if separation between the furfurals and other compounds were to be achieved, the resin would have to periodically be subjected to a regeneration process to remove the furfurals, and such regeneration processes are undesirable.