Ion-exchange process for the recovery of hydroxylamine from Raschig

Apr 1, 1982 - Ion-exchange process for the recovery of hydroxylamine from Raschig ... Journal of Chemical & Engineering Data 2009 54 (7), 2028-2032...
2 downloads 0 Views 896KB Size
Ind. Eng. Chem. Process Des. Dev. 1902, 21, 204-210

204

Ion-Exchange Process for the Recovery of Hydroxylamine from Raschig Synthesis Mixtures Fred W. Koff, Styllanos Slfnlades, and Allen A. Tunlck Chemical Research Center, Allied Chemical Corporation, Morristown, New Jersey 07960

A process for isolating hydroxylamine from technical aqueous acid solutions of hydroxylammoniumand ammonium salts was demonstrated. The solution was fed to a sulfonic acid resin in the ammonium form causing hydroxylammonium ions to replace some of the ammonium ions. Aqueous ammonia was then used to remove hydroxylamine as free base. Emphasis was given to minimizing the use of water in the process and maximizing the purity of the product cut. Hydroxylamine recovery, although it could be made quantitative, was kept at -50% because spent liquors containing hydroxylamine can be effectively utilized in the manufacture of oximes. Due to ion exclusion effects, elution of hydroxylamine from the resin was retarded relative to elution of ionic components. For this reason the resin was rinsed only after the passage of ammonia. Further water conservation was realized by recycling dilute product streams to the resin.

Introduction The Raschig synthesis of hydroxylamine (e.g., eq 1-3) NO + NOz + 2NH4++ C0322NH4++ 2NOZ-+ COz (1)

-

2NH4++ NO2- + SOz

+ HS03-

-

-

HON(S0JZ HON(SO,-),

+ 2NH4++ 2H20

+ 2NH4+ (2)

NH,OH++ 2NH4++ H+ + 2S042-(3) is an economical method for producing hydroxylamine for use as an in situ reagent in the manufacture of oximes (Baker, 1968). The end product of the Raschig synthesis is an aqueous solution of hydroxylammonium ions (-1.8 N), ammonium ions (-4.7 N) and hydrogen ions (-1.9 N) balanced by sulfate and, to a minor extent, by nitrate ions. This is frequently referred to as a “hydrox“ solution. The separation of pure hydroxylamine salts and especially of hydroxylamine base solutions from the mixtures of salts obtained in the Raschig synthesis presents some difficulties. Fractional crystallization of these mixtures generally fails due to the relatively high proportion of other salts. One method (Baker, 1968) has been to use the Raschig synthesis solution (after neutralization) to prepare the oxime of a volatile ketone (e.g., eq 4). The resulting NH30H+ + NH3 + RIR2C-O NH4+ + R,R,C=NOH + HzO (4) ketoxime is separated from the aqueous salt solution and subjected to hydrolysis in the presence of a strong acid. This produces a solution of the hydroxylamine salt of the acid and the ketone which is recycled (eq 5). This method R,R2C=NOH + H+ + H 2 0 NH,OH+ + R1R2C0 (5) has several drawbacks. The hydrolysis of ketoximes is reversible even at low pH; therefore, long periods of heating and continuous removal of liberated ketone are required in order to complete the reaction (and minimize side reactions). Also, hydroxylamine base or salts of hydroxylamine with weak or oxidizing acids cannot be produced directly with this method. Ion-exchange methods have been used in the past for preparing hydroxylamine and hydroxylamine salts (Baker, 1968). For example, a strong cation-exchange resin can be used to convert the commercially available hydroxylamine sulfate to other hydroxylamine salts (Wheelwright, 1977; Thompson, 1970). A solution of hydroxylamine

-

-

sulfate is passed through a bed of sulfonic acid resin in the H+ form. Hydrogen ions are replaced by hydroxylammonium ions, and when another strong acid is passed through the resin bed the hydroxylammonium salt of that acid is eluted, albeit in mixture with some of the free acid. This method is limited to making hydroxylamine salts of strong acids and obviously will not separate hydroxylamine salts from salts of other cations. Alternately, a strong anion-exchange resin can be used to prepare other hydroxylamine salts (Thompson and Golding, 1970) or hydroxyamine base (Baker, 1968; Thompson and Golding, 1970) by passing hydroxylamine sulfate through a bed of resin containing the anion of the desired salt or hydroxide ion. Again, the method does not separate hydroxylamine or its salts from salts of other cations. In addition, it may require the use of strong alkali to regenerate the resin and the disposal of the resulting alkali salts. Results and Discussion We have investigated solvent extraction (Tunick et al., 1979; Sifniades et al., 1982), ion exclusion, and ion-exchange methods for separation of hydroxylamine from hydrolyzed Raschig synthesis (“hydrox”) solutions. The ion-exchange method described herein appears to offer the most promise in terms of an economical and nonpolluting plant operation. Selectivity. When a sulfonic acid resin in the ammonium form, P-S03- NH4+,is contacted with a hydrox solution, hydroxylammonium ions and hydrogen ions partly displace ammonium ions from the resin. P-S03- NH4++ NH30H+ P-S03-NH30H+ + NH4+ (6) P-S03- NH4++ H+ P-SO,-H+ + NH4+ (7) The affinity of the resin for hydroxylammonium ion is close to, but somewhat less, than that for ammonium ion, while the affinity for proton is considerably lower. These affinities can be expressed in terms of relative selectivities of the resin for hydroxylammonium ions vs. ammonium ions, S H , and for hydrogen ions vs. ammonium ions, S,

-

0196-4305/82/1121-0204$01.25/00 1982 American Chemical Society

Ind. Eng. Chem. Process Des. Dev., Vol. 21, No. 2, 1982 205 Table 11. Dissociation Constants of Ammonia and Hydroxylamine at 25 " C

Table I. Ion-Exchange Kinetics and Selectivitya temp, "C

30 50 65 30

loading strippingC

SH

S,,

0.86 0.94 0.97

0.25 0.25 0.20

time t o 90% equilib, s

reporteda Kbr mol L-'

200 100 50 60

Resin in ammonium Dowex 50W-X8, 20-50 Mesh. form in contact with a solution of initial composition 4.72 N NH,', 1.78 N NH30H', 1.936 N H+. Resin equilibrated with a large pool of solution 4.72 N NH,', 1.78 N NH,OH', 1.936 N H', then placed in contact with a solution of initial composition 0.652 N NH,. a

For most of the sulfonic type cation exchange resins we examined, SHranged from 0.7 to 1.0 while S, was in the range 0.15 to 0.35 at room temperature and varied slightly with temperature (e.g., Table I). Thus the amount of hydroxylammonium ion absorbed by a resin in equilibrium with an excess of a typical hydrox solution, where the ratio of [NH4+]to [H+] to [NH,OH+] is -2.5:1.1:1.0, is equivalent to about 20-25% of the total capacity of the resin. The other sites are occupied by ammonium ions (65-75%) and hydrogen ions (5-10%). We have found that a very effective separation of hydroxylamine can be made from this "loaded" cation-exchange resin by taking advantage of the greatly reduced basicity of hydroxylamine (pKbZ5"= 7.97) vs. ammonia (pKbZ5" = 4.74) (Lange, 1961). Contact with aqueous ammonia equivalent to the sum of hydroxylammonium ion and hydrogen ion bound to the resin results in the essentially quantitative release of hydroxylamine as free base to the solution (eq 10) accompanied by conversion of the P-S03-NH30H+

+ NH3

+

P-S03-NH4+

+ NHZOH

(10) The bound hydrogen ions to ammonium ions (eq 11). P-SO,-H+ NH, P-S03- NH4+ (11)

+

-

resin is thus converted completely to the ammonium form, and the resulting solution of hydroxylamine, after separation from the resin, can be used directly or converted to a desired salt by addition of the appropriate acid followed by removal of water. This quantitative release of hydroxylamine by the resin amounts to a dramatic change in selectivity with a change in pH. Actually the real selectivity of the resin for NH30H+vs. NH4+ is not likely to vary much with pH, but the apparent selectivity is altered due to the change in relative concentrations of NH30H+ vs. NH4+ as pH changes. Equation 8 can be restated in terms of the base dissociation constants of NHzOH and NH3 and the concentration of H+ as SH = [NH,OH+Iresin (lo-14/(Kb)NHzOH)+ [H+]soh A X x - (12) H [Nh+Iresin (10-14/(Kb)~~8) + [H+I,,h where A and H are, respectively, the total amounts of ammonia (NH, NH4+)and hydroxylamine (NHzOH NH30H+)in solution. The apparent selectivity defined by eq 13 is therefore related to the selectivity observed at

+

+

a

observed K,, mol L-I

x 10-5

NH3 NH,OH

1.8

1.07 X

1.5 x 1 0 - ~ b 7.0X

ratio of Kb's

1700

2100

Lange (1961).

0.6

1

0.3

1

0.2 0.1

At 4.8 N NH,'.

At 1 . 9 N NH,OH+.

\

1

1 1

0 2

3

4

5

6

7

0

9

PH

Figure 1. Apparent selectivity of Dowex 50W-X8 vs. pH compared to theory (eq 14).

It can be seen that at low pH (high [H+]) (SH)app = S H , while at high pH, eq 12 reduces to

As an experimental check of the relationship expressed in eq 14, the apparent selectivity of Dowex 50W-X8 resin for hydroxylamine vs. ammonia at various pH levels was determined and compared to the selectivity computed with the aid of eq 14. The basic dissociation constants of ammonia and hydroxylamine used in the computation were determined under the same conditions as employed in the selectivity measurements (see Table 11) and were found to differ significantly from the values mentioned above presumably because of the changes in the activity coefficients of these species at the much higher concentrations employed in our study (4.8 N NH4+,1.9 N NH30H+). The experimentally derived selectivities were found to be in good agreement with the theoretical values derived from eq 14 using the dissociation constants observed under these conditions (see Figure 1). A study was undertaken to evaluate various cation-exchange resins for utility in a process based on the phenomena discussed above. The capacity of a given resin for hydroxylamine, H,,depends, of course, on its total exchange capacity,' C,, and the selectivities for both hydroxylammonium ion and hydrogen ion vs. ammonium ion, S H and S,, respectively. For a given hydrox solution, the following relationship can be derived

low pH as The value of Hr for various cation-exchange resins was determined. The more suitable resins were of the sulfonic type and had values of Hr close to 0.50 mequiv/mL for

206

Ind. Eng. Chem. Process Des. Dev., Vol. 21, No. 2, 1982 300

c I

200

-

Theoretical Line for 100%

Theoretleal Line for 100% Retention of NH,OH'

,,

?;

-

NH,OH'

,

D

1

LT

9z E

-

r

IO0

-

~

_

_

_-

H+-

_

~

H'

Breakthrough for NH30H+ 0

500

I000 Total Cahonr Ehkd

1500

- meqwv

2000

Figure 2. Uptake behavior of Dowex 5OW-XB (NHIt form) for NH30Ht and H+ in hydrox (see Experimental Section).

resin in contact with the following hydrox solution: [NH4+] = 4.72 N, [NH30H2-] = 1.78 N, [H+] = 1.936 N. Weakly acidic (carboxylic type) resins have higher overall exchange capacities (C,)but their S , values are so high that neutralization of the hydrox is necessary before the loading operation. This loses the advantage of the low S, values which are obtained with the strongly acidic (sulfonic type) resins. With Amberlite IRC-50, SH was found to be around 1.0, so that the overall capacity for hydroxylamine, H,, in equilibrium with a typical neutralized hydrox solution ([NH,+] = 6.66 N, [NH30H+]= 1.78 N) was -0.7-0.75 mequiv/mL. In practice, however, this increased hydroxylamine capacity was more than offset by inferior column performance resulting from packing, channeling, etc. Apparently the acrylic type resin is not as suitable as the polystyrene type in applications where very large changes in concentration are encountered. Kinetics. The rate of exchange of hydroxylammonium ions and protons from a typical hydrox solution with ammonium ions from a sulfonic resin is relatively rapid and quite temperature dependent. The time required to attain 90% equilibrium during loading of Dowex 5OW-XB was found to range from 200 s at 30 "C to 50 s at 65 "C, from which an apparent activation energy of about 4 kcal/mol can be calculated. The rate of stripping of the hydroxylammonium ion loaded resin by ammonia (initial concentration 0.65 N) was even faster, requiring 60 s at 30 "C for 90% approach to equilibrium (Table I). Considering the high concentrations involved, it is very likely that the exchange is limited by diffusion within the resin particles in both the loading and stripping stages (Helfferich, 1962). Column Operation. Column performance experiments were conducted using Dowex 50W-X8 (20-50 mesh) and Amberlite IR-124 (0.42-0.57 mm). Loading curves were generated for both of these resins (see Experimental Section and Figures 2 and 3). From these results it is apparent that two different modes of the loading operation can be followed. If complete recovery of hydroxylamine from the hydrox solution is desired, the feed to the column per cycle would be limited to the total number of cations eluted before breakthrough of hydroxylammonium ions (for example -550 mequiv with Dowex 50W-XB and -520 mequiv with Amberlite IR-124, under the conditions used). On the other hand, if the recovered amount of hydroxylamine per cycle is to be maximized, the feed can be increased to the extent that the hydroxylamine content of the resin approaches the limiting value (for example 1500 mequiv fed with Dowex 50W-XB and 1700 mequiv fed with Amberlite IR-124, under the conditions used). The latter mode is preferable due to more efficient use of the resin and decreased dilution of the product stream (see below)

0

500

1000 Total Cations Eluted

1500

2000

- mequiv

Figure 3. Uptake behavior of Amberlite IR-124 (NH4+form) for NH30Ht and Ht hydrox (see Experimental Section).

but requires that there be a way of utilizing the unrecovered hydroxylamine values in the partially depleted ionic effluent. This is the case, for example, when the hydroxylamine plant is located near an oxime manufacturing facility. In that case the hydroxylamine values in the partially depleted effluent can be utilized to manufacture oxime. The displacement of hydroxylamine from the resin (and concomitant conversion of the resin to the ammonium form) can also be carried out in two different ways. The hydroxylamine can be completely recovered by employing ammonia in a slight excess of the total of hydroxylammonium plus hydrogen ions bound to the resin. The effluent will then contain a small amount of ammonia in addition to hydroxylamine. If desired, the ammonia can be removed by partial evaporation or fractional distillation of the solution. (The latter avoids losses due to the slight volatility of hydroxylamine.) Alternatively, if hydroxylamine which is free of ammonia is required directly, a slight deficiency of ammonia can be fed to the resin. In this case a small amount of hydroxylamine will remain bound as ions on the resin. These will be displaced in the next cycle by the hydrox feed, but need not be lost, e.g., if the partial recovery mode of loading is employed (see above). The second alternative appears more attractive for a commercial process where unrecovered hydroxylamine in solution is utilized for manufacture of oximes, as it saves the cost of equipment to remove ammonia from the hydroxylamine solution. The alternatives mentioned above add substantial flexibility to this technique. Furthermore, ion exchange recovery of hydroxylamine is not limited to concentrated solutions as are encountered in Raschig process plants. As long as the hydroxylamine salts in solution are an appreciable fraction of the total salt content, recovery and separation of hydroxylamine can be carried out with quite dilute solutions. For example, 3.75 L of aqueous solution containing 15 mequiv each of NH4+and NH30H+and 17 mequiv of H+, all as sulfates, was passed through a 34-mL bed of Dowex 50W-X8 (NH4+form). The adsorbed hydroxylamine was obtained in a 50-mL cut at 95% recovery by passage of aqueous ammonia and water. Wash Requirements. A potential drawback to commercial application of the present approach to isolation of hydroxylamine from hydrox solutions is the necessity of using rinse water to separate the ionic effluent from the nonionic product cut. Any water added t o the process must eventually be evaporated. In practice rinse water and water used to make the aqueous ammonia feed becomes distributed between the hydroxylamine product solution

Ind. Eng. Chem. Process Des. Dev., Vol. 21, No. 2, 1982 207 0 0 1 6 5 cm x 1.5 cm Column A A 107 cm x 1.0 cm Column

00.185 cm x 1.5 em Column; x 50°C A A A 107 cm x 1.0 cm Column

- 0.5 0.6

0.4

a

j0.2

A 0.2 0.1

0

1 10

1 0.1

0

A

99.0% Removal

t

lo 20

40

30 ResidenceTime

-

50

0 X

Min

Figure 4. Rinse requirement (B.V.) for indicated degree of removal of ionic material from Dowex 50W-X8 (loading stage) vs. nominal residence time (bed volume/volumetric flow rate). See text.

and the ionic effluent which is returned to the oxime manufacturing facility. In order to minimize the significant contribution that evaporating this water makes to operating costa and to determine the most efficient manner of operation, studies of elution behavior were made with the ionic and nonionic fractions. The volume of rinse required to achieve a given degree of removal of the ionic and basic fractions from the ionexchange column was determined as a function of flow rate and column size for Dowex 50W-X8. Two columns were used with corresponding bed volumes of 185 cm X 1.5 cm i d . and 107 cm X 1.0 cm i.d. Residence times (defined as the ratio of bed volume to volumetric flow rate) ranged from about 10 min to about 50 min. The rinse requirement for each fraction was defined as the eluted volume required to remove a given percentage of total amount of solutes present in that fraction less the volume of the feed for that particular fraction. The starting point for measuring the volume eluted was taken to be the first appearance of ionic material (in the case of the hydrox feed) or the first appearance of basic species in the case of the ammonia feed. The rinse, expressed in bed volumes, required to remove 99.0% or 99.5% of the ionic fraction (spent hydrox) from each column at 50 "C was plotted as a function of residence time in Figure 4. It is evident from the plot that the two columns do not differ significantly. At residence times over 20 min the rinse requirement levels off at about 0.23 bed volume for 99.0% removal and 0.28 bed volume for 99.5% removal. The rinse required to remove 80%, 90%, and 99% of the nonionic fraction (largely hydroxylamine) was similarly plotted in Figure 5. Removal at the lower levels is worth considering in this case because it may be advantageous to obtain partial hydroxylamine recovery in the stripping stage in order to minimize the evaporative load. The two columns again show no significant differences, but the rinse requirement levels off at somewhat longer residence times, over about 35 min. This is probably due to the lower temperature employed in this case. The effect of temperature was not studied systematically, but raising the temperature to 50 "C at the residence time of 45 min resulted in significant reduction of the rinse requirement ( X marks in Figure 5 ) . Since a commercial process would almost certainly operate a t constant temperature and flow rate, given the small size of the fractions relative to the bed volume of the resin (see below), the results indicate that optimum performance with this resin at 50 "C would be achieved with residence times of around 30 min. Much shorter residence time would erode performance in the basic part of the cycle, while longer residence time would require a larger resin bed (for a given

10

20

30 Residence Time

-

40

50

Min

Figure 5. Rinse requirement (B.V.) for indicated degree of removal of hydroxylamine from Dowex 50W-X8 (stripping stage) vs. nominal residence time (bed volume/volumetric flow rate). See text. nvdmr

0

W?lU

05 Volume Elufed

-EV

10

Figure 6. Uptake behavior of Dowex 50W-X8 (NH4+form) as indicated by elutrate composition. Concentration of individual species (as a fraction of the concentration in the feed) vs. volume eluted (B.V.). Upper ordinate indicates the starting points of feed solution and water (hydrox: 4.72 N NH4+, 1.78 N NH3+, 1.936 N H+as sulfates).

output) without improving performance. When the data used in the loading curve for the Dowex resin is used to construct a concentration profile (see Figure 6), it can be seen that the 0.8 B.V. (bed volume) charge has expanded to -1.1 B.V. on passage through the resin bed. The leading edge of the band has a much smaller slope than the trailing edge. This can be readily understood if one considers that the hydrox feed is highly ionic (density N 1.25, solutes -40-45% by wt). Considerable dehydration of the water-swollen resin beads is to be expected with concomitant dilution of the hydrox solution. Indeed, a freshly filled column can show bed volume shrinkage of up to 10% or more when treated with hydrox solution. The bead shrinkage is reversible, and the shrunken resin reabsorbs water from the rinse solution. Recovery of the bed height is incomplete, however, due to denser packing of the beads. This packing leads to much smaller variation in bed height between the hydrox and rinse phases in columns that have been in use for a short time. These phenomena and the attendant deleterious effect on resin life can be reduced by using a more highly cross-linked polystyrene based resin (e.g., Amberlite IR-124 which is 1 2% cross-linked with divinylbenzene). A concentration profile for the product cut can also be plotted (see Figure 7). In this case the resin loaded in the experiment which generated the data for the loading curve (Figure 2) was treated with 51 mL (0.155 B.V.) of 4.1 N

208

Ind. Eng. Chem. Process Des. Dev., Vol. 21, No. 2, 1982

”’

NH

/water

0.6

C/G,

4

i

Ionic Cut

3 .

I



!

0.4

0.2 1 -

0;

0

02

04

06 Volume Eluted

-

08 BV

10

12

Figure 7. Elutrate composition in stripping stage (Dowex 50W-X8). Concentration of NH20H (as a fraction of [NH,OH] in feed) vs. volume elutad (B.V.). Resin charged as in Figure 6. Upper ordinate indicates starting points of aqueous ammonia feed and water.

NH,OH at 50 O C and 45 min average residence time. Here 96% of the total recovery was obtained in -0.30 B.V. of elutrate. Figure 6 also indicates that the ionic material first appears when -0.27 B.V. of water has eluted from the column from the point of loading the hydrox solution. This is in rough correspondence with the interstitial volume of the resin bed. On the other hand, when the “loaded” resin is treated with aqueous ammonia, the band of product hydroxylamine base does not appear in the elutrate until -0.65 B.V. has eluted after the point of addition (see Figure 7). This behavior has a strong bearing on the process and can readily be explained by the phenomenon of ion exclusion (Wheaton, 1953; Helfferich, 1962). In the loading stage, the electrolytes in solution are effectively excluded from the aqueous gel phase of the resin and thus appear after a forerun approximately equal in volume to the interstitial volume of the resin bed. In the stripping stage, the solution in contact with the resin contains nonelectrolytes (hydroxylamine and/or ammonia) which are not excluded from the aqueous gel phase of the resin and thus appear after a forerun approximately equal to the sum of the interstitial volume and the aqueous gel volume of the resin bed. The net effect is a retardation in the rate of travel through the column of the basic band with respect to the ionic band. In a cyclic process, it is desirable to have a uniform separation between the depleted hydrox and product cuts which is no larger than that necessary to avoid significant cross-contamination. From the loading curves (Figures 2 and 3) it can be calculated that approximately 4.6 mequiv of cations/mL of resin are required for effective loading of either resin. Since a typical hydrox solution is about 8 N, the volume of the charge should be about 0.58 B.V. The volume of the aqueous ammonia feed is dictated by the amounts of hydroxylammonium and hydrogen ions retained by the resin after loading and the ammonia concentration. With the Dowex resin, ammonia concentrations higher than - 4 N did not significantly reduce the volume of the product cut, and the volume of basic feed at this concentration sufficient to recover 90-95% of the adsorbed hydroxylamine (and neutralize the adsorbed hydrogen ions) amounts to -0.15-0.17 B.V. Since the basic product cut lags by -0.3-0.35 B.V. relative to the ionic cut and since only -0.25-0.3 B.V. of rinse is needed

0

35

10

15

-8v Figure 8. Concentration profile of ionic and basic fractions with Amberlite IR-124.Upper ordinate indicates the starting points of feed solutions (H, hydrox, A, aqueous ammonia, W, water). The dotted line represents the basic cut obtained when the aqueous ammonia (A) contained 3.48 N hydroxylamine (see text). Volume Eiwed

to remove >99% of the ionic material, it is apparent that little or no rinse water need be used after the hydrox feed (and before the aqueous ammonia feed). Most or all of the rinse water, therefore, must be addeed after the aqueous ammonia is fed. The volume of rinse water required must be sufficient to overcome the “lag” in the basic band and to separate the product cut from the following band of ionic material (from the next cycle). This amounts to 0.3 plus 0.25-0.30 B.V. (see Figures 4 and 6) or about 0.55-0.6 B.V. Thus a complete cycle would consist of -0.55 B.V. of hydrox followed by -0.16 B.V. of 4 N NHIOH followed by -0.58 B.V. of rinse water. A typical complete cycle is shown in Figure 8. A rather small amount of rinse water might be used to separate the hydrox and ammonia feed solutions in a large-scale operation in order to avoid excessive heat generation due to mixing at the column inlet. Recycling. Since it is important to minimize the amount of water used for rinsing, one can employ the expedient of collecting separately those portions of elutrate where the acidic (ionic) and basic (product) peaks overlap at relatively low concentration and using this dilute solution to replace the first part of the rinse water. The ionic material in this “recycle” portion of the rinse should wind up in the tail of the main ionic peak, having in effect passed through the more slowly eluting product band, while the hydroxylamine base should come out with the trailing edge of the product peak. Experiments with operation of the 185 X 1.5 cm column of Dowex resin over three consecutive cycles were carried out using the above feed and rinse scheme with a simulated “recycle”rinse solution based on the expected composition of the overlap between the main peaks. The column effluent was continuously monitored by pH and conductivity and the main and recycle cuts were analyzed in the usual manner (see Experimental Section). The results of one experiment are summarized in Figure 9 and Table 111. Although fully steady-state operation was not reached and the composition of the “synthetic” recycle part of the rinse did not entirely match the composition of the intermediate cuts removed for that purpose, it is apparent that a workable process can be achieved with minor adjustments. Importantly, the product cuts contained hydroxylamine base in excellent purity and very good recovery.

Ind. Eng. Chem. Process Des. Dev., Vol. 21, No. 2, 1982

209

Table 111. Three Cycle Run with Synthetic Recyclea compn, mequivb cut

cycle 1

A H P so,2base A H P s0,zbase A H P so,2base

main ionic cut

product cut

combined recycle cuts

synthetic recycle

1093 142 305 1540 0.8 172 0.8 173 22.7 14.9

mL

8.7 6.0 14.2 0.5

1094 138 304 1536

89.5

70.1

28.6 8.6

A H P s0,zbase

cycle 2

27 8

1.2 176 1.o 177 19.6 9.8

mL 24 7

89.7

62.0

21.8 7.6 55

8.7 6.0 14.2 0.5

cycle 3 1109 133 31 1 1553 0.7 171 1.o 172 5.2 13.5

mL 248

89.9

81.0

7.7 11.0 55

8.7 6.0

55

14.2 0.5

a Dowex 50W-XS 20-50 mesh 1 8 5 cm x 1.5 cm. Feed hydrox: 4.72 N NH,', 1.78 N NH,OH+ as sulfates, 178 mL. Feed ammonia: 8.04 N NH,, 27 mL. Synthetic recycle: 0.1584 N NH,', 0.0994 N NH,OH+, 0.0095 N N H 2 0 H , 0.2587 N SO,, 55 mL. Code: A, total ammonia (NH, + NH,+); H, total hydroxylamine (NH,OH + NH,OH+); P, proton.

H

1

H

AAW 11 1

I

ARW

iI 1

H

i

ARW

ii i

I 0

2

1

Volume Eluted

- BV

3

4

Figure 9. Conductivity and pH of elutrate vs. volume eluted (B.V.) for three cycle experiment. Feed to resin is indicated on upper ordinate, cuts taken on lower ordinate. Code: A, ammonia (aqueous); H, hydrox solution; I, ionic material; P, product (NH,OH) solution; R, recycle; W, water. See text and Table 111.

A further reduction in the amount of added water needed in the process might be accomplished by replacing the water used for preparing the ammonia solution by a portion of the hydroxylamine product cut. For example, if half the product cut were charged with sufficient anhydrous ammonia to serve as the basic feed for the next cycle, at steady state the concentration of hydroxylamine base in the product cut would be about double the value obtained otherwise, assuming constant feed volume. The theoretical relationship at the steady state is

where Co is the concentration of hydroxylamine obtained with the same volume of basic feed but no hydroxylamine recycle, V , is the basic feed volume, and V, is the volume of the product cut. R represents the fractional recovery of hydroxylamine base and should be close to 1.0 since the main loss of NHpOH product would be the small amount ending up in the ionic band despite the dilute recycle. To test the concept, a basic feed solution was prepared containing 3.82 N NH40H, 3.48 N NH,OH, and 0.21 N NH4+ (as sulfate). A 65-mL portion was used to remove

hydroxylamine from a 185 X 1.5 cm column of Amberlite IR-124 resin which had adsorbed 224 mequiv of hydroxylamine ion and 60 mequiv of hydrogen ion from a plant hydrox solution. The product was obtained in a 120-mL cut which contained 3.46 N NHzOH and 0.02 N NH4+as sulfate. When plotted, (dotted line in Figure 8) the width of the NHzOH product peak was not broadened with respect to a similar run without NHzOH recycle despite the increased height. For the next cycle, 65 mL of this solution could be charged with anhydrous ammonia and used as the basic feed, while 55 mL would be retained as product. The net effect is to save the cost of evaporating the water which would have been used to prepare the aqueous ammonia solution. The ion exclusion phenomenon discussed above might form the basis of a somewhat simpler process. This would involve addition of sufficient ammonia to the hydrox solution to neutralize excess acidity and convert most of the hydroxylammonium ion to hydroxylamine base. Alternative feeding of this solution and rinse water to the ion exchange column (the resin would be almost entirely in the ammonium form) should afford separated peaks of ionic solution and nonionic hydroxylamine product solution. In order to succeed, the volume of neutralized hydrox fed per cycle would have to be sufficiently small so that in the effluent the ionic material from the last part of the solution fed would not overlap the neutral material from the first part of the feed. Also the volume of rinse water would have to be large enough to separate the last of the neutral material emerging in one cycle from the first ionic material emerging in the next cycle. In practice (see below) and due in large part to band spreading, the rinse water requirement for this type of operation is significantly higher for a given amount of hydroxylamine production. Consequently, this ion-exclusion method appears to entail higher operating costs. An experiment employing the above procedure was performed by passing a 50-mL (0.15 B.V.) portion of a solution, 7.65 N in NH4+,0.175 N in NH30H+,as sulfates, and 1.57 N in NH,OH, followed by excess water through a 185 cm X 1.5 cm i.d. column of Amberlite IR-124 in the ammonium form at 50 OC and 6.8 mL/min average flow rate. The separation of the peaks in the effluent was such

210

Ind. Eng. Chem. Process Des. Dev., Vol. 21, No. 2, 1982

that only about half of the hydroxylamine could be isolated relatively free of ionic contaminants. This might be improved on by using a smaller amount of feed or lower flow rate, but both are unattractive from a process economics standpoint. In theory, an anion-exchange resin could also be used for this type of process. An experiment similar to the above was &ied out using a quaternary ammonium resin, Amberlite IRA-900, in the sulfate form. The separation was considerably worse, in line with the expected decrease in the Donnan exclusion effect in this type of system (Helfferich, 1962). Both separations suffer from the fact that the electrolyte concentration in the feed is quite high (Wheaton, 1953).

Experimental Section Chemicals. Hydroxylamine sulfate and ammonium sulfate were MCB reagent grade; sulfuric acid and ammonium hydroxide were B & A reagent grade; water was deionized. Ion-Exchange Resins. The resins were all commercial samples and were washed well with deionized water prior to use. The resins supplied in the sodium form were converted to the acid form by passage of excess 2 N H2S04 followed by deionized water until the elutrate was neutral. Analysis of Aqueous Solutions. Solutions with excess acidity were directly titrated potentiometrically with standard sodium hydroxide. Three breaks in the titration curve were found at approximate pH values of 4,7.5, and 10.5 corresponding to H+, NH30H+,and NH4+. Solutions without excess H+ were brought to pH 2-2.5 with measured standard hydrochloric acid and then titrated as above. The difference between added and remaining H+ gives the total content of free base (NH20H or NH20H plus NH3). Total Exchange Capacity of Resin. A 25-mL sample of resin (settled in water) in the H+ form was treated with an approximately twofold excess (50 mL of 4 N) of standard NH40H solution in a covered beaker. After stirring for 20 min, the resin was removed by filtration, washed well with deionized water, and the combined filtrate and washes were made up to volume and analyzed for NH40H content by titration of an aliquot with standard HC1. Selectivity of Resin. The above resin in NH4+form was drained and treated with 20 mL of a solution 4.72 N in NH4+,1.78 N in NH30H+, and 1.936 N in H+, all as sulfates. After stirring for 20 min, a 1.0-mL portion of solution (free of resin) was withdrawn and analyzed for the three cationic species in the usual manner (see above). The selectivities were calculated (see eq 8 and 9) from the observed aqueous concentrations and the inferred (by difference) resin concentrations. Exchange Kinetics. A 53-mL (102 mequiv) portion of Dowex 50W-X8, 20-50 mesh in the NH4+ form, was drained and placed in a jacketed beaker attached to a constant temperature circulator. A 30-mL portion of hydrox solution (see preceding paragraph for composition) was pre-warmed to the desired temperature and added in one portion to the resin, with good magnetic stirring. Samples of solution were withdrawn at short intervals using pipets containing glass wool plugs, and the resin free samples were analyzed in the usual manner (see above). The concentrations of NH30H+, NH4+, and H+ were plotted vs. time and from these plots the selectivities of the resin and the time required for 90% of the approach to equilibrium were determined. For the kinetics of reaction of loaded resin with ammonia, 53 mL of Dowex 50W-X8 in the NH4+ form was equilibrated with two successive 30-mL portions of hydrox solution (see above)

for 15 min and then was rinsed well with deionized water. A 38.4-mL portion of 0.652 N NH40H was added to this resin in the above-mentioned apparatus, and the experiment was carried out as before. The ratio of ammonia to hydroxylamine in solution was plotted vs. time, and the time required for 90% of the approach to equilibrium was determined. Loading Behavior of Resin. A jacketed column, 240 cm X 1.5 cm i.d. and filled to a height of 185 cm with the ion-exchange resin in the NH4+form, was heated to 50 "C by circulated water, and a pre-warmed 262-mL (0.80 B.V.) portion of hydrox solution of the above-mentioned composition was passed through the resin at -7.5 mL/min followed by water at the same rate. The effluent was monitored by pH and conductivity, and successive 20-mL cuts were taken and analyzed in the usual manner. Deviations from feed composition in these cuts were used to construct the loading curves of Figures 2 and 3 and the concentration profile in Figure 6. Conclusions 1. Hydroxylammonium ions in aqueous acidic solutions displace ammonium ions from sulfonic acid resins with an apparent selectivity factor of -0.9. The hydroxylammonium ions on the resin are almost quantitatively displaced by ammonium ions upon contact with an aqueous solution containing an essentially stoichiometric amount of ammonia. 2. Due to ion exclusion effects, hydroxylamine liberated by ammonia treatment of the ion-exchange resin in a column elutes later than ionic components present in the resin. This eliminates the need of washing the hydroxylammonium ion loaded resin prior to elution with ammonia. 3. The hydroxylamine concentration in the product can be significantly increased by recycling part of the product with the ammonia eluent. Acknowledgment We are indebted to R. R. Hertzog for several useful discussions, to R. H. Belden for suggesting the recycling of the product cut to the regenerant stream, and to J. 0. Tormey for technical assistance. Nomenclature A = total ammonia (NH3plus NH4+),mequiv B.V. = geometric bed volume of ion exchange resin bed C N H ~ O H= final NHzOH concentration with recycle Co = final NHzOH concentration without recycle C, = total exchange capacity of ion-exchange resin H = total hydroxylamine (NH20Hplus NH30H+),mequiv H,= capacity of ion-exchange resin for hydroxylamine Kb = basic dissociation constant P = polymeric substrate R = fractional recovery (of hydroxylamine base) SH= selectivity of ion-exchangeresin for NH,OH+ vs. NH4+ S = selectivity of ion-exchange resin for H+ vs. NH4+ ( = volume of basic product collected Vf = volume of basic material fed Literature Cited Baker, P. J., Jr. "Encyclopedia of Chemical Technology", 2nd ed.;Kirk-0thmer, Ed.; Wliey: New York, 1968; Vol. 11, pp 493-508. Heifferich, F. "Ion Exchange"; McGraw-Hill: New York, 1962; pp 125-151, 431-434. Lange, N. A. "Handbook of Chemistry", 10th ed.;McGraw-Hill: New Ywk. 1961, pp 1202-3. Sifnlades, S.; Tunick, A. A.; Koff, F. W. I d . Eng. Chem. process D e s . D e v . 1982. followina article in this issue. Thompson, W. W ;: Golding, D. R. V. US. Patent 3508864, 1970. Tunlck, A. A.; Koff, F. W.; Sifnlades, S. US. Patent 4 166842, 1979. Wheaton, R. M.; Bauman, W. C. Ind. Eng. Chem. 1953, 45, 228. Wheelwrlght, E. J. Ind. Eng. Chem. Process D e s . Dev. 1977, 76, 220.

Received f o r review September 2 2 , 1980 Accepted September 8, 1981