3060 J. Org. Chem., Vol. 42, No. 18, 1977 Burfield, Lee, and Smithers

Desiccant Efficiency in Solvent Drying. A Reappraisal by Application of a Novel Method for Solvent Water Assay

David R. Burfield,* Kim-Her Lee, and Roger H. Smithers* Department of Chemistry, University of Malaya, Kuala Lumpur 22-11, West Malaysia

Received January 19,1977

The chemical literature, very inconsistent on the subject of the drying of solvents, abounds with contradictory statements as to the efficiency of even the more common desiccants. The recent advent of a novel, highly sensitive method which utilizes a tritiated water tracer for the assay of solvent water content has enabled the first compre- hensive study to be made of the efficiency of various desiccants which pertains unambiguously to solvents. Ben- zene, 1,4-dioxane, and acetonitrile, chosen as model solvents, were wetted with known amounts of tritiated water and treated with a spectrum of desiccants, and the residual water contents were then assayed. The results range from the expected to the highly surprising. Some anomalous results, obtained for benzene and acetonitrile with acidic and basic desiccants, respectively, are discussed in terms of isotopic exchange reactions.

The bench chemist is often confronted by the problem of the selection of desiccants for solvent drying, and although dry solvents are frequently required for use in both prepara- tive situations and in physicochemical studies, there is a paucity of real information in the literature. Some authors2 are content to dismiss drying with statements such as Fre- quently a liquid can be freed from water by shaking with a drying agent such as anhydrous calcium chloride or phos- phorus pentoxide. In the field of organic synthesis, the sit- uation is little better; different reference texts are replete with bewildering contradictions. Thus, magnesium sulfate, de- scribed as either neutral3a.b~~ or acidic,3c,e is alternately an excellent drying agent, rapid in its a c t i ~ n , ~ a , ~ , d , ~ or is ~ 1 0 ~ , 3 h removing only small amounts of water.4 Aluminum oxide is recommended mainly for use in desiccators,3f or as being preferred by many workers for ultimate solvent or reagent drying.Jp Calcium chloride is f a ~ t , ~ ~ - ~ * ~ or alternately not rapid sf in its action, and in any case, the consensus appears to be that calcium sulfate is to be preferred as a and a more e f f i ~ i e n t ~ ~ . ~ desiccant, even though the only existing quantitative comparison for solvents4 shows the complete reverse to be true. Metallic sodium, generally agreed upon3b.d as being efficient, but slow in its drying action, is ridiculed as a desiccant by Ple~ch,~E: who states that the widespread use of sodium as a drying agent by organic chemists is more a ritual than an effective process. Furthermore, there is no doubt that many literature prescriptions for desiccation rely, a t least to some extent, on the chemical intuition of the author, inspired perhaps by the existence of ubiquitous indices of siccative e f f i c i e n ~ y . ~ ~ ~ ? ~ ? ~ ~ , ~ These are usually based on the results of detailed studies of the comparative drying efficiency of desiccants which have been made with regard to the dryness of gases ,3h,6 and direct extrapolation to the condensed phase often gives misleading if not totally erroneous information. For example, phosphorus pentoxide, long considered the ultimate drying standard? is actually quite mediocre in cer- tain situations (vide infra). In summary, no comprehensive study of solvent drying comparable to that made for gases appears to exist, and since the efficiency of a desiccant is de- pendent on the nature of the solvent, this is a serious omis- sion.

R e ~ e n t l y , ~ an extremely sensitive method using a tritiated water tracer for the determination of solvent water content has been developed, where essentially drying efficiency is determined by addition of a specified amount of tritium- labeled water to a rigorously dried solvent and subsequent determination of the decrease in activity of the solvent after treatment with various drying agents. With the limitation that, owing to the problem of isotopic exchange, the method is not applicable to protic solvents, it provides a rapid and

extremely precise assay of solvent water content. This has prompted us to undertake a comprehensive study of the ef- ficiency of drying of a number of desiccants for the solvents benzene, l,4-dioxane, and acetonitrile, representative of a spectrum of others commonly used in the laboratory. Thus, while benzene is a model for a useful range of aromatic and hydrocarbon solvents, and dioxane exemplifies commonly used ethers and bisethers, acetonitrile probably parallels the solvent behavior of a number of other polar, and, on account of its very high dielectric constant, perhaps dipolar aprotic solvents. Although selection of drying agents was generally made on the basis of common usage, some more esoteric ex- amples which have been recommended for use in particular situations were also examined. The results have enabled us not only to present a sensible evaluation of many time-hon- ored solvent drying recipes, but also to advocate the use of novel agents in previously unfamiliar situations.

Results and Discussion Drying of Benzene. Static Drying. Benzene, despite its

carcinogenic properties, is a widely used solvent, which, be- cause of its ease of purification and relative inertness in many chemical systems, has been adopted as a secondary standard. Benzene has a zero dipole moment and on account of its low polarity has little affinity for water, the maximum solubility of water in this solvent being 0.063% wfw at 25 C. Conse- quently, benzene is a relatively easy solvent to dry. Drying has been accomplished in the literatures with the following des- iccants: phosphorus pentoxide, metallic calcium, sodium wire, calcium hydride, and molecular sieves.

In this study benzene containing 100 ppm of water was dried with a selection of the more useful and efficient desic- cants. The results, summarized in Table I, apply to 5% w/v desiccant loadings and to static drying conditions at ambient temperatures (25-29 C). Treatment with molecular sieves, alumina, silica gel, calcium hydride, and lithium aluminum hydride gave super-dryg solvents within 1 day. Alumina in particular was found to be an excellent desiccant for benzene, reducing the solvent water content below 0.01 ppm over this period. These findings thus corroborate earlier conclusionslO that alumina is a particularly effective drying agent for hy- drocarbons, previously exemplified by a-methylstyrene and &pinene. The apparent increase in water content after drying for 7 days is most likely due to exchange or equilibria processes which occur between trace amounts of adventitious mois- ture-released from the surface of the glass vessel or perhaps gaining entry via diffusion through the clearfit stopper seal- and labeled water adsorbed on the desiccant. In any case, it is probably unrealistic to attempt to maintain water contents of below 0.01 ppm outside of totally sealed systems.

Dow

nloa

ded

by U

NIV

OF

PEN

N o

n A

ugus

t 4, 2

009

Publ

ished

on

May

1, 2

002

on h

ttp://

pubs

.acs

.org

| doi:

10.10

21/jo

0043

8a024

marisaHighlight

s-mehriRectangle

s-mehriHighlight

s-mehriHighlight

s-mehriHighlight

Desiccant Efficiency in Solvent Drying J. Org. Chem., Vol. 42, No. 18, 1977 3061

Table I. Efficiency of Various Desiccants for Static Drying of BenzeneC - Residual solvent water content, ppm

Registry no. Desiccant 6 h 1 day 7 days

1344-28- 1

7789-78-8 16853-85-3 7440-23-5 1314-56-3

10043-52-4 1151-82-6

4 A molecular sieves A1203 Silica gel CaH2 LiAlHI Na pzos CaClz Na2S04

2 0.03 0.6 0.006 0.3 0.3 0.2 3 2 (0.03)" 1.5 2 ( 2 ) O 7 12

12 0.1 >28 >28

0.06 0.2 0.1 0.03 0.7

4 ( 4 ) b >28b

>28

Scintillation solution purged with nitrogen and recounted. Distilled sample. Desiccant loading 5% w/v; initial water content 100 ppm (0.01% w/w).

Table 11. Effect of Stirring on the Drying Efficiencies of Desiccants for Benzeneb

Residual solvent water content, ppm -- . -- 6 h 1 day ~~ Desiccant Static Stirred Static Stirred

CaC12 12 0.8 0.1 1 LiAlH4 3 0.7 1.6 0.3 (0.02)"

a Purged with nitrogen and reassayed. Desiccant loading 5% w/v; initial water content 100 ppm (0.01% w/w).

Calcium and lithium aluminum hydrides are both effective desiccants. The values for the complex metal hydride are apparently high through contamination of the solvent with labeled hydrogen resulting from interaction of the hydride with the labeled water. This was confirmed by recounting the sample after purging with nitrogen whereupon the apparent water content was reduced dramatically from 2 to 0.03 ppm. Interestingly, purging had little or no effect on samples dried with calcium hydride and sodium, and this parallels a quali- tative observation that lithium aluminum hydride, perhaps because of its finely divided form, appears to bind the hy- drogen, viz., gas bubbles can still be released from the desic- cant long after the solvent is essentially dry.

Sodium is observed to reduce the water content extensively within the first 6 h , but subsequently the apparent water content is seen to increase significantly. Since purging with nitrogen and distilltition do not reduce the figure it may be speculated that sodium is actually able to metalate benzene, necessarily a t an extremely low rate. Tritiation could then occur by reaction of organosodium intermediates with trace amounts of newly formed tritiated water, whose genesis would be identical with that proposed above.

Phosphorus pentoxide appears to be an ineffective drying agent. However, this conclusion must be tempered by the significant increase in apparent water content with time of drying. An increase of such magnitude can only reasonably be explained by the presence of exchange reactions. Indeed, phosphoric acid catalyzed exchange reactions have been used elsewherell for the synthesis of tritiated aromatic compounds. The presence of exchange reactions thus unfortunately pre- cludes any conclusion as to the efficiency of phosphorus pentoxide as a desiccant for benzene.

Calcium chloride is seen to be an effective drying agent, quite capable of giving super-dry benzene. In contrast, sodium sulfate is completely inept, and the samples obtained after drying were too active for direct counting, indicating little or no drying.

Effect of Stirring. The effect of stirring on rapidity of drying was investigated for calcium chloride and lithium aluminum hydride (Table 11). In both cases stirring has an

\ 3

4 8 I2 16 2 0

Durotion d drying (hr)

Figure 1. Drying of dioxane with various desiccants. Experimental conditions as for Table 111. 1, MgS04; 2, KOH pellets; 3, P205; 4, CaC12; 5 ,4 8, molecular sieves; 6, CaH2.

accelerating effect on drying. This is most likely due to breakdown of particle size which increases the effective des- iccant surface rather than diffusion control of the drying process, since finely divided silica gel is a very rapid desiccant even under static drying conditions (Table I).

Drying by Distillation. Fractionation of benzene with retention of the middle fraction, a time-honored process, has frequently been advocated as a method of drying. In this work it was found that the middle fraction, after discarding the first 20%, contained 15 ppm of water. This is significantly drier than the initial water content, but the drying pales in com- parison with static drying by the majority of desiccants (Table I).

Drying of Dioxane. Static Drying. Dioxane, although not a very polar solvent ( p = 0.45), is completely miscible with water and is consequently far more difficult to dry than ben- zene. Drying is frequently tackled in at least two stages. Pre- liminary drying agent@ include potassium hydroxide, calcium chloride, sodium hydroxide, and magnesium sulfate, whereas final drying8 has been accomplished almost exclusively with sodium and occasionally with sodium-lead alloy.

In this study dioxane with an initial water content of 2300 ppm (0.23% w/w)12 was dried with a selection of both pre- liminary and final drying agents. The initial rate of drying for a selection of desiccants is displayed in Figure 1. It is imme- diately apparent that magnesium sulfate is almost completely ineffective as a drying agent, whereas calcium hydride is both rapid and efficient. It is also interesting that, for the first 24 h at least, the speed of drying parallels desiccant efficiency. Remarkably, phosphorus pentoxide does not excel as a drying

Dow

nloa

ded

by U

NIV

OF

PEN

N o

n A

ugus

t 4, 2

009

Publ

ished

on

May

1, 2

002

on h

ttp://

pubs

.acs

.org

| doi:

10.10

21/jo

00438a

024

3062 J . Org. Chem,., Vol. 42, No. 18, 1977 Burfield, Lee, and Smithers

Table 111. Efficiency of Various Desiccants for Static Drying of DioxaneC

Residual solvent water content, ppm Registry no. Desiccant 6 h 1 day 7 days

Na Na-K alloy CaH2 4 A molecular sieve

CaC12 p205 LiAlH4 KOH (pellets) Silica gel A1203

1118-18-9 Cas04 Na2S04

1481-88-9 MgS04

1310-58-3 KOH (ground)

132 (22) 50

200 204 450

1050 1200 1300 1300

20 22 ( 1 l ) O 30 40

150 300 400 900

1100 1300 1700 1600 1900 2200

6 22 (6)b 23 26 14 (8Ib

300 60

1100 (340)b 200

1200 1400 1500 1800 2200

a After purging to remove H2. b After distillation from desiccant. Desiccant loading 5% w/v; initial water content 2300 ppm (0.23% w/w).

agent, being surpassed in both speed and efficiency by even calcium chloride.

Results for a wider range of desiccants and drying times are summarized in Table 111. Sodium, sodium-potassium alloy (containing 80% by weight of potassium), calcium hydride, and molecular sieves are all seen to be very effective drying agents, although at these initial water loadings none of them dry di- oxane to super-dry levels, which, however, could undoubtedly be achieved by repetitive drying. Sodium-potassium alloy, typically described as possessing a higher drying intensity than metallic s0dium,3~ has been advocated as a siccative in situations where extrcbme desiccation is required,13 where its principal advantage over sodium, i.e., its liquid state at am- bient temperatures, should expedite very efficient drying of solvents boiling below the melting point of the alkali metal. I t is somewhat remarkable therefore that the alloy is not su- perior to sodium granules under static drying conditions. Powdering of KOH pellets is seen to have a dramatic effect on the rapidity and efficiency of drying, and supports a pre- vious report3c that drying of diethyl ether and THF solely by treatment with powdered KOH gives material which is im- mediately suitable for the preparation of reactive organo- metallics. I t is striking that powdered KOH, although slightly slower in action, actually surpasses calcium hydride and mo- lecular sieves in ultimate efficiency.

LiAlH4, though described as highly effective for drying ethers,3h actually appears strangely ineffective. The very high residual water contents cannot be explained in terms of la- beled hydrogen contamination, and it is difficult to invoke any other interferences. A probable conclusion is that unlike CaH2, LiAlH4 is not effective under conditions of high initial water content, and this result, must cast some doubt on its unqual- ified recommendation as a desiccant for THF.I4

The almost complete ineffectiveness of alumina and silica gel for dioxane drying is a complete reversal of their behavior with benzene. This underscores the risks involved in extrap- olating the results of gas drying to the liquid phase.

Magnesium and sodium sulfates are again4 found to be slow and ineffective at these water concentrations. Sodium sulfate, in particular, has earlier4 been shown to be an almost com- pletely inept desiccant for ethers a t much higher water con- centration, viz., the water content of diethyl ether was reduced from 2.07% w/w to merely 1.83% w/w after a period of several weeks. The value of sodium sulfate, even as a preliminary drying agent, must therefore be questionable. In view of its unanimous recommendation, calcium sulfate also rates sur- prisingly poorly.

Effect of Stirring and Refluxing on Drying. Since sol-

Table IV. Effect of Conditions on the Drying of Dioxane Residual solvent water content,

Drying time, ppm Desiccant h Static Refluxed Stirred

CaH2 2 200 110 6 50 29

24 30 14 4 168 23 2

Na 24 20 5 48 3

0 Desiccant loading 5% w/v; initial water content 2300 ppm (0.23% w/w).

Table V. The Use of Visual Indicators in Dioxane Drying Residual solvent

water content Desiccant- after distillation, indicator PPm

Sodium-benzophenone 20 BuLi b-triphenylmethane 22 BuLi-phenanthroline 17 Trityl fluoroborate 650 (800)O

2300 ppm (0.23% w/w). b Registry no., 109-72-8. As determined by the near IR method. Initial water content

Registry no., 341-02-6.

vents are frequently dried by refluxing over desiccants such as sodium and calcium hydride, the effect of refluxing was briefly investigated. I t can be seen (Table IV) that while re- fluxing dioxane over CaHz results in moderate increases in efficiency and speed of drying, stirring is seen to be much more effective.

Refluxing over sodium (Table IV) is seen to give an im- provement in efficiency compared to static drying but pro- longed refluxing is not particularly beneficial.15 I t is worthy of note that stirring over calcium hydride a t ambient tem- peratures is as effective as refluxing over molten sodium.

Drying Agents with Visual Indication. Although desic- cation prescriptions which include a visual indication of sol- vent dryness have become fairly common in recent years, no quantitative measure of their efficiency appears to have been made. These methods generally involve the in situ generation of small amounts of colored, highly moisture-sensitive inter- mediates, often by the action of the desiccant on an added indicator, and the solvent presumed anhydrous when the indicator color persists. Table V, which displays a selection

Dow

nloa

ded

by U

NIV

OF

PEN

N o

n A

ugus

t 4, 2

009

Publ

ished

on

May

1, 2

002

on h

ttp://

pubs

.acs

.org

| doi:

10.10

21/jo

0043

8a024

Desiccant Efficiency in Solvent Drying J . Org. Chem., Vol. 42, No. 18, 1977 3063

Table VI. Efficiency of Various Desiccants for Static Drying of Acetonitrile

Residual solvent water content, ppm Registry no. Desiccant 1 day 7 days

1303-86-2 584-08-7

pzo5 3 A molecular sieves Bz03 K COS 4 X molecular sieves CaC12 Silica gel A1203 CaH2 KOH (powdered) KOH (pellets) Cas04 Ph3C+-BF4

9 ( 1 2 ) Q . b 49 59

250 450

1200 1300 1600 1900 2200sb2d 2500 2500 2700 (2800)7c

5 27

1300 500

2200 1300 1700 1900 (1300),c

1300 2200

Distilled sample. Colored residue. By near IR method. Strong amine smell in distillate. e Desiccant loading 5% w/v; initial water content 2800 ppm (0.28% w/w).

of those investigated for dioxane, reveals that although none of them give super-dry solvent (see, however, discussion below), the first three entries give comparable results to the best of those obtained for dioxane after static drying for 1 day (Table 111). The intense blue sodium ketyl of benzophenone (entry l ) , often used in the preparation of absolute diethyl ether,6 where it presumably also serves to remove peroxides, gives similar results to butyllithium. The appearance of the red triphenylmethyl anion (entry 2) has been advocated as an indicator in the preparation of anhydrous THF.17 We have found that if no special precautions, e.g., anaerobic conditions, are utilized, then the amount of butyllithium required to impart a persistent color to the solvent is excessively high, owing perhaps to corisumption of the indicator by molecules other than water, e.g., oxygen. This shortcoming led to an experiment with 1, LO-phenanthroline (entry 3), previously suggested as an indicator in the alcohol method of assaying BuLi.l8 The formation of the derived rust-red complex re- quired only about half the butyllithium used in entry 2, and since the result was a slightly drier solvent, the use of this in- dicator is to be preferred. In the general context of the bu- tyllithium experimeats, i t must also be pointed out that i t is known that alkyllithiums react relatively slowly with THF,lg to give, initially, 2-lithiotetrahydrofuran, which, if the anal- ogous reaction were to occur with dioxane, may serve to label the ether, and hence raise the apparent water content, by reaction of metalated dioxane with tritiated water. While this reaction has not been reported for dioxane, and in any case would be expected to be extremely slow compared to reaction of the alkyllithium with water, some inflation of the apparent water content by this means cannot be altogether ruled out. Although trityl fluoroborate (entry 4) has not been previously used as a desiccant for ethers, it has been used to dry aceto- nitrile (vide infra), amd this experiment was run to determine its efficiency in a different solvent type. Even though, a t this solvent water concentration, the deep yellow color of the salt solution was not discharged, the recovered ether contained a surprisingly large amount of residual water, and this result was cross-checked by the near IR method. Compared to entries 1 to 3, trityl fluoroborate gives poor ultimate drying, which however, is still better than that obtainable with lithium aluminum hydride (Table 111).

Drying of Acetonitrile. Static Drying. Acetonitrile, a polar aprotic solvent ( f i = 3.44) of high solvating power and favorable physical ]properties, has been widely used as a sol- vent both in the study of chemical reactions and for physical measurements involving spectrophotometric and electro- chemical techniques. However, because of its high affinity for water it is an outstandingly difficult solvent to completely dry.

Drying is conventionally accomplished8 by treatment with preliminary drying agents such as anhydrous sodium or po- tassium carbonate, anhydrous calcium chloride, silica gel, or 3 A molecular sieves, and final drying with calcium hydride, phosphorus pentoxide, or more recently with trityl fluo- roborate.20

The results of static drying with a range of desiccants are displayed in Table VI. In contradistinction to the other sol- vents investigated phosphorus pentoxide is seen to excel in its drying efficiency, but even so super-dry acetonitrile is not obtained. I t is interesting to note that the residual water content is of similar order of magnitude to an earlier result which also utilized P Z O ~ . ~ ~ The only disadvantage to phos- phorus pentoxide drying is the partial loss of solvent through polymerization, and possible contamination by desiccant residues.22 Reasonably effective drying can also be achieved with 3 8, molecular sieves which reduce the water content to less than 30 ppm after 1 week. The relative inefficiency of 4 8, sieves emphasizes the need for careful selection of sieve pore size for effective drying. A hitherto little mentioned3f but useful desiccant is boric anhydride. Direct sampling proved impossible in this case since soluble desiccant residues in- terfered visually with the scintillant, but the sample distilled after 1 day stirring with the anhydride had a water content of 59 ppm. This reagent is advantageous compared to phos- phorus pentoxide since it does not induce polymerization of the solvent nor does it appear to be significantly volatile. It also offers advantages in its ease of handling and disposal.

Silica gel and alumina are again, as with dioxane, largely ineffective. This may reflect their rather low capacity for ef- fective drying at high water ~ontents ,~g but in any case makes them an unlikely choice for preliminary drying. Calcium sul- fate, although generally strongly recommended for efficient drying, is seen to be the least effective of the examined des- iccants, and is clearly surpassed by the underrated calcium chloride.23

The ineffectiveness of the previously excellent desiccants potassium hydroxide and calcium hydride seems anomalous. Careful examination of the results reveals that all the basic siccatives, potassium hydroxide, calcium hydride, and po- tassium carbonate, give apparently little drying. In addition the apparent water content in the presence of the weakly basic potassium carbonate increases very significantly from 251 to 1300 ppm over the course of 1 week. These observations ap- pear indicative of a base-catalysed exchange reaction, viz.

-OH CH3CN + T20 CHzTCN + HOT

Such base-catalyzed exchange reactions of the 1y hydrogens

Dow

nloa

ded

by U

NIV

OF

PEN

N o

n A

ugus

t 4, 2

009

Publ

ished

on

May

1, 2

002

on h

ttp://

pubs

.acs

.org

| doi:

10.10

21/jo

0043

8a024

3064 J . Org. Chem., Vol. 42, No. 18, 1977 Burfield, Lee, and Smithers

have been previous1.y encountered with b-hydroxypropioni- trile24 but rather surprisingly it has been claimed25 that ace- tonitrile itself does not exhibit similar behavior. In an attempt to confirm the presence of exchange reactions, the tracer ex- periment was cross-checked by the near IR method for the calcium hydride case. The near IR value of the water content is significantly lower, and this is suggestive of the presence of exchange reactions. Most unexpectedly, this determination also revealed that calcium hydride is largely ineffective for drying acetonitrile. 'This remarkable observation, together with the resuits for phosphorus pentoxide drying, undermines the intuitive assumption that the relative efficiencies of des- iccants should, barring chemical incompatibility, be inde- pendent of solvent type.

The interference, by exchange reactions, unfortunately makes it impossible to draw many conclusions on the effi- ciency of the basic desiccants save that potassium carbonate is clearly a reasonable desiccant, at least for preliminary drying.

I t is worthy of mention that drying with finely powdered potassium hydroxide gave rise to a colored residue and a sig- nificant amine content in the distilled fractionsz6

Trityl Fluoroborate. The use of this stable, orange car- benium ion salt as a desiccant for acetonitrile would seem to be advantageous; it can be stored in a desiccator for extended periods without deco~npos i t ion ,~~ and is used as a siccativez0 simply by adding it in small batches to the wet nitrile until a strong yellow color persists, thus furnishing a visual indication of dryness. The results obtained by using this and the IR method are displayed in Table VI, and indicate that, at these water concentrations, the carbocation salt is a spectacularly ineffective desiccant. The reason for this impotence seems obscure, although acetonitrile, by virtue of its solvation ability, has a well-known moderating influence on the stability of carbenium ions,28 and indeed, the drying of dioxane by trityl fluoroborate (vide supra) is significantly better than the present solvent. Whatever the true reason, it is clear that, as a desiccant for acetonitrile, the salt is completely worthless.

Merits of t he Study. The present study should be of con- siderable heuristic value, particularly to the bench chemist in the provision of directly relevant data, and it is also worthwhile briefly emphasising again' that for reasons which include (1) contamination of the solvent by desiccant residues and possibly labeled hydrogen, (2) exchange reactions, and (3) the kinetic isotope effect, the apparent water contents, as reported above, will always represent the upper limits of the true content. Of course, this in no way detracts from the value of the work, and, to cite an example, merely means that it is entirely possible that alumina is able to dry benzene to below 6 X lo--" ppm!

Experimental Section Radioactive samples were assayed in a scintillation solution con-

taining 0.4 g of 1,4-bis(5-phenyloxazol-2-yl)benzene (POPOP) and 4.0 g of 2,5-diphenyloxazole (PPO) per liter of toluene with a Beckman Model IJS-lOO liquid scintillation spectrometer. Determination of water content by the near IR method was performed using a Unicam SP700 spec t roph~tometer .~~ Tritium-labeled water was purchased from the Radiochemical Centre, Amersham, England, a t an initial activity of 5 Ci/mL and was diluted with appropriate quantities of inactive water.

Desiccants. Lithium aluminum hydride and phosphorus pentoxide were used as supplied; calcium hydride (99.5%) and reagent grade potassium hydroxide were rapidly powdered immediately prior to use in a mortar and a mechanical blender, respectively. Chromatographic grades of neutral alumina (activity 1) and silica gel, as well as calcium, magnesium, and sodium sulfates, calcium chloride, potassium car- bonate, and 3 and 4 8, molecular sieves were activated for 15 h at 300-320 OC before use. Since hydration occurs rapidly on cooling of these desiccants in moist air, cooling was carried out in a phosphorus pentoxide desiccator, and the samples then used immediately. Sodium

metal, whose oxide crust had previously been removed by melting under xylene, was cut into 2-mm cubes under dry petroleum ether. Sodium-potassium alloy was prepared as detailed elsewhere308 from oxide-free metals. (It is worth noting that the fire hazard associated with destroying excess alloy is completely avoided if the disposal is carried out in two steps. Addition of a little dry ethyl acetate to the alloy in dioxane smoothly consumes potassium-presumably via an acyloin reaction. Unreacted sodium can then be destroyed conven- tionally using ethanol.) Trityl fluoroborateZ7 and boric anhydride3f were respectively prepared from triphenylcarbinol and tetrafluo- roboric acid, and by high-temperature (900 "C) dehydration of boric acid.

Solvents. Benzene. AR grade reagent was stirred for 24 h with finely ground calcium hydride, refluxed, carefully fractionated (bp 80.0 "C), and stored over 4 8, molecular sieves.

1,4-Dioxane. Commercial 1,4-dioxane was purified and dried ac- cording to a method cited by F i e ~ e r , ~ * ~ whereby the glycol acetal impurity is removed by hydrolysis to acetaldehyde, which is itself voided by purging with nitrogen gas. Preliminary drying with potas- sium hydroxide pellets followed by fractionation (bp 101-101.5 "C) from sodium gave material which was stored in a dark bottle over 4 A molecular sieves.

Acetonitrile. Following well-documented procedures: reagent grade material, after being given a preliminary drying with potassium carbonate (24 h), was decanted on to phosphorus pentoxide and stirred at reflux for 2 h. Fractionation gave material of bp 81.5 "C, which was not stored, but used immediately.

Techniques. The procedure used for benzene serves as an example. A stock solution of benzene containing 100 ppm of labeled water was prepared by the addition of 18 WL of tritiated water, specific activity 40 mCi/mL, to 180 g of purified benzene; homogenization was ac- complished by stirring overnight. Aliquots of the stock solution (15.0 f 0.1 mL) were syringed directly onto 0.75 f 0.03 g of activated des- iccant contained in a 25-mL clear-fit round-bottom flask, which was then stoppered. Where appropriate samples were stirred magnetically. Samples (1.00 f 0.02 mL) were taken a t time intervals specified in the text-care was taken to avoid disturbing the desiccant-and syringed directly into the counting vials. Where possible, samples were distilled from the desiccant so as to provide a cross-check against contamination of the solvent by labeled desiccant residues. Samples were accumulated and assayed batchwise.

Similar procedures were used with dioxane and acetonitrile, except that higher water contents were examined and tritiated water of low specific activity (0.5 mCi/mL) was employed.

Registry No.-Benzene, 71-43-2; dioxane, 123-91-1; acetonitrile, 75-05-8.

References and Notes Abstracted in part from the Honours Project Work of K. H. Lee, 1975- 1976. F. Daniels, R . A. Alberty, J. W. Williams, C. D. Cornwell, P. Bender, and J. E. Harriman. "Experimental Physical Chemistry" 7th ed, McGraw-Hill, New York, N.Y., 1970, p 641. (a) T. L. Jacobs, W. E. Truce, and G. R. Robertson, "Laboratory Practice of Organic Chemistry", Macmillan, New York, N.Y., 1974; (b) R. M. Roberts, J. C. Gilbert, L. B. Rodewald, and A. S. Wingrove. "An Introduction to Modern Experimental Organic Chemistry", 2nd ed, Holt, Rinehart and Winston, New York, N.Y., 1974; (c) L. Brandsma, "Preparative Acetylenic Chemistry", Elsevier, Amsterdam, 1971; (d) R. S. Monson, "Advanced Organic Synthesis", Academic Press, New Ywk, N.Y., 1971; (e)L. F. Fieser and M. Fieser, "Reagents for Organic Synthesis", Wiley, New York, N.Y., 1967; ( f ) A. I . Vogei, "A Text-Book of Practical Organic Chemistry", 3rd ed, Longmans, London, 1964; (g) P. H. Plesch, Ed., "The Chemistry of Cationic Polymerisation", Pergamon Press, Oxford, 1963, p 682 f f ; (h) "Drying in the Laboratory", E. Merck, Darmstadt. B. D. Pearson, and J. E. Ollerenshaw, Chem. Ind., (London), 370 (1966). (a) G. Brauer, Ed.. "Handbook of Preparative inorganic Chemistry", Vol. I, 2nd ed, translated by Scripta Technica Inc., Academic Press, New York. N.Y., 1963, p 80; (b) "Handbook of Chemistry and Physics", 53rd ed, Chemical Rubber Publishing Co., Cleveland, Ohio, 1972, p E35. (a) F. Trusell and H. Diehl. Anal. Chem., 35, 674 (1963); (b) J. H . Bower, J. Res. Natl. Bur. Stand., 12, 241 (1934). D. R. Burfield, Anal. Chem., 48, 2285 (1976). See references contained in J. A. Riddlck and W. B. Bunger, "Organic Solvents", 3rd ed, Wiley-lnterscience, New York, N.Y., 1970. The term superdry has been used earlier in a qualitative sense,3f but in this context it denotes solvents containing less than 1 ppm of water. T. H. Bates, J. V. F. Best, and T. F. Williams, Nature (London). 88, 469 (1960). See E. A. Evans, "Tritium and Its Compounds", 2nd ed. Butterworths, London, 1974. This is close to the maximum water content specified for AR grade diox- ane. For example, after standing for 5 days over the alloy, both diethyl- and di-

Dow

nloa

ded

by U

NIV

OF

PEN

N o

n A

ugus

t 4, 2

009

Publ

ished

on

May

1, 2

002

on h

ttp://

pubs

.acs

.org

| doi:

10.10

21/jo0

0438a0

24

Perhydrogenation of 2,8-Diaminopurine J . Org. Chem., Vol. 42, No. 18,1977 3065

methylamine are dehydrated sufficiently that spin coupling of the adjacent N-H and C-H protons is observable in their NMR spectra: K. L. Henold, Chem. Commun., 1340 (1970).

(14) H. Baumgarten, Ed., "Organic Syntheses", Collect. Vol. V, Wiley, New York. N.Y., 1973, p 976.

(15) In a peculiar sense, this result confirms an earlier statement that "One day drying over calcium hydride is found to be as effective as six months re- fluxing over sodium". See A. S. Brown, J. Chem. Phys.. 19, 1226 (1951).

(16) See, for example, G. Kobrich. W. E. Breckhoff, H. Heinemann. and A . Akhtar, J. Organornet. Chem., 3, 492 (1965).

(17) W. R. Purdum and G. .I. Bartling, J. Chem. Educ., 52, 120 (1974). (18) S. C. Watson and J. F Estham, J. Organomet. Chem.. g, 165 (1967). (19) (a) I. Fleming and T. Mah, J. Chem. Soc., Perkin Trans. 7, 964 (1975); (b)

R . B. Bates, L. M . Kroposki, and D. E. Potter, J. Org. Chem., 37, 560 (1972).

(20) Y. Pocker and W . H. VVong, J. Am. Chem. Soc., 97,7097 (1975). (21) J. F. Coetzee and G. R. Padmanabhan, J. Phys. Chem.. 66, 1708

(1962).

(22) Subsequent distillation from potassium carbonate has been recommended for removal of phosphorus pentoxide residues. See P. Walden and E. J. Birr, Z. Phys. Chem. (Leipzig), 144A, 269 (1929).

(23) Drying over calcium chloride for "at least a week" has been advocated as a method for obtaining "thoroughly dry" acetonitrile. A. H. Blatt. Ed., "Organic Syntheses", Collect. Voi. i , 2nd ed, Wiley, New Ywk. N.Y., 1946, pp 5-6.

(24) A. Lapidot. J . Reuben, and D. Samuel, J. Chem. Educ., 41, 570 (1964). (25) See ref 11, p 759. (26) These observations appear to support the earlier contention that "pre-

treatment with potassium hydroxide does more harm than good". See J . F. Coetzee, Pure Appl. Chem., 13, 429 (1966).

127) H. J. Dauben Jr., L. R. Honnen, and K . M. Harmon, J. Org. Chem., 25, 1442 (1960).

Wiley, New York, N.Y., 1968, p 140. (28) N. Lichtin in "Carbonium Ions". Vol. I , G. A. Olah and P. v. R. Schleyer, Ed.,

(29) R. L. Meeker, F. Critchfield, and E. T. Bishop, Anal. Chem., 34, 1510 (1962).

(30) (a) See ref 3e, p 1102. (b) See ref 3e, p 333.

Perhydrogenation of 2,8-Diaminopurine

Mark M. Wegner and Henry Rapoport*

Department o f Chemistry, University of California, Berkely, California 94720

Received March 7, 1977

2,8-Diaminopurine (7) can be hydrogenated over Pt02 in acidic medium to give 2-imino-4-guanidinomethyl-5- imidazolidinone (1 1) which can itself be further hydrogenated to 2-imino-4-guanidinomethylimidazolidine (12). The structures of 11 and 12 were proven by unambiguous synthesis. 2,8-Diamino-6-methylpurine (37) also can be hydrogenated in a similar manner to two analogous compounds, as isomeric mixtures, whose structures are inferred by comparison with 11 and 12. A superior method has been developed for synthesizing the diaminopurines 7 and 37, involving the condensation of the appropriate triaminopyrimidine with N-(methylmercaptochloromethy1)-p - toluenesulfonimide (20) followed by ring closure via the carbodiimide and detosylation with HF.

Saxitoxin is one of the most potent naturally occurring neurotoxins. I t is the sole toxin produced by the marine di- noflagellate Gonyauiax catenella and is a minor constituent of the toxins produced by G. tamarensis. Ingestion of these dinoflagellates by several species of normally edible shellfish is frequently responsible for their toxicity to man. X-ray crystallographic analysis of two derivatives, the bis-p-bro- mobenzenesulfonate3 and the ethyl hemiketal dihydrochlo- ride,4 have established structure 1 for crystalline saxitoxin hydrate, and 13C NMR studies have also established this structure for the molecule in s ~ l u t i o n . ~ Recently5 the major toxins of G. tamarensis, gonyautoxins I1 and 111, also existing as the hydrates, were postulated to have the closely related structures 2 and 3, respectively.

Saxitoxin and the gonyautoxins are unique among natural products in that their structures incorporate a tetrahydro- purine moiety composed of two guanidine units fused together in an azaketal linkage which remains intact under ordinary conditions. We were therefore interested in preparing a simple model of the tetrahydropurine backbone of saxitoxin, devoid of the fused ketone bearing ring and the peripheral carbamate, for both chemical and biological investigations. We chose to study the catalytic hydrogenation of 2,8-diaminopurine (7), which conceivably could lead to 2,8-diiminotetrahydropurine (10) or its tautomers, the simplest possible tetrahydropurine model of saxitoxin. We now report the results of our study of

the heterogeneous catalytic hydrogenation of 2,8-diamino- purines.

The literature relating to the catalytic reduction of purines is relatively meager. 1,6-Dihydropurine ( 5 ) has been pre- pared"' from purine and 6-chloropurine (4), and in weak acid 5 was hydrolyzed to 4(5)-aminomethyl-5(4)-aminoimidazole (6). Similarly a tetrahydropurine is claimed8 to result from catalytic reduction of 2,6,8-trichloropurine. More recently? the catalytic reduction of 2&diaminopurine (7) is reported to yield a compound whose structure was assigned as 2- amino-5-guanidino-l,4,5,6-tetrahydro-6-oxopyrimidine (8). These authors also report the preparation of 2,8-diamino- 4,5,6,9-tetrahydro-l,7,9-trimethylpurine by sodium borohy- dride reduction of 2,8-diamino-1,7,9-trimethylpurine, and claim to have electrolytically reduced 7 to 8 plus tetrahydro- purine 10, obtained as an inseparable mixture with another reduction product 9.

In contrast to that report, we have found that 7 is slowly hydrogenated with a Pt02 catalyst in hydrochloric acid (pH 1.5) a t room temperature and 20 psi pressure to give a single product, A, in quantitative yield. A could be further reduced under more drastic conditions (60 "C, 100 h) to give another product, B, also in quantitative yield. The lH NMR spectrum of A.2HC1 consisted of a doublet (2 H, J = 5 Hz) and a triplet (1 H, J = 5 Hz); its 13C NMR spectrum is tabulated in Table I.

These NMR data suggested that A was not a reduced purine with an intact bicyclic ring system but rather the five-mem- bered monocyclic imidazolidinone 11, The 13C NMR ab- sorption at 6 173 is clearly assigned to the amide carbonyl, and the simple doublet-triplet pattern of the lH NMR spectrum implies the freely rotating methylene group of 11. The alter- native six-membered ring structure 8 previously proposedg for the 2,8-diaminopurine reduction product should display

Dow

nloa

ded

by U

NIV

OF

PEN

N o

n A

ugus

t 4, 2

009

Publ

ished

on

May

1, 2

002

on h

ttp://

pubs

.acs

.org

| doi:

10.10

21/jo

0043

8a024

J. Org. Chem. 1982,47, 3821-3824 3821

under Nz for 3.5 h in an oil bath maintained at 100 "C. The reaction mixture was poured onto an excess of ice, and after the ice had melted HzO was added to bring the total volume to 1500 mL. The solution was extracted with CHzClz (5 X 100 mL), and the combined organic layers were extracted with saturated NaHC03 untiI the extracts were colorless. The combined NaHC03 extracts were back-washed with CHzC12 (100 mL) and then acidified to pH 1 with 12 M HCl. The product was extracted into CHzClz (5 X 100 mL), the organic layer was dried (Na2SO4) and filtered, and solvent was removed on a rotary evaporator followed by a vacuum pump to yield 6.3 g of 3-hydroxy-5-methylbenzo- cyclobutenedione in 63% overall yield from dienophile 5: light yellow crys&, mp 191-193 OC (EtOAc); IR (CHzC12) 3540,1789, 1755; 'H NMR (60 MHz, acetone-ds) 6 7.29 (s, 1 H), 6.95 (8, 1 H), 2.46 (s,3 H). Anal. Calcd for CBH603: C, 66.66; H, 3.73. Found C, 66.56; H, 3.90.

Cycloaddition of l-Methoxy-3-(trimethylsiloxy)-1,3-bu- tadiene (13) to 1,4-Dichloro-3,3,4-trifluorocyclobutene (5). (A) Preparation of Enone 15. 1,4-Dichloro-3,3,4-trifluoro- cyclobutene (5; 3.08 g, 17.4 mmol) and Danishefsky's diene (13; 4.50 g, 26.1 mmol) were placed in a 3-02 Fischer-Porter pressure vessel equipped with a magnetic stirring bar, and the mixture was saturated with dry Nz for 5 min. The tube was sealed, placed in an oil bath maintained at 120 "C, and stirred for 3.5 h. After cooling to 25 OC, the tube was opened, and the contents were transferred to a 100-mL round-bottomed flask with the aid of a small amount of MeOH. To this mixture was added 100 mL of 1:l MeOH/1.2 N HC1, and the solution was stirred at room temperature for 2 h. The dark solution was poured into a sep- aratory funnel, diluted with 200 mL of HzO, and extracted with CHzClZ (3 X 75 mL), and the combined organic layers were dried (Na804), filtered, and condensed on a rotary evaporator to yield 3.11 g of crude 14 as a dark oil. The compound is contaminated with diene decomposition products as well as some enone 15, but the following spectroscopic absorptions of 14 are apparent: IR

-1 H), 3.30 (8, -3 H), 2.6-2.7 (m, -4 H). Without purification, crude 14 (3.11 g) was placed in a 250-mL round-bottomed flask and dissoved in dry benzene (175 mL). After addition of p - toluenesulfonic acid (166 mg, 0.87 mmol), the mixture was refluxed under Nz for 27 h, cooled to room temperature, transferred to a separatory funnel, washed with saturated NaHC03 (2 x 25 mL), dried (Na2SO4), filtered, and condensed on a rotary evaporator, and the residue was chromatographed on silica gel (3 cm x 0.75 m; 3 2 hexane/CHzC12) to yield enone 15: 2.25 g (53% yield from dienophile 5); white needles; mp 44-45 OC (sublimed); IR (CHzCIJ 1695; 'H NMR (60 MHz, CDC1,) 6 6.63 (d, J = 10 Hz, 1 H with smaller splittings), 6.15 (d, J = 10 Hz, 1 H with smaller splittings), 3.90-3.13 (m, 1 H), 2.71 (apparent d, J = 4 Hz, 2 H); mass spectrum, m / e (relative intensity) 244 (M'), 246 (M' + 2, 66).

(CH2ClZ) 1720; 'H NMR (60 MHz, CDC13) 6 4.72 (t, J = 5 Hz,

(B) Aromatization of Enone 15 to 2,2-Difluoro-4- hydroxybenzocyclobutenone (16). A solution of enone 15 (2.25 g, 9.2 mmol) in 60 mL of MeOH was cooled to 0 "C in a 250-mL round-bottomed flask and the solution was saturated with dry N2 for 5 min. The flask was fitted with a pressure equalizing addition funnel and a nitrogen inlet tube and the addition funnel was charged with a solution freshly prepared from sodium (0.846 g, 36.8 mmol) and MeOH (30 mL). The NaOMeIMeOH solution was added dropwise with stirring over 30 min, and after the addition was complete, the reaction mixture was allowed to warm to room temperature and stirred an additional 3 h. An equal volume of 1.2 N HCl(90 mL) was added, and the mixture was refluxed for 3 h to hydrolyze the intermediate ketal. After cooling to room temperature, the reaction mixture was diluted with 200 mL of HzO and the product extracted into CHZClz (3 x 75 mL). The combined CHzClz layers were dried (Na2S04), filtered, and evaporated to dryness on a rotary evaporator followed by a vacuum pump to give 16: 1.5 g (96%); white crystals; mp 163-164 "C (EtOAcIhexane); IR (CHZCl2) 3570, 1795, 1770; 'H NMR (60 MHz, acetone-d6) 6 9.8 (br s, 1 H), 7.70-7.16 (m, 3 H); mass spectrum, m l e 170 (M').

(C) Hydrolysis of 16 to 4-Hydroxybenzocyclobutenedione (17). The sample of 16 prepared above (1.50 g, 8.82 mmol) was placed in a 250-mL round-bottomed flask with 1:l concentrated Hfi04/HOAc (90 mL) and stirred at 90 "C under N2 for 3 h. The reaction mixture was poured onto an excess of ice and diluted with HzO to a total volume of 350 mL. The solution was extracted with Et20 (3 X 100 mL), and the combined organic layers were dried (NafiO,), filtered, and condensed to a crude solid on a rotary evaporator. This material was chromatographed on a silica gel column (3 cm X 1 m; EhO) to yield 4-hydroxybenzocyclo- butenedione: 1.10 g (84% yield, 43% overall from dienophile 5); light yellow crystals; mp 174.5-175 "C (EtOAc) (lit.7 mp 167-170 "C); IR (CHC13) 3600-3000 (br), 3575,1808 (sh), 1790,1769,1755 (sh); 'H NMR (60 MHz, CD,CN) 6 7.81 (d, J = 8 Hz, 1 H), 7.3G7.06 (m, 2 H), 6.23 (br s, 1 H); mass spectrum, m / e 148 (M+).

Acknowledgment is made to the National Cancer In- stitute, DHEW (Grant No. CA 26374), for support of this work.

Registry No. 1, 6383-11-5; 3, 3469-06-5; 4, 82431-14-9; 5, 2927- 72-2; 6a, 63383-46-0; 6b, 73912-36-4; 7a (isomer l), 82431-15-0; 7a (isomer 2), 82468-19-7; 7b (isomer l), 82431-19-4; 7b (isomer 2), 82468-20-0; 8a, 82431-16-1; 8b, 82444-39-1; Qb, 82431-20-7; 108, 82431-17-2; lob, 82431-21-8; lla, 82431-18-3; llb, 82431-22-9; 12a,

82431-25-2; 17, 75833-48-6; anthranilic acid, 118-92-3; 2-carboxy- benzenediazonium chloride, 4661-46-5; 1,l-dichloroethylene, 75-35-4; 1,l-dichlorobenzocyclobutene, 68913-13-3; 3-methyl-2-butenal, 107- 86-8; 2-methyl-3-buten-2-01, 115-18-4.

62416-21-1; 12b, 82431-23-0; 13, 59414-23-2; 15, 82431-24-1; 16,

Deperoxidation of Ethers. A Novel Application of Self-Indicating Molecular Sieves

David R. Burfield Department of Chemistry, University of Malaya, Kuala Lumpur 22-11, West Malaysia

Received January 18, 1982

The removal of peroxides from contaminated ethers by treatment with self-indicating molecular sieves (IMS) is proposed as a safe and facile method of ether purification. Quantitative analysis of peroxide content before and after treatment with IMS show that ethers such as THF, diethyl ether, and diisopropyl ether can be readily decontaminated by an ambient-temperature or reflux process. The deperoxidation process is enhanced under nitrogen and has been safely carried out on a bulk scale and with initial peroxide contents as high as 0.5 M. IMS, in common with most other chemical reducing agents used for ether deperoxidation, are, however, ineffective for the decomposition of unreactive species such as dialkyl peroxides.

Aliphatic ethers, with their characteristic solvation abilities, excel as inert reaction media in numerous syn- thetic procedures. However, in practice this usefulness is

0022-326318211947-3821$01.25/0

often tempered by a n unfortunate proclivity to facile air oxidation at ambient temperatures which leads to peroxide formation.' T h e presence of peroxides is not only po-

0 1982 American Chemical Society

Dow

nloa

ded

by U

NIV

OF

PEN

N o

n A

ugus

t 4, 2

009

Publ

ished

on

May

1, 2

002

on h

ttp://

pubs

.acs

.org

| doi:

10.10

21/jo

0014

1a003

marisaHighlight

3822 J. Org. Chem., Vol. 47, No. 20, 1982 Burfield

Table I. Deperoxidation of Tetrahydrofuran with Indicating Molecular Sieves (IMS) at Ambient Temperatures

peroxide content,c mmol /L sieve scale, loading, run mL 5% w/v initial 1 day 2 d a y s 3 days 4 days 7 days 90 days 1 50 10 7.8 0.9 0.1 N.D. 2 a 50 10 7.8 3.3 0.5 0.26 3 24 00 5 1.1 0.22 0.10 0.06 0.014 d 4 100 5 1.1 0.07 0.007 0.018 0.012 d 5 50 5 22 7.8 5.5 6 b 50 5 22 7.1 4.9 7 50 5 1 2 4 2.8 8 50 10 1 7 6 1 4 9 7 50 5 88 20 6 3 d

10 100 10 96 4

a Performed under an atmosphere of air. IMS activated at 300 "C in air oven. As determined by the spectrophoto- metric method. N o t determined.

tentially hazardous,l12 but also frequently undesirable for chemical reasons, and although possible hazards may be avoided by the routine disposal of time limit expired3 or proven peroxide contaminated4 ethers, the need for rig- orously purified solvents as well as economic and logistic' constraints may frequently necessitate peroxide removal. To this end a bewildering arrayla of safe (and not so safe!6) physical and chemical methods have been proposed.

Column chromatographic purification whereby peroxides are removed by adsorption on substrates such as alumina,1 ion-exchange resins'l or 13X molecular sieved2 have var- iously been reported, and the use of activated alumina in particular has been widely e n d o r ~ e d . ~ ~ J ~ However, such methods, though effective and generally applicable, are disadvantaged by the necessity of using substantial amounts14 of nonregenerable and relatively expensive ad-

(1) See, for example, the brief review by N. V. Steere in 'The Chem- istry of the Ether Link", S. Patai, Ed., Interscience, London, 1967.

(2) See for example: (a) A. G. Davies, J. R. Znst. Chem., 80,386 (1956); (b) "Guide for Safety in the Chemical Laboratory", 2nd Ed., Van Nost- rand-Reinhold, New York, 1972, p 302; (c) 'Safety in Academic Chem- istry Laboratories", 3rd ed., American Chemical Society, Washingtan, DC, 1979; (d) L. Bretherick, "Handbook of Reactive Chemical Hazards", 2nd ed., Butterworths, London, 1979.

(3) It has been proposed6 that unused ethers in opened containers should be disposed of within 1 week of opening (uninhibited grades) or 3 or 6 months (inhibited grades). ElsewhereZc a blanket time limit of 1 month has been suggested for ethyl or isopropyl ethers. In one labora- toe it is standard practice to use only THF from unopened bottles and to discard the remainder within 2-3 days.

(4) It has been recommended that diethyl etherJ containing more than 0.005% peroxide (yellow color with KI test) and THF containing 'larger than trace amounts of peroxiden6 should be discarded.

(5) N. V. Steere, J. Chem. Educ., 41, A575 (1964). (6) H. E. Baumgarten, 'Organic Syntheses", Collect. Vol. V, Wiley,

New York, 1973: (a) p 796, (b) p 695. (7) For laboratories situated outside the main industrial centers,

chemical delivery times exceeding 6 months are not uncommon. As such ethers from unopened bottles are frequently significantly contaminated on receipt.

(8) Potassium hydroxide, which in its finely powdered form is a pow- erful desiccant: has been advocated for peroxide removal. However, serious explosions may occur when treating impure THF with solid or concentrated aqueous potassium hydroxide.E Similarly, LiAlH,, a less efficient p g agent: has proved distinctly hazardous in the purification of ethers. (9) D. R. Burfield, K. H. Lee, and R. H. Smithers, J. Org. Chem., 42,

3060 (1977). (10) W. Dasler and C. D. Bauer, Znd. Eng. Chem. Anal. Ed., 18, 52

(1946). (11) R. N. Reinstein, J. Org. Chem., 24, 1172 (1959). (12) N. Rabjohn, "Organic Syntheses", Collect. Vol. IV, Wiley, New

York, 1963, p 475. (13) (a) A. I. Vogel, "A Text-Book of Practical Organic Chemistry", 4th

ed., Longmans, London, 1978; (b) "Purification of Solvents by Adsor- bents Woelm", Woelm Pharma, Eschwege, West Germany; (c) "Drying in the Laboratory", E. Merck, Darmstadt, West Germany; (d) D. D. Perrin, W. L. F. Armarego, and D. R. Perrin, 'Purification of Laboratory Chemicals", Pergamon, Oxford, 1980.

sorbents and the need for the subsequent safe disposal of the contaminated residue upon which the peroxides are adsorbed chemically unchanged.1

Chemical methods encompass a wide spectrum of which effect reductive decomposition of per-

oxides. These include classical redox systems such as sulfites, bisulfites, amines, metal salts (e.g., FeS04, CuCl, Ce(OH)3, SnC12), traditional reducing couples (e.g., Sn/ HCl, Na/EtOH), and the more recently used complex metal hydrides (LiAlH,, NaBH,). Treatment with aque- oud5 ferrous s ~ l f a t e , ~ ~ * ~ J ~ * , ~ J ~ or sodium sulfite/bi- sulfite'wl6 has received approbation as a safe and effective method of deperoxidation, but their use is restricted to water-immiscible ethers,l* and in any case such procedures necessitate an additional desiccation step. For water- miscible ethers such as dioxane, diglyme, and THF, reflux over solid cuprous6b chloride, stannous ~ h l o r i d e , ~ ~ " , ~ or LiAlH4& has been advocated. The use of these procedures, however, appears to be distinctly hazardous for heavily peroxidized ethers, and preliminary small-scale purifica- tions are normally advisedS6" Interestingly, an earlier methodlg employing solid cerous hydroxide, which is ap- parently effective and widely applicable, does not appear to be generally used probably because of the necessity for preparation of fresh reagent and the inaptitude of the methodology for application to bulk purification.

In the context of these various shortcomings, we propose the use of indicating molecular sieves (IMS) as an effective, safe, and readily available reagent which is generally ap- plicable to the problem of peroxide removal from ethers.

Results and Discussion Deperoxidation of Tetrahydrofuran (THF). The

deperoxidation of THF was most thoroughly studied since of the common ethers discussed below this solvent has the highest proclivity toward peroxide formation.20 Table I

(14) The adsorbent loading required for effective deperoxidation is dependent on the grade of initial alumina and ether type, as well as on the initial water and peroxide content. However, for dry THF or dioxane of moderate peroxide content basic alumina of activity I will only purify less than an equal weight of ether.13bsc

(15) As opposed to cuprous and stannous chlorides, solid ferrous sulfate is apparent1 ineffective16 in peroxide removal, and an explosion has been reported17 on distilling THF from the solid salt. The effec- tiveness of solid inorganic reagents may perhaps be related to their sol- ubilities in ethers since both cuprous and stannous chlorides are appre- ciably more soluble than the ferrous salt.

(16) A. C. Hamstead, Znd. Eng. Chem., 56 (6), 37 (1964). (17) J. Schurz and H. Stubchen, Angew. Chem., 68, 182 (1956). (18) The use of aqueous ferrous sulfate for deperoxidation of THF,'"

is surely impractical as this ether is completely miscible with water. (19) J. B. Ramsey and F. T. Aldridge, J. Am. Chem. Soc., 77, 2561

(1955). (20) This conclusion has been drawn from the analysis of some 30

samples of stored ethers of various types and from monitoring the rate of peroxide formation in purified ether samples exposed to the atmo- sphere.

Dow

nloa

ded

by U

NIV

OF

PEN

N o

n A

ugus

t 4, 2

009

Publ

ished

on

May

1, 2

002

on h

ttp://

pubs

.acs

.org

| doi:

10.10

21/jo0

0141a0

03

Deperoxidation of Ethers J. Org. Chem., Vol. 47, No. 20,1982 3823

Table 11. Deperoxidation of Various Ethers with IMS

sieve peroxide content,c mmol/L

run (I ether % w/v initial 3 h 1 day 2 days 3 days 7 days 90 days conditionsd loading,

1 dioxane 1 0 1.7 0.22 0.44 A 2 10 5.5 4 .0 0.6 A

R R A R A A A

~

3 4 5 6 7 8 9

10 11

5 10.3 0.61 5b 10.3 0.44

5 12.8 0.24 diethyl ether 5 12.8 0.05 e

diisopropyl ether l o c 53 0.3 trimethylene glycol 10 29 5.5

10 109 1.1

dimethyl ether 5 27

3824 J. Org. Chem., Vol. 47, No. 20, 1982

spectrophotometric method used in this study are inef- fective for the detection of dialkyl peroxides. The sig- nificance of this observation is that all the methods re- ported in the literature to date for the removal of peroxides have in fact been proven only for hydroperoxides.

For evaluation of the relative importance of dialkyl peroxide formation, several badly contaminated ether samples were deperoxidized with IMS and the deconta- minated ethers analyzed for both active peroxide (hydro- peroxide) and dialkyl peroxide (Table IV). In every case, except dioxane, the total peroxide content is very much reduced by treatment with IMS. However, the acid-reflux analysis indicates the presence of small but detectable amounts of dialkyl peroxide in these samples. In the case of dioxane there is a very large concentration (>30% of original content) which is not analyzed by the spectro- photometric method and is not decomposed by IMS during the period of treatment. It would appear, therefore, that oxidation of dioxane leads to the formation of much higher concentrations of dialkyl peroxides than other ethers or to hydroperoxides which are much less reactive.

In summary, it appears that predominant contaminants of ethers such as THF and diisopropyl and diethyl ethers are active peroxides which can be readily removed by treatment with IMS. Deperoxidation of dioxane is very much less efficient. Almost certainly, the inapplicability of IMS to the problem of dialkyl peroxides is also a lim- itation with other commonly used redox purification sys- tems such as those based on CuC1, SnC14, FeS04, etc.

Applicability of IMS to Ether Deperoxidation. The prime criteria for deperoxidation methods are effectiveness and safety. Since most purifications are accompanied by distillation, which technique already allows a means of peroxide removal, the key aspect is safety. In this con- nection, IMS have proved effective in the deperoxidation of ethers with peroxide concentrations as high as 0.5 M without hazard. Compared to other methods IMS provide a relatively slow rate of peroxide decomposition, and this feature is almost certainly an important aspect of safe deperoxidation.

Since it has already been shown that molecular sieves are particularly effective in removing water24 and other polar impurities% from ethers, as applied to diethyl ether

Burfield

(24) D. R. Burfield, G. H. Gan, and R. H. Smithers, J. Appl. Chem. Biotechnol., 28, 23 (1978).

for example, the use of IMS provides a safe and effective method of concomitant deperoxidation, deethanolization, and drying of the solvent. In the case of dioxane or other ethers where the presence of significant quantities of alkyl peroxide are suspected, IMS may be used for pretreatment before more rigorous deperoxidation is executed.

Experimental Section Materials. Indicating activated Type 4A molecular sieve,26

(4-8 mesh) were kindly supplied by J. T. Baker Chemical Co. and were used as received. Ethers were typical reagent grade solvents. High peroxide levels were induced by aging purified samples in the presence of air over a period of days or weeks a t ambient temperatures.

Deperoxidation Method. ( i ) Static Deperoxidation. Ethers, in their original containers, were first deoxygenated by bubbling with oxygen-free nitrogen for about 5 min in a fume cupboard. Subsequently, 5% w/v of indicating molecular sieves was added and the ether tightly capped and set aside.

(i i) Reflux Deperoxidation. Ethers were charged into a distillation flask and deoxygenated with a slow stream of nitrogen. Subsequently, 5 % w/v of indicating molecular sieves was added and the ether brought slowly to reflux under nitrogen.

Peroxide Analysis. (i) Spectrophotometric Analysis. Hydroperoxides were analyzed by a modified spectrophotometric method based on the procedure of Wagner et This ferrous ion oxidation method permits quantitative analysis of hydro- peroxides (confirmed for tert-butyl hydroperoxide) but is com- pletely insensitive to dialkyl peroxides such as di-tert-butyl peroxide.

(ii) Total Peroxide Analysis. An acid-reflux method pro- posed by Mair and Graupnera was used and was found to provide a quantitative assay for dialkyl peroxides such as di-tert-butyl peroxide.

Acknowledgment. I acknowledge the assistance of Mr. Lee Meng Lay in conducting the peroxide analyses.

Registry No. Tetrahydrofuran, 109-99-9; dioxane, 123-91-1; di- ethyl ether, 60-29-7; diisopropyl ether, 108-20-3; trimethylene glycol dimethyl ether, 17081-21-9; dibenzyl ether, 103-50-4.

(25) D. R. Burfield and R. H. Smithers, Chem. Ind. (London), 240 (1980).

(26) Indicating molecular sieves (Type 4A) as supplied by Merck and Sigma were found to be equally effective. The product of the latter company is, however, only provided as a mixture (approximately 10% w/w) with nonindicating molecular sieves. Generally smaller bead sizes are more efficient for deperoxidation due to increased surface area.

(27) C. D. Wagner, H. L. Clever, and E. D. Peters, Anal. Chem., 19, 980 (1947).

Dow

nloa

ded

by U

NIV

OF

PEN

N o

n A

ugus

t 4, 2

009

Publ

ished

on

May

1, 2

002

on h

ttp://

pubs

.acs.o

rg | d

oi: 10

.1021/

jo0014

1a003

J. Org. Chem. 1981,46,629-631 629

(50 mL) afforded 0.400 g (80%) of 15: mp 249-250 C (C2H5OH); IR (KBr) v, 1709,1600 cm-; H NMR (Me2SO-ds) 6 2.68 (2 H, t , CH2), 3.04 (2 H, t , H2CC02H), 7.34-7.56 (3 H, m, Ar HI, 7.86 (1 H, s, Ar H), 7.96-8.12 (3 H, m, AI H), 8.50 (2 H, d, Ar H); UV (anhydrous C2H50H) A, 376 nm (e 3800), 357 (4400), 340 (320% 325 (1900), 255 (200000), 247 (90000).

Anal. Calcd for C1,H1402: C, 81.60; H, 5.60. Found: C, 81.33; H, 5.84. 5-(2-Anthryl)pentanoic Acid (16). Ester 6 (0.548 g, 2 mmol)

was hydrogenated over 10% Pd/C in anhydrous CZHSOH (50 mL), and the product obtained was boiled with o-chloranil (0.590 g, 1.2 mmol) for 3 h under NP Hydrolysis of the product from the above reaction with 10 alcoholic KOH solution (20 mL) afforded 0.410 g (71%) of acid 16: mp 191-192 C (C2H50H); IR (KBr) v- 1695,1575 cm-; H NMR (DCCld 6 1.6 (4 H, m, CHJ 2.2-2.3 (2 H, m, CH2), 2.82 (2 H, m, CH2), 7.3-7.5 (3 H, m, Ar H), 7.72 (1 H, m, Ar H), 7.88 (3 H, m, Ar H), 8.00-8.32 (2 H, m, Ar H); UV (anhydrous C2H50H) A, 377 nm (e 5400), 358 (6100), 341 (4400), 327 (2500), 291 (700), 254 (237000), 249 (88900).

Anal. Calcd for C19H1802: C, 82.01; H, 6.44. Found C, 82.13; H, 6.56. 3-(9-Anthryl)propanoic Acid (22).20 Hydrogenation of acid

20 (0.496 g, 2 mmol) over 10% Pd/C (50 mg) in anhydrous C2H50H (20 mL) afforded 0.450 g (90%) of acid 22: mp 188-190 C (C2H50H-H20, lit.20 mp 191-192 C); IR (KBr) 1695,1600 cm-; H NMR (DCClJ 6 2.78-2.96 (2 H, br t, CH2), 3.8-4.0 (2 H, br t, CH2), 7.40-7.55 (5 H, m, Ar H); 7.9-8.1 (2 H, m, Ar H), 8.2-8.4 (2 H, m, Ar H); UV (anhydrous C2H50H) A, 386 nm (e 5000), 361 (5180), 347 (380), 332 (1450), 256 (95700). 5-(I)-Anthryl)pentanoic Acid (23). Hydrogenation of acid

21 (0.556 g, 2 mmol) in anhydrous CzH50H (30 mL) over 10% Pd/C (70 mg) afforded 0.440 g (79%) of acid 23: mp 112-113 C (ether-petroleum ether); IR (KBr) vmar 1695, 1613 cm-; H NMR (DCClJ 6 1.70-1.98 (4 H, d, CHJ, 2.42 (2 H, m, CHJ, 3.58 (2 H, m, CH2), 7.16 (1 H, s, Ar H), 7.26-7.60 (4 H, m, Ar H), 7.94-8.20 (2 H, m, Ar H), 8.10-8.32 (1 H, m, Ar H), 11.14 (1 H, s, C0,H); UV (anhydrous C2HSOH) A,, 387 nm (e 8830), 382 (4640), 367 (8990), 348 (5420), 331 (2490), 318 (1020), 257 (169000), 250 @OW), 236 (21 600), 223 (6700).

Anal. Calcd for C l ~ l s O z : C, 82.01; H, 6.44. Found C, 81.87; H, 6.65.

Acknowledgment. We gratefully acknowledge partial support of this work from the Presidental Challenge Grant Program a t O.S.U. in the form of salary (to K.D.B.) and from the U.S.P.H.S., National Institutes of Health, via a grant from the Institute of General Medical Sciences (Grant GM 25353 to M.G.R.).

Registry No. 1, 2143-81-9; 2, 1099-45-2; 3,42997-19-3; 5,75802- 25-4; 6, 75802-26-5; 7, 75802-27-6; 8, 75802-28-7; 9, 75802-29-8; 10, 75802-30-1; 11, 75802-31-2; 12, 75802-32-3; 13, 75802-33-4; 14, 75802-34-5; 15, 75802-35-6; 16, 75802-36-7; 17,642-31-9; 18, 75802- 37-8; 19,75802-38-9; 20,5335-33-1; 21,75802-39-0; 22,41034-83-7; 23, 75802-40-3; [6-(methoxycarbonyl)hexa-2,4-dien-l-yl]triphenyl- phosphonium bromide, 75802-41-4.

Desiccant Efficiency in Solvent and Reagent Drying. 5.

David R. Burfield, Roger H. Smithem,* and Andrew Sui Chai Tan5

Department of Chemistry, University of Malaya, Kuala Lumpur 22-11, W. Malaysia

Reeeived September 9, 1980

The use of amines in synthesis can be divided into three principal areas: (i) as basic agents for the promotion of

(1) Part 1: D. R. Burfield, K. H. Lee, and R. H. Smithers, J. Org. Chem., 42, 3060 (1977).

0022-326318111946-0629$01.00/0

dehydroeliminations, (ii) as nucleophiles in simple dis- placements, and (iii) as precursors of various metalated derivatives. Because of strong N-H hydrogen bonding, in all these uses, water present in the amine system may exert damaging, i.e., yield lowering, effects by interfering with absolute basicity and nucleophilicity6 and/or reacting either as free water or hydroxide ion with unstable in- termediates or sensitive products. However, despite the existence of an arsenal of desiccants, the presence of water in these systems continues to be a problem for the syn- thetic chemist. This is because the recommended agents for removal of water and polar impuritiesg from other solvent and reagent typed4 may not be suitable for amines. Therefore the radiotracer method for water assay previously developed by uslo has now been applied to ob- tain quantitative data on the drying of some representative amines.

The Pyridine Group. For pyridine, and indeed gen- erally for the amine class, the traditionally recommended siccatives are the alkali and alkali earth hydroxides and 0xides.l Thus, literature prescriptions commonly advo- cate distillation from KOH,12a*b standing over Ba0,lZc or distillation from CaH2,13 the employment of the latter procedure reportedly yielding samples containing 18-20 ppm of residual water. The use of A1203 has also been occasionally reported.14 Our results for pyridine obtained by application of the radiotracer technique are summarized in Table I. The results are largely self-explanatory, and as can be seen, a horizontal line drawn under the entry for KOH sharply demarcates serious desiccants from those which are less efficaceous. Surprisingly perhaps, alumina is seen to be rather unimpressive; however, this ineffec- tiveness in the drying of polar reagents has been noted previously.J It is also worth noting that the use of sodium is to be avoided; it is not particularly efficient and con- tributes to material loss by a wasteful side reaction which produces bipyridyls.

Alkylated derivatives of pyridine are more basic and often less nucleophilic than pyridine itself, and these at- tributes are considered advantageous in synthesis. We therefore thought it of interest to compare the difficulty

(2) Part 2: D. R. Burfield, G. H. Gan, and R. H. Smithers, J. Appl. Chem. Biotechnol., 28, 23 (1978).

(3) Part 3: D. R. Burfield and R. H. Smithers, J. Org. Chem., 43,3966 (1978). (4) Part 4 D. R. Burfield and R. H. Smithers, J. Chem. Technol.

Biotechnol., in press. (5) Abstracted in part from the Final Year Project of Andrew S. C.

Tan, 1978-1979. (6) As is well-known, solvation effects play a vital part in determining

both basicity and nucleophilicity; aee, for example, C. Reichardt Solvent Effects in Organic Chemistry, Verlag Chemie, Weinheim, Germany, 1979, pp 55-60, 148-155. (7) In a pertinent example from our own laboratories, the literature

preparation of methyl diphenylphosphinite calls for reaction between chlorcdiphenylphosphine and methanol in the presence of pyridine and gives a reported yield of 5270.8 In our hands, the use of rigorously dried pyridine and methanol increased the yield to 75%.

(8) A. E. Arbuzov and K. V. Nikonorov,, Zh. Obsch. Khim., 18,2008 (1948). (9) D. R. Burfield and R. H. Smithers, Chem. Ind. (London), 240

(1980). (10) D. R. Burfield, Anal. Chem., 48 2285 (1976). (11) See, for example: (a) D. Todd Experimental Organic

Chemistry, Prentice-Hall Inc., NJ, 1979; (b) R. S. Monson, Advanced Organic Synthesis, Academic Press, New York, 1971; (c) J. A. Riddick and W. B. bunger, Organic Solvents, 3rd ed., Wiley-Interscience, New York, 1970.

(12) See, for example: (a) G. A. Olah and M. Watkins, Org. Synth., 68, 75 (1978); (b) W. H. F. Sasse, Organic Synthesis, Collect. Vol. V, Wiley, New York, 1973, p 102; (c) R. F. Evans, H. C. Brown, H. C. Van der Plas, ibid., p 977.

(13) D. Jerchel and E. Bauer, Angew. Chem., 68, 61 (1956). (14) D. N. Glew and N. S. Rath, Can. J. Chem., 49, 837 (1971).

0 1981 American Chemical Society

Dow

nloa

ded

by U

NIV

OF

PEN

N o

n A

ugus

t 4, 2

009

Publ

ished

on

May

1, 2

002

on h

ttp://

pubs

.acs

.org

| doi:

10.10

21/jo

0031

6a030

marisaHighlight

marisaHighlight

marisaHighlight

marisaHighlight

marisaHighlight

630 J. Org. Chem., Vol. 46, No. 3, 1981 Notes

Table I. Desiccant Efficiency in the Dryinga,b of a PyridineC Series residual water content,d ppm

desiccant CaH, CaC, BaO 4A sieves 3A sieves benzene azeotrope KOH powder Na CaO silica gel - 4 1 2 0 3

2-methyl- pyridine pyridine 39 (14)e 84 44 ( i o j e

101 106 (0.3)f 1 1 7 125 152 388 962 926

1306

71 27

55 40

176

2,g-dimethyl- pyridine

248 (138) 519 360 268 (1 26) 200 (128) 20 7 325

93 5

2,4 ,g-trimethyl- pyridine

132 8

33

47 390

27

a Static drying modes unless specified otherwise. Water content assayed by the radiotracer technique. Desiccant 24-h drying times unless specified otherwise. e 168-h loading 5% w/v; initial water content 2500 ppm (0.25% w/w).

drying time. f Sequentially dried sample, 24 h.

of drying of some of the more commonly used alkylated derivatives. A survey of the literature revealed that al- though the alkylpyridines are usually subjected to similar drying procedures as pyridine itself, fractionation alone and fractionation from BF3 (!) have been advocated for 2-meth~lpyridine~ and 2,6-dimethylpyridine,16 respec- tively. The results summarized in Table I exhibit a clear-cut trend; whereas one alkyl group at the 2-position gives results of similar overall order to pyridine, when N is flanked by two such groups (as in the lutidine) there is a marked increase in difficulty of drying, with 2-10-fold greater water levels being observed. Interestingly, the final member of series, the collidine, gave rise to some of the lowest residual water levels recorded. Not only is 2,4,6- trimethylpyridine the most basic of the series examined,17 it is also the easiest example to dry. Clearly then, there seems to be no relation between increasing basicity and drying difficulty, and instead this property appears to be determined by the interplay of water-solubility factors and of steric crowding about the N atom. Thus, for 2,6-di- methylpyridine, an infinitely water-miscible base, an ob- vious speculation is that the two flanking methyl groups present a serious impediment to the close approach of a siccative to the water-coordinating site. For trimethyl- pyridine, a drop in residual water concentrations is par- alleled by a corresponding large decrease in water solu- bility.18

Triethylamine. Triethylamine, commonly used as a mild base in dehydrohalogenations, has been dehydrated by the alkaline earth KOH,lgb CaH2,1gC and molecular sievedgd as well as by alumina and sodium metal.11c The results summarized in Table I1 require little comment beyond the fact that, despite being considerably more basic than the pyridines, triethylamine (pKb = 3.1) appears very much easier to dehydrate.

Diisopropylamine. Diisopropylamine is the precursor of lithium diisopropylamide, a powerful highly hindered

(15) L. A. Walter, Organic Syntheses, Collect. Vol. 111, Wiley, New York, 1955, p 757.

(16) A. N. Sharpe and S. Walker, J. Chem. SOC. 2974 (1961). (17) The relevant pKb values for pyridine, 2-methylpyridine, 2,6-di-

methylpyridine, and 2,4,6-trimethylpyridine are 8.8, 8.0, 7.3, and 6.6, respectively.

(18) While pyridine itself, as well as ita mono- and dialkyl derivatives, is essentially completely water miscible, the solubility of the trimethylated pyridine is only about 3%. See Beilsteine, 4th ed., 20, 164 (1953).

(19) (a) R. Breslow and J. Posner, Organic Syntheses, Collect. Vol. V, Wiley, New York, 1973, p 514; (b) H. Rinderknecht and M. Guten- stein, ibid, p 822; (e) M. E. Jung and C. A. McCombs, Org. Synth., 58, 163 (1978); (d) T. J. Atkins, R. E. Richman, and W. F. Oettle, ibid., 58, 87 (1978).

Table 11. Desiccant Efficiency in the Dryinga of Various Aminesb

residual water content,c ppm

(Me,CH),- desiccant Et,Nd NHe NH,(CH,),NH,e

KOH powder 37 ( 23)f 4A sieves 33 ( 28)h 3A sieves 34 CaH, 68 ( 34)f Na 83 BaO 89 (53)f CaC, 98 ( 80) f CaO 165 ( 56)f -4120, silica gel CaSO,

223 (223) j 451

750g 1370 (3700)g < 25 < 25 < 25 < 25 150 500 < 25 150

50 1100

J. Org. Chem. 1981,46, 631-632 63 1

1,3-Propanediamine. This is another example of an amine which has sprung to prominence in synthesis be- cause of unique reactions brought about by a derivative, in this case its monopotassium SalkB Desiccation has been achieved by vacuum distillation alone,% distillation from alkali metal or distillation from KOH.24c Our results using the near-IR method are displayed in Table I1 and indicate that, not surprisingly, this amine with two coordinating centers is one of the more difficult examples to dry. It is worth noting that freshly ground KOH gives a drier sample than aged material stored in a desiccator, although as may be seen, in comparison with several other siccatives, e.g., CaCz or sieves, KOH is not particularly good. In addition, although seldom recommended for use in these situations, CaS04 (Drierite) was investigated in view of some of the more extravagant claims which have been made on its behalf.25 As may be seen, the results are rather dismal.

Desiccation of Amines. General Recommendations. Perhaps not surprisingly, the results from this study lend support to the current usage of molecular sieves and CaH2 as serious and widely applicable siccatives for amines. Apart from this, it is fitting to draw attention to the high efficiency of CaC2 in these studies, which, though infre- quently prescribed as a desiccant, is often seen to surpass CaH2 in potency and is more desirable than the hydride from consideration of cost and safety in storage. On the other hand, the performance of alumina is uniformly disappointing and it cannot be advocated as a serious desiccant for amines.

Experimental Section Details of the water assay techniques as well as the source,

activation, and handling of most of the desiccants have already been described.14 Determination of water content by the near-IR method22 was carried out on a Varian Cary 17 Instrument. Calcium carbide was of industrial grade and was crushed in a mortar immediately prior to use. Appropriate venting was prc- vided for desiccants producing gases, e.g., CaC2, CaH2.

Amines were of laboratory reagent grade and purified by standard methods.llc The pyridines and triethylamine were stood over KOH for 24 h, decanted, fractionated, and stored in dark bottles over 20% w/v 3A molecular sieves. Pyridine had bp 115-116 OC, 2-methylpyridine 128-129 "C, 2,6-dimethylpyridine 145-146 "C, and 2,4,6-trimethypyridine 176-178 "C. EtaN had bp 89-89.5 "C. Diisopropylamine was first allowed to stand over 20% w/v 3A molecular sieves, decanted, stirred overnight with CaH2, fractionated, and finally stored as above, bp 84 OC. 1,3- Propanediamine was mixed with 20% v/v benzene, fractionated, stirred overnight with CaH2, and fractionated again, bp 135-136 "C.

Acknowledgment. We thank the Department of Chemistry, University of Malaya, for support of this work and Miss Lim Siew Heong for her able technical assistance.

Registry No. Pyridine, 110-86-1; 2-methylpyridine, 109-06-8; 2,6-dimethylpyridine, 108-48-5; 2,4,64rimethylpyridine, 108-75-8 EhN, 121-44-8; (Me2CH)2NH, 108-18-9; NH,(CH2)3NH2, 109-76-2; CaH2, 1789-78-8; CaC2, 15-20-1; BaO, 1304-28-5; benzene, 71-43-2; KOH, 1310-58-3; Na, 7440-23-5; CaO, 1305-78-8; A1203, 1344-28-1; CaS04, 7778-18-9.

(23) Potassium 3-aminopropylamide possesses the singular property of bringing about the conversion internal alkyne - 1-alkyne in essentially quantitative yields. See C. A. Brown and A. Yamashita, J. Am. Chem. soc., 97,891 11975).

(24) (a) A. Gero, J. Am. Chem. SOC. 76,5159 (1964); (b) L. R. Dalton, J. L. Dve. E. M. Fielden. and E. J. Hart, J. Phvs. Chem., 70,3358 (1966); (c) K. k'Badri and L. Y. Goh, Inorg. Chim. Acta, in press.

pany, Xenia, OH. (25) W. A. Hammond, "Drierite", the W. A. Hammond Drierite Com-

0022-3263/81/1946-0631$01.00/0

Trapping of Intermediates in the Thermolysis of a-Azidochalcone. Insight into the "Zwittazido

Cleavage" Reaction

Benjamin A. Belinka, Jr., Alfred Hassner,* and J. M. Hendler

Department of Chemistry, State University of New York at Binghamton, Binghamton, New York 13901

Received July 28, 1980

The "zwittazido cleavage" reaction (Scheme I) has re- cently been the subject of considerable study' because this novel rearrangement possesses both interesting mecha- nistic and advantageous synthetic characteristics. Al- though a large amount of work with cyclic and acyclic ketoazides has been reported,lV2 the mechanism of this reaction is still somewhat obscure, especially the question of whether a concerted or stepwise process (or both) occurs.

In the rearrangement of cyclic keto vinyl azides op- portunity exists for an intramolecular ring closure following an initial ring cleavage (Scheme I). In acyclic systems formation of azirines, indoles, or rearranged cyano ketones can take place.2 In transformation 1-2, the reaction has been postulated to occur via intermediate 3. An alternate pathway would involve a cleavage to ions 4 and 5, which may recombine to 2 (Scheme 11).

Our interest in the chemistry of vinyl azides3 led us to investigate this reaction and to show that even in the acyclic case cleavage can occur and that the intermediates can be trapped with alcohols or amines.

Thermolysis of a-azidochalcone (1) in o-dichlorobenzene for 24 h produced a-cyano-a-phenylacetophenone (2) in 70% yield. When the reaction was carried out in the presence of 10 equiv of either ethanol or benzyl alcohol, the yield of 2 decreased while ethyl benzoate (6) and benzyl benzoate (7), respectively, were isolated. Thermolysis with 10 equiv of benzylamine yielded N-benzylbenzamide (8). In all three trapping experiments, considerable quantities of 2 were also obtained along with benzyl cyanide (9) and polymeric materials. The results are recorded in Table I.

The formation of compounds 6-9 indicates that at least part of the thermolysis of the acyclic a-azidochalcone proceeds by a cleavage mechanism as shown in Scheme I1 which allows for an acyclic cation intermediate 4 to react with the alcohols and amine. Protonation of 5 leads to benzyl cyanide (9).

We showed that the azido ketone 1 did not react with ethanol in boiling toluene to produce ethyl ben~oa te .~ Furthermore, no ethyl benzoate (6) was detected by GC when either the cyano ketone 2 or the crude mixture re- sulting from heating 1 in o-DCB was subjected to trapping reaction ~ondi t ions.~

(1) H. W. Moore, Acc. Chem. Res., 12,125 (1979), and references cited there in.

(2) D. Knittel, Hemetsberger, R. Leipert, and H. Weidman, Tetrahe- dron Lett., 1459 (1970).

(3) For instance (a) G. L'Abb6, and A. Hassner, Angew. Chem., Int. Ed. Engl., 10, 98, (1971); (b) A. Hassner, E. S. Ferdinandi, and R. J. Isbister, J. Am. Chem. SOC., 92,1672 (1970); (c) A. Hassner, Acc. Chem. Res., 4, 9 (1971).

(4) Azidochalcone (1) reads with nucleophiles such as sodium sulfide or ylides at the azide rather than at the carbonyl function; B. A. Belinka, Jr., and A. Hassner, J. Org. Chem., 44, 4712 (1979), and unpublished results in this laboratory; G. Mathys, S. Toppet, and G. L'abbd, Chem. Ind. (London), 6,278 (1975).

(5) G. L'abbg and A. Hassner, J. Org. Chem., 36, 258 (1971). 0 1981 American Chemical Society

Dow

nloa

ded

by U

NIV

OF

PEN

N o

n A

ugus

t 4, 2

009

Publ

ished

on

May

1, 2

002

on h

ttp://

pubs

.acs

.org

| doi:

10.10

21/jo

0031

6a030

3966 J. Org. Chem., Vol. 43, No. 20,1978 Notes

Other acetals which are commonly used as protecting groups for alcohols are the tetrahydropyranyl ether (ref 4, p 104), ethoxyethyl ether (S. Chladek and J. Smtt, Chem. Ind. (London), 1719 (1964)), 2-methoxyisopropyl ether (ref 4, p 107), and 4-methoxytetrahydropyranyl ether (ref 4, p 108). Removal of all of these groups can be accornpllshed with dilute, aqueous acid. They are prepared from the corresponding vinyl ether with acidcataiysis. The methoxymethyl ether,' as well as our tert-butoxymethyl ether, is prepared from the chloro ether with base catalysis. See, for example, (a) the synthesis of chloromethyl methyl ether: C. S. Marvel and P. K. Potter, "Organic Synthesis", Collect. Vol. I, 2nd ed, A. H. Blatt, Ed., Wiley, New York, N.Y., 1941, p 377; (b) the preparation of methoxyethyl chloromethyl ether: E. J. Corey, J.-L. Gras, and P. Ulrich, Tetrahedron Lett., 809 (1976); and (c) a review on a-halo ethers: L. Sum- mers, Chem. Rev., 55, 301 (1955). Ethers are halogenated on the (Y carbon: M. L. Poutsma in "Methods in Free Radical Chemistry", Vol. 1, E. S. Huyser, Ed., Marcel Dekker, New York, N.Y., 1969, p 137. Methods for the free-radical halogenation of organic compounds have been reviewed: E. S. Huyser, Synthesis, 7 (1970). Sulfuryl chloride has been used to chlorinate tetrahydrofuran (THF): C. G. Kruse, N. L. J. M. Broekhof, and A. van der Gen, Tetrahedron Lett., 1725 (1976). All attempts to isolate the chloro ether by concentration have led to de- composition. J. F. Norris and G. W. Rigby, J. Am. Chern. SOC., 54, 2088 (1932). There is no reaction at 0 O C after 4 h but a satisfactory reaction rate is obtained at room temperature. The water bath is used for cooling purposes only. The 'H NMR spectra of the tert-butoxymethyl ethers show singlets at 6 4.1-4.7 (2 H) and 1.2-1.25 (9 H). The corresponding l3C NMR spectra are also consistent with the proposed structures. For example, the acetal and quaternary carbons are found at 6 89.248 and 74.251, respectively, for the benzyl alcohol acetal and 6 90.038 and 74.068, respectively, for the 1-hexanol acetal.

Desiccant Efficiency in Solvent Drying. 3. Dipolar Aprotic SolventslJ

David R. Burfield* and Roger H. Smithers

Department of Chemistry, University of Malaya, Kuala Lumpur 22-11, Malaysia

Received May 2, 1978

I t is generally acknowledged that dipolar aprotic solvents are the media of choice in some reactions and are unique in facilitating others.3 The special solvent effects of molecules such as DMF and Me2SO are attributable to their large di- electric constants coupled with the absence of solvation by hydrogen bonding and typically manifest themselves in properties such as poor anion solvation, voracious cation solvation, and a marked hydrophilicity. For the chemist, this latter feature is unfortunate since small amounts of water in these systems can diminish314 their nucleophilicity and may even be hazardous to some operations.5 The drying of these solvents is thus of paramount importance, but in these cases, as previously,' the chemical literature contains little reliable quantitative data.

We have recently developed a method of solvent water assay

which utilizes a tritiated water tracer for the determination of water content.6 The method circumvents many of the problems encountered in other assay methods and has pro- vided some new correlations on the efficiency of desiccants.lS* For example, i t has been shown that, rather surprisingly, the efficiency of a given desiccant is strongly dependent upon the solvent type,' and there is thus much uncertainty in extrap- olating generalizations from one solvent type to another.

The method has now been applied to the desiccation of the dipolar aprotics acetone, DMF, MeZSO, and HMPT. Since the dielectric constants of these solvents range between 20.7 (acetone) and 46.7 (MeZSO), their rigorous desiccation is ex- pected to be difficult.

Results and Discussion Drying of Hexamethylphosphoric Triamide (HMPT).

Caution! HMPT is a suspected carcinogen. Although in re- cent years the favored desiccant for HMPT appeared to be calcium h ~ d r i d e , ~ drying has also been previously accom- plished with alkali metal^,^^^ alkali metal earth oxides,s and 4A9* and 13Xgb molecular sieves.

The results with the siccatives summarized in Table I are largely self-evident, but the following points are worth noting. The extreme resistance to desiccation is demonstrated by the impossibility of obtaining super-dry lo H M P T under any of the conditions used here. Even sequential drying,ll which was previously found to be effective with acetonitrile,2 falls short in this case. The use of sodium-potassium alloy as a drying agent seems questionable in view of the thermal instability of soluti