kilomentor

The sulfate salt is the second most common pharmaceutical salt behind the hydrochloride. Bisulfate salts are quite acidic so the base from which one is made needs to be acid stable.

Sulfuric acid is a diprotic acid. It can form two different stoichiometric salt types: the 1:1 bisulfate salt and the 2:1 sulfate salts in which two moles of amine are protonated by each of the two protons of H₂SO₄. The pKas of sulphuric acid are –3 and 1.92 with almost five orders of magnitude difference between the acidity of the first and second hydrogen. Most pharmaceutical salts are of the 1:1 bisulfate type. Sulfates are most often made by the addition of an, at least partially aqueous, solution of acid because neat acid is not soluble in apolar solvents and it has some dehydrating capability which can lead to by-products when sufuric acid is in excess. Typical organic solvents used in making sulfates are methanol, ethanol, 1-propanol, 2-propanol, acetone and mixtures thereof. Acetone however is not recommended because an excess of acid causes the oligomerization of acetone creating color in the solution.

Kilomentor anticipates that by providing some examples of pharmaceutical sulfate salt preparations with some commentary to draw attention to important aspects of the methods a skilled experimentalist should have no difficulty making others.

US7230016 PREPARATION OF PIOGLITAZONE SULFATE

24.g of sulfuric acid was added slowly, at room temperature, to 250 ml of methanol followed by addition of 80 g of pioglitazone base with stirring. The mixture turned into a clear solution. 250 ml of ether was slowly added followed by 500 ml of heptane. A solid precipitated, and the suspension was stirred for 3 hours. The solid (98.4 g, yield was 96.5%) was collected by filtering and washed once with ether. The solid had a mp: 1113.5-116.5° C. (recrystallized from methanol).

This example illustrates the addition of the base to the organic solution of sulphuric acid in methanol. Although a small amount of methyl hydrogen sulphate might form this not a problem because MeOSO2OH is a pharmaceutically acceptable counterion. Note also that in the procedure the chemistry provided three opportunities to obtain crystals. Pioglitazone hydrogen sulfate might have precipitated from the methanol solution itself after partial dissolution. The salt might have crystallized when the methanol was diluted 50:50 with diethyl ether. The final opportunity occurred when the solution was diluted 1:1 with heptane and this was successful. Notice that the methanol could not be diluted with heptane directly. Two phases would have resulted. This is an example of a well designed approach to getting crystalline solid. If crystals still had not formed the solution would have been concentrated.

WO06040728A1: Preparation of 1-(2-(4-benzyl-4-hydroxy-piperidin-1-yl)-ethyl)-3-(2-methyl-quinolin-4-yl)-urea

Example 1
1-(2-(4-benzyl-4-hydroxy-piperidin-1-yl)-ethyl)-3-(2-methyl-quinolin-4-yl)-urea

(1 equivalent) is dissolved in ethanol at a concentration of 25% w/w and the mixture is heated at 50°C. Aqueous sulfuric acid (1M, 1.1 equivalents) is added. Optionally, the crystallization is initiated by a wet seed of Example 1 (0.5%). The suspension is cooled to 0°C with a cooling rate of 15 C°/h and maintained at this temperature at least 1 hour before filtration and washing with aqueous ethanol (50 % W/V). The solid is dried at 30°C under a wet stream of nitrogen (50% RH) to provide the title compound with a purity of 97.7% with a yield of approximately 90%.

The example illustrates the addition of the acid to an excess of base. The addition is performed warm. An aqueous sulphuric acid reagent is used and it is added to a water miscible solvent in this case ethanol. Using seeds of the salt product is optional here. The example prescribes a cooling rate that will lower the temperature to the final filtration temperature over somewhat more than 3 hours. This is followed by a hold time to ensure that all the material that can crystallize has come out before the filtration. The wash solution is a mixture of solvents similar to that from which the solid is crystallized. Often a slightly less polar wash solution is used than the mixture from which the crystals are produced. This gives some assurance that the wash will not redissolve the solid. Although it is not reported the wash solvent is usually cooled to the temperature of the slurry that was filtered originally. On scale, this is done simply by loading the wash solvent mixture into the crystallizer. Because the solvent is a mixture with water, there is no danger of condensing damaging moisture into the wash solution. The example illustrates using a moist gas stream to dry the solid without dehydrating it.

Example 2
1-(2-(4-benzyl-4-hydroxy-piperidin-1-yl)-ethyl)-3-(2-methyl-quinolin-4-yl)-urea sulfate trihvdrate.

To a suspension of 1-(2-(4-benzyl-4-hydroxy-piperidin-1-yl)-ethyl)-3-(2-methyl-quinolin-4-yl)-urea (21.36 kg) in CH₃OH (178 L) is added aqueous H2SO4 (6 L, 9.91%) during 10 min. The clear solution is filtered and further aqueous H2SO4 (33.8 L, 1.07 M) is added during 45 min. The solution is cooled to -2°C during 1.5 h and stirred at -5 to -9°C for 1 h. The formed precipitate is filtered, washed with cooled CH30H (- 5°C, 54 L) and dried under a stream of nitrogen provide 1-(2-(4-benzyl-4-hydroxy-piperidin-1-yl)-ethyl)-3-(2-methyl-quinolin-4-yl)-urea sulfate of formula l as a non-defined hydrate. A slurry of the so obtained salt in H2O (16.2% w/w) is stirred for 3 days at 25°C. Filtration and drying at 30°C under a wet stream of nitrogen (50% RH) provides the title compound.

This example replaces the ethanol with methanol and is in most particulars very much the same. Here unhydrated gas was used in the drying ad there apparently was some dehydration. Stirring a slurry in water for an extended period recreates the hydrate illustrating a method of preparing a pseudopolymorph hydrate. Drying to the trihydrate was successful when the relative humidity was controlled at 50%.

Example 4
1-(2-(4-benzyl-4-hydroxy-piperidin-1-yl)-ethyl)-3-(2-methyl-quinolin-4-yl)-urea sulfate dihvdrate

1-(2-(4-benzyl-4-hydroxy-piperidin-1-yl)-ethyl)-3-(2-methyl-quinolin-4-yl)-urea (15.4 kg, 1 equivalent) is dissolved in ethanol (78 L) and the mixture is heated at 50°C. Aqueous sulfuric acid (1M, 1.1 equivalents) is added during minutes. The crystallization is initiated by a wet seed of Example 1 (1%) as described below. The suspension is cooled to 1°C with a cooling rate of 14C°/h and maintained at this temperature at least 11 hours before filtration and washing with aqueous ethanol (50 % W/W, 50 L). The solid is dried at 30°C under a wet stream of nitrogen (33-40% RH) to provide the title compound with a purity of 99.4% with a yield of approximately 79%.
The wet seed used in the above procedure is prepared by mixing - 1-(2-(4-benzyl-4-hydroxy-piperidin-1-yl)-ethyl)-3-(2-methyl-quinolin-4-yl)-urea sulfate dihvdrate (Example 1,) with a saturated solution (421 9) of 1-(2-(4-benzyl-4-hydroxy-piperidin-1-yl)-ethyl)-3-(2-methyl-quinolin-4-yl)-urea sulfate dihvdrate (Example 1, 73.9 9) in aqueous ethanol (50 % W/W, 810 9).

Example 5

1-(2-(4-benzyl-4-hydroxy-piperidin-1-yl)-ethyl)-3-(2-methyl-quinolin-4-yl)-urea sulphate dihvdrate

1-(2-(4-benzyl-4-hydroxy-piperidin-1-yl)-ethyl)-3-(2-methyl-quinolin-4-yl)-urea (1.01 kg, 1 equivalent) is dissolved in ethanol (3.05 kg) under stirring (200±20 rpm) and the mixture is heated at 50°C. Aqueous sulfuric acid (1 M, 1.1 equivalents) is added during 20 minutes. The crystallization is initiated by a wet seed of Example 1 (1 %) as described below. The obtained mixture is maintained at 50°C for about 15 minutes, then it is cooled to 0°C with a cooling rate of 15°C/h and maintained at this temperature for least 1 hour before filtration and washing with aqueous ethanol (50 % W/W, 3 kg). The solid is dried in a conductive agitated dryer at a temperature of 35± 3°C under a wet stream of nitrogen (45±5% RH), optionally under stirring (max. rpm) in case the cake humidity is below 25%, to provide the title compound with a purity of 99.8% with a yield of approximately 94%.

The wet seed used in the above procedure is added in two shots and is prepared by mixing 1-(2-(4-benzyl-4-hydroxy-piperidin-1-yl)-ethyl)-3-(2-methyl-quinolin-4-yl)-urea sulfate dihvdrate (Example 1, 6.5 9) with a saturated solution (13.9 9 for the first shot, plus 15.6 9 for subsequent rinsing and second shot) of 1-(2-(4-benzyl-4-hydroxy-piperidin-1-yl)-ethyl)-3-(2-methyl-quinolin-4-yl)-urea sulfate dihvdrate (Example 1, 7.0 9) in aqueous ethanol (50 % W/W, 50.0 9) for about 2 minutes. The first shot of wet seed is prepared at least 5 minutes before use to ensure that the seed is correctly wetted.

Example 5 illustrates that in a process description of crystallization the mixing times, cooling rates and stirring need to be precisely controlled. The need to properly moisten the seed crystals with the solvent is also illustrated. If the seeds’ surface does not wet properly they cannot catalyze crystal growth properly.

US20060194833A1: Crystalline 1H-imidazo[4,5-b]pyridin-5-amine, 7-[5-[(cyclohexylmethylamino)-methyl]-1H-indol-2-yl]-2-methyl, sulfate (1:1), trihydrate and its pharmaceutical uses

According to the method, ER807447 is first suspended in water to form an aqueous suspension. Sulfuric acid is added to the aqueous suspension to form a solution while keeping the internal temperature of the solution below 25° C. The solution typically has a yellow color. The solution may optionally be filtered to remove particulates from the solution. Other techniques for removing particulates known in the art, centrifuging, etc. may be used as the filtering step. The solution is then slowly warmed until E6070 crystallizes from solution. The solution may be warmed to about 100° C. Typically crystal formation occurs at temperatures of about 70° C. Preferred rates of warming typically range from about 30 minutes to 5 hours. Longer or shorter times may be used, particularly depending upon the batch size. E6070 may not crystallize as readily from highly dilute solutions.

To enhance crystallization, an anti-solvent may be used in the method of making the crystalline E6070 or to recrystallize crystalline E6070. The recrystallization procedure is described in Example 5. In the above method, the anti-solvent may be added to the aqueous suspension before sulfuric acid addition or to the solution after sulfuric acid addition and the optional filtration step. Useable anti-solvents and their use are known in the art. Typical anti-solvents include water-miscible anti-solvents such as, for example, methanol, ethanol, 1-propanol, 2-propanol, acetone and mixtures thereof. When an anti-solvent is used, the solution may become cloudy. It is generally not necessary to warm the solution to as high of temperatures as when just using an aqueous solution.

The procedure above illustrates forming a bisulfate from water. Filtration or other clarification of the formed solution is illustrated. Removing insolubles removes nuclei that can catalyze improper nucleation. The example illustrates that a bisulfate salt in water may actually be supersaturated but the rate of nucleation may be impracically slow. Heating the solution increases the rate of nucleation and causes the insoluble salt to come out. If the sulphate is a high molecular weight molecule that should give an insoluble sulphate in water, perhaps heating will enhance the rate of seed formation as here.

The use of an antisolvent is also illustrated. Note that water miscible organics are often antisolvents for sulphate salts because sulphate salts are so hydrophilic.

Tizanidine Monosulfate (5-chloro-4-f2-imidazolin-2-ylamino')-2,l,3-benzothiadiazole monosulfate )

In a second preparation, tizanidine monosulfate was prepared by the following method: to solid tizanidine (9.957 g; 39.24 mol) was added a solution of sulfuric acid (5.438 g; 55.45 mol) in acetonitrile (175 mL). The yellow solid rapidly converted to a white crystalline solid. The mixture was heated to 60°C and stirred for 90 minutes. The mixture was cooled to room temperature and the solid was subsequently filtered and washed with additional acetonitrile (50 mL). The solid was collected via filtration and air-dried.

Tizanidine monosulfate comprises a 1:1 ratio of ionized tizanidine to sulfate counterion. In this example sulfuric acid in acetonitrile is used

Rosiglitazone Sulfate

Example 1:

5-[4-[2-(N-methyl-N-(2-pyridyl)amino) ethoxy]benzyl] thiazolidine-2,4-dione sulfate

5-[4-[2- (N-Methyl-N-(2-pyridyl)amino) ethoxy]benzyl]thiazolidine-2,4- dione (20.0 g) in glacial acetic acid (50 ml) was stirred and heated to 75°C until a clear solution 5 was observed. Concentrated sulfuric acid (1. 5 ml) was added and the stirred solution cooled to 21 °C. After evaporation of solvent under reduced pressure, methanol (100 ml) was added and the mixture stirred at 21°C for 48 hours. The solid was collected by filtration, washed with methanol (50 ml) and dried under vacuum to give 5-[4-[2-(N-
methyl-N-(2-pyridyl)amino)ethoxy]benzyl]thiazolidine-2,4- dione sulfate (10.7 g) as a crystalline solid.

Melting point: 184 - 189°C.

DSC: Tosser = 184.4°C, Tpeak = 189.1 °C Elemental Analysis: 15 Found: C; 52.96 H; 4.94 N; 10.23 S; 11.78 Theory: (C36H40N6O,oS3) C; 53. 19 H; 4.96 N; 10.34 S; 11.83

In this example glacial acetic acid is used as solvent for the free base and heating is require to get a clear solution. The experimentalist apparently expected the sulfate to precipitate at about ambient but it did not. It is important that in this example the basic group was tertiary because the combination of acetic acid and sulfuric acid could cause acetylation with primary or secondary amines. Undeterred the experimentqalist removed some acetic acid under vacuum and replace it with the antisolvent methanol. This time cooling gave the desired solid derivative.

Example 2:

5-[4-l2-(N-methyl-N-(2- pyridyl)amino) ethexylbenzyl] thiazolidine-2,4 20 dione sulfate

5- [4-[2-(N-Methyl-N-(2- pyridyl)amino)ethoxy]benzyl] thiazolidine-2,4-dione (40.0 g) in glacial acetic acid (100 ml) was stirred and heated to 70°C until a clear solution was observed. Concentrated sulfuric acid (3.1 ml) was added and the mixture stirred for 10 minutes at 70°C, then cooled to 21°C with stirring. The solvent was evaporated under reduced pressure, followed by the addition of methanol (100 ml) and the mixture was stirred at 21 °C until crystallization was complete. The product was collected by filtration, washed with methanol (200 ml) and dried under vacuum over phosphorus pentoxide for 4 hours at 50°C to give 5-[4-[2-(N-methyl-N-(2 pyridyl)amino)ethoxy] benzyl]thiazolidine-2,4-dione sulfate (36.9 g) as an off white crystalline solid.

Example 3:

5-[4-[2-(N-methyl-N-(2-pyridyl)amino) ethexy]benzyl] thiazolidine-2,4 dione sulfate

Concentrated sulfuric acid (1.94 ml) was added to a stirred suspension of 5-[4- [2 35 (N-methyl-N-(2-pyridyl)amino) ethoxy]benzyl] thiazolidine-2,4-dione (25.0 g) in methanol (1000 ml) at 56°C. The reaction mixture was stirred at 60°C until a clear solution was observed, then cooled to 21 °C and stirred at this temperature for 16 hours.
The product was collected by filtration, washed with methanol (100 ml) and dried under vacuum at 21°C for 3 hours to afford 5-[4-[ 2-(N-methyl-N- (2 pyridyl)amino)ethoxy]benzyl]thiazolidine-2,4- dione sulfate (19.5 g) as a white crystalline solid.

By far the most frequently successful pharmaceutical salt is the hydrochloride. In fact the hydrochloride salt is selected 50% of the time when chemists look for an acceptable salt. Typically, there must be a very good reason for not adopting this salt. If a hydrochloride crystallizes, one typically needs a n excellent reason not to use the hydrochloride.

The hydrochloride is a preferred choice because chloride does not have any activity of its own, unlike bromide, nitrate and others. Hydrochloric acid is a very significant acid in the stomach. By salt exchange hydrochlorides are formed to some extent no matter what the counter ion of an API is in the pharmaceutical product.

Hydrochloric acid is a strong mineral acid strong enough to quantitatively protonate even weak bases.

Hydrochlorides characteristically are substantially more soluble than the free bases used to make them, so the hydrochloride typically improves the bioavailability.

Hydrochlorides can be prepared in aqueous solution, in protic organic solvents, in aprotic organic solvents, and in non-polar solvents because hydrogen chloride can exist in both a covalent form in apolar solvents or as ionized protons and chloride ions in more polar solvents. The actual acidity varies being equal to the acidity of the conjugate acid of the solvent molecule. That is to say hydronium ions exist in water, protonated alcohol ions in alcohol, protonated acetic acid in glacial acetic acid or protonated ethyl acetate molecules in ethyl acetate. The multiple forms of HCl result in multiple techniques for the addition of the hydrogen and chloride ions to the pharmaceutical base we need to make into a salt.

Hydrogen chloride gas can be passed into neat organic solvents to prepare titratable molar solutions that are quite stable. Hydrogen chloride in lower alcohols is not stable for a long time and must e used soon after it is formed. More often the gas is added to the base dissolved in the lower alcohol. HCl forms quite stable solutions in IPA which can be sotred at ambient temperatures for several days. Hydrochloric acid solutions in non aqueous solution can be made by adding acetyl chloride into ethanol wher a quantitative reaction occurs to give hydrogen chloride and an equamolar amount of ethyl acetate.

A recent PCTpatent teaches the creation of hydrogen chloride in situ from trimethylsilyl chloride and any solvent with a silylated functionality or any inert solvent containing a slight excess over the silyl chloride of a silylable group.

Hydrochlorides can be made by reaction of the organic base with an equivalent of ammonium chloride. The stronger organic base preferentially takes the hydrogen chloride and the ammonia gas is liberated and may exit the reaction mixture.

If the free base of concern has some solubility in water, the hydrochloride can be made from aqueous hydrochloric acid and the base in water. Often heating is required to get the free base dissolved and the hydrochloride salt separates on cooling. In aqueous solution the solubility of the amine hydrochloride can be decreased by the addition of additional inorganic water soluble chloride to increase the common chloride ion. The addition of inorganic salts also increases the precipitation by the salting out effect.Excess hydrochloric acid can be used to decrease the solubility of the desired salt so long as the pharmaceutical chloride is stable in strong aqueous acid.

Standard aqueous solutions of hydrochloric acid can be added directly into the base dissolved in a water miscible polar solvent such as methanol, ethanol, propanols, butanols, acetone, 2-butanone, acetonitrile,

The most powerful and widely practiced method of making a hydrochloride salt in the laboratory is to add gaseous hydrogen chloride into a diethyl ether solution of the free base. If you think a hydrochloride salt might not be crystalline, this technique is likely to provide evidence one way or the other. It is not a practical process method to make the slat but it will give evidence that the solid salt is possible and will provide some seed crystals for other preparative methods.

If a solid pharmaceutical hydrochloride is formed, the next goal is to obtain it in a satisfactory recovery. Pharmaceutical bases are typically expensive moieties and losing material in a low recovery hydrochloride formation is undesirable. When a solution of a hydrochloride and the pharmaceutical base has been formed but only a small amount or no crystals at all precipitate three strategies are possible

1. cool the solvent to decrease the overall solubility in the solvent volume
2. all an antisolvent which changes the quality of the solvent system and lowers the solubility
3. add an antisolvent and then cool

As a rule of thumb if the recovery is 80% or more at ambient temperature, simply cooling the solvent can be expected to give an excellent recovery; if the recovery is 40-80% at ambient cooling should be applied and then an altisolvent judiciously added; but when the starting recovery is less than 40% an antisolvent should be added to just the cloud point and then cooling should be applied.

Treatment is completely reversible. What you cool down you can rewarm so this is usually tried first.

If treatment 2 is used it is useful for the analysis of the results to take a sample of the solid obtained just by cooling in order to measure the purity at that point, then add the antisolvent to give a practical recovery and compare the purity of the product when using an anti-solvent with the purity before that addition.

When the solid formation in the single solvent condition is low or none, an anti-solvent is added to the cloud point either to a hot or ambient solution and then controlled cooling is applied to try for crystallization.

Mixtures of solvents are not preferred in scale up processes because it introduces the nedd for an in process test to guarantee the proper solvent ratio. Second crops are more difficult to obtain from a mixed solvent in a simple procedure. Nevertheless situations where a mixed solvent gives the best purification and recovery do occur. It is advantage if two solvents differ substantially in boiling point. This allows recovery of pure solvents from the filtrate.

Solvents that form an azeotrope with water have the advantage that it is easier to be sure that the crystallization is under anhydrous conditions.

An article entitled “Pharmaceutical Manufacturing Goes Green” appeared in the recent March 1 2008 issue of Genetic Engineering & biotechnology News (GEN). It was stated that “advances in enzymatic catalysis of synthetic reactions, solvent substitutions, and recycling of by-products and waste may not only reduce the environmental impact of pharmaceutical processes but, with the potential to have a positive effect upon synthetic efficiency and overall productivity, can also decrease waste streams, lessen energy input, and minimize the need for hazardous reagents”.

It is only human nature that academic scientists who have hitched their career aspirations to green chemistry will advocate the greenness of a process as the essential basis for process evaluation; but realistically all that can be hoped for is that industrial process chemists will increasingly include green technology and its potential advantages to the corporation among all the basis upon which a preferred process is selected. There is, however, good reason to expect that this will happen.

As the GEN article notes realistically, ‘This may be a particular challenge in organic chemistry, in which many of the classical chemical reactions have been in use for decades to produce small molecule drugs”.

The industry does have an advantage in this re-education program since at this time there is a particularly rapid turnover of chemists because so many of them are reaching retirement age. The ranks must necessarily be quickly filled up with younger more environmental conscious and hopefully trained in universities very aware of the need for green chemistry.

Because of the patent laws and health regulations in the main, world pharmaceutical markets, the emphasis in the pharmaceutical industry must be on the synthetic route which is overall the most rapid, inexpensive, scaleable and rugged. The high risk of failure and the cost of even a successful development necessitate that pharmaceutical companies maximize the number of years that they have a monopoly in the marketplace before patent expiry. This forces these organizations towards using the first dependable, economic and scaleable process that will deliver a product that can satisfy the health regulatory authorities. Once product from this early process has been used in successful clinical trials, there are bureaucratic hurdles created that make it very expensive to “green up” steps of the process or to replace whole portions of the process with ecologically more friendly alternatives. Pharmaceutical companies will never introduce process retrofits that require the repeating of clinical trials because these human tests are the most expensive part of drug development.

Green chemistry will typically be unsuccessful as a retrofit strategy for old products but they can be immediately introduced into all new processes. All that is required is awareness that it is wanted. When a process step is optimized, the entire reaction space is not explored. Decisions are made at the beginning of the project concerning the values of discrete variables (variables that cannot change continuously like time or temperature) that will be included in the optimization study. An example is useful to understand this. Suppose in a process step an alcohol is converted to an acetate ester. Only a few acetylating agents will be chosen to examine: say acetic anhydride and acetyl chloride. If transesterification with ethyl acetate or isopropenyl acetate is not included at this stage, no transesterification process has any chance of becoming the optimized method. To exemplify again, an auxiliary base may be include in the optimization study and the candidates will be preselected: say pyridine, and triethylamine. If a recyclable base such as polyvinylpyridine is not included among these discrete variables in the study, it has no chance of becoming part of the optimized process step. If chloroform or methylene chloride are included among the discrete solvents selected for the optimization, they have a good chance of becoming part of the optimized process step. If the choices at the outset for building the process step are greener, the resulting optimized step must be greener. It is only when this ‘greener’ study cannot come up with an adequate process step that chemists may need to fall back on the traditional ungreen methodologies.

In the GEN report, David Constable of GlaxoSmithKline teaches that the cost of new solvent, combined with the cost of disposal of both the used solvent and the water contaminated with organic solvent constitute about 75-80% of the environmental impact and energy use in the life cycle of a pharmaceutical compound. If Kilomentor translates this into actionable terms, it means: learn the effect the solvent/substrate ratio upon yield and purity during process optimization.

The selection of the reagents, catalysts and solvents to be tested in the optimization of new process steps is going to be done within industry. The work is being performed on drug intermediates that are secret at the time the work is going on. Where academia can contribute to the green revolution is to provide a literature, which suggests green alternatives that industrial project manager can use in their optimization studies. Publish a more atom economic acetylation. Publish and popularize a readily recyclable trap for hydrogen chloride gas to replace triethylamine. Put it in your text books. Teach it in your lectures. Make it top of mind.

The GEN article reported that David Constable of GlaxoSmithKline iinformed a conference that in the past five years his company had achieved a 50% per kg API reduction in use of methylene chloride for final processing, where it had been replaced with either methyl isobutyl ketone or ethyl acetate. He also reported a 95% decline in DMF use.

The GEN article also mentions the solvent tetrahydrofuran as a solvent of concern to EPA for emission of toxins to air and water. Tetrahydrofuran is an important solvent for organometallic reagents and 2-methyltetrahydrofuran appears to be a good replacement. This needs to be made more widely known. 2-methyltetrahydrofuran is not completely miscible with water and so can be recovered more easily.

Atom economy is being popularized as an important aspect of green chemistry. Atom economy measures the amount of a reagent that gets incorporated into the final product versus the portion that goes into waste. For example in making an acetate product from an alcohol using Ac-OAc, the Ac goes to make the acetate product and the OAc goes into waste. Using AcCl, the Ac goes into product, the Cl goes into waste. The goal of atom economy is to reduce the percentage weight that goes into waste for any reagent.

It is important for process chemists to understand the concept of atom economy but Kilomentor’s assessment is that the idea is overhyped. Reagents typically contribute very little to the waste burden compared to solvents, and reagents are very much more likely to be critical to the particular reactivity of a substrate because the reagent is a very important part of the activated complex for the reaction transition state. The result is likely to be that a small positive influence of reagent structure is going to be difficult to counterbalance by arguments about relative atom economy. If a less atom efficient reagent gives a more cost effective process, the latter will win out.

Neal Anderson of Anderson’s Process Solutions identifies oxidation/reduction reactions as among the least atom economic which is rather automatic given that most oxidations or reductions only add or remove a couple of hydrogens. Kilomentor has already provided a blog discussing the more promising oxidation technologies, including a method using only catalytic amounts of chromium. Anderson’s good suggestion to design processes avoiding unnecessary changes in oxidation state is a worthy repetition of what he taught in his superb book, Practical Process Research & Development, Academic Press. , This book is still the best process chemistry book available.

An earlier rendition of this blog was cut short by an electronic glich.

This replaces it. Kilomentor

There are drug substances do not contain a functional group that can form a stable salt but many others do. Drug discovery chemists frequently plan to incorporate a salt forming functional group into their candidate structures because making pharmaceutical salts can modulate the critical bioavailability a successful drug product.

Because finding highly preferred salt forms of drug candidates is a frequent undertaking, efficient protocols for identifying preferred compositions are in place among the firms that search for new drugs. Many of the steps in the screening have been automated. The evidence makes it difficult to argue against the proposition that the tools, steps and essential considerations for deciding upon the best pharmaceutical salt candidates are well known to skilled salt selection practitioners and are taught in the primary and secondary literature for all who are interested.

P. Heinrich Stahl and Camille G. Wermuth have edited a book, Handbook of Pharmaceutical Salts: Properties, Selection and Use. International Union of Pure and Applied Chemistry, Wiley-VCH 2002 hereafter H&W, which bring together a great deal of material about pharmaceutical salts already in the literature, particularly in patents.

The editorial stance however may be a little annoying to readers from the generic drug industry because the authors imply an exaggerated idea about the difficulties encountered in making them. In particular at pg. 250 they write, “The preparation of pharmaceutical salts is usually not a matter of university teaching, and so most of the organic chemists are not trained to prepare salts.” Taken strictly literally what they say is true but I do not think the authors’ purpose is sarcasm. The authors are implying a requirement for inventive ingenuity only accessible to graduates when in reality with perhaps rare exceptions, preparing pharmaceutical salts is too simple to be the subject matter of university teaching.

Pharmaceutical salts typically are more soluble and more rapidly soluble in stomach and intestinal juices than non-ionic species and so are useful in solid dosage forms. Furthermore, because their solubility often is a function of pH, selective dissolution in one or another part of the digestive tract is possible and this capability can be manipulated as one aspect of delayed and sustained release behaviours. Also, because the salt-forming molecule can be in equilibrium with a neutral form, passage through biological membranes can be adjusted.

It is true that selection patents for particular pharmaceutical salts are used to extend the monopoly on many important medicines even though it is difficult to imagine the inventive step in the development of these salts. In the patent literature a great fuss is made about the millions of possible permutations of process variables that may need to be explored in order to devise a practical procedure for making a particular salt. Indeed, there are unusual cases where making any pharmaceutical salt turns out to be difficult. In such an instance, after the exhaustive trials, which would be the result of such an instance, it would be a simple matter to document, the difficulties and to justify a patent for the solution to that particular problem. In general however, once a compound is known, its pharmaceutical salts become readily available without further inventive steps to persons of ordinary skill in the art.

The work-horse cited document concerning pharmaceutical salts is S.M. Berge, L.D. Bighley, D.C. Monkhouse, J. Pharm. Sci. 1977, 66, 1-19. This work listed the pharmaceutical salts from which a pragmatic choice might be made. This work was updated by L.D. Bighley, S.M. Berge, D.C. Monkhouse, in “Encyclopedia of Pharmaceutical Technology’. Eds. J. Swarbrick and J.C. Boylan, Vol. 13, Marcel Dekker, Inc., New York, Basel, Hong Kong 1995, pp. 453-499. In this most recent compilation they found 113 different anions (13 inorganic) and 38 different cations (11 inorganic). About 75% of the basic drugs had been combined with one of just eight anions: chloride, sulfate, bromide, mesylate, maleate, citrate and phosphate. About 50% of all pharmaceutical salts were just hydrochlorides. There was an even more significant concentration for cations with acid drugs. Nearly 90% of the pharmaceutical salts were made with sodium, calcium, potassium, magnesium, meglumine or ammonia with more than 55% made with sodium.

An important point that Stahl and Wermuth’s book brings out is that finding an appropriate pharmaceutical salt has become easier because the choices are to-day more limited. Referring to the Berge, Bighley and Monkhouse references the statement is made on pg. 331.

“While these authors presented the results of a survey on the approval status of drug salts 25 years ago, the present day situation is different. Accumulated knowledge and experience has led to a reduction of the number of acids and bases regarded as innocuous. Moreover, national health authorities reacted in different ways to certain findings in the area. Therefore, it was deemed timely to put up a revised list of useful salt-forming acids and bases.

In the following tables, an attempt has been made to group the salt-forming acids and bases into classes of first, second and third choice. The following criteria for assignment to the respective classes were applied.

1. First Class salt-formers are those of unrestricted use for that purpose because they form physiologically ubiquitous ions, or because they occur as intermediate metabolites in biochemical pathways. The first group is typically and quite impressively represented by the past and present use frequency of hydrochloride/chlorides and sodium salts. The second group comprises many acids present in food or vegetable origin, or those generated in the body’s metabolic cycles.

2.Second Class salt formers are considered those that are not naturally occurring, but, so far, during their profuse application have shown low toxicity and good tolerability.

3. Third Class salt-formers might be interesting under particular circumstances in order to achieve special effects such as ion-pair formulation, or for solving particular problems. some of them are assigned to this class because they have their own pharmacological activity. Also some of the acids and bases were used much less frequently in the past….

…It is recommended to search for the latest safety records in the RTECS inventory and in literature at the time when a Class 3 acid or base would be considered for salt formation with a NCE.”

There are just 30 First Class and 27 Second Class acids listed. There are only 9 First Class bases and 10 Second Class bases listed.

The First Class Acids are alphabetically: acetic acid, adipic acid, L-ascorbic acid, L-, capric, carbonic, citric, fumaric, galactaric, D-glucoheptanoic, D-gluconic, D-glucuronic, Glutamic, glutaric, glycerophosphoric, hippuric, hydrochloric, DL-lactic, lauric, maleic, (-)-L-malic, phosphoric, sebacic, succinic, sulphuric, (+)L-tartaric, and thiocyanic. Glycolic aspartic, palmitic and stearic are First Class acids also but they are used almost exclusively to make ester derivatives which are actually pro-drugs. Glycolic acid is used to make ether pro-drugs not a pharmaceutical salt per se.

The Second Class acids are alphabetically: alginic, benzenesulfonic, benzoic, (+)camphoric, caprylic, cyclamic, dodecylsulfuric, ethane-1,2-disulfonic, methanesulfonic, ethanesulfonic, 2-hydroxy-, gentisic, 2-oxo glutaric, isobutyric, lactobionic, malonic, methanesulfonic, naphthalene-1,5-disulfonic, naphthalene-2-sulfonic, 2-napthoic 1-hydroxy, nicotinic, oleic, orotic, oxalic, pamoic, propionic, (-)-L-pyroglutamic and p-toluenesulfonic acids.

The First Class acids, which are also among the most frequently used 15 acids are: hydrochloride, sulfate, tartrate, maleate, citrate, phosphate, acetate, lactate, and fumarate. Those which are not First Class acids but are among the top 15 salt formers are: hydrobromide (3), mesylate (2), pamoate (2), hydroiodide (not listed), nitrate (3), and methylsulfate(not listed). The class is listed in brackets. The pamoate salt is frequently quite insoluble in water. It finds particular use in making sustained release formulations. It also can be used to make quite insoluble salts to dibasic materials. Nitrate salts in former times were popular but are now recognized to have their own physiological effects and so are unlikely to be accepted today. S&W states at page 298 that the nitric acid salts should no longer be considered for formation of salts for internal use. Methyl sulfate salts are exclusively salts of quartenary ammonium ions with at least one methyl. The salt is created by methylation of the tertiary amine with dimethylsulfate. Kilomentor could find no other structures in which it was the pharmaceutical salt form.

What is evident from this is that there are only 9 acids which are both First-Class and in the top 15 historically. Among the top 15 acids there are some used exclusively in special situations and which need not be considered at all for regular screening applications.

Aspartate is characterically used to make salts with other amino acids. Kilomentor found no salts of drug substances.

Glycolic acid is not used to make pharmaceutical salts; covalent ether derivatives have been made to improve water solubility.

Palmitic and caproic acids are used only to make steroid esters.

D-glucoheptanoic: bisguanidine sebacic stearic

There are other sources of advice on preparing crystalline salts of complex basic substances. R. H. F. Manske writing about the isolation of alkaloids in sources of Alkaloids and their isolation, wrote at pg. 12,

“Should both fractional crystallization and distillation fail [to get the crystalline free base] in the resolution of these mixtures then they may be converted into any one of a number of salts in the hope that one of the component salts may be insoluble. There are a number of cases where certain special salts crystallize remarkably well but preliminary trials should be limited largely to the use of such acids as hydrochloric, hydrobromic, perchloric, picric, and oxalic, although sulphuric acid frequently affords acid or neutral sulfates that are sparingly soluble in alcohol or water. Instead of aqueous hydrochloric or hydrobromic acid absolute methanolic solutions of the reagents are recommended, since methanol is a good solvent for many bases. The methanol solutions offer the added advantage that the excess hydrogen halide is readily removed by precipitating the salt with an excess of dry ether. Hydrochlorides, thus prepared, often crystallize readily from boiling acetone, or acetone containing just enough methanol to effect solution”.

It must be born in mind that Manske is trying to get one alkaloid to precipitate from a mixture of alkaloids and he is not constrained to making pharmaceutically acceptable salts. this is why he advocates perchloric, picric and oxalic acid but his recommendations of other preferred salts that are pharmaceutical and his solvent recommendations based on a massive alkaloid experience are worth noting.

Hydrochloric acid

Hydrochloride salts frequently exhibit less than desirable solubility in gastric and other physiological fluids because of the common ion effect. Because hydrogen chloride is a volatile gas, salts with weak bases may lose acid over time when combined with weak bases. Hydrochlorides can be corrosive to machine surfaces, when somewhat hygroscopic.

Sulfuric acid

Sulfuric acid can make two kinds of salts a sulfate and a bisulfate. The second pKa is 1.92. Hélène Perrier and Marc Labelle in J.Org. Chem. 1999, 64, 2110-2113 had the goal of choosing a salt form to be used to precipitate or crystallize a large number of different substrates whose only common feature was the presence of a quinoline base. Their first choice was the bisulfate salt using a standard procedure or a modification of it. There procedure was for precipitating the quinoline substrates from a reaction mixture in one of the solvents:ethyl acetate, methylene chloride, chloroform, dimethoxyethane, acetonitrile, dimethyl formamide, methanol, ethanol, and tetrahydrofuran. A solution in one of these solvents was diluted to 0.2 molar with ether and one equivalent of sulphuric acid was slowly added with vigorous stirring. With a few exceptions this produced a solid phase. Difficulties were experienced with compounds dissolved in DMF or alcohol solvents. This problem was solved in two different ways. In a first procedure, an extraction method was applied where the mixture was diluted with an ethyl acetate-water mixture, the organic phase was separated, and the compound was precipitated from that phase after dilution with ether. a second procedure applied to DMF simply involved a 4-fold dilution with methylene chloride (from 0.5M in DMF to 0,12M) followed by the standard ether dilution to 0.08M and acid precipitation.

(+)-L-Tartaric acid

The pharmaceutical form, (+)-L-tartaric acid has pKas of 3.02 and 4.36. A mixture of forms might be formed in bitartrates. Tartrates as a group show augmented solubility. where solubility is a problem the tartrates may be candidates for solution for the problem. A problem might arise using L-tartaric acid as the counterion with a racemic drug substance, because partial resolution might occur by selective crystallization of one enantiomer of the API. Among the compounds in USAN 1993, only metraprolol is a racemic free base. The other partners were either single enantiomers or achiral. There seems to be a preference for the stronger bases as partners of tartaric acid such as guanidines, amidines, thiuronium (in furazolium tatrate) and the predominant form is the hydrogen tartrate (1:1 stoichiometry). Tartrates are also more frequent when the basic structure contains alcohol and phenol as additional functionality. Kilomentor hypothesizes that there may be other hydrogen bonds between cation and anion. The amine functionality can be without other hydrogen bond donors such as in the compounds:ditrimeprazine , phendimetrazine and altanserin (all tert-alkylamines).

Maleic

Maleic acid has two pKas 1.92 and 6.23. They are distinctly different because of the rigid structure which holds the first anion close to the site of the second deprotonation. For comparison the pKas of the geometric isomer fumaric acid are 3.03 and 4.38. In arecent report maleic acid could be made responsible for acute tubular necrosis in dogs after a single peroral dose of a test substance supplied as a maleate (pravadoline maleate) corresponding to a dose of 9 mg/kg maleic acid. [R.M. Everett, G. descotes, M. Rollin, Y. greener, J.C. Bradford, P.D. Benziger, S.J. Ward, Fundam. Appl. Toxicol. 1993, 21, 59-65.]

Maleic acid as a counter ion can be reactive with nucleophilic primary and secondary amines when heated strongly together or for an extended duration. The amines can undergo a Michael addition to the activated double bond. Nucleophiles can open any small amounts of maleic anhydride that might form making a conjugate with the maleic acid. These problems are more frequently encountered in the preparation of the API itself.Because maleic acid is a diprotic acid, there is the possibility of producing chains of cations and anions associated together by reaction with free bases that have more than one basic site. In fact examination of the compositions in the USAN 1993 that form salts with maleic acid 26 of them have a second basic site at least as basic as pyridine and 23 of them are effectively mono basic APIs. Although the second pKa of maleic acid is not going to protonate something like a pyridine substructure, there is a good chance for a strong stabilizing hydrogen bond.

Citric acid

Looking into ASAN 1993 to see the structures of the free base form of APIs that form citrate salts there is no primary or secondary amine in any of the structures. Each structure has a tert-alkylamine with occasionally a n additional aryl heterocyclic amine. The pipeazine substructure is frequent. Kilomentor would not recommend trying to make a citrate salt with an organic base containing any hydrogens on a basic amine functionality. Citric acid binds magnesium and calcium ions, which may appear in the formulation excipients. Because it complexes polyvalent metals which can operate catalytically, citric acid may have some antioxidant properties

Fumaric acid

Fumaric acid has both its pKas close together: 3.03 and 4.38. Because the pKas are close together, a mixture of 1:1 and 2:1 salts is possible. The same concern about Michael addition impurities arises as it did with maleic acid.

Phosphoric acid

Phosphates of aliphatic sec-and tert- amines and of heterocyclic bases are likely to exhibit low water solubility but phosphoric acid is a syrupy liquid and is difficult to work with. In addition, phosphates have a tendency to form hydrates. Perrier and Labelle considered phosphates the second best salt to consistently precipitate from organic structures containing the substructure quinoline.

Kilomentor thinks that it is important to point out that simple salts of dihydrogenphosphate mono anion are actually rare. Clindamycin, metronidazole, rosaramicin, etoposide, fludarabine, tricirabine phosphates are actually phosphate esters. Other phosphates are often disodium phosphate esters. Where regular phosphates have been selected as a preferred pharmaceutical salt, the API is almosr always a structure with two or three basic groups, for example, clomacran (2 groups), chloroquine (3 groups), venpiroline (3 groups), primaquine (3 groups),disopyramide (2 groups) or histamine (3 groups). Usually one of these basic groups is a heterocycle. Only octryptoline is monobasic from among the drugs in USAN 1993. Klomentor recommends that phosphate salts be preferentially attempted only of substrates more than monobasic or that contain the quinoline substructure tested by Perrier and Labelle.

It may be that phosphates are insoluble in aqueous organic media. Easily handled sources of phosphate may be mono and dibasic ammonium phosphates: NH₄ H₂PO₄, (NH₄)₂ HPO₄. These compounds have good water solubility and the former has some alcohol solubility. In the presence of stronger less volatile bases, it may be possible to drive out the ammonia.

Acetic acid

Because the acid is a weak one good salts are only formed with strong bases. The free acid is a liquid and excess can easily be removed. It is volatile again explaining the need for a strong base partner to keep its stoichiometric integrity. The low molecular weight could be useful in high load solid dosage forms where the size of the drug product could be an issue.

Lactic acid

Both (+)-L-Lactic acid and racemic (±)-DL-lactic acid can be used for salt formation as the enantiomers of lactic acid are interconvertible in biological systems. The pKa is 3.86. Although the solids are known they are most readily available as aq. solutions. It is reported [P.H. Stahl, Ciba-Geigy AG, Basel, Switzerland unpublished ] that otherwise sparingly soluble and weak bases can be advantageously dissolved with these acids. Aqueous lactic acid is a complex solution with varying amounts of oligomeric esters present such as lactoyllactic acid depending upon the concentration and age of the solution. This may make the preparation of pure salts difficult not just in the crystallization but in stoichiometric preparation. Pure (+)-L-lactic acid should be used to form salts with a chiral base.

Methanesulfonic

Although it is not a First Class acid, methanesulfonic acid is among the top 15 acid salt formers. It deserves special consideration because is a strong acid with a low molecular weight and excellent aqueous solubility properties Methanesulfonic acid has the advantages of a low molecular weight, a high acidity and it is a liquid miscible with some solubility in organic solvents even as non-polar as toluene as well as being totally soluble in water and a liquid at ambient temperatures. It can be obtained inexpensively in an anhydrous form. There has been a warning to be careful about the possible formation of methyl, ethyl or isopropyl mesylate from the use of the acid in these alcohol solvents. The main risk is from small amounts of methanesulfonyl halide in the acid that can react with alcohols. Methanesulfonate salts have no tendency to form hydrates.

Among basic salt-forming substances as designated in the First Class bases are alphabetically: ammonia, L-arginine, calcium hydroxide, choline, N-methylglucamine, lysine, magnesium hydroxide, potassium hydroxide, sodium hydroxide.

Among basic salt-forming substances the Second Class bases are alphabetically: Benethamine, benzathine, betaine, deanol, diethylamine, 2-diethylaminoethanol,hydrabamine, 4-(2-hydroxyethyl) morpholine, 1-(2-hydroxyethyl)- pyrrolidine, and tromethamine.

Before moving on to discuss the most common salt forming bases Kilomentor thinks it might be useful to provide some teaching about how to best obtain the free base form from the most common salt form the hydrochloride.

Recovering the Free Base Form from the Hydrochloride Salt

So predominant is the hydrochloride salt among pharmaceutical salts that it is useful to know the easy methods to regenerate the free base from the hydrochloride. The most frequent and least expensive method is to mix the hydrochloride in a mixture of water and a water immiscible organic solvent to which aqueous alkaline sodium hydroxide is added to neutralize the hydrogen chloride irreversibly. The free base is extracted into the organic phase where after optional drying it is recovered.

Sometime there is some reason to neutralize without contact with water. In the laboratory this can be done very simply by passing an organic solution of the hydrochloride through a plug of basic alumina. The material eluted will be the free base. The acid is retained by the adsorbant. On larger scale the hydrochloride salt is reacted with an equivalent of the epoxide of propylene. The 1-chloro-2-propanol can be removed by evaporation.

Example Belgium Patent 775,082 May 9 1972 F. Hoffmann-LaRoche

Insoluble anion exchange resin in the hydroxide form can also be used to neutralize hydrochloride salts. The excess resin can be filtered for removal. These resins however typically contain some residual water.

An ammonium salt upon evaporation to dryness and/or drying under vacuum hydrolyzes and the ammonia can be removed leaving the free acid.

Today’s Kilomentor blog pertains to a laboratory technique, which is particularly appropriate at a scale of 5-10 litres where it becomes problematic to pick them up and pour from them. It is sometimes necessary either to filter a solution without sucking the solvent dry and exposing the slurried solid or to draw away the liquid phase from a reactor flask to leave behind the solid phase. This can be particularly useful if the solid is sensitive to the air or moisture in the atmosphere. it can also be used when the slurry material would block the filter, make the flow impractical, and prevent a filtration from being completed. Removing a liquid phase from a vessel to leave the solid inside is called inverted filtration.

An apparatus for inverted filtration is described in Organic Synthesis Coll. Vol. 2 1966. A figure is printed on pg. 610. The solution is drawn up from the bottom of the reactor through a tube using gentle suction and carried over through the tube to a second vessel. A filter entirely encloses the end of the tube that is inserted into the reaction mixture. The filter is prepared from an extraction thimble and a rubber stopper, which has a single hole drilled through the top to accommodate the glass tube. The thimble is stuffed with glass wool to hold the thimble centered around the glass tube.

During the reaction the filtering assemblage is held above the reaction solution but inside the reactor. When the time for the inverted filtration arrives, the assemblage is pressed down by pressing the glass tube into the solution sliding it further down on the stopper which blocks a neck of the reactor flask. When the filter assembly is at least partially dipping into the solution, clear liquid will pass through the extraction thimble By the application of pressure to the reactor or gentle suction on the tube (by way of the filtrate receiving flask) the liquid which has been filtered through the extraction thimble will be forced up the tube and over into the receiving flask. Thus only liquid, which passes through the filter is transferred.

A reaction may proceed quite well to give an 80% yield of the desired product but still be very difficult to work up when it is a mixture of neutral compounds. In this situation acid-base extraction cannot help to obtain some partitioning between organic and aqueous phases. Furthermore, most often the two compounds making up the reaction mixture are both essentially insoluble in water. When there is 20% by weight of an impurity, even when you can find a solvent which gets the major compound to selectively crystallize, the recovery is usually quite poor simply because by the time you have crystallized 60% of the product, the mother liquors are a 1:1 mixture of desired and undesired compounds. At this point the rate of crystallization normally becomes impractically slow and for practical purposes the crystallization has stopped.

Usually thin layer chromatography in more than one solvent system can quickly tell you whether the main impurity, which most probably is the one blocking the crystallization, is, by-and-large, less polar or more polar than the desired major component. When the minor component is the more polar, what we intuitively would like to do is triturate with water, modified so that it can dissolve more of the mixture, hoping that the additional material dissolved into the water rich phase will be disproportionately the more polar impurity component.

A cosolvent for water to be effective must prefer to mix with the water rather than forming an oil phase with the products. Only experimentally can we find something guaranteed to work, but perhaps kilomentor can propose a rule of thumb, which could increase the likelihood of success. This aqueous phase modifier should be completely miscible in all proportions with water. If a diluent is only partially miscible with water it is more likely that when mixed with the neat reaction oil it will simply migrate into the oil.

The most lipophilic solvents that are completely miscible in all proportions with water are: acetone, methyl ethyl ether, methyl acetate,and t-butanol. The lower homologues of each of these function group types will also be completely miscible. That is: methanol, ethanol, propanol, isopropanol are also completely miscible and could be used as diluents. For esters, ethyl formate is not completely stable in water so it cannot be used. Acetonitrile is completely miscible but propionitrile is not. Nitromethane is not completely miscible, while Dimethylformamide, N-methylformamide, formamide, DMSO, pyridine, and acetic acid are.

In addition to adding small quantities of these solvents to a large excess of water to increase the leaching power of the polar phase, recrystallization from the less polar of these at least: acetone, t-butanol, pyridine or methyl acetate by the gradual addition of water could be fruitful.

Once the level of the impurity is reduce below 10% from the 20% range, crystallization in general can be expected to give a superior recovery. From a mixture containing just 10% impurity one could crystallize 80% before the mother liquors would be 50:50 product : impurity.

Even on scale a reaction mixture can be freed of organic solvent by concentration in the presence of a water phase to give a reaction product oil as an oil in water. The aqueous phase modifier could be added into this mixture.

Kilomentor has already written about steam distillation. Steam distillation is however just a special case of the technique of co-distillation. What makes the steam case unique is that the substance that is being co-distilled with the components of interest is water, which is practically free. Therefore, if one needs to distil a large amount of water making a lot of waste water, in order to recover a small amount of product the cost of materials is still not high. On scale any co-distillate liquid can readily be recovered and does not need to be purified before it is recycled.

The other essential criterion for a liquid used in co-distillation is that it must be easily separated from the compounds you are interested in purifying. Usually this means that the substance you are trying to purify must have a low solubility in the co-distillation liquid. Another usual requirement is that the liquid have a high, but not inaccessible boiling point, which at the same time not higher than the desired material. Kerosene and nujol (paraffin) fractions would qualify because many compounds are essentially insoluble in hydrocarbons. Silicone oil is also a possibility. This particular possibility came to my attention once when I had to recover a compound that had been mixed into a heating bath when a flask accidentally broke. I realized how easy it was to recover the lost product by extraction from the silicone oil. At the other end of the polarity range solvents like ethylene glycol, propylene glycol and glycerine can be used as codistillation liquids for high boiling non-polar constituents.

With these codistillations the liquid is not typically totally vaporized and injected into the still pot as is done in steam distillation but more simply a mixture of the liquids is heated to reflux in the normal way used in regular distillation. As the distillate is collected fresh (or recycled)co-distillation liquid is added into the still pot by some addition device.

What is necessary for efficient co-distillation is to wrap the vapour path to the condenser well and to try to supply supplemental heating. Fractionation is not possible in co-distillation and for energy conservation reasons one does not want to vaporize the distillate more than once in its journey to the intended condensing surface.

If all that is needed is a higher temperature than is provided by the typical steam distillation, then you may be better served by superheated steam distillation where the water vapor is kept at a temperature distinctly different from 100 C. Vacuum steam distillation will also serve to purify very high boiling compounds.

Keywords: polymorphs, polymorphism, solvates, hydrates, crystal habits, digestion, flowability, powder mixing, dissolution, solubility, bioavailability, API

Specifying a three dimensional connectivity table for a chemical substance does not specify a single physical form of a substance. Such a uniquely bonded covalent molecular array will very often order itself in multiple ways in the solid state. This is often but not always related to different conformations (rotational isomers) any one of which can end up being the major conformer when the covalent substance is packed into a crystal lattice. Such different physical forms are called polymorphs if they have the same three dimensional connectivity and the same elemental analysis but different powder x-ray diffraction patterns.

Synthetic chemists from predominantly academic backgrounds, when they begin to regularly prepare organic substances in hundreds of grams or more, often see but do not recognize the significance of different physical properties for solids of the same structure. Most often these differences arise from different polymorphs that have crystallized in different crystal habits. Although these differences are not significant in terms of the success of a project as defined in synthetic terms, they are tremendously important in sao far as formulation difficulties are concerned, when the product is a pharmaceutical product.

Kilomentor vividly remembers such a situation in the first project he took into the plant. The first intermediate when produced on scale precipitated either as a smooth mud that took many hours to filter or (less often) as a coarse sandy material that seemed to filter in minutes. Although which one was obtained in the laboratory was of minor importance, the plant operators you can imagine had a strong preference!

It was once thought that the melting point of a solid was an invariant characteristic of a particular covalent atomic arrangement (molecular structure) but the existence of polymorphic forms shows that this is not true. Different polymorphic forms of the same basic molecular structure can have different melting points. Very often however when a melting point is being determined by watching the behaviour of a solid in a melting point tube, two polymorphs will appear to have the same melting point when they actually do not, because the lower melting form may t convert to the higher melting form, without the observer detecting it, during the melting point determination. Or sometimes the two polymorphs may have different melting points but which are very close to each other.

Crystal Habits

If two samples have the same three dimensional covalently bonded array and the same powder x-ray diffraction (XRD) pattern and the same elemental analysis, but look different; then, at the unit cell scale the two substances are the same but they are said to have different crystal habits. Crystal habits are characterized by the relative dimensions of the macroscopic crystal forms. For example, a substance may crystallize as needles (essentially in a one dimensional line), plates (in a two dimensional plane), or as three dimensional rombahedra. A crystal habit difference occurs when two or more faces of the same crystal class grow at different relative rates. This is a macroscopic difference in relative dimensions not just the difference between large and small crystals of the same overall shape (ie large and small needles).

It is a well known teaching from inorganic gravimetric analysis that if a solid is too fine to allow rapid quantitative filtering this condition can often be improved by what is called digestion.

For example, in A Textbook of Quantitative Inorganic Analysis including Instrumental Analysis, Arthur I. Vogel, Third Edition, John Wiley and Sons, New York. N.Y. 1961, at page 111-112 there is the teaching:

“This [digestion] is usually carried out by allowing the precipitate to stand for 12-24 hours at room temperature, or sometimes by warming the precipitate for some time, in contact with the liquid from which it was formed: the object is, of course, to obtain complete precipitation in a form which can easily be filtered. During the process of digestion or the ageing of precipitates, at least two changes occur. The very small particles, which have a greater solubility than the larger ones, will, after precipitation has occurred, tend to pass into solution, and will ultimately redeposit on the larger particles; co-precipitation on the minute particles is thus eliminated and the total co-precipitation on the ultimate precipitate reduced. The rapidly formed crystals are probably of irregular shape and possess a comparatively large surface; upon digestion these tend to become more regular in character and also more dense, thus leading to a decrease in the area of surface and a consequent reduction of adsorbtion. The net result of digestion is usually to reduce the extent of co-precipitation and to increase the size of the particles, rendering filtration easier”.

It is well known that pronounced variations in the crystallization conditions: temperature, rate of temperature change, intensity of stirring, the initial level of super-saturation, solvent type and polarity, water content, the type and concentration of impurities (particularly structurally related impurities), solute concentration and the solution viscosity all can change crystal habit. Further complicating the crystallization operation, many of these factors vary as the crystallization proceeds. Crystal habits probably will not affect solubility, dissolution rate or bioavailability. Crystal habits can be important for the flow properties of powder mixtures, but as skilled practitioners know, problems in powder flow can be addressed by forming the powder into lumps (called granulation) or by pressing, grinding, micronizing or other well known mechanical aggregation or disintegration methods.

The core factors that affect crystal habit also affect the crystal size because they modify differently the rates of crystal nucleation and crystal growth. Synthetic chemists typically are most experienced in the wide variety of conditions that may promote crystal nucleation, because without any crystal nucleation an otherwise solid product remains a troublesome oil. The optimal crystal nucleation temperature is rarely the best temperature to increase the rate of crystal growth. That is why on laboratory scale, crystallization is often promoted by alternately raising and lowering the temperature or having different parts of the oil at different temperatures and/or stirring and scratching with a glass rod to create discontinuities on the vessel’s wall where nucleation has a better chance to begin.

Hydrates and other Solvates

Two or more chemical substances can also crystallize together in an organized relationship within the crystal lattice as co-crystals. This is much more common than is commonly realized because all racemic compounds are co-crystals of the two enantiomeric (mirror image) forms. Co-crystals wherein one of the chemical species is a volatile substance are called solvates. Hydrates are just a special subclass of solvates where the solvent is water.

A synthetic process chemist who prepares a three dimensional covalent structure different from the target structure has failed in the project. If the skeleton and stereochemistry are correct the synthetic organic process chemist has succeeded no matter what polymorph, solvate or hydrate is recovered from the final synthetic step. This is because polymorphs are routinely and simply interconverted and solvates and hydrates are readily desolvated. In the case of solvates or hydrates this is usually done by some combination of vacuum, heat and chemical reaction either alone or severally. The use of dehydrating agents is one common example of this.

Although polymorphs, solvates and hydrates are rather unimportant to the synthetic chemist, they are very important to formulators who work to make pharmaceutical dosage forms like tablets, powders or capsules and or to patent chemists who try to create intellectual property that provides a legal monopoly for pharmaceutical companies. Although polymorphs can be found by applying routine screening strategies, patenting these new polymorphs of medicinally importance compounds can extend the legal monopolies of the ‘inventors’ by a dozen years or more. The anti-cholesterol drug, atorvastatin, first discovered by Pfizer, is the most prescribed medicine in the world; there are 23 known polymorphic forms, most of which have been patented.

Although the greatest importance of polymorphs is that they can be used to extend pharmaceutical patent monopolies, the differences between polymorphs, hydrates and other pharmaceutically acceptable solvates can sometimes actually be important when these forms are incorporated with excipients into a drug product such as a tablet or capsule. One crystalline polymorph might formulate to produce a stable suspension while another might deteriorate on storage. A case is known where a polymorph is claimed to have up to ten times the absolute solubility of another and this can affect the bioavailability. Different polymorphs have different tendencies to retain solvent and this can be important for the removal of impurities during the washing of a crystalline API. Different polymorphs of a particular pharmaceutical can have different tendencies to be created in different crystal habits and crystal habit and crystal size are key determinants of the flow properties and manufacturability of API in solid powder mixtures, although poor flow as has already been noted can be changed by mechanical processing.

In summary there are, in a minority of instances, significant advantages to using a particular polymorph in a pharmaceutical product, but usually the claims to their importance are really about monopoly patent rights. Moreover, discovering polymorphs does not require ingenuity or inventiveness. The literature contains loads of suggestions for simple crystallization conditions that can give rise to polymorphs. It has been said that the number of polymorphs of a pure substance is probably directly proportional to the time spent looking for them. There are even automated robotic systems that can be used to search for polymorphs. No wonder that scientists that author polymorph patents don’t subsequently publish their work in peer journals. It’s not creative, not surprising and not unusual. It’s not work you can expect admiration for doing.

Getting crystals to consistently form with a chunky crystal habit on the other hand might require some if the solid did not give you what you want simply by old fashioned digestion. Avoiding needles and plate morphologies really can help to avoid demixing of the powder mixture of active pharmaceutical and the inactive excipients when it is flowing into the punches of a tablet press. The problem here is that the problem can also be overcome most of the time by granulating (lumping) components or conversely by grinding chopping and sieving them.

Solvates are discovered auromatically during the search for polymorphs. all one needs is a proper characterization of the solid that is isolated. A thermogravimetric analysis, an NMR, a n elemental analysis and a weight loss on drying. That is just careful classical measurement of properties.

What is the impurity from Hell? It is the impurity which probably wasn’t present in the early samples of the laboratory synthesis, but which appeared during the process modification, development or optimization and which cannot be removed by all the normal purification methods without a large loss of the final product.

Can we deduce anything generally true of this worst kind of impurity? The impurity does not have a difference in functional groups. The difference from the main constituent is not important for the stereochemistry of the ring structure. It probably is a structural isomer or homologue of some hydrocarbon or at least apolar substituent, which most likely is conformationally flexible or floppy. The reasons I would offer these hypotheses is because the difference between the impurity from hell and the pure substance is not substantial and the impurity and the desired substance probably fit into the same crystal lattice because the crystal allows some disorder in this side chain. The difference is more likely quite far away from any functionality so that it cannot change the functional properties because of its inductive or steric effects.

Where do these impurities most likely come from? Not from preparative by-products, but from impurities in the starting materials, I would postulate. In particular, most likely from the starting materials which actually incorporate carbon atoms into the total structure. This is the reason we are so unsuccessful at figuring out what the impurity is- we never consider the component to be present in the first place.

This is part of the reason that process chemists are always warned to perform their process development using the same quality of materials that will be available upon scale up. It is not just that different grades of material may behave differently, but they may contain different impurities, which upon transformation may give rise to these impurities from hell. Remember that organic compounds are carbon compounds and their ultimate source is sunlight, either the ‘geologic sun’ that made petroleum or the recent sun that produced natural products. Nature just naturally produce mixtures that have been purified by us by mechanical separation or reactive transformation and the separation is never perfect.

On a COA most of us look to see the % purity but the more important question may be what the identities of the major impurities are, rather than how much of them there is. This problem of the impurity from hell may be the source of the adage that the most frequent source of difficulties in all of chemistry arises from inadequately pure starting materials.

When developing a process, a useful mental exercise is to try to imagine what the impurity from hell is likely to be in your synthesis. Imagine the possible impurities in your reactants that get incorporated into the product. Identify where these atoms end up and the minute difference that substituting the impurity for the desired building block makes. Does the structural difference occur far from the functionality in the final product or is it close to reactive centres and likely to affect them?.

How can you learn what the structures of the impurities in your building blocks are? You could ask the supplier? Or measure the MS from a GC or HPLC. Or search for the different syntheses in the literature.

Remember that as you perform the process development, the opportunity to encounter the impurity from hell increases, because you are removing operations and it is these extra supposedly unnecessary isolations performed at the end of each step that may be excluding precursor of the impurity from hell at an intermediate stage. When you telescope reactions one of the deciding criteria for proceeding is that you are not removing a useful purification opportunity. But the isolation that you think removes nothing, may be the one that discriminates between product precursor and impurity precursor. Typically we are not so worried about impurities of a few percent in intermediates because we hope and expect that the remaining steps of the synthesis will separate them by differential rates of reaction or because, even more fortunately, the impurity may not be able to undergo a subsequent reaction at all. But the character of the impurity from hell is that it differs little from the product and as the overall molecular weight of intermediates increases, the difference between it and the desired product shrinks.

If we cannot actually isolate and identify the impurity from hell is there some way to prove rather than just hypothesize that it comes from the co-reactant/starting material? Suppose we purify a buildin block/reactant but instead of using the bulk of the purified material (I think of it as the middle cut of a distillation fraction or of a chromatographic peak) take the head and the tail of the purification and use them in the synthesis. If the level of the impurity from hell jumps, it most likely comes from the impure reactant. If the reactant is a crystallized solid, the head and tail of the purification could be the first small crop of crystals formed combined with the mother liquors.

By identifying a precursor of an impurity from hell in a starting material and performing an initial purification of that building block so that you don’t need to do extensive final product purification with the attendant losses of more expensive product, then a more linear less convergent synthesis can be made more convergent with all the attendant benefits of that change.

Be particularly concerned about building blocks that are added close to the end of the synthesis. With these there are the fewest additional process steps and process isolations to clean up the by-products from the impurities in starting material.

We all know what it means to put something "on ice", but what is a chemical reaction "on water"? This new expression has been coined by a team headed by K. Barry Sharpless, winner of the 2001 Nobel Prize for chemistry, to describe reactions of organic substances that are not water-soluble, yet react well or even considerably faster in the presence of water than in organic solvents. If water could also replace organic solvents more often on the industrial scale, it would save money, increase the safety of chemical facilities, and reduce stress on the environment. Another advantage is that after the reaction, the organic and aqueous phases separate, eliminating the need for complex isolation steps to obtain the product.

Until now, a central aspect in the area of aqueous organic chemistry has been the effort to improve the water- solubility of the substances involved. Has this been the wrong approach? Is the axiom that has been passed on from the days of alchemy, corpora non agunt nisi soluta (substances do not interact with each other if they are not dissolved), no longer valid? Do reactants not need to be water-soluble at all in order to react in an aqueous environment? It seems that the situation bears some rethinking. Says Sharpless, "In contrast to prior assumptions, it seems that in many cases the immiscibility of the organic and aqueous phases is a considerable advantage."

So what exactly does "on water" mean? The expression simply refers to the fact that the essentially insoluble reactants and the water are vigorously stirred together. This forms a suspension, meaning that the immiscible liquids are finely divided into tiny drops. The contact surface between the aqueous and organic phases is thus especially large.

Why certain important categories of reactions, such as the Claisen rearrangement, work so well in aqueous suspension is not yet clear. Particularly astonishing is the fact that the reactions occasionally go faster "on water" than in a mixture of the pure reactants (without any solvent). "Molecules at the interface between two different phases often behave differently than molecules within the phase." Sharpless speculates: "It is possible that the unique properties of molecules at the interface between the water and the hydrophobic, oily organic phase play an important role in speeding up the reactions."

This is not the first report of reactions “on water” Henry Shaw, Howard D. Perimutter, and Chen Gu with Susan D. Arco and Titos O. Quibuyen reported in J. Org. Chem. 1997, 62, 236, that free radical brominations can be advantageously performed by photolysis in a heterogeneous nature with a water phase. Water was advantageous in this situation because it is an excellent medium for free radical reactions and because the oxygen-hydrogen bond is so strong that it is inert. They also report running the brominations in the neat starting material liquid. Yields of brominated products were the same in the absence as in the presence of water. The advantages of the water noted by the authors were:

· the reaction on water can replace the reaction in carbon tetrachloride which is disfavored because of the environmental toxicity of carbon tetrachloride

· the hydrogen bromide was removed from the organic layers into the water or lost to the atmosphere

· by providing a diluting medium the free radicals are dispersed over the reaction volume reducing the recombinations near the window where the light enters the reactor

· by providing a heat sink the water maintains a more uniform temperature making the product distribution more dependable

· partitioning of the heavier than water products from the lighter than water starting materials creating a three phase mixture

An advantages that was not mentioned was the increase in reactor volume that may be required to reach the stirrer when working on scale.

Quality synthetic chemists these days are more easily differentiated from the average by their ability to devise efficient isolations, particularly isolations that are rugged enough to work on scale up. Substructure, reaction and citation searching have simplified the art of the constructing the synthetic path itself.

Although there exist many methods to separate a mixture into acid, base and neutral fractions, and even to separate mixed bases or acids using of their relative proton donor/proton acceptor abilities, the vast majority of organic substances are essentially neutral. Therefore, methods that can separate the neutral fraction into sub-fractions in a simple fashion are valuable.

The only separations of aldehydes and ketones from other neutral functional group classes which is quickly recalled by the average chemist is sodium bisulfite for aldehydes and Girard’s P and T reagents for all carbonyls. Kilomentor, in another blog article, has discussed the use of the Okomoto reagent for aldehydes.

R.P. Singh, H. N. Subbarao and Sukh Dev, Tetrahedron 37, 843 (1981) have written a paper subtitled, Silica-Gel Supported Reagents for the Isolation of Aldehydes and Ketones. This technique, as they teach, works only for neutral carbonyl containing fractions that are fully soluble in hexane, toluene or other non-polar media, because it is necessary that the non-aldehyde/non-ketone fraction remain dissolved in the non-polar solvent during the method. This requirement is easily met, since the neutral fraction can be first partitioned between the non-polar solvent (preferable hexane or cyclohexane) and methanol/water or acetonitrile.

In the technique the neutral fraction to be separated is dissolved in the non-polar solvent and treated with an appropriate amount of 10%w/w semicarbazide on silica-gel. The mixture is heated and stirred at 70 C for 12-18 hours. As the carbonyl components in the mixture react with the semicarbazide they become immobilized on the insoluble solid silica gel phase. The end of reaction is detected by the absence of carbonyl compound in the solution phase as measured by TLC developed with 10% 2,4-DNPH in aq. Aq. HCl. When the reaction is complete the mixture is cooled, filtered and the solid washed with the same solvent used in the adsorption step. The combined filtrate and washing that contain the non-aldehyde/non ketone fraction are processed or discarded as the overall isolation process requires. The solid phase containing the semi-carbazone (if it contains about 1 mmole) is added to a solution of about 10 mmoles of oxalic acid in 16 ml of water, covered with a layer of immiscible organic solvent and the mixture stirred and refluxed for 4-5 hours. The solid is separated, washed; the aqueous phase is extracted and all the organic layers combined. The aldehydes and ketones can be found within this phase.

The authors report that this method has been used with good effect to separate almost a kilogram of neutral natural product extract containing 90 gm of carbonyl fraction.

The method can also be used to separate a small amount of carbonyl impurity from a large amount of non-carbonyl product. Such a separation would be even more applicable to large scale since the amount of reagent adsorbed on silica gel would be smaller.

The Semi-Carbazide on Silica Gel Reagent is prepared as follows:

Semicarbazide hydrochloride (5.0g; 0.045 moles) was added to a solution of sodium hydroxide (2.0 g: 0.05 mole) in water-methanol (1:1; 60 ml) and to the resulting clear solution, silica gel (45 gm) was introduced with stirring. The whole mixture was mechanically shaken (1 hr) at room temperature (3- 35 C; India) and water-methanol removed on a rotary evaporator (about 90 C/80-90 mm; 30-45 min) to get a free flowing powder. This material should weigh 60-63 g. The product is stored in a brown bottle at room temp. A two-year old product did not undergo deterioration.

This very widely applicable methodology has been only little applied. A citation search would show how little. The only reference that I am very familiar with is the Masters thesis of Tarcisia Khomasurya from the University of Toronto Canada. Khomasurya applied the reagent to the separation of the ketone from the non-carbonyl components of cedar oil. For natural product mixtures the preferred reaction solvent is cyclohexane because it can be easily thoroughly purified so it does not put impurities into the fractions.

Expired US patents 4,332,968 and 4,306,068 contain a chemical trick that can be used to remove water from many wet liquids. Apparently if mesityl oxide, (Me)₂C=CH-CO-Me, is added to the wet liquid in the stoichiometry of 1 mole of mesityl oxide for each mole of water and a catalytic amount of a primary amine is added and the liquid is then warmed up two equivalents of acetone will form as the acetone is distilled out of the mixture.

The mechanism of the process has been studied by Ralph M. Pollack and David Strohbeen, J. Am. Chem.. Soc. (1972), 94(7), 2534-5. One can imagine that the method could be used to de-water liquids that are difficult to dry by other means such as DMSO, DMF, etc. The method would certainly efficiently dry lower molecular weight water-miscible primary amines, where the catalyst would be present in enormous excess.

The two patents are directed to other very practical uses. Reaction of a mixture of a primary and a non-primary amine by heating the mixture together with mesityl oxide results in a mixture of the imine of the primary amine with acetone, and unreacted non-primary amine. The patents teach that where the amine mixture cannot be separated by distillation, the new mixture of imine and non-primary amine usually can be.

Another use claimed by these lapsed patents is a means to make the imine quantitatively as a protecting gro