CROSS-REFERENCE TO RELATED APPLICATIONS This application claims the benefits of U.S. Provisional Patent Application Nos. 60/583,641, filed June 28, 2004, 60/609,716, filed September 14, 2004, 60/622,065, filed October 25, 2004, 60/652,330, filed February 115 2005, 60/583,644, filed June 28, 2004, 60/652,332, filed February 11, 2005, 60/583,643, filed June 28, 2004, 60/652,331, filed February 11, 2005, 60/666,666, filed March 30, 2005, 60/675,369, filed April 26, 2005, Application No. Not Yet Known (Attorney Docket No. 12670/46803), filed June 9, 2005, and Application No. Not Yet Known (Attorney Docket No. 12670/47001), filed June 14, 2005, the contents of all of which are incorporated herein by reference.

FIELD OF THE INVENTION The present invention relates to an isolated impurity of Atomoxetine hydrochloride, its preparation as well as the preparation of other impurities, and their use as reference standards.

BACKGROUND OF THE INVENTION Atomoxetine (ATM), known as (R)(-)-N-methyl-3-(2-methylphenoxy)-3- phenylpropylamine , has the following structure:

In addition, Atomoxetine has the formula C17H21NO, a molecular weight of 255.35, and a composition of 79.96 percent C, 8.29 percent H, 5.49 percent N, and 6.27 percent O, by weight. The hydrogen chloride salt of Atomoxetine, Atomoxetine HCl, is marketed as STRATTERA®, which is prescribed as oral capsules having dosages of 10 mg, 18 mg, 25 mg, 40 mg, and 60 mg for the treatment of Attention-Deficit-Hyperactivity Disorder (ADHD). Atomoxetine is a competitive inhibitor of norepinephrine uptake in synaptosomes of rat hypothalamus, 2 and 9 times more effective than the racemic mixture and the (+)- enahtiomer, respectively, ofN-methyl-3-(2-methylphenoxy)-3-phenylpropylamine (Tomoxetine, "TMX"), disclosed in EP 0 052 492. Atomoxetine is the (R)-(-) enantiomer of Tomoxetine.

Racemic Tomoxetine as well as many other aryloxyphenylpropylamines, e.g., FLUOXETINE® and NISOXETINE®, are disclosed in U.S. Patent No. 4,018,895, which also discloses a psychotropic effect of the compounds. Atomoxetine, including its pharmaceutically acceptable addition salts, e.g., the hydrochloride, is disclosed in EP 0 052 492, which also discloses their use as antidepressants. The use of atomoxetine in the treatment of ADHD was disclosed in EP 0 721 777.

Manufacturing processes for Atomoxetine Hydrochloride, known in the art, include those disclosed in European Patent Publication No. EP 0 052 492, U.S. Patents Nos. 4,868,344 and 6,541,668, International Patent Application Publication No. WO 00/58262, the teachings of which are incorporated herein by reference.

It is well known in the art that, for human administration, safety considerations require the establishment, by national and international regulatory authorities, of very low limits for identified, but toxicologically uncharacterized impurities, before an active pharmaceutical ingredient (API) product is commercialized. Typically, these limits are less than about 0.15 percent by weight of each impurity. Limits for unidentified and/or uncharacterized impurities are obviously lower, typically, less than 0.1 percent by weight. Therefore, in the manufacture of APIs, the purity of the products, such as Atomoxetine Hydrochloride, is required before commercialization, as is the purity of the active agent in the manufacture of formulated pharmaceuticals.

It is also known in the art that impurities in an API may arise from degradation of the API itself, which is related to the stability of the pure API during storage, and the manufacturing process, including the chemical synthesis. Process impurities include unreacted starting materials, chemical derivatives of impurities contained in starting materials, synthetic by-products, and degradation products.

hi addition to stability, which is a factor in the shelf life of the API, the purity of the API produced in the commercial manufacturing process is clearly a necessary condition for commercialization. Impurities introduced during commercial manufacturing processes must be limited to very small amounts, and are preferably substantially absent. For example, the ICH Q7A guidance for API manufacturers requires that process impurities be maintained below set limits by specifying the quality of raw materials, controlling process parameters, such as temperature, pressure, time, and stoichiometric ratios, and including purification steps, such as crystallization, distillation, and liquid-liquid extraction, in the manufacturing process.

The product mixture of a reaction is rarely a single compound with sufficient purity to comply with pharmaceutical standards. Side products and by-products of the reaction and adjunct reagents used in the reaction will, in most cases, also be present in the product mixture. At certain stages during processing of an API5 such as Atomoxetine Hydrochloride, it must be analyzed for purity, typically, by HPLC or GC analysis, to determine if it is suitable for continued processing and, ultimately, for use in a pharmaceutical product. The API need not be absolutely pure, as absolute purity is a theoretical ideal that is typically unattainable. Rather, purity standards are set with the intention of ensuring that an API is as free of impurities as possible, and, thus, is as safe as possible for clinical use. As discussed above, in the United States, the Food and Drug Administration guidelines recommend that the amounts of some impurities be limited to less than 0.1 percent.

Generally, side products, by-products, and adjunct reagents (collectively "impurities") are identified spectroscopically and/or with another physical method, and then associated with a peak position, such as that in a chromatogram, or a spot on a TLC plate. (Strobel p. 953, Strobel, H.A.; Heineman, W.R., Chemical Instrumentation: A Systematic Approach, 3rd dd. (Wiley & Sons: New York 1989)). Thereafter, the impurity can be identified, e.g., by its relative position in the chromatogram, where the position in a chromatogram is conventionally measured in minutes between injection of the sample on the column and elution of the particular component through the detector. The relative position in the chromatogram is known as the "retention time." The retention time varies daily, or even over the course of a day, based upon the condition of the instrumentation, as well as many other factors. To mitigate the effects such variations have upon accurate identification of an impurity, practitioners use the "relative retention time" ("RRT") to identify impurities. (Strobel p. 922). The RRT of an impurity is its retention time divided by the retention time of a reference marker. In theory, Atomoxetine Hydrochloride itself could be used as the reference marker, but as a practical matter it is present in such a large proportion in the mixture that it can saturate the column, leading to irreproducible retention times, as the maximum of the peak can wander (Strobel, Fig. 24.8(b), p. 879, illustrates an asymmetric peak observed when a column is overloaded). Thus, it may be advantageous to select a compound other than the API that is added to, or present in, the mixture in an amount sufficiently large to be detectable and sufficiently low as not to saturate the column, and to use that compound as the reference marker.

Those skilled in the art of drug manufacturing research and development understand that a compound in a relatively pure state can be used as a "reference standard." A reference standard is similar to a reference marker, which is used for qualitative analysis only, but is used to quantify the amount of the compound of the reference standard in an unknown mixture, as well. A reference standard is an "external standard," when a solution of a known concentration of the reference standard and an unknown mixture are analyzed using the same technique. (Strobel p. 924, Snyder p. 549, Snyder, L.R.; Kirkland, JJ. Introduction to Modern Liquid Chromatography, 2nd ed. (John Wiley & Sons: New York 1979)). The amount of the compound in the mixture can be determined by comparing the magnitude of the detector response. See also U.S. Patent No. 6,333,198, incorporated herein by reference.

The reference standard can also be used to quantify the amount of another compound in the mixture if a "response factor," which compensates for differences in the sensitivity of the detector to the two compounds, has been predetermined. (Strobel p. 894). For this purpose, the reference standard is added directly to the mixture, and is known as an "internal standard." (Strobel p. 925, Snyder p. 552).

The reference standard can even be used as an internal standard when, without the addition of the reference standard, an unknown mixture contains a detectable amount of the reference standard compound using a technique known as "standard addition." In a "standard addition," at least two samples are prepared by adding known and differing amounts of the internal standard. (Strobel pp. 391-393, Snyder pp. 571, 572). The proportion of the detector response due to the reference standard present in the mixture without the addition can be determined by plotting the detector response against the amount of the reference standard added to each of the samples, and extrapolating the plot to zero. (See, e.g., Strobel, Fig. 11.4 p. 392).

As is known by those skilled in the art, the management of process impurities is greatly enhanced by understanding their chemical structures and synthetic pathways, and by identifying the parameters that influence the amount of impurities in the final product. BRIEF DESCRIPTION OF THE FIGURES Figure 1 shows a typical chromatogram of ATM-HCl impurities obtained via GC on a HP-35: 35% Phenyl Methyl Siloxane (Hewlett Packard cat. N° 19091G-113) column. Figure 2 shows a typical chromatogram of 2-fluoro toluene impurities obtained via GC on a Zebron ZB-WAX 100% Polyethylene glycol (Phenomenex cat. N° 7KK-G007-22) column. Figure 3 shows a typical chromatogram of D-ATM obtained via achiral chromatography on a YMC-Pack ODS-AQ, S-5μm column. Figure 4 shows a typical chromatogram of atomoxetine HCl obtained via chiral chromatography on a CHIRALCEL OD-R cellulose tris(3,5-dimethylphenylcarbamate) column.

SUMMARY OF THE INVENTION m the first embodiment, this invention is directed to an isolated Atomoxetine HCl impurity; the isolated hydrochloride salt of N-methyl-3-(3-methylphenoxy)-3- phenylpropylamine (3-ATM HCl).

In another embodiment, this invention is directed to methods for preparing 3-ATM HCl, by reacting 3-fluorotoluene (3FT) with the alkolate of N-methyl-3-hydroxy-3- phenylpropylamine, under the process conditions defined below.

hi yet another embodiment, this invention is directed to methods of preparing N- methyl-3-(4-methylphenoxy)-3-phenylpropylamine (4-ATM HCl), another Atomoxetine HCl impurity, by reacting 4-fluorotoluene (4FT) with the alkolate of N-methyl-3-hydroxy-3- phenylpropylamine, under the process conditions defined below.

hi yet another embodiment, this invention is directed to methods of preparing N- methyl-3-phenoxy~3-phenylpropylamine (D-ATM HCl), yet another Atomoxetine HCl impurity, by reacting fluorobenzene (FB) with the alkolate of N-methyl-3-hydroxy-3- phenylpropylamine, under the process conditions defined below.

In one embodiment, the invention is directed to the use of 3-ATM HCl, as well as of the two Atomoxetine HCl impurities, 4-ATM HCl and D-ATM HCl as reference markers in a qualitative analysis of Atomoxetine HCl. In another embodiment, the invention is directed to a method of using 3 -ATM HCl, A- ATM HCl and D-ATM HCl as reference standards to analytically quantify the purity of Atomoxetine Hydrochloride.

hi yet another embodiment, the invention is directed to a method to define the limits for the amounts of 3FT, 4FT, and FB impurities in the 2FT starting material for the synthesis of Atomoxetine Hydrochloride of a desired purity, comprising the use of 3-ATM HCl, A- ATM HCl and D-ATM HCl as reference standards .

hi a further embodiment, the invention is directed to a method for the quantification of the purity of Atomoxetine Hydrochloride, comprising the use of 3-ATM HCl, 4-ATM HCl and D-ATM HCl as reference standards, where the reference standards may be either external standards or internal standards.

hi one embodiment, the invention is directed to an analytical method for the analysis of the purity of 2-fluorotoluene.

In another embodiment, the invention is directed to a method comprising limiting the amounts of the impurities 3FT, 4FT, and FB in the 2FT starting material used in the synthesis of Atomoxetine Hydrochloride to ensure the purity of the Atomoxetine Hydrochloride product by determining the amounts of 3FT, 4FT, and FB in the 2-fluorotoluene starting material, using at least one of 3-ATM HCl, 4-ATM HCl and D-ATM HCl as a reference standard.

In yet another embodiment, the invention is directed to a novel GC method, used for the measuring of the three impurities in an atomoxetine HCl sample, hi a further embodiment, the invention is directed to a novel GC method, used for the measuring of 3FT, 4FT and FB in a 2-fluorotoluene sample.

DESCRIPTION OF THE INVENTION As used herein, unless the context requires otherwise, the three impurities referred to in the application include both the racemic mixture and the enantiomerically pure forms. Thus, for example, N-methyl-3-(3-methylphenoxy)-3-phenylpropylamine (and its acronym "3-ATM HCl") refers to either (±)-N-methyl-3-(3-methylphenoxy)-3-phenylpropylaniine or (R)(-)-N-methyl-3 -(3 -methylphenoxy)-3 -phenylpropylamine. As used herein the term "aromatic solvent" refers to a C6-1O aromatic hydrocarbon such as, but not limited to, benzene, xylene, or toluene.

As used herein, the term "predetermined level", in reference to the level of the impurities 3-ATM HCl, 4-ATM HCl and D-ATM HCl, means a level of 0.15% or less, as measured by GC or HPLC.

As used herein, the term "isolated" refers to a compound that is at least 80%, preferably at least 90%, even more preferably at Ieast5 95%, and most preferably at least 99% pure, as judged by GC or HPLC.

A "reference marker" is used in qualitative analysis to identify components of a mixture based upon their position, e.g., in a chromatogram or on a Thin Layer Chromatography (TLC) plate (Strobel pp. 921, 922, 953). For this purpose, the compound does not necessarily have to be added to the mixture if it is present in the mixture. A "reference marker" is used only for qualitative analysis, while a reference standard may be used for quantitative or qualitative analysis, or both. Hence, a reference marker is a subset of a reference standard, and is included within the definition of a reference standard.

As used herein, the term "reference standard" refers to a compound that may be used both for quantitative and qualitative analysis of an active pharmaceutical ingredient. For example, the HPLC or GC retention time of the compound allows a relative retention time to be determined, thus making qualitative analysis possible. The concentration of the compound in solution before injection into an HPLC or GC column allows the areas under the HPLC or GC peaks to be compared, thus making quantitative analysis possible.

Reference standards are described in general terms above. However, as will be understood by those skilled in the art, a detector response can be, for example, the peak heights or integrated peak areas of a chromatogram obtained, e.g., by UV or refractive index detection, from the eluent of an HPLC system or, e.g., flame ionization detection (FID) or thermal conductivity detection, from the eluent of a gas chromatograph, or other detector response, e.g., the UV absorbance of spots on a fluorescent TLC plate. The position of the reference standard may be used to calculate the relative retention time for Atomoxetine Hydrochloride and impurities of Atomoxetine Hydrochloride.

As used herein, the term "substantially", in reference to RRTs being substantially the same, refers to a relative standard deviation that is equal to or less than 5% for a population of 6 injections. The present invention is directed to an impurity of Atomoxetine Hydrochloride, which was previously unidentified, its preparation as well as of other known impurities, and to the use of these impurities as reference standards for the analytical quantification of Atomoxetine Hydrochloride purity, as required in the manufacture of high purity Atomoxetine Hydrochloride.

Three hydrochloride impurities have been identified in the (±)Tomoxetine solutions and the final product. The three hydrochloride impurities are N-methyl-3-(3~ methylphenoxy)-3-phenylpropylamine (3-ATM HCl), a previously unidentified compound, as well as the hydrochloride salt of N-methyl-3-(4-methylphenoxy)-3-phenylpropylamine (4-ATM HCl), and the hydrochloride salt of N-methyl-3-phenoxy-3-phenylpropylamine (D- ATM HCl). These three hydrochloride impurities are formed by reacting 3-fluorotoluene (3FT), 4-fluorotoluene (4FT), and fluorobenzene (FB), respectively, with the alkolate of N- methyl-3-hydroxy-3-phenylpropylamine. The hydrochloride impurities have been found to be surprisingly difficult to remove using normal purification steps.

The present invention provides an isolated Atomoxetine Hydrochloride impurity, N- methyl-3-(3-methylphenoxy)-3-phenylpropylamine (3-ATM HCl), having the structure:

The 3-ATM HCl impurity is a regio-isomer of Atomoxetine.

The present invention provides a process for the preparation of 3-ATM HCl by reaction of 3-fluorotoluene, an impurity of 2-fluorotoluene, with the alkolate of N-methyl-3- hydroxy-3-phenylpropylamine. This process comprises: a) combining N-methyl-3-hydroxy-3-phenylpropylamine with DMSO in the presence of a strong base at a temperature of at least 9O0C; b) adding 3-fluorotoluene and maintaining the reaction mixture for at least 5 hours; c) adding a first organic solvent and water to the reaction mixture; d) recovering crude (±) 3-methyl Tomoxetine base; e) combining crude (+) 3-methyl Tomoxetine base with (S)-(+)-Mandelic acid in the presence of a Ci-4 alcohol and an aromatic solvent, and heating to a temperature of about 65°C to about 7O0C; f) recovering 3-methyl Atomoxetine (S)-(+)-Mandelate salt; g) combining 3-methyl Atomoxetine (S)-(+)-Mandelate salt with a second organic solvent, water and a base; h) recovering 3-methyl Atomoxetine base; i) converting the 3-methyl Atomoxetine base to the corresponding hydrochloride salt.

DMSO in step a) may be added in a relatively small amount, and may even be considered a catalyst. Preferably, the amount of DMSO is about 0.1 to about 20 moles per moles of N-methyl-3-hydroxy-3-phenylpropylamine. Most preferably, the amount of DMSO is about 3 to about 4 moles per moles of N-methyl-3-hyά^oxy-3-phenylproρylamine.

The strong base in step a) may be any one of NaOH, KOH, Ca(OH)2 or Ba(OH)2. Preferably, the strong base is KOH. The base is preferably present in an amount of about 3 to about 4 moles per moles of moles of N-methyl-3-hydroxy-3-phenylpropylamine.

The 3-fluorotoluene added in step b) is preferably at an amount of at least 2 moles per moles of N-methyl-3 -hydroxy-3 -phenylpropylamine.

Preferably, the first organic solvent in step c) is selected from the group consisting of C5-10 aliphatic and aromatic hydrocarbons, which aromatic hydrocarbons maybe substituted with one or more (preferably one to three) C1-3 alkyl groups, C3-8 alkyl esters and C3-8 alkyl ethers. More preferably, the organic solvent is selected from the group consisting of toluene, benzene, xylenes, di-isopropyl ether, methyl-tert-butyl, ethyl acetate n-butylacetate, and isobutylacetate. Most preferably, the organic solvent is toluene.

Preferably, the C1-4 alcohol in step e) is added in an amount of about 0.1 mL per Ig of the (±) 3-methyl Tomoxetine base. Most preferably, the Ci-4 alcohol is methanol. The aromatic solvent can be an aromatic hydrocarbon which may be substituted with one or more (preferably one to three) C1-10 alkyl groups, such as toluene, benzene, xylenes. A preferred aromatic solvent is toluene. Preferably, following heating, the reaction mixture is cooled to a temperature of about 00C to about 200C. Preferably, the reaction mixture is cooled to a temperature of about 5° to about 10°. Preferably, the base in step g) is selected from an alkali metal hydroxide, such as NaόH or KOH, or an alkali metal carbonate such as Na2CO3 or K2CO3. Most preferably, the base is NaOH

Preferably the second organic solvent in step g) is selected from the group consisting of aliphatic or aromatic hydrocarbons such as C5-8 alkanes, toluene and xylene, C1-4 alkyl esters such as methyl acetate, ethyl acetate, n-butyl acetate and iso-butyl acetate, ketones such as methyl-ethyl ketone, linear or branched C4-8 alcohols such as n-butanol, 2-butanol and n- pentanol and mixtures thereof. More preferably, the second organic solvent is selected from the group consisting of ethyl acetate, n-butyl acetate and iso-butyl acetate. Most preferably, the second organic solvent is n-butyl acetate.

The second impurity is N-methyl-3-(4-methylphenoxy)-3-phenylpropylamine (4- ATM), is disclosed in U.S. Patent No. 4,018,895, and has the structure:

The 4-ATM compound is a regio-isomer of Atomoxetine.

The present invention provides a process for its preparation by reaction of 4- fluorotoluene, an impurity of 2-fluorotoluene, with the alkolate of N-methyl-3-hydroxy-3- phenylpropylamine, under the process conditions described above for the preparation of 3- ATM HCl.

The third impurity is N-methyl-3-phenoxy-3-phenylpropylamine or "Des-methyl Atomoxetine" (D-ATM). This compound is disclosed in U.S. Patent No. 4,018,895, and has the structure:

The present invention provides a process for the preparation of the D-ATM impurity by reaction of fluorobenzene, an impurity of 2-fluorotoluene, with the alkolate of N-methyl- 3-hydroxy-3-phenylpropylarnine, under the process conditions described above for the preparation of 3-ATM HCl.

These three impurities are identified through separate syntheses, wherein (±)TMX is purposely prepared from 2FT (2-fluorotoluene) containing known amounts of 3FT (3-fluorotoluene), 4FT (4-fluorotoluene), and FB (fluorobenzene), in an amount of about 2 percent by weight each. Relative amounts of the corresponding by-products formed in the reaction, as measured by GC, area percent, are in agreement with the expected aromatic activation towards nucleophilic displacement. That is, the amount of D-ATM HCl from FB is greater than the amount of 3-ATM HCl from 3FT, which is greater than the amount of 4-ATM HCl from 4FT. The relative amounts of the impurities found in excess of the 2FT recovered after reaction are also in agreement. That is, the amount of FB is less than the amount of 3FT, which is less than the amount of 4FT. The results are provided in Table 1

Table 1

The (±)Tomoxetine solution from the stepwise process for the synthesis of ATM hydrochloride and the intermediates are analyzed by gas chromatography to separate the three impurities from the ATM, which, being GC achiral, always produces peaks including both enantiomers.

In a purification process, given an analytical technique producing peaks corresponding to impurities in samples, a purification factor of impurity X can be defined as follows: (peak area% of X in starting material sample) / (peak area% of X in product sample)

This calculation implies that the higher the purification factor, the easier the removal of the impurity, and that purification factors below 1 indicate that it is likely to be extremely difficult, if not impossible, to remove the impurity.

Very low purification factors are generally obtained for the 3 -ATM, 4- ATM, and D- ATM impurities, which had a maximum value of 4.56. Values of greater than 50 are found for other impurities, and the removal of those impurities is found to be effective. Purification factors of less than 1 are also found for the D-ATM impurity in the Mandelate salt stage. A detailed analysis is provided in the following tables, where GC area percent data are reported together with calculated purification factors, i.e., the ratio of A% for the starting material to A% for the product, for each impurity in each step in the synthesis.

Table 2 Step 2: Crude Atomoxetine (S)-(+) mandelate salt (crude ATM-SMA) formation

Table 3 Step 3: Purified Atomoxetine (S)-(+) mandelate salt (purified ATM-SMA) formation

Table 4 Steps 4 and 5: Atomoxetine Hydrochloride (ATM HCl) formation

The purification factors calculated for the final amounts of D-ATM HCl, 3-ATM HCL, and 4-ATM HCL may be used to back-calculate the amounts of FB, 3FT and 4FT in the 2-fluorotoluene starting material. The calculation is as follows:

Final: D-ATM HCI 3-ATM HCI 4-ATM HCI GC A% 0.15 0.15 0.15

Purified intermediate: pD-ATM-SMA p3-ATM-SMA p4-ATM-SMA GC A% 0.34 0.68 0.53

Crude intermediate: cD-ATM-SMA C3-ATM-SMA C4-ATM-SMA GC A% 0.30 0.94 0.61

(+)TMX: (±) D-TMX (±) 3-TMX (+) 4-TMX GC A% 0.24 1.32 0.68 Corresponding 2FT starting material: FB 3FT 4FT GC A% 0.059 0.636 0.990

Despite a process that includes five purification steps, i.e., two extractions and three crystallizations, the removal of 3-ATM HCl, 4-ATM HCl and D-ATM HCl is surprisingly low. Therefore, Atomoxetine Hydrochloride of the desired purity is typically not obtained, as commonly expected, by the repetition of purification steps used typically in the art. The desired purity may be obtained by limiting the amounts of the impurities 3FT, 4FT, and FB in the 2-fluorotoluene starting material. The limit amounts of the 3FT, 4FT, and FB impurities may be defined through the purification factor calculations described above, only if a suitable reference standard is available.

The present invention provides various methods involving the use of 3 -ATM HCl, 4- ATM HCl, and D-ATM HCl as reference markers or reference standards.

When using these impurities enantiomerically pure, they will be eluted at the same retention time as their racemates.

Provided is a method of identifying an impurity in a sample of Atomoxetine Hydrochloride comprising: (a) providing a reference sample comprising a reference marker and Atomoxetine Hydrochloride; (b) carrying out HPLC or GC on the reference sample to determine the relative retention time of the reference marker compared to Atomoxetine hydrochloride; (c) carrying out HPLC or GC on the sample of Atomoxetine hydrochloride to determine the relative retention time of the impurity compared to Atomoxetine hydrochloride; (d) comparing the relative retention times determined in steps (b) and (c);

where, if the relative retention times determined in steps (b) and (c) are substantially the same, the impurity is identified as being the same as the reference marker;

where the reference marker is selected from the group consisting of 3-ATM HCl, 4-ATM HCl and D-ATM HCl.

Also provided is a method of determining the amount of an impurity in a sample of atomoxetine hydrochloride comprising: (a) adding a known amount of a reference standard to the atomoxetine hydrochloride sample; (b) subjecting the atomoxetine hydrochloride to HPLC or GC; (c) identifying and measuring the area of an HPLC or GC peak associated with the impurity; (d) identifying and measuring the area of an HPLC or GC peak associated with the reference standard; (e) calculating the amount of the impurity in the atomoxetine hydrochloride sample based on the results of steps (c) and (d);

where the reference standard is selected from the group consisting of 3-ATM HCl, 4- ATM HCl and D-ATM HCl.

Also provided is a method of determining the amount of an impurity in a sample of atomoxetine hydrochloride comprising: (a) providing a sample of atomoxetine hydrochloride containing an unknown concentration of the impurity; (b) providing a sample of a known concentration of the impurity; (c) subj ecting a portion of the sample of atomoxetine hydrochloride and a portion of the sample of the impurity to HPLC or GC; (d) measuring the area of the impurity peaks obtained from the sample of atomoxetine hydrochloride and from the sample of the impurity; and (e) calculating the concentration of the impurity in the sample of Atomoxetine Hydrochloride from the measurements of step (d);

where the impurity is selected from the group consisting of 3-ATM HCl, 4- ATM HCl and D- ATM HCl.

Also provided is a method of ensuring the purity of a preparation of Atomoxetine Hydrochloride comprising: (a) choosing a predetermined level of the impurities 3-ATM HCl, 4-ATM HCl and D- ATM HCl. (b) measuring the purification factor for each impurity formed during each step of the method of synthesizing Atomoxetine Hydrochloride; (c) calculating the highest level of 3-fluorotoluene, 4-fluorotoluene, and fluorobenzene that may be present in the 2-fluorotoluene such that the levels of 3-ATM HCl, 4-ATM HCl and D-ATM HCl in the Atomoxetine Hydrochloride produced are below the predetermined levels;

where step (c) involves the use of 3-ATM HCl, 4-ATM HCl and D-ATM HCl as reference standards.

Suitable methods of synthesizing Atomoxetine Hydrochloride that involve reacting 2- fluorotoluene with N-methyl~3-hydroxy-3-phenylpropylarnine are known in the art and include those disclosed in European Patent Publication No. EP 0 052 492, U.S. Patents Nos. 4,868,344 and 6,541,668 and International Patent Application Publication No. WO 00/58262. Typically, the synthetic path for the production of Atomoxetine comprises:

I. Tomoxetine synthesis, using 2-fluorotoluene as a reactant; II. Tomoxetine optical resolution using (S)-(+)-Mandelic acid as a resolving agent; and III. Atomoxetine Hydrochloride formation and isolation.

Analytical Methods

Atomoxetine Hydrochloride impurities may be determined using the following gas chromatography apparatus and procedures:

Column & Packing: HP-35: 35% Phenyl Methyl Siloxane (Hewlett Packard cat. N° 19091G-113) or equivalent Length: 30 m Diameter: 0.32 mm Film thickness: 0.25 μm

Injector temperature: 250°C Detector temperature: 250°C Oven temperature: Time (min) Temperature 0 180°C 20.0 180°C 23.5 250°C 33.5 250°C Equilibrium time: 5 min Injection volume: l μl Carrier gas: Nitrogen Flow: 1.5 ml/min. Detector: FID Split ratio: 10/1 Typical retention times that may be obtained are as follows: RT RRT (±)D-ATM 12.8' 0.81 ATM 15.9' 1.00 (±)3-ATM 16.9' 1.06 (±)4-ATM 18.3' 1.15 A sample chromatogram is shown in Figure 1.

Sample solutions may be prepared as follows: About 15 mg of ATM HCl is dissolved in 1 ml of water. Then, 10 ml of n-hexane and 2 ml of a 5 percent by volume ammonia solution are added. The sample is mixed vigorously, and, after phase separation, about 1 ml of the upper organic phase is transferred into an injection vial.

2-fluorotoluene impurities may be determined using the following gas chromatography apparatus and procedures :

Column & Packing: Zebron ZB-WAX 100% Polyethylene glycol (Phenomenex cat. N° 7KK-G007-22) or equivalent Length: 60 m Diameter: 0.53 mm Film thickness: 1.0 μm Injector temperature: 180°C Detector temperature: 2000C Oven temperature: Rate °C/min Temperature Hold min. 600C 24 10.0 200°C 10 Equilibrium time: 3 min Injection volume: l μl Carrier gas: Nitrogen Flow: 3.0 ml/min. Detector: FID Split ratio: 10/1 Typical retention times that maybe obtained are as follows:

RT RRT F-Benzene 14.4' 0.65 2-Butanol 16.1' 0.73 2-F Toluene 22.1' 1.00 3-F Toluene 22.9' 1.04 4-F Toluene 23.5' 1.06

A sample chromatogram is shown in Figure 2.

Quantitative analysis of atomoxetine may be performed using the following achiral HPLC method:

Column & Packing: YMC-Pack ODS-AQ, S-5μm , 12nm 250mmX4.6mmX5.0μm, cat n.042574458 (W) or equivalent Buffer: NaH2 PO4 monohydrate pH 3.0: 2.8 g in 100OmL of deionized water, adjust pH at 3.0 with H3PO4 85% (w/w). Filter on a 0.45 μm filter. Eluent A: Acetonitrile : Water 90:10 Gradient Time (min) % Buffer % Eluent A 0 85 15 20 48 52 30 48 52 Equilibrium time: 8 minutes Flow Rate: 1.5 mL/rnins. Detector: UV at 215 nm Column temperature: 4O 0C Diluent BuffeπAcetonitrile (60:40) Injection 5 μL

Standard Solution preparation Weigh accurately about 20mg of Desmethylatomoxetine (D-ATM) standard into a 100 ml volumetric flask and dilute to volume with diluent solution. Transfer 0.5 ml of the obtained solution into 100 ml volumetric flask and dilute to volume with diluent solution.

Sample Solution preparation : Prepare a solution of aboutl.O mg/ml solution of ATM*HC1 sample in diluent.

Procedure: Inject blank run of diluent in order to get a good column stabilization and recognize the system peaks, 3 blank runs are suggested. Inject standard and sample solution into chromatograph continuing the chromatogram of samples up to the end of gradient program. A sample chromatogram is shown in Figure 3.

Enantiomeric purity can be established by the following chiral HPLC method:

Column & Packing: CHIRALCEL OD-R cellulose tris(3,5-dimethylphenylcarbamate) 250 mm x 4.60 mm x 10 urn (Daicel chemicals cat. NA DAIC 14625) or equivalent Mobile phase: KPF6 10OmM / ACN - 60 /40 Note: continue the chromatogram up to 30 minutes Sample volume: 5.0 μL Flow Rate: 0.8 mL/min. Detector: UV at 215 nm Column temperature: 35°C Diluent mobile phase

Mobile phase composition and flow rate may be varied in order to achieve the required system suitability. A sample chromatogram is shown in Figure 4.

EXAMPLES While the present invention is described with respect to particular examples and preferred embodiments, it is understood that the present invention is not limited to these examples and embodiments. The present invention, therefore, includes variations from the particular examples and preferred embodiments described herein, as will be apparent to one of skill in the art.

EXAMPLE l Synthesis of N-methyl-3-(3-methylphenoxy)-3-phenylpropyIamine (3-ATM HCI)

76g (0.969 mol) of dimethylsulfoxide, 4Og (0.242 mol) of N-methyl-3-hydroxy-3- phenylpropylamine (100%, mw 165.12), and 75.5g (1.211 mol) of potassium hydroxide (bulk industrial grade, 90% assay) are heated under stirring at 100°C ± 3°C for 1 hour. The resulting mixture is cooled to 800C, and, with good stirring, 79.95g (0.726 mol) of 3-fluorotoluene are added over a period of about 1 hour, maintaining the temperature at 83°C ± 30C, resulting in the precipitation of salts. The mixture is then stirred at 83 °C ± 3°C for about 5 hours, and 300 ml of water and 300 ml of toluene are added. The mixture is stirred for a few minutes, and the phases are separated. The aqueous phase is extracted with 50 ml of toluene, and the organic phases are collected and washed three times with 80 ml of water. The washed organic phase is concentrated under vacuum at about 60° to about 8O0C to produce about 70 g of a liquid brown residue, which is crude (±) 3-methyl Tomoxetine base (3-TMX) containing about 0.2 mol of 100% base. 400 ml of toluene, 3 ml of methanol, and 19.0 g (0.125 mol) of (S)-(+)-Mandelic acid are added at 25°C, producing a slurry that is heated to a temperature of from about 65° to about 70°C until completely dissolved to form a solution. The resulting solution is then cooled to from about 5° to about 10°C, resulting in crystallization of solid 3-methyl Atomoxetine (S)-(+)-Mandelate salt (3 -ATM-SMA), which is isolated by filtration, which is found to be difficult and slow, and washed with toluene, which also is found to be difficult and slow. After drying under vacuum at about 50° to about 60°C, 30 g of product are recovered. The 30 g of dried 3 -ATM-SMA are mixed under stirring with 150 ml of n-butyl acetate and 150 ml of water at room temperature. About 1 Ig of 30% aqueous sodium hydroxide are added to adjust the pH to about 12.5, and the phases are separated. The organic phase, containing 3 -ATM base, is washed twice with 30 ml of water, filtered on paper, and used in the subsequent hydrochloride preparation as follows.

Under stirring, while maintaining the temperature between about 15° and about 25°C with an ice-water bath, 8.55g of 36% aqueous hydrogen chloride are dropped on the filtered organic phase, resulting in the crystallization and suspension of the 3-ATM Hydrochloride (mw 291.82). The suspension is stirred at about 20°C for 1 hour, the solid is collected by filtration, washed twice with 35 ml of n-butyl acetate, and dried for 18 hours under vacuum at about 50° to about 600C, yielding about 19.6 g of product, having a melting point of about 159° to about 160°C, and 1H-NMR data of 9.58 ppm, bs, 2H; 7.35-7.20 ppm, m, 5H; 7.01 ppm, t, IH; 6.68-6.65 ppm, m, 2H; 6.59 ppm, dd, IH; 5.30 ppm, dd, IH; 3.10 ppm, quint, 2H; 2.57 ppm, t, 3H; 2.41 ppm, m, 2H; 2.22 ppm, s, 3H. The yield is 28 percent by weight based on the starting N-methyl-3-hydroxy-3-phenylpropylamine. The 3-ATM HCl is determined to be a mixture of enantiomers, with relative ratio: 99/1 by chiral HPLC.

EXAMPLE 2 Synthesis of N-methyl-3-(4-methylphenoxy)-3-phenyIpropylamine (4- ATM)

76g (0.969 mol) of dimethylsulfoxide, 4Og (0.242 mol) of N-methyl-3-hydroxy-3- phenylpropylamine (100%, mw 165.12), and 75.5g (1.211 mol) of potassium hydroxide (bulk industrial grade, 90% assay) are heated under stirring at 1000C ± 30C for 1 hour. The mixture is then cooled to from about 8O0C, then, under good stirring, 79.95g (0.726 mol) of 4-fluorotoluene are added in about 1 hour, maintaining the temperature at 830C ± 3°C, resulting in the precipitation of salts. The mixture is then stirred at 830C ± 3°C for about 5 hours. 300 ml of water and 300 ml of toluene are then added, the mixture is stirred for a few minutes, and the phases are separated. The aqueous phase is extracted with 50 ml of toluene, and the organic phases are collected and washed with three times with 80 ml of water. The washed organic phase is concentrated under vacuum at about 60° to about 80°C, producing about 70 g of a liquid brown residue of crude (±) 4-methyl Tomoxetine base (4-TMX) (about 0.2 mol of 100% base). 400 ml of toluene, 3 ml of methanol, and 19.Og (0.125 mol) of (S)- (+)-Mandelic acid are added at 250C, producing a slurry that is heated to a temperature of from about 65° to about 700C. The resulting solution is then cooled to 0°C for 2 hours. The solid 4-methyl Atomoxetine (S)-(+)-Mandelate salt (4- ATM-SMA) is isolated by filtration, washed with toluene, optionally washed with 40 ml of MTBE, and dried under vacuum at about 50° to about 600C, yielding 50 g of product.

The 50 g of dried 4-ATM-SMA are mixed under stirring with 250 ml of n-butyl acetate and 250 ml of water at room temperature. About 21 g of 30 percent aqueous sodium hydroxide are added to adjust the pH to about 12.5, and the phases are separated. The organic phase, containing the 4- ATM base, is washed twice with 30 ml of water, filtered on paper, and used in subsequent hydrochloride preparation as follows.

Under stirring, while maintaining the temperature between about 15° and about 25°C with an ice-water bath, 14.25g of 36 percent aqueous hydrogen chloride are dropped on the filtered organic phase. 4- ATM Hydrochloride (mw 291.82) crystallizes in bulk, and 40 ml of n-butyl acetate are added to allow stirring. The resulting suspension is stirred at about 200C for 1 hour, the solid is collected by filtration, washed twice with 40 ml of n-butyl acetate, and dried for 18 hours under vacuum at about 50° to about 600C, yielding 23.5 g of product, having a melting point of about 164° to about 1670C, for a 33 percent yield, based on the N- methyl-3-hydroxy-3-phenylpropylamine starting material. The 4-ATM HCl is found by chiral HPLC as a mixture of enantiomers, with a relative ratio of 58:42.

EXAMPLE 3 Synthesis of N-methyl-3-phenoxy-3-phenylpropylamine (D-ATM)

76g (0.969 mol) of dimethylsulfoxide, 4Og (0.242 mol) of N-methyl-3-hydroxy-3- phenylpropylamine (100%, mw 165.12), and 75.5g (1.211 mol) of potassium hydroxide (bulk industrial grade, 90% assay) are heated under stirring at 1000C ± 3°C for 1 hour. The mixture is then cooled to 80°C, and, under good stirring, 69.77 g (0.726 mol) of fluorobenzene are added in about 1 hour, maintaining the temperature at 830C ± 3°C, precipitating salts. The mixture is stirred at 83°C ± 3°C for about 5 hours, 300 ml of water and 300 ml of toluene are added, and the mixture is stirred for a few minutes. The phases are then separated, and the aqueous phase is extracted with 50 ml of toluene. The organic phases are collected, and washed three times with 80 ml of water. The washed organic phase is concentrated under vacuum at about 60° to about 8O0C, yielding about 7Og of liquid brown residue, i.e., crude (±) Desmethyl Tomoxetine base (D-TMX) (about 0.2 mol of 100% base).

400 ml of toluene, 3 ml of methanol, and 19.0 g (0.125 mol) of (S)-(+)-Mandelic acid are added at 25°C, producing a slurry that is heated to a temperature of from about 65° to about 7O0C (incomplete solubilization), followed by cooling to from about 5° to about 100C. Solid Desmethyl Atomoxetine (S)-(+)-Mandelate salt (D-ATM-SMA) is isolated by filtration, washed with toluene, and dried under vacuum at about 50° to about 6O0C, yielding 48 g of product. The 48 g of dried D-ATM-SMA are mixed under stirring with 250 ml of n-butyl acetate and 250 ml of water at room temperature. About 16 g of 30 percent aqueous sodium hydroxide are added to adjust the pH to about 12.5, and the phases are separated. The organic phase, containing D-ATM base, is washed twice with 35 ml of water, filtered on paper, and used in subsequent hydrochloride preparation as follows.

Under stirring, while maintaining the temperature between about 15° and about 250C with an ice-water bath, 12.3g of 36 percent aqueous hydrogen chloride are dropped on the filtered organic phase, resulting in the crystallization of D-ATM hydrochloride (mw 277.80). The resulting suspension is stirred at about 200C for 1 hour, the solid is collected by filtration, washed twice with 40 ml of n-butyl acetate, and dried for 18 hours under vacuum at about 50° to about 600C, yielding 32.7 g of product, a 48 percent yield based on the N-methyl-3- hydroxy-3-phenylpropylarnine starting material, having a melting point of about 170° to about 174°C. The D-ATM HCl is found by chiral HPLC to be a mixture of enantiomers, with a relative ratio of 69:31.