2.2. UV-VIS Spectroscopy and Fluorimetry

Sdndor G6r6g

2.2.1. Applications Without Chromatographic Separation [1]

2.2.1.1. Methods Based on Natural UV Absorption

The use of UV-VIS spectroscopy and spectrofluorimetry as tools for the identification and structure elucidation of impurities in drugs without chroma- tographic separation is of very little importance. The applicability of UV-VIS spectroscopy for this purpose is restricted to those instances where the drug material itself is spectrophotometrically completely inactive and the impurity absorbs selectively in the ultraviolet region above 220 nm. Due to the non- selective nature of the light absorption below 220 nm it is impossible to draw valuable conclusions regarding impurities from this spectral region: many of the functional groups occurring even in the spectrophotometrically inactive drug materials are chromophoric groups (isolated double bond, carboxyl group and its derivatives, etc.), which have their 7r-er* band in this region. Even the n-o-* band of auxochromic groups such as amines and halogen derivatives falls into the region below 220 nm.

Although in principle the situation is more favourable in the spectral range above 220 nm, in fact one has to be very cautious with drawing conclusions regarding impurities from the spectra in this range also. For example, if the drug to be tested contains an isolated carbonyl group, its n-er* band is around 280-290 nm. Although this band is very weak (e < 100) it can easily be confused with a strong band of an impurity falling into this spectral range. To draw any conclusion regarding impurities from the spectrum of a drug material requires its comparison with that of a sample of high purity. The spectra published in the literature should be treated with great caution. As an example the spectra of lynestrenol are shown in Fig. 2.2.A. Curve 1 is taken from a spectrum atlas [2] while curve 2 is recorded in the author's laboratory. Taking into consideration the lack of the shoulder around 240 nm in curve 2 and that the chromophoric groups in the molecule of lynestrenol are restricted to an isolated double and triple bond it can be stated that this shoulder in curve a was due to a conjugated impurity in the sample used for taking the spectrum published in the spectrum atlas. Although it is naturally impossible to say anything about the other parts of the molecule on the basis of the UV spectrum,

Organic Impurities 85

0.4

0.2

2 \ ! �9 w

240 280

, ! , ,

320 nm

Figure 2.2.A. Spectra of lynestrenol. (1) Spectrum taken from the literature [2]; (2) Spectrum taken in the author's laboratory, c - 0.05%; solvent: methanol

it is probable that it was the 3-oxo derivative of lynestrenol (about 0.25% norethisterone) but the possibility of the presence of some 3,5-diene derivative cannot be excluded either.

CH-, OH H " r l 3 ~ ~ -C~CH

(, Lynestrenol 3-oxo impurity 3,5-diene impurity

(norethisterone)

The use of UV spectrophotometry as a tool for the quantitative determi- nation or at least limitation of certain impurities in bulk drug materials or other compounds of pharmaceutical interest is of somewhat greater importance: even the latest editions of the pharmacopoeias contain several tests based on absor- bance measurements. The following examples have been taken mainly from the European Pharmacopoeia [3-5], British Pharmacopoeia 1998 [6] and the US Pharmacopoeia 24 [7].

86 Chapter 2

As the first example the paragraph "Conjugated compounds" in the monograph of ethynodiol diacetate is mentioned" the absorbance at 236 nm of the 0.05% solution should not exceed 0.47 [8] or 0.5 [7a]. In this case the term "conjugated compounds" can cover the 3,5-diene derivative (Amax = 236 nm) which is the product of the acid-catalysed decomposition of ethy- nodiol diacetate and also the 4-ene-3-oxo derivative (norethisterone acetate, which is an intermediate of the synthesis; ~max - - 240 nm). Fig. 2.2.B shows the spectra of ethynodiol diacetate spiked with increasing quantities of the 3,5-diene. It can be seen that at 236 nm at the given concentration the absorbance of pure ethynodiol diacetate is 0.15, about 0.1% of the impurity can be detected and above 0.2% it is easily measurable, moreover, at this level even the very characteristic fine structure in the spectrum of the impurity is detectable. The absorbance limit of 0.47 corresponds to about 1% of the 3,5-diene impurity.

0.60

0.45

A 5

0.30 4

0.15

, I ,,

220 240 260 280 ~. rim

Figure 2.2.B. Spectra of ethynodiol diacetate (0.05% w/v in ethanol) spiked with the impur- ity 17a-ethinylestra-3,5-diene-17-ol diacetate; (a)0%; (b)0.1%; (c)0.2%; (d) 0.4%; (e) 0.8%)

Organic Impurities 87

0 II

O--C~CH3 CH3 ~iC~C H

I C = o Ethynodiol diacetate I CH3

3,5-diene impurity 3-oxo impurity

Another example for the measurement of spectrophotometrically active impurities in inactive materials is the testing of dimeticone (methyl siloxane polymer) for phenylated compounds [3a] on the basis of the absorbance in the range of 250-270 nm, where the a-band of the phenyl moiety can be found. The selectivity and the sensitivity of the determination of the unsubstituted phenyl group can be improved by using derivative spectrophotometry. For example, the benzene content in 96% ethanol [6a] is determined by second- derivative ultraviolet spectrophotometry. The measurement of the amplitude of the peak at 253 nm in the 2D spectrum enables as low a benzene concentration as 2 ppm to be determined. The highly absorbing 5-hydroxymethylfurfural (8283 nm= 16.300), degradation product of sugars (e.g. fructose) is determined at its absorption maximum [3b].

The intense band of the conjugated system of a phenyl ring and a double bond, e.g. in apohyoscine at 245 nm is the basis for the determination of this impurity in hyoscine hydrobromide [3c] which contains only an isolated phenyl group poorly absorbing at this wavelength. The selectivity can be improved by using the absorbance ratio method (A246/A263) in the course of the determina- tion of apo-ipratropium in ipratropium bromide [3d]. The determination of as little as 0.1% of 1-isopropylamino-3-(2-prop-l-enylphenoxy)propan-2-ol impurity in alprenolol hydrochloride or benzoate (the isomeric non-conjugated 2-allyl derivative) is based on the same principle (Amax = 297 nm) [3e]. The problem of the direct determination of related aromatic ketones (phenones) in benzyl alcohol derivatives is quite similar. If there are no hydroxyl groups on the phenyl ring (phenylpropanonamine impurity in phenylpropanolamine hydrochloride [3f,7b] or piperidylpropiophenone impurity in trihexyphenidyl hydrochloride [6b]) the wavelength of the measurement is 285 and 247 nm, respectively, while in the case of phenol derivatives (isoprenalone in isoprena- line sulphate [3], adrenalone in epinephrine [7c], noradrenalone in noradrena- line hydrochloride [3g,7d], related phenones in terbutaline sulphate [3h,7e], phenylephrine [3i], metaraminol tartrate [6c], isoxsuprine hydrochloride [3j], isoetharine hydrochloride [7f], isoproterenol hydrochloride [7g], metaproter- enol sulphate [7h]) it is in the range of 310-330 nm. Because of the not

88 Chapter 2

completely selective absorption of the phenones at these wavelengths the measurement is suitable for the limitation rather than exact determination of the impurities.

/CH3 CH3 OH OH

c c = o o-c- H 9' ' CH2 CH2 CH2 I

I NH NH OH I I

CH TH Hyoscine Apohyoscine H3C/ \CH 3 H3C CH3

Isoprenaline Isoprenalone

Due to its biphenyl moiety the dimeric derivative of droperidol, 4,4 ~- bis[ 1,2,3,6-tetrahydro-4-(2-oxo- 1 -benzimidazolinyl)- 1 -piridyl] butyrophe- none absorbs selectively at 330 nm. This is the basis for the limit test (1.5%) of this impurity [7i]. The dimeric derivative of haloperidol can be measured in a similar manner at 335 nm [7j].

In some cases the selection of a suitable pH for the solvent enables the selective ionisation of the impurity and on the basis of this its selective deter- mination. For example, the absorption maxima of estramustine sodium phos- phate where the phenolic hydroxyl at position 3 is in the carbamoyl form exhibits maxima at 267 and 275 nm. This spectrum is not changed upon alkalinisation while that of free estradiol 17fl-phosphate impurity undergoes bathochromic shift in 0.1 M sodium hydroxide thus enabling its selective determination at 300 nm [6d]. The determination of phenol in phenoxyethanol [6e] and of (3-hydroxyphenyl)trimethylammonium bromide impurity in neos- tigmine bromide [3k] in alkaline medium at 287 and 294 nm, respectively, is based on the same principle.

In other instances the selectivity of the determination of the impurity can be assured by selective extraction prior to the spectrophotometric measure- ment. As the first example the determination of free chloramphenicol in chlor- amphenicol palmitate is presented. The sample to be tested is dissolved in xylene, the water-soluble impurity is extracted with water. After purification of the aqueous phase by extracting it with toluene the absorbance of the aqueous phase is measured at 278 nm thus enabling the selective determination of low quantities of free chloramphenicol impurity: the limit is set at 450 ppm [3]. Another example is the determination of non-quaternerised 3-dimethyla- minophenol as an impurity in edrophonium chloride (ethyl(3-hydroxyphenyl)- dimethylammonium chloride). The impurity is extracted with chloroform from

Organic Impurities 89

the aqueous solution of the drug buffered to pH 8. Following this, the impurity is re-extracted with 0.1 M aqueous sodium hydroxide and measured as the phenolate at 293 nm [6f]. The principle of the determination of non-quater- nerised impurities (expressed as 4-aminoquinaldine) in dequalinium chloride is similar. The impurity is extracted from alkaline solution with ether and after re- extraction with 1 M aqueous hydrochloric acid its quantity is calculated from the absorbances of the latter measured at 319 and 326.5 nm [6g]. When dimethylaminophenol impurity is measured in its quaternary ammonium deri- vative edrophonium chloride, the extraction is carried out with chloroform and the absorbance of the extract is directly measured at 252 nm [7k]. The free prednisolone content of prednisolone sodium phosphate can be measured at 241 nm after extraction with dichloromethane from an aqueous solution [71]. A limit test for lumifiavin in riboflavin and its 5~-phosphate sodium salt includes the treatment of the test substance with alcohol-flee chloroform, filtration and absorbance measurement of the filtrate at 440 nm [7m].

Ultraviolet spectrophotometry is suitable for the solution of even more delicate problems. For the determination of tetraene-type minor component (nystatin, amphotericin A) in the heptaene-type macrocyclic antibiotic drug amphotericin B the absorbances are measured at two wavelengths (282 and 304 nm) and calculation is based on the classical Vierordt method [ 1 ]. The limit is set to 10% [6h], 5% (parenteral preparations) or 15% (oral and dermatological preparations) [7n]. Another example is the limit test for bacitracin F and related substances in bacitracin. This comprises the absorbance measurement at two wavelengths: A290 nm/A252 nm should be less than 0.20 [3m].

Quite often the aim of the absorbance measurement above the cut-off wavelength of the drug substance to be tested is not the determination of a well defined impurity but the limitation of impurities absorbing in that spec- trum range in general. For example, the paragraph "Light-absorbing impuri- ties" in the monograph of oxytetracycline (Amax -- 353 nm, A 1%, 1 cm 300) prescribes the measurement of the absorbance at 430 and 490 nm in the 99:1 mixture of methanol and 1 M hydrochloric acid. The specific absorbances at these wavelengths must not exceed 1.25 and 0.2, respectively [3n]. The absor- bance of warfarin sodium (1.25 g in 10 ml of 1 in 20 solution of sodium hydroxide; 1-cm cell) should be less than 0.1 [3o]. The limit test of non- hydrogenated alkaloids in ergoloid mesylates is based on the absorbance ratio A317.5/A280 [7o]. Sometimes the requirements are related to wavelength ranges rather than well-defined wavelengths. For example the absorbance of a 1% w/v aqueous solution of betadex (fl-cyclodextrin) must not exceed 0.10 at any wavelength between 230 and 350 nm and 0.05 between 350 and 750 nm [3p].

90 Chapter 2

2.2.1.2. Methods Based on Colour Reactions

As seen from the examples discussed in this section, the use of ultraviolet spectroscopy in the direct determination of related impurities in drugs is restricted to a limited number of advantageous cases. These possibilities can be somewhat expanded by using selective chemical reactions to shift the spec- tra of the impurities towards the longer wavelengths, usually to the visible spectrum range. In the early period of pharmaceutical analysis these colour reactions were the most important means for the detection and as far as possi- ble determination of impurities in drugs [1,9-13]. As a consequence of the introduction and spreading of chromatographic methods their importance has been greatly decreased, b u t - as it will be demonstrated on the examples taken from the principal pharmacopoeias [3-7] - many of them are still in use. In some cases these colour reactions are followed by absorbance measurements enabling the limitation of the quantity of the impurity on an objective basis. In other cases they are used as limit tests based on the visual comparison of the colours of the test and reference solutions.

The classical colour reaction based on complex formation of salicylic acid with iron(III) reagents is used among others for the determination of free salicylic acid impurity in carbasalate calcium (equimolecular compound of calcium-di(acetylsalicylate) and urea). The reagent is iron(III)nitrate and the analytical wavelength is 525 nm. The absorbance limit of 0.115 of the test solution with a carbasalate calcium concentration of 0.2 g/100 ml is equivalent to 0.5% [4a]. In the limit tests for salicylic acid in aspirin [3r,7p] and benolirate [6i] iron(III)ammonium sulphate or iron(III)chloride is used as the reagent and visual comparison of the colours is carried out. The same method is used for the estimation of 5-chlorosalicylic acid in niclosamide. The high sensitivity of the visual method can be characterised by the low limit for the impurity (60 ppm) [3s]. Complexation with iron(III)chloride is used for the determination of meconic acid (3-hydroxy-4-oxo-4H-pyran-2,6-dicarboxylic acid) impurity in morphine ( / ~ m a x - - 480 nm) [4b] and streptomycin impurity in dihydrostrepto- mycin (Amax - 550 nm) [7q]. Other methods based on metal complex formation are, e.g. the determination of peroxides in copovidone and crospovidone [5a] or in ether [7r] with the aid of titanium(IV)chloride reagent (Amax- 405 and 410 nm, respectively), the determination of non-tertiary amine impurities in mebe- refine hydrochloride using the copper(II)chloride-pyridine reagent (Amax -- 405 nm) [6j], determination of glycerol in gold sodium thiomalate with copper(II)chloride reagent [7s] and the determination of halogenated compounds in benzyl alcohol after splitting the carbon-halogen bond by nickel-aluminium alloy followed by reaction with the mercury(II)thiocyanate reagent [3t].

Another group of methods is based on the classical colour reaction of

Organic Impurities 91

diazotisation with nitrous acid of primary aromatic amine-type impurities in acylamino type drug materials followed by azo coupling to form azo dyes. A reaction of this type is Fig. 2.2.C taking the determination of free sulfathiazole in succinylsulfathiazole as the example. The coupling agent is N-(1-naphtyl)- ethylenediamine [3u]. The absorbance of the forming dye is compared at 550 nm with that of a standard solution of sulfathiazole. Sulfathiazole is measured in phthalylsulfathiazole in a similar manner [3v]. The same reaction is used also in the limit tests with visual comparison of the intensities of the colours, e.g. reddish-blue colour in the estimation of 4-chloroaniline impurity in chlor- hexidine [3w] or pinkish-violet in the estimation of 2-chloro-4-nitroaniline impurity in niclosamide. The high sensitivity of the method enables to set the limit in the latter case to as low as 100 ppm [3x]. Further applications of the same reaction are the estimation of volatile diazotisable impurities in 4- aminobenzoic acid [7t], diazotisable impurities in hydroflumethazide [7u] as well as free aromatic amines in diatrizoate [7v], iopamidol [7w], iothalamic acid [7x] and iohexol [7w]. Another coupling agent is ce-naphtol which is used for the determination of free aromatic amines in amidotrizoic acid (/~max - - 485 nm) [5b].

An even more sensitive method is the estimation of 4-aminophenol impurity in benorilate [6i] and acetaminophen [7y] by using the sodium nitroprusside-sodium carbonate reagent (visual comparison of the colour intensities; limit: 20 ppm [6i] or absorbance measurement at 710 nm; limit 50 ppm [7z]). The same reagent is used for the estimation of primary amines and ammonia in piperazine [7z] and N-aminohexamethyleneimine in tolazamide [7a~]. An example for the limit test for non-aromatic primary amines in tertiary amines is their estimation in benzydamine hydrochloride

NH2 N~N ci 0 NH2

N - N NH NaNOz

O = S = O ~-- O = S = O I HCI I HN--CH2~CH2~NH2 NH NH

S/j'% N S/J'% N O = S = O

NH

Figure 2.2.C. I'ransformation of sulfathiazole to an azo dye

92 Chapter 2

based on the yellow colour of the Schiff's base formed with the 4-dimethy- laminobenzaldehyde reagent [6k].

The limit tests of aldehyde-type impurities can be based on their conden- sation with phenylhydrazine, e.g. in the test for nicotinaldehyde in nicotinyl alcohol tartrate (hmax of the forming hydrazone is 370 nm) [61] or their redu- cing properties can be utilised. The first example for the latter is the limit test in betadex [3p] for reducing sugars, where the latter reduce the copper(II)-tartrate reagent and the forming copper(I) is measured by the molybdenum blue reac- tion using a sodium arsenate-ammonium molybdate reagent. The absorbance at 740 nm is compared with that of a glucose reference solution (limit: 0.2%). Free fructose in inulin is measured with the aid of the blue tetrazolium reaction ( h m a x - - 530 nm) [7b~]. Another example is the enzymatic oxidation of alde- hydes in povidone using NAD (nicotinamide-adenine-dinucleotide) as the oxidising agent and aldehyde dehydrogenase enzyme as the catalyst. The test is based on the absorbance at 340 nm of the formed NADH (limit: 500 ppm expressed as acetaldehyde) [3y,7c~]. Formaldehyde is measured in tyloxapol with the phenylhydrazine-potassium ferricyanide reagent (limit: 75 ppm, / ~ m a x -- - 520 nm) [7d~]. For the determination of the same in oxidised cellulose chromotropic acid reagent is used (limit 0.5%, /~max = 570 rim) [7e~].

The classical detection methods for free morphine in codeine are still in use: formation of 2-nitromorphine with nitrous acid followed by the estimation of the orange-red colour in ammoniacal solution (limit 0.1%) [3z] and its oxidation with the iron(Ill)chloride-potassium ferricyanide reagent and esti- mation of the blue colour [7f~]. The latter method is in use for the estimation of morphine in noscapine [7g~]. Another classical reagent (for reducing substances) is tetrazolium blue. It is used for the very sensitive detection of unspecified reducing substances in ergocalciferol (Amax -- 525 nm; limit 2 ppm expressed as hydroquinone) [3a~,7h~]. The limit test (300 ppm) for oxalate in sodium citrate begins with the reduction of the latter to glyoxalic acid with zinc/hydrochloric acid followed by condensation with phenylhydrazine and terminated with oxidation of the phenylhydrazone with potassium ferricyanide reagent to form a pink reaction product [3b~]. Another reagent for oxalate, iron salycilate is applied for its estimation in cromolyn sodium ( /~max = 480 rim) [7i~]. The reagent for the detection of iminodibenzyl in desipramine hydro- chloride [7j ~] is furfural-hydrochloric acid.

The chemistry behind the estimation of two impurities in fosfestrol sodium (bis-orthophosphate tetrasodium salt of stilbestrol) is quite interesting [ 14]. Unesterified free stilbestrol is measured after extraction of this impurity from an aqueous solution with dichloromethane, followed by transformation by short-wave UV irradiation to a tricyclic conjugated tetraene-dione structure measurable at 418 nm (limit 0.15%); see Fig. 2.2.D. The reducing properties of the free phenolic moiety of the semi-esterified stilbestrol sodium monopho-

Organic Impurities 93

/•OH H5C27c=~ C UV light

H O / ~ ~C2H5

O O

Figure 2.2.D. UV-light induced transformation of stilbestrol

sphate impurity are the basis for the estimation of this impurity in the aqueous phase of the above extraction using the phosphomolybdotungstic reagent (Amax = 660 nm; limit 1%) [6m].

There are many more methods for the spectrophotometric estimation of related impurities in drugs in other pharmacopoeias and in the (usually early) literature [1,9-13]. Generally speaking, however, the importance of these methods is expected to further decrease in the future: the TLC or HPLC methods are suitable limit tests even for those named impurities which are now still measured by spectrophotometric-colorimetric methods (in the major- ity of cases in parallel with the chromatographic purity tests). It is to be noted that in addition to the above discussed related impurities colorimetric methods are widely used as the limit tests for inorganic impurities (arsenic, various metals, sulphite, hydrazine, etc.). These are described in Section 4.1.

2.2.1.3. Fluorimetric Methods

Fluorimetry without preliminary chromatographic separation is only seldom applied for the estimation of impurities in drugs. Two examples are shown; in both instances the measurement is carried out after derivatisation reactions. The limit test (0.5%) for hydroxylamine in acetohydroxamic acid is based on derivatisation with pyridoxal 5-phosphate and measurement of the forming aldoxime; excitation at 350 nm, emission at 450 nm [7k~]. The quan- tity of aminobutanol in ethambutol hydrochloride is limited to 1%. Its primary amino group is selectively derivatised with fluorescamine followed by fluori- metric measurement (excitation: 385 nm, emission: 485 nm) [71 ~] (see Fig. 2.2.E).

2.2.2. Applications after Chromatographic Separation

UV-VIS spectroscopy and fluorimetry are important tools for the identi- fication, moreover, structure elucidation of impurities after their planar chro-

94 Chapter 2

NH2 I

H O - - C H 2 ~ C H I

CH2 I

CH 3

CH3 I

/ ~ c.2

% d o o= .Oo

fluorescamine ~

Figure 2.2.E. Derivatisation of aminobutanol for the fluorimetric measurement

matographic separation from the main component and from each other. Spot elution techniques for these purposes can be regarded to be obsolete. However, in situ UV-VIS spectra (especially in the reflection mode) and fluorescence spectra without or after colour development with suitable spray reagents are widely used in this field. The evidence for the identity of a separated material with a reference standard can be greatly strengthened by adding to the identity of the Rf values the identity of these spectra. In addition to this the spectra can contribute also to the structure elucidation of unknown impurities. These aspects are discussed in Sections 2.1.2-2.1.4.

More or less the same applies to the use of UV-(VIS) spectroscopy and fluorimetry after column chromatographic separation. Before the introduction and spreading of high-performance liquid chromatography (HPLC) this was the most reliable and widely used method for the separation of impurities in drugs for their identification, structure elucidation and quantitative determina- tion. Of the immense literature on this technique two monographs [ 15,16] and a review are mentioned [17].

At present the technique of classical column chromatography has been almost completely superseded by its modem variant, the HPLC method. Using this technique the identity of the spectra easily obtainable with the aid of diode- array UV detectors greatly improves the value of identification of the impu- rities on the basis of retention matching with reference standards (see Section 2.7.3). As for the application of the spectra for structure elucidation of unknown impurities, it is usually unnecessary to take the spectra after (semi)- preparative column chromatographic separation (usually preparative HPLC). The reason for this is that in the majority of cases a good-quality diode-array UV spectrum is available already at the beginning of the search for the struc- ture of the impurity. The role of these spectra in this research is discussed in Section 2.1.4 while in Section 2.7.3 several examples are presented to demon- strate the possibilities and limitations of these spectra in the structure elucida- tion of related impurities in drug materials. The quantitative aspects of HPLC

Organic Impurities 95

(where almost exclusively UV or fluorimetric detectors are used for the deter- mination of impurities in drugs) are discussed in Section 2.7.1. What is described for the role of UV spectroscopy and spectrophotometry as well as fluorimetry in high-performance liquid chromatography, almost exactly applies to capillary electrophoresis and capillary electrochromatography, too (see Sections 2.8 and 2.9, respectively).

References

1. S. G6r6g, Ultraviolet-Visible Spectrophotometry in Pharmaceutical Analysis, CRC Press, Boca Raton, FL (1995)

2. H.-W. Dibbern, UV and IR Spectra of Some Important Drugs, Vol. III, Edition Cantor, Aulendorf, No. 2221 (1980)

3. European Pharmacopoeia, 3rd edn, Council of Europe, Strasbourg (1997). Page numbers: a, 753; b, 881; c, 993; d, 1048; e,368; f, 1055; g, 1242; h, 1608; i, 1322; j, 1064; k, 1225; 1, 593; m, 436; n, 1272; o, 1721; p, 465; r,345; s, 1229; t, 457; u, 1547; v, 1329; w, 599; x, 1229; y, 1370; z 673; a I, 802; b I, 1483

4. European Pharmacopoeia, 3rd edn, Supplement 1998, Council of Europe, Strasbourg (1998). Page numbers: a, 222; b, 389

5. European Pharmacopoeia 3rd edn, Supplement 1999, Council of Europe, Strasbourg (1999). Page numbers: a, 399; b, 251

6. British Pharmacopoeia 1998, The Stationery Office, London (1998). Page numbers: a, 548; b, 1330; c, 863; d, 542; e, 1022; f, 522; g, 432; h, 94; i, 147; j, 847; k, 1312; 1, 929, m, 619

7. The United States Pharmacopoeia 24, USP Convention Inc., Rockville (2000). Page numbers: a, 698; b, 1319; c, 645; d, 1196; e, 1604; f, 919; g, 928, h, 1051; i, 614; j, 804; k, 624; 1, 1384; m, 1480; n, 135; o, 655; p, 161; q, 568; r, 692; s, 783; t, 110; u, 835; v, 533; w, 899, x, 908; y, 17; z, 1340; a/, 1674; b/, 888; c/, 1372; d I, 1727; e/, 358; ( , 461; g/, 1207; h/, 651; i/, 475; j/, 505; k/, 43; 11, 689

8. British Pharmacopoeia 1993, p 270, The Stationery Office, London (1993)

9. M. Pesez and J. Bartos, Colorimetric and Fluorimetric Analysis of Organic Compounds and Drugs, Marcel Dekker, New York (1974)

10. B. Kakac and Z.J. Vejdelek, Handbuch der Kolorimetrie, I, II (Kolorime- trie in der Pharmazie), III (Kolorimetrie in der Biologie, Biochemie und Medizine), Gustav Fischer Verlag, Jena (1962-1966)

11. B. Kakac and Z.J. Vejdelek, Handbuch der photometrischen Analyse organischer Verbindungen, Verlag Chemie, Weinheim (1974)

12. Z.J. Vejdelek and B. Kakac, Farbreaktionen in der spektrophotome-

96 Chapter 2

trischen Analyse organischer Verbindungen, I, H Ergiinzungsband I, H, Gustav Fischer Verlag, Jena (1969-1973, 1980-1982)

13. F.D. Snell and C.T. Snell, Colorimetric Methods of Analysis, III, IV (Organic Compounds)IliA, IVA, Van Nostrand, Princeton (1948-1970)

14. T.D. Doyle, W.R. Benson and N. Filipescu, J. Am. Chem. Soc. 98, 3262- 3267 (1976)

15. O. Mikes (Ed.), Laboratory Handbook of Chromatographic and Allied Methods, Ellis Horwood, Chichester (1979)

16. Z. Deyl, K. Macek and J. Jan~k, Liquid Column Chromatography, Else- vier, Amsterdam (1975)

17. T.D. Doyle and J. Levine, J. Assoc. Off. Anal. Chem. 61, 172-191 (1978)