Journal of Chromatographic Science, Vol.

34, January 1996

Rapid Direct Analysis of p-Xylene Oxidation Products

by Reversed-Phase High-Performance Liquid
Chromatography
Antonello Viola and Giacomo Cao*
D i p a r t i m e n t o di Ingegneria C h i m i c a e Materiali, U n i v e r s i t a di Cagliari, P i a z z a D A r m i , 09123 C a g l i a r i , Italy

Abstract acetic acid by gas chromatography and high-performance liquid

chromatography (HPLC), respectively. Solid products were an­
A reversed-phase high-performance liquid chromatographic ( H P L C ) alyzed after esterification by gas chromatography. Very few in­
method is developed for the determination of the products of liquid dications about the analytical conditions have been given; in ad­
phase oxidation of p-xylene to terephthalic acid catalyzed by cobalt dition, two different sets of equipment and a time-consuming
naphtenate w i t h methyl benzoate as a solvent. Analyses are pretreatment of the solid samples were employed. Jacobi and
performed on a 2 5 0 - × 4 . 6 - m m octadecylsilica ( O D S ; C ) bonded1 8
Baerns (4) analyzed the reaction products of p-xylene oxidation
phase column w i t h a 1 0 - × 3 - m m guard column in 22 min. A three- in the absence and presence of solvent (i.e., chlorobenzene and
component mobile phase (methanol, acetonitrile, water) is used for
benzoic acid methylester) by H P L C . Depending upon the
gradient elution, and ultraviolet detection is selectively performed at
number of nuclei present in the products, different mixtures of
220 and 254 n m . T h e internal standard method (using cumene as the
internal standard) is employed. T h e developed method avoids
eluent have been used by maintaining a constant mobile phase
filtration, extraction, and derivatization pretreatment of the samples. flow rate (i.e., 1 mL/min) at a temperature of 34°C. No infor­
mation about the duration of the analysis has been reported.
More recently, Cao and co-workers (5,6) investigated the liquid-
phase catalytic oxidation of p-xylene to p-toluic acid and to
Introduction
terephthalic acid using methyl benzoate as a solvent by ana­
lyzing the reaction products with HPLC. For the analysis of ox­
Liquid-phase catalytic oxidation of p-xylene to terephthalic idation products up to terephthalic acid formation, a relatively
acid plays an important role in the petrochemical industry (1). rapid and easy-to-use technique that represents the natural
This process has received a great amount of attention in the lit­ upgrade of the one developed for the detection of the products
erature, with particular emphasis on the effects of the catalyst up to p-toluic acid formation (5) has been employed (6).
and promoter concentrations, the nature of the solvent, the re­ The aim of this work was to describe in detail the develop­
action temperature, and other conditions on the oxidation rate ment and the performance of this technique that allows the
(1,2 and references therein). Few studies have addressed the de­ rapid and simultaneous analysis of p-xylene, p-tolualdehyde,
velopment of kinetic schemes of this process that are simple but 4-methylbenzyl alcohol, p-toluic acid, 4-hydroxymethylben-
yet provide a description of the product distribution sufficient zoic acid, 4-carboxybenzaldehyde, terephthaldicarboxaldehyde,
to evaluate the performance of industrial processes (3-6). In terephthalic acid, methyl benzoate, and cumene. The method is
these studies, the complex radical chain mechanism of this free of filtration, extraction, and derivatization steps.
process has been typically lumped into appropriate kinetic
schemes that account for the most important intermediates and
final products, that is, p-tolualdehyde, 4-methylbenzyl alcohol,
p-toluic acid, 4-hydroxymethylbenzoic acid, 4-carboxyben- Experimental
zaldehyde, terephthaldicarboxaldehyde, and terephthalic acid.
As lumped kinetic schemes of this type can be proposed only on Principle
the basis of experimental evidence, a key point is represented in For the determination of the oxidation products of p-xylene
general by the analytical techniques employed. Cavalieri d'Oro within a relatively wide range of concentrations, a reversed-
and co-workers (3) analyzed reactants and intermediate prod­ phase HPLC method with ultraviolet detection was developed.
ucts obtained from low-temperature oxidation of p-xylene in It is based on the gradient elution technique in which a mobile
phase consisting of methanol, acetonitrile, and water and an
* Author to whom correspondence should be addressed. octadecylsilica (ODS; C ) column were used.
18

Reproduction (photocopying) of editorial content of this journal is prohibited without publisher's permission. 27
 Journal of Chromatographic Science, Vol. 34, January 1996

Selective detection of the various compounds at 220 and 254 mobile phase and reduce the analysis run time, the methanol and
nm was performed. Analyses were carried out by the combination acetonitrile content were then progressively increased. Small
of an isocratic period and a series of linear elution gradients of the quantities of phosphoric acid were added into the mobile phase
mobile phase. The starting composition of the mobile phase re­ to avoid asymmetrical effects and peaks tailing.
sulted in a highly polar mixture that allowed effective interaction The internal standard method was adopted due to its capa­
of 4-hydroxymethylbenzoic acid, terephthalic acid, tereph- bility of minimizing errors deriving from system and procedure
thaldicarboxaldehyde, and 4-carboxybenzaldehyde with the sta­ variations (7–9). In addition, this method is particularly useful
tionary phase. In order to enhance the solvent strength of the since it avoids frequent recalibrations because the response
factors of all analytes remain practically constant for relatively
Table I. Maximum Absorption Wavelengths of all long periods of time (10).
Analytes for HPLC Detection in the Range 195–320 nm

Analyte Wavelength (nm) Apparatus

The equipment used included a Hewlett-Packard 1050 liquid
p-Xylene 207, 263 chromatograph equipped with a quaternary pump (Model
p-Tolualdehyde 254 79852A, Hewlett-Packard; Palo Alto, CA), a helium degassing
4-Hydroxymethylbenzoic acid 231 system for solvents, a multiple wavelength ultraviolet detector
4-Carboxybenzaldehyde 249 (Model 79854A) with an 8-pL volume and a 6-mm optical path
Cumene 263 internal flow cell, an autosampler (Model 79855A), and an inte­
4-Methyibenzyl alcohol 215-225, 259
grator (Model 3396A) connected to an IBM PC-XT 8086 con­
p-Toluic acid 246
trolling a data acquisition system (Woodward McCoach, Inc.;
Terephthaldicarboxaldehyde 257
237
West Chester, PA). A Chrompack Chromsep Spherisorb 5 ODS-
Terephthalic acid
Methyl benzoate 225
2 column (250×4.6-mm; particle size, 5μm;pore size, 8 nm;
2 2
superficial area, 220 m /g; surface coverage, 2.2 μmol/m ; carbon
content, 10.5%) with a 10-×3-mm guard column was used.

Table II. Working Concentration Range of all Analytes

Reagents
Analyte Concentration(%) HPLC-grade water (Art. n. 412142, Farmitalia Carlo Erba;
Milano, Italy), HPLC-grade acetonitrile (Art. n. 412411, Farmi­
p-Xylene 5-44 talia Carlo Erba), HPLC-grade methyl alcohol (Art. n. 412142,
p-Tolualdehyde 1–3 Farmitalia Carlo Erba), and 75% phosphoric acid (Art.
4-Hydroxymethylbenzoic acid 0–1 n. 304051, Farmitalia Carlo Erba) were used as solvents for the
4-Carboxybenzaldehyde 0–1 mobile phase preparation. Standard solutions for calibration
Cumene 13–16
were prepared using the following: 4-hydroxymethylbenzoic
4-Methylbenzyl alcohol 0–1.5
acid (Art. η. H5892, Sigma; St. Louis, MO), terephthalic acid
p-Toluic acid 0–7
Terephthaldicarboxaldehyde 0–1
(Art. n. 18,536-1, Aldrich-Chemie; France), 4-carboxybenzalde­
Terephthalic acid 0–0.5 hyde (Art. n. 12,491–5, Aldrich-Chemie), terephthaldicarbox­
Methyl benzoate 45–58 aldehyde (Art. η. Τ 220-7, Aldrich-Chemie), 4-methylbenzyl al­
cohol (Art. n. 15.592.72, Janssen Chimica; Gardena, CA),

Table III. Routine Chromatographic Parameters of all Analytes for the Quality Test of the Analytical Method

Retention Capacity Asymmetry No. theoretical

Analyte time (min) Resolution* factor †
factor plates

4-Hydroxymethylbenzoic acid 10.517 3.844 1.01 64,667

Terephthalic acid 11.615 12.76 4.35 1 128,839
4-Carboxybenzaldehyde 12.77 11.49 4.88 1 50,478
Terephthaldicarboxaldehyde 13.204 2.56 2.56 1 18,100
5.08
4-Methylbenzylalcohol 14.272 7.19 5.57 1 259,054
p-Toluic acid 15.229 15.2 6.04 1.15 280,132
p-Tolualdehyde 16.543 18.83 6.62 1.21 358,848
Methyl benzoate 17.049 7.84 6.85 1.3 393,140
p-Xylene 20.617 57.55 8.5 1.01 654,121
Cumene 21.346 12.25 8.83 1 725,168

* Corresponds to the analyte in the row and to the previous one.

†
Calculated using the retention time of uracil (2.171).

28
Journal of Chromatographic Science, Vol. 34, January 1996

p-toluic acid (Art. n. 13.906.35, Janssen Chimica), p-tolualde- under a helium flux before and during the analyses (7). The
hyde (Art. 13.900.29, Janssen Chimica), p-xylene (Art. n. 492504, volume of the samples injected by the autosampler was 20 μL,
Farmitalia Carlo Erba), cumene (Art. n. 18.101.59, Janssen and the analyses were carried out with a flow rate of 1.5 mL/min
Chimica), and methyl benzoate (Art. m 12.634.24, Janssen by applying the following elution gradient program: isocratic
Chimica). For sample dilution, HPLC-grade methyl alcohol conditions with 95% water and 5% acetonitrile for 5 min; gra­
(Artn. 412533, Farmitalia Carlo Erba) was used. All the reagents dient elution from 5 to 12 min until a mobile phase composi­
were pure for analytical use. tion of 55% water, 35% acetonitrile, and 10% methanol was
reached; a second linear gradient from 12 to 20 min, thus
Working conditions reaching the final composition of 15% water, 75% acetonitrile,
The column was first conditioned for about 30 min with a and 10% methanol. T o compensate for mobile phase com­
mobile phase containing 95% water and 5% acetonitrile. The pressibility, a value of 60E-6/bar and automatic stroke volume
solvents of the mobile phase were degassed and maintained setting on the pump module were selected, thus obtaining an

Table IV. Relative Standard Deviations (%) of Peak Areas and Retention Times

Analysis number

†
Compound* 1 2 3 4 5 6 7 %RSD Tr

p-Xylene
Cone. 11 21.5 55 135 270 540 1085 0.018
%RSD 2.2 2.1 1.1 0.9 1 0.8 0.9

4-Methylbenzyl alcohol
Cone. 0.41 0.82 2 5.1 10.2 20.5 41 0.042
%RSD 7.6 3.1 2.6 1.2 2 0.7 0.6

p-Tolualdehyde
Conc. 0.78 1.6 3.9 9.7 19.5 39 77.9 0.037
%RSD 3.3 0.7 1.2 1.1 0.8 0.8 0.7

p-Toluic acid
Conc. 2 4 9.9 24.8 49.6 99.2 198 0.041
%RSD 8.3 1.4 1.1 1 1.9 0.9 1.8

4-Hydroxymethylbenzoic acid
Cone. 0.33 0.67 1.7 4.2 8.4 16.7 33.4 0.087
%RSD 7.8 3.8 3 3.3 3.1 3 1.9

Terephthaldicarboxaldehyde
Conc. 0.14 0.28 0.7 1.8 3.5 7.1 14.1 0.052
%RSD 5.5 3.8 2.6 0.7 1.6 0.9 1.9

4-Carboxybenzaldehyde
Conc. 0.33 0.67 1.7 4.2 8.4 16.9 33.7 0.052
%RSD 2.5 2.8 0.8 0.8 1.3 0.9 0.8

Terephthalic acid
Conc. 0.18 0.36 0.9 2.3 4.5 9.1 18.2 0.058
%RSD 9 3.1 1.3 0.9 1.7 1.1 1.1

Cumene
Conc. 2 4 28.5 70 140 285 565 0.019
%RSD 5.7 3.4 2.1 1.1 1 0.9 0.7

Methyl benzoate
Conc. 17.5 35.5 90 220 445 885 1770 0.03
%RSD 2.4 1 1.2 1 0.9 0.8 0.7

* Concentrations are given in micrograms per milliliter of methanol. Percent relative standard deviations are for peak area counts.
†
Percent relative standard deviation of retention times.

29
 Journal of Chromatographic Science, Vol. 34, January 1996

optimum pressure ripple within the whole analysis duration. Standard preparation
After each run, the column was conditioned for 15 min at the Several standard solutions of the species involved during the
starting mobile phase composition (95% water, 5% acetoni­ oxidation of p-xylene to terephthalic acid were prepared to
trile). The detection of various compounds was achieved by cover a wide calibration range of concentrations, as reported in
varying the wavelength as follows: 254 nm from 0 to 13.8 min; Table II. In order to avoid solubility problems, solid species
220 nm from 13.8 to 15.9 min; 254 nm from 15.9 to 22 min. were weighted and then mixed with about 50 mL methanol in
Note that the wavelength variation was selected in order to a 100-mL Pyrex Erlenmeyer flask. The solution was shaken
achieve a high degree of reproducibility and detection (good and heated until the complete solubility of all components was
peak resolution and elution gradient characteristics). In fact, if achieved. It was then thoroughly transferred in a 100-mL vol­
a wavelength program in which the values of maximum ab­ umetric flask where all liquid species (p-tolualdehyde,
sorption of all species (Table I) were used, quite low precision p-xylene, cumene, and methyl benzoate) had been previously
results due to signal drift and background noise would occur. In weighted. The resulting solution was finally diluted up to
particular, detection was at 254 nm for 4-hydroxymethylbenzoic 100 mL with methanol. For the analysis, 0.1 mL of each cali­
acid, terephthalic acid, terephthaldicarboxaldehyde, and 4-car- bration standard was diluted to 5 mL with methanol.
boxybenzaldehyde. Detection occurred at 220 nm for 4-methyl­
benzyl alcohol and p-toluic acid. Finally, the second change of Sample preparation
wavelength (i.e., 254 nm) allowed p-tolualdehyde, p-xylene, Determinations were performed on samples that were ob­
methyl benzoate, and cumene to be detected. On the detector tained directly from a batch oxidation laboratory reactor (5,6)
module, the following bandwidths were used: 8 nm for the se­ where the oxidation of p-xylene to terephthalic acid (catalyzed
lected wavelengths; a reference wavelength of 400 nm with a by cobalt naphtenate with methyl benzoate as a solvent) was
bandwidth of 80 nm; and a multiplication factor of 0.5 for re­ studied. About 0.5 mL of solution was taken from the reactor
ducing the signal noise. A peak width with 0.1 s response time by means of a syringe or a pipette that was brought up to a
was selected for the signal to obtain a good detection of all temperature sufficiently high to prevent the precipitation of
peaks, which were narrow due to the high system efficiency. All p-toluic acid and terephthalic acid. Each sample was accurately
analyses were carried out at room temperature. weighted and added with a precise quantity (15% of the sample
weight) of cumene as an internal standard, and finally the mix­
ture was diluted with 25 mL methanol. Only 0.1 mL of the
sample obtained as described was diluted up to a volume of
1 mL with methanol in 1.5-mL vials that were finally used for
analyses. Each sample was run in triplicate.

Results and Discussion

The system suitability was verified by checking all significant

parameters for method validation, according to the guidelines
and recommendations available in the literature (11–13). The
resolution, R , of all components was calculated according to
S

the following relationship (14):

Eq 1

where t is the retention time of each component as reported in

R

Table III, and w is the mean peak width. All species displayed op­
timal resolution values (greater than or equal to 1.5), as indi­
cated in Table III between each pair of components, corre­
Figure 1. Chromatogram of a standard solution. Conditions: column,
Chrompack Spherisorb 5 ODS-2 (250×4.6 mm); temperature, room; mo­
sponding to a complete baseline separation of all peaks.
bile phase, H O – C H C N – C H O H + 0 . 1 % H P O ; flow rate, 1.5 mL/min;
2 3 3 3 4
The capacity factors for all compounds were estimated by in­
injection, 20 μL; detection, ultraviolet at 254, 215, and 254 nm. The fol­ jecting a Uracil standard solution (Chrompack, test mix 201) as
lowing gradient elution program was used: 0–5 min, 95% H O and 5% 2
the unretained component on the analytical stationary phase.
C H C N ; 5–12 min, 55% H O , 35% C H C N , and 10% C H O H ; 12–20
3 2 3 3
The calculated values of the capacity factors were all acceptable
min, 15% H O , 75% C H C N , and 10% C H O H . Peaks: 1, 4-hydrox-
2 3 3

ymethylbenzoic acid (10.52 min); 2, terephthalic acid (11.64 min); 3,4-car- (14) and are indicated in Table III. All peaks were found satis­
boxybenzaldehyde (12.75 min); 4, terephthaldicarboxaldehyde (13.35 factorily symmetrical with good asymmetry factors, which are
min); 5, 4-methylbenzyl alcohol (14.38 min); 6, p-toluic acid (15.4 min); also reported in Table III.
7, p-tolualdehyde (16.71 min); 8, methyl benzoate (17.24 min); 9, p-xylene The system efficiency was evaluated in terms of number of
(20.81 min); 10, cumene (21.55 min).
theoretical plates of the column for each compound with which

30
Journal of Chromatographic Science, Vol. 34, January 1996

satisfactory results could be obtained (Table III). Calculations system stability. The system stability, which should be guaran­
were made according to the following relationship (14): teed for at least 24 h (16), was checked up to 6 days by analyz­
ing standard solution samples that were properly stored in a
refrigerator during that time (Figures 2A-2C). The results show
Eq 2
good stability for standard samples. The same conclusions are
also valid for samples taken from the experimental oxidation
where Ν represents the number of theoretical plates, t is the R reactor (5,6).
retention time, and is the peak width at half height. Detection and quantitation limits were evaluated with respect
The precision of the developed analytical method was studied to signal-to-noise ratios of 2 and 10, respectively, through the
using several standard solutions in a wide range of concentra­ following relationships (11,13):
tions. The results summarized in Table IV were achieved by run­
ning each standard six times, according to the literature guide­
lines (11,13); a typical chromatogram of a standard solution is Eq 3
also shown in Figure 1. Note from Table IV that the values of
percent relative standard deviation are always less than 3.8%,
except for the lowest level of concentration of each species
where the maximum allowed limit of 15% indicated in the lit­
erature (15) is not exceeded. The values of the percent relative Eq 4
standard deviations of the retention time of each species are re­
ported in Table IV. These results are extremely low, and in gen­ where DL and QL represent the detection and the quantitation
eral, they come within 0.1%. limits, respectively, I is the injected amount, and S/N is the
The accuracy of the method was tested by comparing the signal-to-noise ratio, which can be calculated as follows (11):
measured concentration values with the target concentrations.
A series of standard solutions were prepared in a wide range of
concentrations, and the highest values of the relative percent Eq 5
error are reported in Table V for each species. A value of 4% was
never exceeded. Accuracy checks were performed by means of where h represents the signal height and R =6Σ, where Σ is
n

a multilevel calibration with linear fit on the Hewlett-Packard the standard deviation of the noise. The obtained values, which
HP 3396A integrator. The level of accuracy of the method de­ are reported in Table V, are satisfactory and refer to a standard
pends not only on good precision and resolution but also on solution containing very low quantities of those species, i.e.

Table V. Analytical Parameters for Method Validation

Accuracy Detection Quantitation Selectivity Concentration Correlation

(% relative limit limit (purity in methenol coefficient
Compound error) (S/N=2) (S/N=10) index) (μg/mL) (linearity)

4-Hydroxymethylbenzoic acid 0.9 4.2 20.8 1.06 0–17 0.997

0–10 0.9996

Terephthalic acid 1.9 1.2 6.3 1.06 0–10 0.998

0–5 0.9998

4-Carboxybenzaldehyde 3.9 0.5 2.8 1.2 0–17 0.997

0–10 0.9998

Terephthaldicarboxaldehyde 4 2.6 13.1 1.1 0–10 0.998

0–5 0.9999

4-Methylbenzyl alcohol 1.8 2.4 12.3 1.08 0–20 0.997

0–10 0.9998

p-Toluic acid 3.8 0.57 2.8 1.03 0–200 0.9994

p-Tolualdehyde 0.5 1.3 0–80 0.998

0–40 0.999

Methyl benzoate 2.4 1 0–1800 0.998

0–900 0.9997

p-Xylene 0.5 1.05 0–1100 0.997

0–300 0.999

Cumene 1.07 0–600 0.9991

31
 Journal of Chromatographic Science, Vol. 34, January 1996

4-hydroxymethylbenzoic acid (0.64 ng/μL), terephthalic acid centrations and measuring the ratio of the peak areas at 220
(0.3 ng/μL), 4-methylbenzyl alcohol (0.69 ng/μL), and p-toluic and 254 nn. This ratio remained fairly constant and the values
acid (0.73 ng/μL), that can display low concentrations in the of the purity index reported in Table V were always less than the
samples from the experimental oxidation reactor. maximum limit of 1.5.
The specificity or selectivity of the method was checked by in­ The linearity of the developed method was also verified. This
jecting, for each species, several samples with different con- was done by plotting the injected amount for each analyte as a
function of the peak area counts. The corre­
sponding regression lines w e r e fitted
through the least-squares method. As shown
in Figure 3 for the case of terephthalic acid,
the graph is linear over the concentration
range presented in Table V. This result is also
valid for the other analytes, as seen from the
values of the regression coefficients listed in
Table V, which were always greater than
0.997 over the concentration ranges pre­
sented in the same table. Note that the ana­
lytical method showed good precision, accu­
racy, and linearity over these ranges.
However, as expected, better values were ob­
tained by reducing the concentration range
(Table V).
As noted in the introduction, this analytical
method has been used recently (5,6) to de­
velop a lumped kinetic model for liquid-phase
catalytic oxidation of p-xylene to terephthalic
acid. A typical chromatogram of a sample
from the oxidation reactor is shown in Figure
4. The analytical method allowed the deter­
mination of the products of p-xylene oxida­
tion under a relatively wide range of oper­
ating conditions, i.e., seven temperature
levels (80, 90, 100, 105, 110, 120, and 130°C),
three different initial values of p-xylene (4, 6,
and 9 mol/kg sol) and p-tolualdehyde con­
centrations (0.11, 0.17, and 0.22 mol/kgsol),
and two partial pressure levels of oxygen
(1 and 0.21 atm).

Conclusion

A reversed-phase HPLC method for the rel­

atively quick determination of p-xylene and
its catalytic liquid-phase oxidation products
within a wide range of concentrations with
limited sample manipulation was developed.
The characteristics of the analytical method
were described, and its reliability was suc­
cessfully tested with respect to the recom­
mended parameters available in the litera­
ture. The developed method has been used
recently (5,6) to study the kinetics of p-xylene
liquid-phase oxidation to terephthalic acid
catalyzed by cobalt naphtenate using methyl
Figure 2. Representation of sample stability over 6 days. benzoate as a solvent under a wide range of
operating conditions.

32
Journal of Chromatographic Science, Vol. 34, January 1996

Acknowledgments
The financial support of the Consiglio
Nazionale delle Ricerche (Progetto Finaliz-
zato Chimica Fine), Rome, Italy, is gratefully
acknowledged. The assistance of Ing. M. Pisu
and S. Sulis is also acknowledged.

References
1. P. Raghavendrachar and S. Ramachandran.
Liquid phase oxidation of p-xylene. Ind. Eng.
Chem. Res. 31: 453–62 (1992).
2. S. Carrá and E. Santacesaria. Engineering as­
pects of gas-liquid catalytic reactions. Catal.
Rev. Sci. Eng. 22: 75–140 (1980).
3. P. Cavalien D O r o , E. D a n o c z y , and P. Roffia.
O n the l o w t e m p e r a t u r e o x i d a t i o n o f p-
x y l e n e . Oxidation Comm. 1: 153–62 (1980).
Figure 3. Linear regression curve of terephthalic acid.
4. R. Jacobi and M. Baerns. T h e effect of o x y g e n
transfer limitation at the gas-liquid interface.
Erdoel Kohle Erdgas Petrochem. 36: 322–26
(1983).
5. G . C a o , A . Servida, M . Pisu, and M. Morbidelli. Kinetics of p-
x y l e n e liquid-phase catalytic oxidation. A.I.Ch.E. J. 40: 1168–78
(1994).
6. G . C a o , M . Pisu, and M . Morbidelli. A lumped kinetic model for
liquid-phase catalytic oxidation of p-xylene to terephthalic acid.
Chem. Eng. Sci. 49: 5775–88 (1994).
7. F. Savoia. Cromatografia Liquida ad Alta Pressione, 1 st e d . O E M F ,
Milano, Italy, 1980, pp 42, 65, 84, 109, 110.
8. R.W. Yost, L.S. Ettre, R . D . C o n l o n , and S. G a l l o . Introduzione
Pratica Alia Cromatografia in Fase Liquida (HPLC), 2nd e d .
Morgan edizioni tecniche, Milano, Italy, 1990, pp 205–212.
9. D.J. Runser. Maintaining and Troubleshooting HPLC Systems, A
User s Guide, 1st ed. John W i l e y & Sons, N e w York, N Y , 1981,
pp 53,54.
10. J . Miller. Chromatography: Concepts and Contrasts, 1st ed. John
W i l e y & Sons, N e w York, N Y , 1988, ρ 105.
11. L. Huber. Good Laboratory Practice. Hewlett-Packard, G e r m a n y ,
1993, publication no. 12–5091–6259E.
12. C o d e of Federal Regulations 21, Part 58. G o o d Laboratory Prac­
tice for Nonclinical Laboratory Studies. T h e Office of the Federal
Register National A r c h i v e s and Record Administration, W a s h ­
ington, D C , 1991.
13. T h e United States Pharmacopeia XXII, National Formulary, X V I I . Val­
idation of Compendial Methods. T h e United States Pharmacopeial
Convention, lne, Rockville, M D , 1990, General Chapter 1225.
14. M.T. Gilbert. High Performance Liquid Chromatography, 1st e d .
Figure 4. Chromatogram of an oxidation reaction sample. Conditions:
Wright, I O P Publishing Limited, Bristol, England, 1987, pp 5–11.
column, Chrompack Spherisorb 5 ODS-2 (250 × 4.6 mm); temperature,
15. V.P. Shaw, K.K. M i d h a , S. Dighe, et al. Analytical method valida­
room; mobile phase, H O - C H C N - C H O H + 0.1% H P O ; flow rate, 1.5
2 3 3 3 4
tion: bioavailability, bioequivalence and pharmacokinetic studies.
mL/min; injection, 20 μL; detection, ultraviolet at 254, 215, and 254 nm.
* Eur. J. Drug Metabol. Pharmacokin. 16: 249–55 (1991).
The following gradient elution program was used: 0–5 min, 95% H O , 5%2
16. J.A. Adamovics. Chromatographic Analysis of Pharmaceuticals, 1 st
C H C N ; 5–12 min, 55% H O , 35% C H C N , 10% C H O H ; 12–20 min,
3 2 3 3
ed. Marcel Dekker, N e w York, N Y , 1990, pp 3–26.
15% H O , 75% C H C N , 10% C H O H .
2 3 3

Manuscript accepted April 13, 1995.

33