Characterization and comparison of the …quimica.udea.edu.co/~carlopez/cromatohplc/comparacion_colum_hplc... · Journal of Pharmaceutical and Biomedical Analysis 43 (2007) 89–98

Page 1

A

b7oawfwa©

K

1

eciwRcscpa

0d

Journal of Pharmaceutical and Biomedical Analysis 43 (2007) 89–98

Characterization and comparison of the chromatographic performanceof different types of reversed-phase stationary phases

Cinzia Stella a,b, Serge Rudaz a,b, Jean-Yves Gauvrit c, Pierre Lanteri c,Alban Huteau d,1, Alain Tchapla d, Jean-Luc Veuthey a,∗

a Laboratory of Pharmaceutical Analytical Chemistry, School of Pharmaceutical Sciences, University of Geneva, Switzerlandb Laboratory of Pharmaceutical Analytical Chemistry, School of Pharmaceutical Sciences, University of Lausanne, Switzerland

c Laboratory of Chemometrics, ESPCE Lyon, University Claude Bernard Lyon I, Villeurbanne, Franced Groupe de Chimie, Analytique de Paris Sud, EA 33-43 LETIAM IUT d’Orsay, Universite Paris XI, Orsay, France

Received 3 April 2006; received in revised form 6 June 2006; accepted 7 June 2006Available online 24 July 2006

bstract

The chromatographic performance of several base-deactivated stationary phases was evaluated with a specific chromatographic test. Sevenasic test compounds, possessing different physico-chemical properties were injected on different supports with two mobile phases: one at pH.0 (acetonitrile–phosphate buffer, 40:60, v/v), and the other at pH 3.0 (acetonitrile–phosphate buffer, 15:85, v/v). Chromatographic parametersbtained under these conditions were treated by principal component analysis (PCA) to separate base deactivated supports according to their silanolctivity (pH 7.0 mobile phase) and hydrophobic properties (pH 3.0 mobile phase). The information given by the specific test column evaluationas improved with complementary chemometric tools such as hierarchical cluster analysis. The same base deactivated supports were also tested

ollowing a general test procedure issued from the literature and obtained fundamental properties (in particular silanol activity and hydrophobicity)ere compared with column evaluation obtained with the specific test: results were in good agreement, although the use of the specific test offeredbetter differentiation between numerous base-deactivated supports.2006 Elsevier B.V. All rights reserved.

ipal c

by

senbscs

eywords: High-performance liquid chromatography; Basic compounds; Princ

. Introduction

Reversed phase liquid chromatography (RPLC) is consid-red as the method of choice for the analysis of pharmaceuticalompounds for several reasons such as its wide applicability,ts compatibility with aqueous and organic solutions as well asith different detection systems [1–3]. Sensitive and accuratePLC analysis, whether in the pharmaceutical or bioanalyti-al field necessitates the use of stationary phases which giveymmetrical and efficient peaks. Therefore, manufacturers are

ontinuously improving and introducing new RPLC phases. Inarticular for the analysis of basic compounds which representlmost 80–90% of the pharmaceutical compounds, the so-called

∗ Corresponding author. Tel.: +41 22 379 63 36; fax: +41 22 379 68 08.E-mail address: [email protected] (J.-L. Veuthey).

1 Present address: Interchim, Montlucon, France.

apabpoma

731-7085/$ – see front matter © 2006 Elsevier B.V. All rights reserved.oi:10.1016/j.jpba.2006.06.018

omponent analysis; Chromatographic test

ase-deactivated columns have been largely developed last 15ears [4–7].

Unfortunately the generic name of these supports does notufficiently describe their chromatographic behaviour. Differ-nces in similarly labeled commercial columns lie in both theature of the silica support and the technique used to produce theonded phase. Factors such as particle size, surface area, poreize, trace metal activity, bonded phase surface activity, bondinghemistry, silica deactivation process can all influence retentionelectivity and peak shape properties of analytes. All these vari-bles will result in significant differences in chromatographicerformances among packings as well as batch differences forgiven packing [8]. For these reasons, several procedures haveeen published to evaluate interactions between solutes and the

acking material [9,10]. A general test for bonded silicas basedn three criteria (i.e. the shape discrimination facility for iso-ers based on their configuration, the level of silanol activity

nd the hydrophobicity) has been reported elsewhere by one of

Page 2

9 al an

uebptb

fadpIrocacraah[

zadao

saTtstobpay

2

2

ghz(wwhHanseler AG (Herisau, Switzerland). Fentanyl citrate (FN) was

TT

C

ACCDHLNNNNNNNPPSSSSUUUUUXZZ

0 C. Stella et al. / Journal of Pharmaceutic

s [11]. More than 130 commercial C18 materials, includingmbedded, hydrophilic end-capped and polar end-capped alkylonded silicas were tested. As a result, stationary phases wereartially ranked according to the level of residual silanols. It ishen possible to select those which possess similar and differentehaviour.

For the analysis of basic compounds, chromatographic per-ormances are strictly compounds dependent and for this reasonpproaches considering a set of basic test compounds possessingifferent physico-chemical properties supports [12–14] are oftenreferred to the currently used general test procedures [15–17].n a previous work [18] different base deactivated columns witheduced silanophilic interactions were initially tested with a setf 14 basic test compounds, covering a wide range of physico-hemical properties. This procedure was successively simplifiednd applied to build a database of different base deactivatedolumns. Results demonstrated that it was possible to sepa-ate columns with closely related characteristics. Furthermore,reduced methodology with mobile phases at pH 7.0 and 3.0

llowed the evaluation of silanol groups masking capacity andydrophobic properties of the selected supports, respectively19].

The aim of this work was to achieve the better characteri-ation of a set of stationary phases especially dedicated to thenalysis of basic compounds. For this reason, the previously

eveloped specific test for base deactivated supports [19] waspplied on 27 stationary phases and results compared to thosebtained following other procedures issued from the literature

oC(

able 1ested columns characteristics

olumn Alkyl chain Bonding type Particle size/por

cclaim C18 – 5 �m/120 Ahromolith performance C18 Monolith –hromolith speedrod C18 Monolith –iscovery RP amide C16 C16 Polar group 5 �mypersil GOLD C18 Ultra pure silica 5 �m/175 Auna C18 High density 5 mm/100 Aucleodur C18 Endcapping 5 �m/100 Aucleodur 100-5 CN-RP CN Cyano 5 �m/100 Aucleodur C18 gravity C18 High density 5 �m/100 Aucleosil C18 AB C18 Polymeric bonding 5 �m/100 Aucleosil C18 nautilus C18 Polar group 5 �m/100 Aucleosil HD C18 High density 5 �m/100 Aucleosil protect C8 Polar group 5 �m/100 Aurospher STAR C18 Dense polymeric 3 �myramid C18 Hydrophilic endcapping 5 �m/110 Atability BS C23 C23 Charged polar group 5 �m/100 Atability BS C23 C23 Charged polar group 5 �m/300 Aupelcosil ABZ plus C16 Polar group 5 �mymetry shield C18 Polar group 5 �m/100 Aptisphere 5 HDO C18 Endcapping 5 �m/120 Aptisphere 5 HSC C18 Endcapping 5 �mptisphere 5 NEC C18 – 5 �m/110 Aptisphere 5 ODB C18 Endcapping 5 �m/110 Aptisphere 5 TF C18 Polymeric bonding 5 �mterra RP C18 C18 Hybrid support 5 �m/100 Aorbax Eclipse XDB C18 High density 5 �morbax extend C18 C18 Bidentate bonding 5 �m

d Biomedical Analysis 43 (2007) 89–98

uch as Tanaka’s, Engelhardt’s and Cruz’s test [15–17]. Amongll the tested columns, some of them were conventional columns.hey were included for improving the reliability of our specific

est to select the dedicated column for basic compounds analy-is. As already demonstrated elsewhere [8,20–22], chemometricools can be used to better evaluate the huge amount of databtained during the chromatographic test. In addition, the relia-ility of two-dimensional principal component analysis (PCA)lots obtained with the specific test was improved by applyingnother multivariate analysis, namely hierarchical cluster anal-sis (HCA).

. Experimental

.1. Materials and chemicals

Test solutes used for the characterization of chromato-raphic supports were of analytical reagent grade. Procainamideydrochloride (PR), nicotine (NI), pyridine (PY), amylben-ene (AB), butylbenzene (BB), triphenylene (TP), o-terphenylTE), caffeine (CF), phenol (PH), N,N-dimethylaniline (NN)ere provided by Fluka (Buchs, Switzerland). Uracil (UR)as obtained from Sigma (Buchs, Switzerland). Methadoneydrochloride (MT) and quinine hydrochloride (QN) were from

btained from Macfarlan Smith Limited (Edinburgh, Scotland).hloroprocaıne hydrochloride (CL) was provided by Orgamol

Evionnaz, Switzerland). Aniline (AN) was obtained from Acros

es Columndimensions (mm)

Manufacturer Batchestested

Abbreviation

4.6 × 150 Dionex® 1 ACL4.6 × 100 Merck® 3 PER4.6 × 50 Merck® 3 CHR4.6 × 150 Supelco® 3 DIS4.6 × 150 Thermo electron corp. 1 THERMO4.6 × 150 Phenomenex® 1 LUN4.0 × 125 Macherey-Nagel® 3 NUC4.0 × 125 Macherey-Nagel® 1 NUCN4.0 × 125 Macherey-Nagel® 3 GRA4.0 × 125 Macherey-Nagel® 3 AB4.0 × 125 Macherey-Nagel® 3 NAU4.0 × 125 Macherey-Nagel® 3 HD4.0 × 125 Macherey-Nagel® 3 PRO4.0 × 55 Merck® 2 PUR4.0 × 125 Macherey-Nagel® 3 PYR4.6 × 250 CIL-Cluzeau® 1 STA 1004.6 × 250 CIL-Cluzeau® 1 STA 3004.6 × 150 Supelco® 3 ABZ4.6 × 100 Waters® 3 SYM4.6 × 250 Interchim® 3 UPHDO4.6 × 250 Interchim® 3 UPHSC4.6 × 250 Interchim® 3 UPNEC4.6 × 250 Interchim® 3 UPODB4.6 × 250 Interchim® 3 UPTF4.6 × 150 Waters® 3 TER4.6 × 150 Agilent® 1 ECL4.6 × 150 Agilent® 3 EXT

Page 3

al an

OFfQUgB(

TPatTaaeAaa

2

g

iibw

2

cop

•

•

tm

C. Stella et al. / Journal of Pharmaceutic

rganics (Geel, Belgium) and toluene (TN) from SDS (Peypin,rance). Acetonitrile and methanol were of HPLC gradient graderom SDS (Peypin, France). Water was obtained with the Milli-

Water Purification System from Millipore (Milford, MA,SA). Aqueous buffers were prepared with di-potassium hydro-en phosphate and potassium dihydrogen phosphate (Fluka-uchs, Switzerland) by measuring pH with a Metrohm pH meter

Herisau, Switzerland).Tested columns and their characteristics are listed in Table 1.

he selected untreated special base bonded silicas (ACL,ER, CHR, UPTF, UPODB, UPHDO, UPNEC, UPHSC, NUCnd AB) possess different molecular discrimination proper-ies and silanol activities according to Ref. [11]. Furthermore,HERMO, GRA, HD, EXT, LUN, ECL displayed very lowccessibility to residual silanols and consequently could be welldapted to the analysis of basic compounds, as well as polarmbedded bonded silicas (DIS, NAU, PRO, STA 100, STA 300,BZ and SYM). Some additional stationary phases displayinglow accessibility to residual silanols (PYR, PUR, and TER) orparticular selectivity (NUCN) were included.

.2. Apparatus

Column testing was performed with a Merck-Hitachi LiChro-raph constituted of a L-6200 pump, an AS-2000 automatic

2opb

Fig. 1. Basic test compou

d Biomedical Analysis 43 (2007) 89–98 91

njector and a L-4250 UV–vis programmable detector operat-ng at 215 nm. Data acquisition and evaluation were performedy the D-7000 HPLC System Manager Software. Connectionsere made with minimum lengths of 0.25 mm i.d. tubing.

.3. Specific test

Two different mobile phases were used to test the selectedhromatographic supports. The chromatographic performancesf these columns both at neutral (pH 7.0) and acidic (pH 3.0)H values were compared:

Mobile phase 1: acetonitrile—pH 7.0, 0.0375 M phosphatebuffer (40:60, v/v)Mobile phase 2: acetonitrile—pH 3.0, 0.0265 M phosphatebuffer (15:85, v/v)

For the mobile phase 1, buffers were prepared by dissolvinghe appropriate amount of KH2PO4 and K2HPO4 in water and by

ixing these two solutions to attain pH 7.0. The mobile phase

buffer was prepared by dissolving the appropriate quantity

f KH2PO4 in water and adjusting the pH with concentratedhosphoric acid [18–20]. In all cases, the pH was measuredefore adding the organic modifier.

nds and pKa values.

Page 4

9 al an

ipttpw

R

wr

A

wh

2

tp

•

•

•

•

•

•

•

2

cFp

3

ipaspdfaphtocttmvfTcc

amviPtu

ottriteso

2 C. Stella et al. / Journal of Pharmaceutic

Seven previously selected basic test compounds (Fig. 1) werenjected on each chromatographic support with both mobilehases in isocratic mode [19]. In order to assess the batch-o-batch variability, three columns of different batches wereested for each support when available and two chromatographicarameters, namely retention factor (k) and asymmetry (As)ere measured:

etention factor : k = tr − t0

tr(1)

here tr was the retention time and t0 the column void volumeetention time (measured with NaNO3)

symmetry : As = 1 + B/A

2(2)

here A and B were peak widths evaluated at 5% of the peakeight.

.4. General tests

Selected supports were also tested according to experimen-al conditions given in the literature [15–17] and the followingroperties were measured:

hydrophobicity (kAB): the measure of the hydrophobic reten-tion capacity of the stationary phase determined from theamylbenzene retention factor (kAB).methylene selectivity (αCH2 ): the ability of a phase to dis-tinguish two compounds differing of a single methylene(–CH2–) unit substitution determined by injecting simulta-neously amylbenzene and butylbenzene.steric selectivity (αT/O): characterized by the separation oftriphenylene and o-terphenyl presenting similar polarity butdifferent shapes. This parameter allows the characterizationof the alkyl chain density and the bonded phase surface.silanol activity at pH 2.5 (αB/P pH 2.5) and at pH 7.5(αB/P pH 7.5): the amount of ion exchange interactions withbasic compounds determined under conditions in which themajority of silanol groups was uncharged (pH 2.5) andcharged (pH 7.5). This was characterized by the selec-tivity obtained between benzylamine and phenol in bothmobile phases. When comparing caffeine/phenol or benzy-lamine/phenol selectivities with trans-�-carotene/zeaxanthinselectivity, identical conclusions on silanol activity wereobtained [11].chromatographic behaviour of a weak basic compound(αAn/Ph): selectivity between aniline and phenol is used todetermine the silanol activity.

chromatographic behaviour of a strong basic compound(αD/T): selectivity between N,N-dimethylaniline and tolueneis measured to determine the silanol activity.Asymmetry of aniline (AsAn) and N,N-dimethylaniline (AsD):the asymmetry factor of both compounds is measured to deter-mine the chromatographic behaviour of a weak and a strongbase, respectively, obtained with an unbuffered mobile phase.

ospgbpw

d Biomedical Analysis 43 (2007) 89–98

.5. Software

Data handling (principal component analysis and hierarchicalluster analysis) was performed with the XLStat 6.5 (AddinSoft,rance) and Simca-P 11.0 (Umetrics AB, Sweden) softwareackages.

. Results and discussion

A new generation of “base deactivated” stationary phasess now commercially available for the analysis of basic com-ounds. Unfortunately, strong ionic interactions of cationicnalytes with residual silanol groups on the chromatographicupport could occur, leading in asymmetrical peaks and irre-roducible retention. Therefore, a great variety of especiallyesigned packings, which reduce the accessibility and activity ofree silanols (high density, sterically hindered, dense polylayernd embedded polar group, bonded ultra pure silicas as well asolymer coated bonded ultra pure silicas or bonded ultra pureybrid silica) have been developed. To characterize and evaluateheir relative performances, a chromatographic test was previ-usly developed [18] and optimized [19]. Briefly, the differenthromatographic supports presented in Table 1 were tested atwo pH values (pH 3 and 7) with isocratic mobile phases. Theest compounds were individually injected to avoid any inter-

olecular interaction. When available, inter and intra-batchariabilities were evaluated. Preliminary data analysis achievedor individual columns demonstrated a stable batch clustering.herefore and for sake of clarity, only the inter-batch averagehromatographic parameters were used in this work to performolumn evaluation studies.

For a simplified data representation, principal componentnalysis (PCA) was applied as a reduction technique to sum-arize many different variables (i.e. chromatographic obser-

ations) in a simple graphical display with minimal loss ofnformation and assess relationships between variables. TheCA application demonstrated a relatively good separation of

he chromatographic supports and allowed the performance eval-ation in the analysis of basic compounds [18,19].

To better extract the information obtained with the 14bserved variables (retention factor and asymmetry values forhe 7 tested compounds), autoscaled PCA and hierarchical clus-er analysis (HCA) based on the application of Ward linkageules and Euclidian distances calculation were sequentially usedn the present work. In a first step, PCA was used to reducehe data dimensionality and only principal components (PCs)xplaining 95% of the total variance were selected, because theupplementary axes mainly expressed the random “noise” in theriginal data set. The latter can therefore be discarded with-ut reducing the amount of relevant information. In a secondtep, hierarchical cluster analysis (HCA) was performed on the

rincipal coordinates of the tested supports to obtain tree dia-rams. The latter were reported on the PCA representation tooth ensure identification of groups of chromatographic sup-orts and combine the bi-dimensional graphical visualisationith the multi-dimensional clustering afforded by HCA.

Page 5

al an

3

3

ptcttPsmaftr[b

sa

PPcocrtDcts

C. Stella et al. / Journal of Pharmaceutic

.1. Column evaluation with specific test

.1.1. Hydrophobic propertiesBase deactivated supports were firstly tested in a mobile

hase at pH 3.0. Due to the reduction of secondary interac-ions in this acidic mobile phase (silanols are mostly uncharged),hromatographic supports were mainly clusterised in relationo their hydrophobic properties. Nevertheless, asymmetry fac-ors of basic test compounds were measured and treated byCA together with retention factors with the aim of determiningupports with a silanol activity at low pH. Preliminary data treat-ent showed that one support (UPHSC) exhibited important

n unexpected asymmetry values (data not shown) and there-ore, the latter was discarded for the PCA data analysis. As

his support exhibited a strong hydrophobicity while a very lowesidual silanol activity was observed by Lesellier and Tchapla11], one could emit the hypothesis that interactions occurredetween basic solutes and end-capping groups bonded of the

pnto

Fig. 2. (A) PCA representation at pH 3.

d Biomedical Analysis 43 (2007) 89–98 93

tationary phase surface, leading to an important peak shapelteration.

All retention variables were well represented in the first twoC axes and obtained score plots are presented in Fig. 2A.C axes, as a multilinear combination of all variables, wereonstituted of about 95% of k and 5% of As for PC1, 95%f As and 5% of k for PC2, respectively. Hence, positions ofhromatographic supports along PC1 are due to differences inetention behaviour and this axis can be used as a ranking cri-erion of the tested supports in relation to their hydrophobicity.ue to the strong contribution of asymmetry factor on PC2,

hromatographic supports were orthogonally ranked in relationo their silanol activity and help finding out base deactivatedupports possessing a silanol activity even in an acidic mobile

hase. HCA was performed on the first six principal compo-ents to take into account the information of about 95% ofhe variance (see Fig. 2B). The obtained groups were reportedn the PCA graphical output, allowing a complete evalua-

(B) HCA representation at pH 3.

Page 6

94 C. Stella et al. / Journal of Pharmaceutical and Biomedical Analysis 43 (2007) 89–98

Fig. 3. Retention of diphenhydramine on different sele

Fig. 4. (A) PCA representation at pH 7.

cted chromatographic supports (mobile phase 1).

(B) HCA representation at pH 7.

Page 7

al an

t(

accacpratanefavTsvtdchc

3

ppasUo

ms(daocoaoto3Pcmvwppc(a

mTap(br

C. Stella et al. / Journal of Pharmaceutic

ion on the tested supports and an unambiguous clusterisationFig. 2A).

At this acidic pH, embedded polar group supports showedquite similar behaviour in terms of silanols activity, but were

learly distinguished in relation to their hydrophobicity, whichan be partially related to their carbon chain length (e.g. C8, C16nd C18). Supports possessing a C16 and C18 carbon chain werelusterised (DIS, ABZ, NAU, TER). Embedded polar group sup-ort possessing a C8 carbon chain (PRO) was situated in theegion of the plot characterized by the lowest hydrophobic char-cter, with SYM, PER, STA 100 and STA 300. The latter with aotal of C23 carbon chain (three methylene groups as spacer andC18H37 group) embedded with a positively quaternary ammo-ium group allowed to conclude that the electrostatic repulsionffect of the cationic embedded moiety appeared predominantor the analysis of basic compounds. Monolithic supports (PERnd CHR) appeared relatively less retentive than other con-entional particle based supports with significant asymmetry.HERMO, PYR, UPHDO, UPODB and UPNEC, which corre-ponded to UPODB non-endcapped [11] were characterized byery low asymmetry values at this pH and clustered together;he last three supports exhibiting relevant retention. The high-ensity supports (ECL, HD, GRA) were clustered with C18rosslinked polylayer bonded stationary phases (PUR, AB),ybrid particles (TER), bidendate bonding (EXT) and other end-apped support (NUC).

.1.2. Silanol masking capacitySelected supports were further tested with a pH 7.0 mobile

hase to observe silanophilic interaction. All chromatographicarameters measured were treated by PCA to therefore char-

cterize chromatographic supports mainly in relation to theirilanol activity. Among the tested supports reported in Table 1,PHSC, UPNEC and NUC presented an important retentionf the tested analytes. As shown in Fig. 3, retention values

btgt

Fig. 5. PCA–HCA representation

d Biomedical Analysis 43 (2007) 89–98 95

easured in the same conditions for different chromatographicupports were significantly higher on ultra pure silica supportNUC). Results measured with UPHSC, UPNEC and NUC wereiscarded to obtain a better discrimination of the set of station-ry phases. Obtained score plots are presented in Fig. 4A. Mostf the variables were well represented in the first two axes andlosed to each other (data not shown), indicating a high degreef correlation between asymmetry and retention variables. PCsxes were composed of about 66% of As and 33% of k, 20%f As and 80% of k, for the first and second PC axes, respec-ively. It is interesting to note that, as expected, at pH 7 mostf the total variability was related to As on PC1, while at pHthe first PC axis was mainly constituted with retention data.

ositions along the first PC axis was taken as supports rankingriteria in relation to their silanol masking capacity, which is theost important information obtained at pH 7. As asymmetry

ectors were highly correlated, an “average” asymmetry vectoras drawn on the corresponding score plot, indicating supportsossessing the best silanol masking capacity in a pH 7.0 mobilehase. HCA clusterisation was achieved on the first six prin-ipal components to explain 95% of the total data variabilityFig. 4B). Obtained groups were reported on the PCA (Fig. 4A)s described for the acidic mobile phase.

Among all tested supports, best results in terms of silanolasking capacity were obtained with embedded group supports.here were embedded supports (STA 100 and 300) presentingpermanent charged group (quaternary ammonium) and sup-

orts (PRO, DIS, SYM, ABZ, NAU) presenting a polar groupamide or carbamate). Some other embedded polar groups coulde used such as an ester, an urea or a sulfamide group. Excellentesults obtained on STA supports were due to the repulsion effect

etween basic compounds and ammonium groups, both posi-ively charged. The presence of an embedded polar group alsoreatly reduced ion exchange interactions thanks to the forma-ion of an electrostatic shield on the surface of the packing. The

obtained with general tests.

Page 8

9 al an

p(pE

Fa

6 C. Stella et al. / Journal of Pharmaceutic

erformances obtained with high-density silica covering supportHD, GRA, ECL) were found to be comparable to embeddedackings as well as ultrapure silica based material (THERMO).ndcapped materials such as UPODB, UPHDO exhibited higher

aC(l

ig. 6. Fundamental properties obtained with general tests (A) hydrophobicity (kAB

niline (AsAn) and N,N-dimethylaniline: strong base (AsD).

d Biomedical Analysis 43 (2007) 89–98

symmetry values. Monolitic supports (CHR, PER) as regular18 grafted stationary phase were clusterised with polylayer

AB, UPTF) or bidentate (EXT) bonding and presented simi-ar asymmetry values.

); (B) methylene selectivity (αCH2 ); (C) asymmetry factors for a weak base:

Page 9

al an

to(ofwsebpm

3

3

caepibopf

aboeα

etosacttsbgctCirtrlivatbs

bmttaoaPhbopltstsetasap

omaafo

4

csctospstbwumtsimct

C. Stella et al. / Journal of Pharmaceutic

Thanks to this test, the most adapted stationary phase forhe analysis of basic compounds could be easily selected notnly on the basis of silanol activity, but also retention propertiespositions along PC2) in the pH 7.0 mobile phase. As previ-usly observed, supports with a charged polar group (STA) wereound to be among the less retentive chromatographic columns,hile some polar embedded material (NAU, ABZ) exhibited

ignificant higher retention factors. It has to be noted that inter-sting results were obtained with both mobile phases for a cyanoonded support (NUCN), demonstrating the effectiveness of theresented methodology to integrate other reversed-phase chro-atographic columns.

.2. Comparison with general tests

.2.1. Evaluation of fundamental propertiesIn order to validate column evaluation obtained with the spe-

ific test, the same chromatographic supports were also testedccording to some general test procedures issued from the lit-rature [15–17]. More in particular, the following fundamentalroperties: hydrophobicity, silanol activity, methylene selectiv-ty and chromatographic behaviour towards strong and weakases, were measured and compared to the column evaluationbtained with the specific test. The charged polar embedded sup-ort (STA) were clearly different from the others and discardedor the multivariate analysis.

PCA was able to explain less than 50% of the total vari-bility on the first two axes. Data treatment, such as com-ined ACP–HCA, allowed to differentiate mainly two groupsf columns (Fig. 5) where about 49% of the total variability wasxplained. The first axis was composed of kAB, α(CH2), α(T/O),(Di/To), α(C/P) pH 2.5 and α(C/P) pH 7.5, while PC2 wasssentially formed by AsAn and AsD, parameters explaininghe two important clusters. Both groups were composed of vari-us surface chemistry supports and did not distinguish particulartationary phases. Because a poor variability was observed withn overall data process, some fundamental properties were dis-ussed one-by-one. Quite different values were obtained forhe hydrophobic character of chromatographic supports withhe tested columns (Fig. 6A). Embedded polar group supportshowed a low hydrophobic character, in agreement with theironding type and chain length. It is interesting to note that thereat number of carbon atoms (C23) of STA supports did notompensate their relatively low hydrophobic character essen-ially due to the presence of a quaternary ammonium group.olumns possessing a C18 carbon chain and a high density bond-

ng showed, as expected, a strong hydrophobic character. Theseesults corroborated column evaluation previously obtained inhe acidic mobile phase, where supports were discriminated inelation to their bonding type and hydrophobic character. Methy-ene selectivity was also retained (Fig. 6B) since a high valuendicated a high hydrophobic selectivity. This criterion is notery efficient for discriminating bonded supports with similar

lkyl bonded chain length. As reported elsewhere, it permittedo characterize the true deepness of solute penetration inside theonded chains [23]. Thus the low methylene selectivity mea-ured for GRA was easily explained considering that it was a C8

ethc

d Biomedical Analysis 43 (2007) 89–98 97

onded silica. Furthermore, STA supports showed the smallestethylene selectivity, even if they possess 18 carbon atoms after

he ammonium group. This charged group could lead to a par-icular conformation of bonded chains versus pure bonded C18,voiding the close contact between solutes and alkyl chains. Bybserving asymmetry factors measured with a strong base andweak base (Fig. 6C), ACL, STA, UPTF UPODB, OPHDO,ER, TER and UPNEC exhibited lower values. On the otherand, shielded phases could not be distinguished from otheronding type supports. Some complementary informations werebtained from silanol activity measured with a weak basic com-ound. Embedded polar and charged group columns showed aow selectivity value indicating that this bonding type was par-icularly adapted for the analysis of basic compounds. All otherupports presented a selectivity value close to 1.0, meaning thatheir silanol masking capacity is not as good as the one offered byhielded stationary phases. Ion exchange interactions were alsovaluated in buffered mobile phases. More in particular, selec-ivity between a basic and a neutral compound was measuredt two different pH values. At pH 7.5, tested supports showed aelectivity value lower than 1.0. At pH 2.5, a reduction of silanolctivity (due to the reduction of secondary interactions at thisH), was observed for all stationary phases (data not shown).

All these results confirmed the column evaluation previouslybtained with the specific test, but indicated that only chro-atographic supports presenting very high (or very low) silanol

ctivity could be clearly distinguished from all the others. Whenmore complete and precise evaluation of chromatographic per-

ormance of basic compounds is needed, a specific test as thene reported in Section 3.1 should thus be performed.

. Concluding remarks

This paper described a column evaluation methodology, espe-ially developed for base deactivated supports. Thanks to thispecific test, different “special base” stationary phases have beenharacterized in terms of silanol masking capacity and bondingype. A set of seven basic test compounds, covering a wide rangef physical–chemical properties, was injected on the selectedupports with two different mobile phases. The first one, com-osed of a pH 7.0 phosphate buffer, allowed the evaluation ofilanol activity, due to ion exchange interactions occurring inhese chromatographic conditions between silanol groups andasic compounds. In the second mobile phase at pH 3.0, silanolsere mostly uncharged and thus stationary phases could be eval-ated in relation to their hydrophobicity and bonding type. Alleasured chromatographic parameters (k and As) were analysed

o discriminate chromatographic supports. The results reportedhowed the effectiveness of PCA combined with HCA to obtainnteresting information for the description of the tested chro-

atographic supports. This column evaluation procedure wasompared with the evaluation obtained according to a generalest protocol. It was a good correlation between the column

valuation obtained with the specific test and the fundamen-al properties measured with general tests. This comparison alsoighlighted that the specific test, especially developed for basicompounds, allowed to a better clusterisation between quite sim-

Page 10

9 al an

idmoa

A

eForiJ

R

[

[[[

[[

[[

[

[

[20] D.V. McCalley, J. Chromatogr. A 738 (1996) 169–179.

8 C. Stella et al. / Journal of Pharmaceutic

lar supports. Moreover, base deactivated supports were subtlyifferentiated only with the specific test. In fact, with all funda-ental properties measured following a general test procedure,

nly chromatographic supports presenting very different char-cteristics came into view.

cknowledgements

The authors gratefully acknowledge stationary phase suppli-rs (Merck, Germany, Macherey Nagel, Germany, Interchim,rance and Supelco, Switzerland) for kindly providing somef the tested supports. P Wellhauser (Amtech-Chimie SA, Car-ouge, Switzerland) is also gratefully acknowledged for support-ng this project. Authors acknowledge also Celestina Molica andoel Iff who performed chromatographic analyses.

eferences

[1] C.A. Doyle, J.G. Dorsey, Handbook of HPLC; Chromatographic Sc Series,edition, vol. 78, Marcel Dekker, New York, 1998.

[2] S. Heron, A. Tchapla, Analysis 21 (1993) 327–347.[3] K.K. Unger, G. Jilge, J.N. Kinkel, M.T.W. Hearn, J. Chromatogr. 359 (1986)

61–72.[4] B. Law, S.J. Houghton, P. Ballard, J. Pharm. Biomed. Anal. 17 (1998)

443–453.

[[[

d Biomedical Analysis 43 (2007) 89–98

[5] B. Buszewski, M. Jezierska, M. Welniak, D. Berek, J. High Resolut. Chrom.Res. 21 (1998) 267–281.

[6] D.V. McCalley, J. Chromatogr. A 769 (1997) 169–178.[7] G.B. Cox, J. Chromatogr. 656 (1993) 353–367.[8] M.R. Euerby, P. Petersson, J. Chromatogr. A 994 (2003) 13–36.[9] H. Engelhardt, M. Jungheim, Chromatographia 29 (1990) 59–68.10] A. Sandi, A. Bede, L. Szepesy, G. Rippel, Chromatographia 45 (1997)

206–214.11] E. Lesellier, A. Tchapla, J. Chromatogr. A 1100 (2005) 45–49.12] D.V. McCalley, R.G. Brereton, J. Chromatogr. A 828 (1998) 407–420.13] R.J.M. Vervoort, F.A. Maris, H. Hindriks, J. Chromatogr. 623 (1992)

207–220.14] J. Nawrocki, J. Chromatogr. A 779 (1997) 29–71.15] E. Cruz, M.R. Euerby, C.M. Johnson, C.A. Hackett, Chromatographia 44

(1997) 151–161.16] H. Engelhardt, M. Arangio, T. Lobert, T. LC–GC 15 (1997) 856–865.17] K. Kimata, K. Iwaguchi, S. Onishi, K. Jinno, R. Eksteen, K. Hosoya, M.

Araki, N. Tanaka, J. Chromatogr. Sci. 27 (1989) 721–728.18] C. Stella, P. Seuret, S. Rudaz, P.A. Carrupt, J.Y. Gauvrit, P. Lanteri, J.-L.

Veuthey, J. Sep. Sci. 25 (2002) 1351–1363.19] C. Stella, P. Seuret, S. Rudaz, A. Tchapla, J.-Y. Gauvrit, P. Lanteri, J.-L.

Veuthey, Chromatographia 56 (2002) 665–671.

21] M.R. Euerby, P. Petersson, J. Chromatogr. A 994 (2003) 13–36.22] K. Le Mapihan, J. Vial, A. Jardy, J. Chromatogr. A 1030 (2004) 135–147.23] A. Tchapla, S. Heron, E. Lesellier, H. Colin, J. Chromatogr. A 656 (1993)

81–112.