Optimization of the synthesis of a hyper-crosslinked stationary phases: A new generation of highly efficient, acid-stable hyper-crosslinked materials for HPLC

Research Article

Optimization of the synthesis of a hyper-crosslinked stationary phases: A newgeneration of highly efficient, acid-stablehyper-crosslinked materials for HPLC

A new family of hyper-crosslinked (HC) phases was recently developed for use under very

aggressive conditions including those encountered in ultra-fast, high-temperature two-

dimensional liquid chromatography (2-DLC). This type of stationary phase has improved

acid stability compared with the most acid-stable, commercial RPLC phases. Kinetic

studies are here reported that allow optimization of reaction time and crosslinking reagent

concentrations used to prepare such HC phases. We have determined that the Friedel-

Crafts chemistry used to prepare HC phases is nearly complete within about 15 min.

Thus, reaction time for each step of the synthesis was greatly reduced from the multihour

reactions used previously without sacrificing the stationary phases’ acid stability and

separation performance. Results from elemental analysis of the finished particles were

combined with LC data to provide insights regarding the properties of these HC phases.

This new generation of acid stable HC phases, with their attractive chromatographic

properties, should be very useful in the separations of bases or biological analytes in acidic

media, especially at elevated temperatures.

Keywords: Acid stable / LC / RPLC / Stationary phase / Two dimensionalDOI 10.1002/jssc.201100252

1 Introduction

Higher separation efficiency and faster speed have always

been of great interest in HPLC and have become increas-

ingly important in recent years mainly driven by the

challenge of either more complex samples or growing

numbers of samples [1]. Many approaches have been

developed as potential solutions including sub-2 mm parti-

cles at ultra-high pressure [2, 3], monolithic columns [4],

superficially porous stationary phases [5, 6], and high-

temperature liquid chromatography (HTLC) [7–9]. Among

these techniques, HTLC has recently drawn a lot of

attention. In fact, the advantage of elevated temperature

was predicted almost two decades ago [10] and that a 20-fold

decrease in analysis time could be achieved when the

column is operated at high temperatures (150–2001C).

Interest in this type of separation was greatly enhanced by

recent advances in instrumentation ovens which could

reach very high temperature (150–2001C) and became

commercially available. Due to the five- to tenfold decrease

in solvent viscosity [11, 12] and the concomitant increase in

analyte diffusivity [13], HTLCs can dramatically improve

analysis speed without compromising separation efficiency.

In addition, reduced retention allows for significantly less,

and in some cases complete elimination of the organic

modifier used in the mobile phase. This enables the

development of ‘‘green chromatography’’ where RPLC

separations could be performed in pure water [14–16].

Furthermore, it has been shown in recent studies that

temperature can be very helpful in method development

especially for charged basic analytes, that the changes in

temperature can be used to not only fine-tune selectivity [17,

18], but also improve the peak symmetry [19–21] by

enhancing the slow kinetics of ‘‘strong’’ sites such as

silanols on stationary phase surface [21].

Although the advantages of doing LC at higher

temperatures are well-established, HTLC is not commonly

used in routine work. One of the major barriers is the lack of

Yu Zhang1�

Yuming Huang2

Peter W. Carr1

1University of Minnesota,Minneapolis, MN, USA

2Southwest University, Collegeof Chemistry & ChemicalEngineering, Chongqing,P. R. China

Received March 19, 2011Revised March 19, 2011Accepted March 23, 2011

Abbreviations: CMME, chloromethylmethyl ether;DMCMPES, dimethyl-chloromethylphenylethylchlorosilane;

EPG, embedded polar groups; HC, hyper-crosslinked; HSM,hydrophobic subtraction model; HTLC, high-temperatureliquid chromatography; IPA, isopropanol; ISEC, inversesize exclusion chromatography; MeCN, acetonitrile; MeOH,methanol; PAH, polycyclic aromatic hydrocarbons;

SH, styrene heptamer; TBM, 2,4,6-tris(bromomethyl)-mesitylene; TPM, triphenylmethane

�Additional correspondence: Professor Yu Zhang

E-mail: zhang317@@umn.edu

Correspondence: Professor Peter W. Carr, University of Minne-sota, 207 Pleasant St. SE, Minneapolis, MN 55455, USAE-mail: [email protected]: 11-612-626-7541

& 2011 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.jss-journal.com

J. Sep. Sci. 2011, 34, 1407–1422 1407

reliable and thermally stable stationary phases [9]. High

temperature can result in significant reduction in column

life due to increased rate of siloxane bond hydrolysis. Thus

traditional silica-based stationary phases allow only a rela-

tively small increase (50–601C) in temperature over ambi-

ent. It should not exceed 601C especially when acidic- or

basic-buffered eluents are present. In response, extensive

work has been done to enhance the stability of silica-based

stationary phases. In general, three strategies have been

employed: (i) to better shield the underlying siloxane bond

from hydrolysis by steric protection, such as the use of a

bulky silane (i.e. C18 versus C8 versus C4 [22], or isobutyl

side groups [23, 24]) or by coating silica with polymers

(i.e. a self-assemble monolayer [25–29]); (ii) to decrease the

rate of phase loss by the formation of multiple bonds

between the stationary phase and the silica substrates [24,

28, 30, 31]; (iii) to change the chemistry of the surface

bonding by replacing the surface Si–O–Si bonds with

chemically more stable Si–C bonds [32, 33], or use pH inert

materials such as metal oxides [34–36] and polymers instead

of silica [37, 38]. These efforts have been comprehensively

discussed and compared in several excellent books and

reviews [39–41].

In previous articles [42–45], we developed a novel type of

acid-stable, silica-based platform. Orthogonal polymeriza-

tion reactions were used to prepare a monolayer of hyper-

crosslinked (HC) aromatic network that is entirely confined

to the surface. Although the chemistry involved in making

the crosslinked functionalized aromatic layer is similar to

that employed by Davankov [46–48], our study is funda-

mentally different in that Davankov’s materials are totally

organic polymers, whereas our materials are based on silica

modified with a surface confined layer of a highly cross-

linked aromatic polymer. The fully connected polymer

network prevents the loss of bonded phase, which leads to

superior hydrolytic stability in acid when these new phases

are compared with conventional silica-based phases [44, 45].

At the same time, the excellent mass transfer properties of

monomeric-bonded phases were preserved and none of the

adverse kinetic effects [42–45] of depositing a polymer has

been observed. The stability and efficiency afforded by these

novel phases make them good candidates for use at elevated

temperature. However, the Friedel-Crafts chemistry we

use to synthesize the HC phases is complicated and involves

many experimental variables (reaction temperatures,

catalyst concentration, reaction times, catalyst types, and

crosslinker types). Optimization of the synthesis is

needed to achieve further improvements in column effi-

ciency of the HC phases without compromising its acid

stability.

Previously, the effects of catalyst type and silica

substrate were studied in this lab [44, 45]. The very high

silanophilicity of our first generation of HC phases evident

in the poor plate numbers for basic solutes [44, 45]

synthesized by using aluminum trichloride as the Friedel-

Crafts catalyst and a relatively active silica as the substrate

was primarily due to the activation of some silanol groups

caused by contaminating the silica with traces of aluminum

(III) and the intrinsic high activity of the type of silica used.

The problem was essentially solved by replacing aluminum

trichloride with tin tetrachloride (a less reactive but still

effective Friedel-Crafts catalyst [44]), and by using a better

grade of HPLC silica. The plate count of organic bases on a

5.0 cm� 0.46 cm column increased from 2100 to 3700 upon

use of the tin catalyst, and from 3700 to 4600 upon the

additional change to the less active silica substrate. The

phase thus prepared is denoted as the second generation of

HC phases. Although more than twice the plate count for

basic drugs were achieved on these second generation of HC

phases than on the first generation materials, studies in this

lab suggested that tin(IV) contamination still takes place

during the synthesis [49, 50]. This became evident when a

mixture of peptide standards was separated on the HC

phases, considerably wider peak widths (20–30%) and lower

plate counts were observed on these phases when compared

with conventional bonded C18 columns [49]. In addition, the

positive charges due to the presence of metal ions also affect

the performance of ion exchange phases based on the HC

platform [50], producing peak tailing and irreproducible

retention of analytes which could chelate metals (e.g. cate-

cholamines). In this study, kinetic studies were performed

to investigate each step of the Friedel-Crafts processes to

further optimize the HC phases. The effects of the reaction

times and crosslinking reagents were investigated to mini-

mize the silanolphilicity of the HC phases by improving the

synthesis conditions. The resulting third generation of HC

phases is denoted as the ‘‘HC-T’’ phase since in the last step

the HC platform was derivatized with toluene in contrast to

the second generation phase which is called the ‘‘HC-C8’’

phase because it is derivatized with an octyl phenyl group. It

should be pointed out that the purpose of this study is to

identify a set of synthetic conditions that will allow us to

further reduce metal contamination during the develop-

ment of HC phases.

2 Materials and methods

2.1 Chemicals

All solvents used in this study were of HPLC grade.

Acetonitrile (MeCN) was obtained from Burdick and

Jackson (Muskegon, MI). Dichloromethane was obtained

from Mallinkrodt-Baker (Paris, KY). 1,2-Dichloroethane was

obtained from Sigma (St. Louis, MO). Tetrahydrofuran

(THF) was obtained from EM Science (Gibbstown, NJ).

Acetone, methanol (MeOH), and isopropanol (IPA) were

obtained from PharmCo (Brookfield, CT). TFA, toluene,

triphenylmethane (TPM), chloromethylmethyl ether

(CMME), 2,4,6-tris(bromomethyl)-mesitylene (TBM), and

SnCl4 (99%) were obtained from Aldrich (Milwaukee, WI).

Dimethyl-chloromethylphenylethylchlorosilane (DMCMPES)

was obtained from Gelest (Tullytown, PA). Styrene heptamer

(SH) is a polymer standard (number average molecular

J. Sep. Sci. 2011, 34, 1407–14221408 Y. Zhang et al.


weight Mn: 800) purchased from Scientific Polymer Products

(Ontario, NY). HPLC water was prepared by purifying house

deionized water with a Barnstead Nanopure II deionizing

system with an organic-free cartridge and run through an

‘‘organic-free’’ cartridge followed by a 0.45 mm Mini Capsule

filter from PALL (East Hills, NY).

All chromatographic solutes were obtained from

Aldrich or Sigma. Polystyrene standards were obtained from

Polyscience (Warrington, PA) and dissolved in pure THF at

a concentration of approximately 0.5–2 mg/mL.

Type-B Zorbax silica particles and SB C18 particles were

gifts from Agilent Technologies (Wilmington, DE). The

particle diameter, surface area, and pore diameter of the

particles are 5.0 m, 180 m2/g, and 80 A respectively.

2.2 Kinetic studies

2.2.1 Primary crosslinking

One gram of Type-B Zorbax silica that had been silylated

with DMCMPES was slurried in 10 mL of 1,2-dichlor-

oethane and dried before use. The slurry was sonicated

under vacuum in a 50 mL round-bottom flask for 15 min to

fully wet the pores. After sonication, 1.2 mmol SH or

2.4 mmol TPM was added to the stirred slurry together with

0.1 mmol nitrobenzene as an internal standard. The

reaction mixture was then refluxed at 801C. Before adding

any catalyst, a small amount of sample was taken by syringe

to serve as a blank before adding any catalyst. Then 5 mmol

of SnCl4 was added to the slurry and served as the Friedel-

Crafts catalyst. An activated alumina column was used to

close the condenser to prevent water contamination from

lab air. After 0.5, 5, 15, and 60 min, samples were taken and

quenched with MeCN–water (70:30). The mixture was then

filtered and analyzed by HPLC.

2.2.2 Secondary crosslinking

One gram of starting material (Type-B Zorbax silica first

silylated with DMCMPES and then reacted with SH for

15 min) was slurried and sonicated in 10 mL of fresh 1,2-

dichloroethane as described above. An aliquot of 5 mmol

CMME or 1.8 mmol TBM was added to serve as the

secondary crosslinking reagent. The reaction vessels were

then refluxed at 501C. Before adding any catalyst, a small

amount of sample was taken as a blank. Then, 5 mmol

SnCl4 was added to serve as the Friedel-Crafts catalyst. An

activated alumina column was used to cap the condenser.

Samples were taken and filtered at 15, 30, 45, 60, and

90 min. The silica particles were washed sequentially with

350 mL aliquots of dichloromethane, THF, MeOH, MeOH/

water, and acetone. After washing, the silica was dried

under vacuum at 601C. A small fraction (�0.1 g) of the silica

sample was sent for elemental analysis. It should be pointed

out that CMME is a highly volatile, strong alkylating agent

with known toxicity. Thus, all the reactions were performed

in a hood with great caution exercised when handling this

chemical.

2.2.3 Third derivatization

One gram of starting material (Type-B Zorbax silica that had

been silylated with DMCMPES, reacted with TPM for

15 min and then with TBM for 45 min) was slurried and

sonicated in either 10 mL of 1,2-dichloroethane or toluene.

Then, 5 mmol octylbenzene was added to the slurry. The

reaction vessels were then refluxed at 801C. Before adding

5 mmol SnCl4 as the catalyst, a blank sample was taken. An

activated alumina column was used to close the condenser

during the reaction. Samples were taken and filtered at 10,

20, and 30 min. Before elemental analysis, the silica was

washed and dried as describe above.

2.3 Stationary phase synthesis

Both HC phases were prepared by a series of three SnCl4catalyzed Friedel-Crafts alkylations on Type B Zorbax silica

that had been silylated with DMCMPES. The detailed

reaction conditions to prepare HC-C8 are summarized in

Table 1 and can also be found in our previous publications

Table 1. Summary of synthetic conditions and separation efficiencies of various HC phases

Stationary phasea) Silica substrate Catalyst Reaction reagent Reaction time (min) Reaction temperature (1C) N/meterb)

Step 1 Step 2 Step 3 Step 1 Step 2 Step 3 Step 1 Step 2 Step 3 Neutral Basic

SB C18 Zorbax, 5 mm N/A N/A N/A N/A N/A N/A N/A N/A N/A N/A 90 000 71 000

HC-C8 Zorbax, 5 mm SnCl4 SH CMME Octylbenzene 180 90 30 80 50 80 102 000 59 600

HC-T Zorbax, 5 mm SnCl4 TPM TBM Toluenec) 15 45 10 80 50 80 105 000 78 600

a) For all crosslinking reactions, tenfold SnCl4 relative to surface DM-CMPES was used to catalyze all reactions. All reactions were done in

1,2-dichloroethane.

b) Chromatographic conditions: All columns (5.0�0.46 cm) are packed with 5 mm particles. 401C, F 5 1.0 mL/min. 0.1–0.2 nmol of

acetophenone (neutral), and nortriptyline (basic) were injected to assess column efficiency. However, %ACN was varied to make k0 similar

on two phases. For SB C18 phase, mobile phase: 0.1% formic acid in 38:62 ACN/water; for HC-C8 phase, mobile phase: 0.1% formic acid in

31:69 ACN/water; for HC-T phase, mobile phase: 0.1% formic acid in 29:71 ACN/water.

c) Toluene was used as solvent.

J. Sep. Sci. 2011, 34, 1407–1422 Liquid Chromatography 1409


[42, 45, 49]. To synthesize the HC-T phase (Fig. 1), the first

reaction used TPM at a molar ratio of 6:1 TPM/initial

surface DMCMPES groups (i.e. 18:1 phenyl rings/surface

chloromethyl groups) to crosslink the surface-bound chloro-

methyl groups on the silylated silica. The reaction was

stopped after 15 min of refluxing at 801C. In the second

step, we used a fourfold molar excess of TBM relative to the

amount of surface chloromethyl groups to further crosslink

the surface aromatic groups and simultaneously provide

additional chloromethyl functionalities for further derivati-

zation. This reaction was done at 501C for 45 min. In the last

step, toluene was used as both solvent and reagent

to react with the residual chloromethyl groups for 10 min

at 801C.

Before each Friedel-Crafts reaction, the slurry was

sonicated under vacuum for 15 min to fully wet the particle

pores. To prevent atmospheric water from deactivating the

catalyst, we used an alumina column to close the condenser

during the synthesis. After each Friedel-Crafts step, the

particles were filtered and washed sequentially with 350 mL

of fresh 1,2-dichloroethane, THF, THF with 10% concen-

trated hydrochloride, THF, MeOH, and acetone. At the end

of the washing sequence, the stationary phase was air dried

overnight at room temperature before the next step was

performed.

2.4 Acid pretreatment

After synthesis, the HC-T phase was preconditioned by

gradient acid (5% TFA) washing at 1501C as described in

our previous publications [45, 49]. The purpose

of this aggressive gradient acid washing is to: (i) remove

residual tin(IV) introduced during the Friedel-Crafts

reactions, (ii) hydrolyze residual chloromethyl groups and

the labile Si–O–Si bonds to prevent their slow hydrolysis

over time during use, and (iii) eliminate incompletely

crosslinked organic surface moieties. After gradient

washing, the column was flushed with 50:50

IPA/H2O thoroughly and then unpacked to dry the

particles.

2.5 Elemental analysis

After each step a small amount of stationary phases was

sent for carbon, hydrogen, bromine, and chlorine analysis

conducted by Atlantic Microlabs (Norcross, GA).

2.6 Column packing

HC particles were packed into a 5.0� 0.46 cm column for

further characterization. The particles were slurried in IPA

(1.0 g/8.0 mL) and sonicated for 20 min prior to packing.

Columns were packed by the downward slurry technique

using helium gas to drive a high-pressure pump (Haskel

16501, Haskel International, Costa Mesa, CA) and push IPA

through the column. The packing pressure was increased

from 3000 to 6000 psi within the first 5 s and then kept at

6000 psi until about 90 mL of solvent was collected before

the pressure was released.

2.7 Acid stability test

Dynamic stability tests were performed on a 5.0 cm� 0.21

cm HC-T column. A 50:50 v/v MeCN/H2O with 5% TFA

(pH 5 0.5) mobile phase flowed through the column at

1501C for more than 1000 column volumes. The relative

retention of a neutral probe was monitored and used to

evaluate the stability of the phase.

Primary Crosslinking

Secondary Crosslinking

Third Derivatization

Post Synthesis Acid Aging

Figure 1. Synthesis scheme for the HC-T phase.



2.8 Inverse size exclusion chromatography

Inverse size exclusion chromatography (ISEC) was

performed on a 5.0 cm� 0.46 cm HC-T column and a

5.0 cm� 0.46 cm column packed with bare Zorbax silica. A

series of low polydispersity polystyrene standards (Mw 5

300, 2727, 4000, 13.7� 103, 18.7� 103, 30� 103, 50� 103,

10� 106, and 20� 106 Da) were run on both columns with

100% THF as the eluent. The columns were thermostated at

401C and the flow rate was 1.0 mL/min. Toluene was

injected as the marker to measure the void volumes of the

two columns. The diameter of the polystyrene probes was

calculated using the method of Halasz and Martin [51].

2.9 SEM experiments

SEM analysis was performed on bare Zorbax silica, HC-T

phase, and HC-T phase after HF digestion. The last sample

was prepared by adding 0.5 mL of 48% HF (ultra-high

purity) into a slurry of 0.1 g of HC-T particles fully wetted by

10 mL of MeOH/water (50:50). The slurry was shaken and

then allowed to settle for 5 min before 0.1 g of boric acid was

added with 9.5 mL water. These treated particles were

washed extensively with 50:50 v/v MeOH/water and pure

MeOH before they were dried at 601C under vacuum. In all,

0.1 g of bare silica, untreated HC-T particles, and HF treated

particles were then evenly deposited onto carbon tape and

coated with 100 A of platinum. The SEM images were taken

with JEOL 6500 (characterization facility, University of

Minnesota, MN) with 10.5 mm working distances (WD) and

5 kV accelerating voltages. Magnification was set to be

5000� for all samples.

2.10 Chromatographic conditions

All chromatographic experiments were carried out on a

Hewlett-Packard 1090 system, equipped with a binary

pump, an autosampler, a temperature controller, and a

diode array detector (Hewlett-Packard S. A., Wilmington,

DE). Data were collected and processed using Hewlett-

Packard Chemstation software. The solutes were prepared

in ca. 1 mM 50:50 MeCN/H2O solutions and the injection

volume was 0.5 mL.

3 Results and discussion

Previous studies [44] suggest that the synthetic conditions of

the three Friedel-Crafts steps are critical to the performance

of the final HC phases. In particular, the changes in the

specific Friedel-Crafts catalyst and the specific silica

substrate used can significantly reduce the effect of the

metal contamination on silica surface and thus improve the

separation efficiency of the HC phases as discussed above

[44]. To further reduce tin(IV) contamination and optimize

the synthesis conditions, a study of all permutations of other

synthetic variables (e.g. reaction time, the concentration,

and the type of reagent) is clearly necessary to understand

the key experimental variables. As a result, a detailed kinetic

study was done to determine the effects of the crosslinking

reagents and reaction times on the performance of final

stationary phases while the catalyst amount was held

constant. However, it is important to note that all

optimizations must be done without sacrificing the cross-

linking efficiency and acid stability of the HC phases.

3.1 Effect of crosslinking and derivatization reagents

3.1.1 Optimization of new crosslinking reagents

As summarized in Table 1, in the first Friedel-Crafts step,

SH was initially used as the primary crosslinker based on its

high reactivity and multiple reactive phenyl groups available

for alkylation. However, one of our major concerns with this

chemical is the fact that it is not a pure compound but

rather a mixture of styrene oligomers with different chain

lengths (n 5 4, 5, 6, 7, 8y). This chemical complexity,

together with batch-to-batch differences, from the manu-

facture can lead to irreproducibility of the HC phases thus

prepared. Moreover, the relatively low solubility associated

with the higher-molecular-weight fractions limits the high-

est concentration that we can use and thus limits the overall

speed of reaction. In an effort to solve these problems, an

alternative crosslinkable aromatic compound, triphenyl

methane (TPM), was chosen based primarily on its chemical

simplicity and high solubility. As shown in Fig. 2, TPM is

also a multivalent reagent with the same reactive function-

ality, i.e. phenyl rings, as SH. Therefore, we expected to see

comparable reactivity with TPM as for the long-chain

oligomers. Indeed, very similar patterns were observed in

n=4,5,6, 7,8….

Triphenyl methane (TPM) Styrene heptamer (SH)

2,4,6-Tris(bromomethyl)-mesitylene(TBM)

Chloromethylmethylether(CMME)

Br

Br

Br

n

o o

A

B

Figure 2. The structures of the crosslinking reagents used in theFriedel-Crafts reactions. (A) Primary crosslinker: TPM versus SH;(B) secondary crosslinker: TBM versus CMME.



both kinetic plots as shown in Fig. 3A, suggesting that TPM

is as good as SH but with higher purity and better solubility.

Although the secondary crosslinker CMME has shown

exceptional alkylating activity and provided highly efficient

crosslinking in the second step of Friedel-Crafts reactions, it

is a highly volatile reagent with known carcinogenicity. This

strongly restricts the use of CMME in practice, especially at

elevated temperatures and high concentrations. In addition,

when handling this type of chemical great caution needs to

be taken due to its toxicity and volatility. With all these

concerns, we decided to test 2,4,6-tris(bromomethyl)-mesi-

tylene (TBM) as an alternative secondary crosslinker. The

major advantage of TBM over CMME is that it is a nonvo-

latile thus less hazardous but still a strong alkylating

reagent. As shown in Fig. 2B, there are three reactive

–CH2Br groups in each molecule and thus it might provide

a higher extent of crosslinking to generate a more exten-

sively crosslinked polymer network than the di-functional

analog CMME. In addition, the higher reactivity of –Br

versus –Cl [52, 53] makes it possible that a reasonable

crosslinking degree could be achieved under milder reaction

conditions (e.g. shorter reaction times), thereby reducing

the effect of tin(IV) contaminations. However, to our

surprise, considerably slower reaction kinetics were

observed with TBM when compared with that of CMME

(Fig. 3B). For example, after an hour of reaction with surface

phenyl rings, there were only 1.070.1 mmol/m2 of TBM

attached to the aromatic network based on carbon content

analysis, which was 60% less than the loading density of

CMME (i.e. 2.770.1 mmol/m2). One potential cause of this

slow reaction kinetics is the low solubility of TBM in the

solvent 1,2-dichloroethane. In fact, the amount of TBM that

can be dissolved and used in the kinetic study was only one-

third compared with CMME. The dramatically decreased

concentration of TBM versus CMME can cause significantly

lower reactivity for TBM as measured by the number of

molecules reacted per unit time. Another possible reason is

the size difference between the two molecules. As shown in

Fig. 2, TBM is clearly bulkier with 12 carbons compared

with CMME. Such structure can generate steric problems in

reactions (especially at solid surfaces) and thereby lead to

slower kinetics. Nevertheless, we felt that the reactivity of

TBM was sufficient for the formation of a surface polymer

network; ultimately, this was confirmed and will be

discussed below. Thus, we decided to use TBM in place of

CMME in our modified synthetic scheme.

3.1.2 Optimization of new derivatization reagents

For the last step, we also employed Friedel-Crafts chemistry

mainly because it allows us to easily introduce different

functionalities into the surface aromatic groups. Thus, we

can potentially produce a wide variety of stationary phases

with diverse chromatographic selectivities. During the

development of the second generation of HC phases,

octylbenzene was attached to the surface polymer scaffold

to produce a reversed type phase (i.e. the HC–C8 phase).

However, as noted above, a chemical reagent with higher

reactivity should be more favorable since it might facilitate

the completion of reaction under milder reaction conditions.

Thus, we decided to replace octyl benzene with toluene

since it can not only be used as a reagent, but more

importantly, it can also be used at very high concentration

by taking it as the solvent for the third Friedel-Crafts step.

The kinetics of the modified reaction is shown in Fig. 3C,

together with that of the original Friedel-Crafts reaction

0.00

0.05

0.10

0.15

0.20

0 15 30 45 60Reaction time (min)

load

ing

dens

ity

of p

rim

ary

cros

slin

kers

(µm

ol/m

2 )

0.0

1.0

2.0

3.0

4.0

0 30 60 90Reaction time (min)

load

ing

dens

ity

of s

econ

dary

cros

slin

kers

(µm

ol/m

2 )

A

B

C

0

0.3

0.6

0.9

1.2

1.5

0 10 20 30Reaction time (min)

load

ing

dens

ity

of d

eriv

atiz

atio

nre

agen

ts (

µmol

/m2 )

Figure 3. Kinetic studies of crosslinking reagents in Friedel-Craftsreactions. (A) Primary crosslinker: TPM (�) versus SH (3);(B) secondary crosslinker: TBM (m) versus CMME (n); (C) thirdderivatization reagents: toluene (& ) versus octyl benzene (&).



using octylbenzene. As expected, due to the dramatically

increased concentration (at least an order of magnitude) of

toluene versus octylbenzene, the optimized reaction is

much more effective as indicated by almost a twofold

increase in the number of molecules reacted.

3.2 Effect of reaction time

During the kinetic studies, we made a very important

observation. Even with a mild Lewis acid such as tin(IV)

chloride as the catalyst, to our surprise, the kinetics of the

Friedel-Crafts reactions were much faster than we had

expected. Specifically, for the primary Friedel-Crafts cross-

linking and the tertiary Friedel-Crafts derivatization, the

reactions were nearly complete within the first 15–20 min,

whereas for the secondary Friedel-Crafts crosslinking step,

although both CMME and TBM were still reacting at

90 min, the bulk of the carbon added (�80%) was loaded

onto the surface within the first 45 min. This discovery

allowed us to synthesize our HC phases using much shorter

reaction times thereby greatly decreasing the exposure of

the silica to metal ions without sacrificing the degree of

crosslinking, that is a network polymer was still formed and

acid stability preserved (see below).

3.3 Synthesis and elemental analysis of the HC-T

phase

The synthesis scheme for the HC-T phases and the phase

structure at each step are shown in Fig. 1. In order to

understand this complex, multistep synthesis, the product

of each stage in the reaction was characterized by elemental

analysis and compared with that of the HC-C8 phase

(Table 2). Since the chemicals and reaction time in each step

are now different, consequently the absolute carbon loading

on these two HC phases is different. In order to make a fair

comparison of the relative success of each step of the

synthesis for both the HC-C8 and the HC-T phases, ‘‘loading

density’’ was calculated based on the number of moles of

carbon added during the reaction [44],

mmol=m2 ¼ ð%2C�%1CÞ � 106

12� NC � ð100�%2C�%2H�%2ClÞ

� 1

Sð1Þ

where %1C is the percentage of carbon before crosslinking,

%2C the percentage of carbon after crosslinking, %2H the

percentage of hydrogen after crosslinking, %2Cl the

percentage of chlorine after crosslinking and NC the

number of carbon atoms per crosslinker molecule.

The results are summarized in Table 3. After primary

crosslinking, there were 0.25 mmol/m2 of TPM loaded onto

silica on the HC-T phase, whereas 0.20 mmol/m2 of SH

groups was added to the HC-C8 phase. The result obtained

here agrees very well with observations from the kinetic

study, suggesting TPM is indeed a highly efficient cross-

linker. Note that although the reaction time was now greatly

reduced from 180 to 15 min, no loss in crosslinking effi-

ciency was found since the reaction was essentially

completed within the first 15 min.

In the secondary crosslinking reaction, less TBM was

reacted compared with CMME. This was anticipated and

could be attributed to the lower reactivity observed for TBM

in the kinetic studies as well as the shortened reaction time

used in the modified synthetic scheme. Nevertheless, the

amount of TBM loaded on the surface was adequate to

ensure the formation of a network polymer on the silica

surface (see below).

For the third derivatization, to our surprise, the loading

density of toluene is only 0.44 mmol/m2, approximately 50%

less than the loading of octylbenzene found in our previous

study [44]. Based on the results of the kinetic studies shown

in Fig. 3C, we expected to see more toluene loaded onto the

surface of silica than octylbenzene. However, the toluene-

loading results in the large-scale derivatization process

(0.44 mmol/m2) in 10 min do not agree well with the results

found in the kinetic study (Fig. 3C) where almost

0.80 mmol/m2 was loaded in 10 min. The large-scale deri-

vatization was repeated twice with nearly the same result. As

far as we can state that all conditions (i.e. reaction time,

Table 2. Summary of elemental analysis at each stage in the

synthesis of the HC-T and HC-C8 phasea)

Silica substrate HC-Tb) HC-C8b)

C% H% Cl% C% H% Cl%

DM-CMPES 6.61 0.92 1.51 6.30 0.79 1.53

Primary crosslinking 7.53 0.94 0.35 8.27 0.91 o0.25c)

Secondary crosslinking 8.73 1.04 0.77 8.94 1 1.61

Derivatization 9.32 1.07 o0.25c) 11.49 1.27 0.61

after gradient washing 9.72 1.24 o0.25c) 11.31 1.25 0.27

a) All results are represented in w/w.

b) Both of these two columns are based on Zorbax silica.

c) Detection limit is 0.25% Cl (w/w).

Table 3. Summary of loading density at each stage in the

synthesis of the HC-T and HC-C8 phase based on

elemental analysis

Reagents Loading density of crosslinking reagenta)

(mmol/m2)

HC-T HC-C8

TPM/SH 0.25 0.18

TBM/CMME 0.52 3.51

Toluene/octylbenzene 0.44 0.97

a) Calculated based on carbon content (w/w) from elemental

analysis.



temperature, and relative amount of reagents) are the same,

the only difference is the scale of the reaction – 1 g of silica

was used in the kinetic study and 10 g of silica was used in

the large-scale derivatization. We have no explanation for

this difference other than reaction scale.

3.4 Characterization of the stability of HC-T polymer

network by SEM

To verify that an HC network polymer had been formed on

the interior surface of the silica particle by the new synthetic

scheme, we took SEM pictures of the coated silica particles

before and after removal of the silica skeleton by exhaustive

dissolution of the silica by hydrofluoric acid. The resulting

micrographs together with the images of bare Zorbax

particles are shown in Fig. 4. If we first compare the

particles before and after the hyper-crosslinking modifica-

tions, the morphologies of the two cases are essentially

the same, both of which are spherical particles with an

average diameter of 5 mm. Since silica can be totally

dissolved in hydrofluoric acid, there was nothing left when

we treated the bare Zorbax particle by hydrofluoric acid

digestion. However, the result is clearly different for the

HC-T particles as indicated by the residual microsphere of

polymer beads shown in Fig. 4C. This is because, even after

the silica substrate was fully removed, the polymer network

remains intact due to the extensive crosslinking and

network formation. These results confirm the formation of

HC network polymers on the interior surface of the HC-T

phase. The existence of such a network can prevent phase

loss even under very aggressive conditions [44, 45]; thus, the

HC-T phase is expected to have high chemical stability

when compared with conventional ODS phases. Note that

the size of the polymer beads is significantly smaller than

the HC-T phase before HF digestion. We believe that the

decrease in image size is mainly due to the removal of the

silica support and the drying process during sample

preparation before SEM.

3.5 Characterization of the stability of HC-T phase by

dynamic aging

The chemical stability of the HC-T phase was further tested

and compared with the second HC-C8 phase and sterically

protected C18 phases under more chromatographic relevant

conditions. We used the same aging conditions as in our

previous study [44, 45]. The results are shown in Fig. 5 in

which the relative retentions of a neutral probe (i.e.

hexadecanophenone) were plotted against the volume of

mobile phases flushed though the columns. As shown in

Fig. 5, all of the plots followed the same pattern with

normalized retention factors decreasing over time.

However, it is evident that the decrease of retention is

more rapid on the sterically protected C18 phase, which is

the one of the best acid-stable commercial phases, whereas

the two HC phases showed much less tendency to change.

For example, after 1000 column volumes, the retention

factor of hexadecanophenone dropped by 35% on the

sterically protected C18 phase, whereas it dropped only

15% on the two HC phases. It should be stressed that the

Figure 4. SEM images of the bare Zorbax particle, HC-T phases before and after removal of the silica substrate by HF digestion. (A) BareZorbax particles; (B) the HC-T-coated silica particles before HF treatment; (C) the HC-T-coated silica particles after HF removal of the silicasubstrate.

Figure 5. Stability comparison of the HC-T phase (m), the HC-C8

phase (n), and the SB C18 phase (�). The stability test conditionsare 5% TFA in 50:50 ACN/water (v/v), T 5 1501C. Data of SB C18

and HC-C8 are adapted from [44].



decrease in retention of both HC phases is not due to the

decreases in the phase ratio as it is for conventional, i.e. non-

HC phases. This was verified by the minimal loss in the

carbon content of the HC-T (4%) phase and HC-C8 phase

(2%) (Table 2) after this very aggressive acidic washing

procedure. This agrees well with our previous studies of all

the other HC phases [44, 45] and further confirms the

existence of HC networks on silica surface. We believe that

the changes of the k0 value on HC phases are primarily due

to the hydrolysis of the siloxane bonds and chloromethyl

groups. It is clear that the two HC phases possess

outstanding stability even compared with sterically

protected C18 phases. More importantly, the new HC-T

phase is just as stable as the second generation of HC-C8

phase, suggesting the improvement in synthetic scheme

was not achieved at the price of diminished chemical

stability.

3.6 Characterization of pore size distributions of

HC-T phase

It is very important to point out that the two sequential

Friedel-Crafts coupling reactions did not compromise the

pore accessibility of the underlying silica substrate. This was

confirmed by ISEC. The results of the ISEC study are shown

in Fig. 6. It is clear that the pore accessibility of the HC-T

phase is even better than that of the monomeric silanized

Extend C18 phase, which is also based on Zorbax silica. This

suggests the Friedel-Crafts polymer forming reactions are

confined to the surface and there is no pore blockage which

is believed to be the primary cause for poor chromato-

graphic efficiency of many polymer-coated stationary phases

[54–57]. The results here agree well with the previous

conclusions based on our earlier HC phases [44, 45].

3.7 Separation efficiency for basic analytes

The silanolphilicity of different HC phases was studied and

compared by testing the phases with several representative

basic drugs as probes; the resulting chromatogram is shown

in Fig. 7. The observed plate counts, retentions, and peak

shapes of the basic analytes on the two HC phases are

compared in Table 4. As it is clear from the table, the

observed plate counts and peak shapes for all three basic

analytes are much better on the HC-T phase than on

the HC-C8 phase, whereas these two phases gave similar

plate counts for neutral compounds such as aceto-

phenone. These results strongly suggest that the effect of

the metal contamination was reduced significantly when the

reaction times during phase preparation were properly

optimized.

Next, the separation efficiency of the HC-T phase was

compared with that of three commercial stationary phases.

One of them is the sterically protected C18 phase, which is

also based on the same silica used to make the HC phase

(Zorbax silica). The other two are the ACE C18 and the

Inertsil ODS-3 phase, both of which are made on high-

quality silica substrates with endcapping and thus provide

high efficiencies for basic solutes as long as the columns are

not overloaded [58, 59]. Small amounts of solutes were

injected in all cases and the resulting chromatograms are

0

0.2

0.4

0.6

0.8

1

1.2

0 100 200 300 400Approximate Pore Diameter (A)

Fra

ctio

n of

por

e vo

lum

e ac

cess

ible

Figure 6. Pore accessibility study by ISEC with polystyrenestandards. Mobile phase: 100% THF; T 5 401C; F 5 1.0 mL/min,l5 254 nm. & Bare Zorbax silica; m Extend C18; � HC-T.

0

5

10

15

20

25

30

35

time (min)

Abs

orba

nce

(mA

U)

0

5

10

15

20

25

30

0 0.5 1 1.5 2 2.5 3 3.5 4 4.5

0 0.5 1 1.5 2 2.5 3 3.5time (min)

Abs

orba

nce

(mA

U)

Alprenololk’ = 1.07N = 4210

Amitriptylinek’ = 4.51N = 3560

Nortriptylinek’ = 3.75N = 3940

Alprenololk’ = 0.92N = 2830

Nortriptylinek’ = 3.57N = 2980

Amitriptylinek’ = 4.53N = 2720

A

B

Figure 7. Silanophilicity comparison of the HC-T and the HC-C8

phases. A mixture of alprenolol, nortriptyline, and amitriptylinewas separated on: (A) HC-T, 4.6� 50 mm (Nacetophenone/meter 5

102 000) and (B) HC-C8 phase, 4.6� 50 mm (Nacetophenone/meter 5 102 000) in 0.1% v/v formic acid-buffered mobile phases,%ACN was varied to make k0 similar on both phases. For HC-TfACN 5 29%; for HC-C8 fACN 5 31%. T 5 401C, F 5 1 mL/min,l5 254 nm. 0.1–0.2 nmol of samples were injected.



shown in Fig. 8. As clearly shown, the new HC-T phase gave

the highest plate counts for all three basic analytes using

formic acid-buffered mobile phase, which was deliberately

chosen. Formic acid is a very weak ion-pairing agent [60, 61]

and thus using it as the buffer constitutes a more strenuous

test media as compared with TFA or perchloric acid buffers

[60, 61]. It is very important to note that the HC-T phase was

in fact prepared from Zorbax silica. Previous study in this

lab had already demonstrated that this type of silica typically

shows a relatively strong silanol activity and thus it did not

perform as well as did the other lower activity silica when

used as the substrate on which to form an HC phase [44].

Thus, the excellent peak shapes observed in this study

support our view that the decreased period of exposure to

the Friedel-Crafts metal catalyst (tin(IV)) in the present work

decreased the contamination of the surface with metal and

gave a less silanophilic product.

3.8 Separation efficiency for peptide standards

Previously, when a mixture of peptide standards was

separated on the second generation of HC phases,

considerably wider peaks (20–30%) were observed on these

phases when compared with conventional-bonded C18

columns [49]. We believe that this is mainly due to the

strong silanophilicity caused by tin(IV) contamination

during the three steps of Friedel-Crafts reactions. Since

our newly developed HC-T phases showed significantly

improved efficiency for basic analytes in acidic conditions, it

was of great interest to evaluate the separations of peptides

on the new HC-T phases, to find out the effect of synthesis

optimization on the separation of peptides on HC phases.

The sample chromatograms are shown in Fig. 9. The

measured retentions and peak widths of the nine peptides

on the two HC phases together with the results of the

Table 4. Efficiency comparison of three basic drugs in formic acid-buffered mobile phasea)

Stationary phase Acetophenone Alprenolol Nortriptyline Amitriptyline

k0 N/meter k0 N/meter k0 N/meter k0 N/meter

HC-Ta) 1.65 105 800 1.07 84 200 3.75 78 800 4.51 71 200

HC-C8a) 4.43 102 600 0.92 56 600 3.57 59 600 4.53 54 400

SB C18b) N/A 90 000 1.31 18 000 3.74 71 000 4.28 55 000

ACE C18b) N/A 90 000 1.16 29 000 3.62 75 000 4.06 68 000

Inertsil ODS3b) N/A 80 000 0.83 33 000 3.8 70 000 4.22 60 000

a) The chromatographic conditions are the same as described in Fig. 7.

b) Data were obtained from [44]. The chromatographic conditions are the same as described in Fig. 8.

Time (min)

0

2

4

6

8

10

1 23

A B

C D

Time (min)

0

2

4

6

8

10

1

23

Time (min)0.0 0.5 1.0 1.5 2.0 2.5 3.0 0.0 0.5 1.0 1.5 2.0 2.5 3.0

0

2

4

6

8

10

12

3

0

5

10

15

20

25

30

35

0.0 0.5 1.0 1.5 2.0 2.5 3.0 0 0.5 1 1.5 2 2.5 3 3.5 4.5Time (min)

Abs

orba

nce

(mA

U)

Abs

orba

nce

(mA

U)

Abs

orba

nce

(mA

U)

Abs

orba

nce

(mA

U)

1

2

3

4

Figure 8. Comparison of separationefficiencies of the HC-T, the SB C18,the ACE C18, and the Inertsil ODS-3phases with three basic drugs.Separation of alprenolol, nortripty-line, and amitriptyline was done on(A) SB C18, 4.6� 50 mm; (B) ACE C18,4.6�50 mm; (C) Inertsil ODS-3,4.6�50 mm; (D) HC-T, 4.6�50 mm in0.1% v/v formic acid-buffered mobilephases, %ACN was varied to make k0

similar on all phases. For SB C18

jACN 5 38%; for ACE C18 jACN 5 34%;for Inertsil ODS-3 jACN 5 16.3%; forHC-T jACN 5 29%. T 5 401C, F 5 1 mL/min, l5 254 nm. In total, 0.1–0.2nmol of samples was injected. (1)Alprenolol, (2) nortriptyline, and (3)amitriptyline. SB C18, ACE C18, andInertsil ODS-3 data are adapted from[44].



sterically protected C18 phase and the ACE C18 are listed and

compared in Table 5. Under gradient elution conditions, the

quality of separation is usually evaluated by the peak width

at half height in time unit, W1/2. As it is clear from the table,

among the stationary phases tested the peptide standards

clearly showed the widest peaks on the HC-C8 phase. On the

other hand, all nine peptides have similar if not exactly the

same observed W1/2 values on the HC-T phase when

Figure 9. Comparison of separation efficiencies of the HC-T, the HC-C8, the SB C18, and the ACE C18 phases with peptide standards.Chromatographic conditions: All columns (5.0� 0.46 cm) are packed with 5 mm particles. Mobile phases: solvent A: 0.1% formic acid in5:95 v/v ACN/water, solvent B: 0.1% formic acid in 55:45 v/v ACN/water. Gradient profile: 0.00–15.00 min 10–80% B, 15.00–15.01 min80–10% B, 15.01–22.00 min 10% B. F 5 1.4 mL/min, T 5 401C, l5 214 nm. Nine peptides mixture (0.01–0.08 mg/mL) 5 mL injection. Solutes:1, Gly-Phe; 2, Phen-Phe; 3, LHRH human; 4, angiotensin II; 5, [Val5] angiotensin; 6, substance P; 7, renin substrate; 8, insulin chain B; 9,melittin; (A) HC-T; (B) HC-C8; (C) ACE C18; (D) SB C18.

Table 5. Separations of nine peptides on various stationary phases in TFA-buffered mobile phasesa)

Samples Retention time (min) Peak width W1/2 (min)

HC-T HC-C8 SB C18 ACE C18 HC-T HC-C8 SB C18 ACE C18

Gly-Phe 2.47 1.88 3.46 3.59 0.069 0.065 0.069 0.077

Phe-Phe 6.52 5.52 7.36 7.37 0.086 0.097 0.079 0.088

LHRH human 8.00 7.75 7.87 7.78 0.072 0.082 0.064 0.069

Angiotensin II 8.52 8.10 8.96 8.91 0.076 0.086 0.069 0.073

[Val5]-Angiotensin I 8.97 8.62 9.60 9.59 0.080 0.086 0.073 0.076

Substance P 10.10 9.88 10.58 10.64 0.075 0.084 0.073 0.076

Renin substrate 11.08 10.83 11.47 11.46 0.080 0.084 0.076 0.077

Insulin chain B 12.66 12.62 12.70 12.74 0.075 0.081 0.073 0.070

Melittin 15.62 15.75 17.22 17.37 0.104 0.109 0.155 0.110

Average 0.080 0.086 0.081 0.080

Standard deviation 0.010 0.012 0.028 0.013

a) Chromatographic conditions are the same as described in Fig. 9.



compared with the other two C18 phases. This result agrees

very well with the observations from the separations of basic

analytes, suggesting that the surface silanols were indeed

involved in the poor peak shape of the bases and peptides on

the HC-C8 phase. The silanolphilicty of the HC-T phase has

been greatly reduced by optimizing the synthesis condi-

tions. As a result, the separation efficiency of basic drugs

and peptides on the HC-T phase was significantly improved

and is now comparable if not actually superior to the SB C18

and ACE C18 phases.

3.9 Stationary phase selectivity

3.9.1 Characterization by hydrophobic subtraction

model

The selectivity of the new HC-T phase was studied by the

hydrophobic subtraction model (HSM) [62–70] using the

equation

logðk0=k0EBÞ log a ¼ Z0H � s0S�1b0A1a0B1k0C ð2Þ

(i) (ii) (iii) (iv) (v)

In this method, the relative retention of a solute k0=k0EB

on a reversed-phase column can be separated into contri-

butions from various types of intermolecular interactions.

According to the definition given by Snyder et al. [62–70],

these are hydrophobic interactions (i), steric resistance of

the stationary phase to insertion of the bulky molecules (ii),

hydrogen bonding of acidic solutes with a basic column

group or basic solutes with an acidic column group (iii)–(iv),

and coulombic interaction (v). A set of 16 judiciously

selected but chemically simple probe solutes were used to

characterize the chromatographic system by these interac-

tions, wherein H, S�, A, B, and C are the column selectivity

parameters and Z0, s0, b0, a0, and k0 are the corresponding

solute properties.

The new HC-T phase was studied to compare it with the

reversed-phase HC-C8 and commercial phases in Table 6.

As indicated by the regression results, all of the HC-based

phases are well fit by the HSM.

It is clear from the results summarized in Table 6 that

the hydrophobicity of the HC-T phase is almost the same as

the average of commercial embedded polar groups (EPG)

phases; this is higher than the H for the average phenyl

phases but lower than that for the C8 phases. The phenyl

phases show the lowest H due to the absence of a hydro-

phobic alkyl chain. The lower H of the HC-T phase

compared with the average H of C8 phases (either on our

HC platform or commercial silica) results from a significant

decrease in ligand length (C1 versus C8) and ligand density

(0.44 versus 0.97 versus 2–3.5 mmol/m2). This is clearly

indicated by the relatively lower carbon content measured by

elemental analysis as discussed above.

The S� coefficient, which indicates the stationary phase

steric resistance, is clearly negative and large in magnitude

on the new HC-T phase. This agrees well with our previous

studies of all the other HC phases [43, 44, 71], suggesting

the two solutes (i.e. trans-chalcone and cis-chalcone), which

are the major determinants of S�, are also more strongly

retained on the HC-T phase than on conventional RPLC

phases. We believe that this is mainly attributed to the

strong p–p interaction between the two p-active probes and

the highly aromatic HC-T phase as shown in Fig. 1. The

presence of the p–p interaction is further confirmed by this

study in which a set of compounds with various aromati-

cities were tested on the HC phases (Section 3.9.2). Another

potential cause of high retention of these two solutes on

HC-T phase is the rather low surface density and the

absence of the long alkyl chain on HC-T surface as discus-

sed above; thus, the steric repulsions on the new materials

are significantly reduced.

The A-values, which represent the stationary phase H-

bond acidity, are higher for all the HC phases as compared

with the averages of the other types of reversed phases and

are among the top 10% of all the RPLC columns studied. As

previously reported [43–45], the strong H-bond acidity is

primarily caused by the large number of active hydrogen

bond donors generated upon hydrolyzing the siloxane bonds

Table 6. Comparison of HSM coefficients on different stationary phasesa)

Stationary phases Column coefficients

H S� A B C(2.8) log k0EB r2 SE

HC-T 0.7270.02 �0.2270.02 0.3870.03 �0.0270.01 0.4470.02 0.2070.01 0.998 0.02

HC-C8, Zorbax 0.8170.02 �0.2870.02 0.2870.03 0.0270.01 0.2670.02 0.4670.01 0.999 0.02

HC-C8, Hichrom 0.7570.02 �0.2770.03 0.1370.05 0.0470.02 0.2470.03 0.6970.02 0.997 0.03

Average of type A C8b) 0.80 �0.05 0.01 0.07 0.59 0.60 N/A N/A

Average of type B C8b) c) 0.81 �0.01 �0.13 0.02 0.02 0.36 N/A N/A

Average of Phenylb) 0.63 0.10 �0.23 0.02 0.18 0.71 N/A N/A

Average of EPGb) 0.73 �0.01 �0.32 0.11 �0.41 0.38 N/A N/A

a) The chromatographic conditions: 50:50 v/v ACN/60 mM phosphate (pH 5 2.8), T 5 351C, F 5 1.0 mL/min.

b) The data for different commercial phases were obtained from [43].

c) The type B C8 phases are the C8 phases synthesized on type B silica.



on the surface of the HC platform during the post-synthesis

hot acid washing. Comparing the A-value of the three HC

phases, a decrease in A was expected on HC-T upon redu-

cing exposure to Sn(IV) and thus decrease silanol activation.

However, this is not observed, suggesting that the H-bond

acidity is mainly controlled by the intrinsic acid activity of

the underlying silica substrate.

The B-term, which represents H-bond basicity, is lower

on the HC phases than the average polar-embedded phase

but similar to the other classes of conventional phase

(Table 6). Clearly, the B-values of the three HC phases are

essentially the same. According to Snyder, the B coefficient

tracks the amount of water adsorbed on the stationary phase

[62]. This suggests that the amount of water adsorbed on the

HC phases is mainly controlled by the surface silanol

groups and possibly the benzyl alcohol groups generated by

the hot acid washing step.

Finally, it is clear from the results summarized in

Table 6 that the cation exchange capability (represented by

the C-term) follows the order: EPGoType-B C8oPhe-

nyl�HC-C8oHC-ToType-A C8. As expected, Type-A C8

phase showed the highest C due to the greatest acidity of the

underlying silica substrate. The higher C of the HC phases

compared with the average type B C8 phase indicates their

stronger cation exchange ability at pH 2.8. We believe this

results from (i) the large number of surface silanol groups

released by the hot acid treatment; (ii) the HC phases are

not endcapped because they are designed to be used in

strong acid media. For both basic test compounds, the HC-T

phase is more retained than either of the two HC-C8 phases.

This is surprising as we expected to see reduced silanol

activity on the new material because we greatly decreased its

exposure to the activating effect of metals during synthesis.

However, considering that the HC-T phase is based on a

more acidic silica substrate with less shielding from short

alkyl chain, the increase in C of the new phase may be

attributable to easier access to free silanols. For the EPG

columns, the C-term is mostly negative and larger in

magnitude at pH 5 2.8 as shown in Table 4. As noted

previously [65], this can be attributed to the presence of

positively changed amino groups at low pH, leading to near-

zero or even ‘‘negative’’ retention of ionized cations because

of cation exclusion.

3.9.2 Characterization with aromatic solutes

The synthetic scheme shown in Fig. 1 clearly indicates that

a large portion of the phase is aromatic. In addition to

ensuring the high chemical stability of the HC phases, the

aromatic HC network also provides enhanced selectivity for

aromatic compounds. This is indicated by the negative S�

coefficient in the HSM study in which the retention of two

p-active probes is greatly enhanced on the HC phases

(Section 3.9.1).

To further assess such selectivity, the HC-T phase was

studied with a set of 28 nonelectrolyte solutes and compared

with the HC-C8, SB-C18, and SB-Phenyl phases. The 28

solutes were selected to span a wide range in aromaticity and

thus can be divided into three groups: Group (I) benzene

and alkylbenzene homologs; Group (II) p-active probes with

strong electron-donating or electron-withdrawing groups

activating the phenyl ring; Group (III) polycyclic aromatic

hydrocarbons (PAH) with highly conjugated structures. The

k–k plots of the HC-T phase versus the other three phases

are shown in Fig. 10.

A k–k plot is a plot of log k0 from one chromatographic

system versus the log k0 from another chromatographic

system. Under the same elution conditions, the difference

between the two chromatographic systems reflects directly

the differences between the two stationary phases. Accord-

ing to Horvath and coworkers [72], a good linear correlation

between two sets of log k0 indicates similar retention

-0.40

0.00

0.40

0.80

1.20

logk' / HC-C8

logk

' / H

C-T

-0.40

0.00

0.40

0.80

1.20

logk' / SB C18

logk

' / H

C-T

0.20

0.60

1.00

1.40

1.80

logk' / SB C18

logk

' / H

C-C

8

-0.40

0.00

0.40

0.80

1.20

0.20 0.60 1.00 1.40 1.80 0.20 0.60 1.00 1.40 1.80

0.40 0.80 1.20 1.60 2.00 0.10 0.50 0.90 1.30logk' / SB Phenyl

logk

' / H

C-T

A B

C D

Figure 10. Selectivity comparisonbetween different phases for aromaticcompounds based on k–k plots. Chro-matographic conditions: 5.0� 0.46 cmcolumn, 50:50 MeCN/H2O, T 5 401C,F 5 1.0 mL/min. Solutes: Group I (~):benzene; o-, m-, p-xylene; i-, n-propylbenzene; t-, s-, i-, n-butylben-zene; 1,2,3,4-, 1,2,3,5-, and 1,2,4,5-tetramethylbenzene; Group II (3): 1,3-dimethyl-2-nitrobezene, 1,2-dimethyl-3-nitrobenzene, 1,4-dimethylnitroben-zene; cis-, trans-chalcone; cis-, trans-stilbene; o, m-, p-terphenyl. Group III(�): naphthalene; anthracene; pyrene;triphenylene. (A) HC-T versus HC-C8;(B) HC-T versus SB-C18; (C) HC-C8

versus SB C18; (D) HC-T versus SB-Phenyl.



mechanisms of the two stationary phases. A slope close to

unity indicates almost identical energetics of retention. On

the other hand, a poor linear correlation of the k–k plot

implies different selectivities. As expected, the HC-T shows

very good correlations with the HC-C8 phases. All three

groups of compounds fall on a single straight line with the

correlation coefficient close to one. This suggests that the

two phases based on the same HC platform are indeed very

similar in the separation of these investigated solutes. On

the other hand, selectivity of the HC-T phase is clearly

different from the commercial C18 phase as illustrated by

the distinct retention patterns. As shown in Fig. 10B and C,

there are three straight lines well separated in the plots. This

indicates the differences in selectivity between the two

stationary phases toward the different types of compounds.

In particular, the slopes on both HC phases relative to the

C18 column follow the order: benzene and alkylbenzene

homologs (least)op-active probesoPAH (most), suggesting

the preferred retention of solutes with higher aromaticity on

these HC materials. The same trend was reported previously

where the separation of several aliphatic and aromatic

solutes was compared between a C18 and a phenyl column

[73, 74]. As pointed out by Tanaka et al. [74], this can be

attributed to the presence of p–p interaction between the

unsaturated solutes and the phenyl ligands. Figure 10D

shows the same selectivity comparison made between the

HC-T and the SB-Phenyl phase. It is clear that the HC-T

phase exhibits a greater difference in selectivity from the

aliphatic phase SB-C18 and it certainly more closely resem-

bles the conventional aromatic SB-Phenyl phase. This

further confirms the p–p interactions involved in the

retention mechanism of HC platform and strongly suggests

that the HC-phase has a special selectivity toward unsatu-

rated compounds.

It is also noteworthy that the intercept of k–k plots can

be used to measure the differences of the phase ratios

between two columns [72]. A negative intercept indicates the

lower phase ratio of the stationary phase on the y-axis when

compared with the one on the x-axis. Based on the average

intercepts of three solute groups as summarized in Table 7,

the relative magnitude of the phase ratios follows the order:

HC-ToSB-PhenyloHC-C8oSB-C18, which in general

agrees with the order of hydrophobicity measured by the

H-value from the HSM model. As expected, the phase ratio

of the HC-T phase is substantially lower than that for the

HC-C8 phase but more closely resembles the phenyl phase.

We believe that this is primarily due to the significantly

reduced carbon load on the HC-T phase as discussed above.

4 Concluding remarks

Kinetic studies have been carried out on the effect of

reaction time and cross-linking reagents used in preparing

HC phases. We conclude that:

(i) The Friedel-Crafts chemistry used to prepare HC

phases is nearly complete within about 15 min. Thus,

a new generation of HC materials for HPLC was

developed by greatly reducing the reaction time for

each step of the synthesis from the multihour

reactions used previously.

(ii) The acid stability of the new HC phase is not

compromised by these synthetic changes. Stability

tests at 1501C in 5% TFA indicate that the new phase

is just as stable as the previously developed HC

phases, both are more than an order of magnitude

more stable than the benchmark conventional-type

C18 phase.

(iii) Preliminary results also showed a much better

efficiency for both small basic analytes and peptides

on the new HC-T phase compared with both

the HC-C8 phase and the commercial stationary

phases.

(iv) Given its superb acid stability and outstanding

chromatographic efficiency, this phase will be useful

for ultra fast high-temperature chromatography [75]

or as the second-dimension separation media in high-

temperature fast 2-DLC [76]. The excellent peak

shapes indicate that it should be useful for LC-MS

analysis of bases in acidic media.

The authors thank the National Institute of Health for thefinancial support (Grant GM54585). The authors thank Mac-Mod analytical (Chadds Ford, PA) for the donation of theHiChrom silica and ACE C18 stationary phase, Agilent Tech-nologies (Wilmington, DE) for the donation of Zorbax silicaand the SB C18 phase, and Varian (Palo Alto, CA) for thedonation of Inertsil ODS 3 columns.

The authors have declared no conflict of interest.

Table 7. Comparison of slope, intercept, and r2 coefficients of k–k plots between different phases for aromatic compoundsa)

Solutes HC-C8 versus HC-T SB C18 versus HC-T SB C18 versus HC-C8 SB-Phenyl versus HC-T

Slope Intercept r2 Slope Intercept r2 Slope Intercept r2 Slope Intercept r2

Alkylbenzene 0.9270.01 �0.4570.01 0.999 0.7070.01 �0.4470.01 0.997 0.7670.02 0.0270.02 0.996 1.0770.04 �0.2970.02 0.988

p-Acceptor or p-donor 0.9770.02 �0.5170.03 0.995 0.9370.04 �0.5070.04 0.988 0.9570.05 0.0170.07 0.974 1.3470.04 �0.4270.03 0.993

PAH 0.8970.03 �0.3970.04 0.998 1.1770.06 �0.6770.07 0.995 1.3170.06 �0.3170.07 0.996 N/A N/A N/A

a) Chromatographic conditions are the same as in Fig. 10.



5 References

[1] Majors, R. E., LC-GC North Am. 2008, 26, 16–28.

[2] MacNair, J. E., Lewis, K. C., Jorgenson, J. W., Anal.Chem. 1997, 69, 983–989.

[3] Mazzeo, J. R., Neue, U. D., Kele, M., Plumb, R. S., Anal.Chem. 2005, 77, 460a–467a.

[4] Wu, N. J., Thompson, R., J. Liq. Chrom. Rel. Technol.2006, 29, 949–988.

[5] DeStefano, J. J., Langlois, T. J., Kirkland, J. J.,J. Chromatogr. Sci. 2008, 46, 254–260.

[6] Zhang, Y., Wang, X. L., Mukherjee, P., Petersson, P.,J. Chromatogr. A 2009, 1216, 4597–4605.

[7] Thompson, J. D., Carr, P. W., Anal. Chem. 2002, 74,4150–4159.

[8] Yan, B. W., Zhao, J. H., Brown, J. S., Blackwell, J., Carr,P. W., Anal. Chem. 2000, 72, 1253–1262.

[9] Heinisch, S., Rocca, J. L., J. Chromatogr. A 2009, 1216,642–658.

[10] Antia, F. D., Horvath, C., J. Chromatogr. 1988, 435, 1–15.

[11] Li, J. W., Carr, P. W., Anal. Chem. 1997, 69, 2550–2553.

[12] Chen, H., Horvath, C., Anal. Method Instrum. 1994, 1,213–222.

[13] Wilke, C. R., Chang, P., Am. Inst. Chem. Eng. J. 1955, 1,264–270.

[14] Yang, Y., J. Sep. Sci. 2007, 30, 1131–1140.

[15] Smith, R. M., J. Chromatogr. A 2008, 1184, 441–455.

[16] Hartonen, K., Riekkola, M. L., Trac. Trend Anal. Chem.2008, 27, 1–14.

[17] Mao, Y., Carr, P. W., Anal. Chem. 2000, 72, 110–118.

[18] Vanhoenacker, G., Sandra, P., J. Sep. Sci. 2006, 29,1822–1835.

[19] McCalley, D. V., J. Sep. Sci. 2003, 26, 187–200.

[20] McCalley, D. V., J. Chromatogr. A 2002, 965, 51–64.

[21] Mao, Y., Carr, P. W., Anal. Chem. 2000, 72, 2788–2796.

[22] Hetem, M. J. J., De Haan, J. W., Claessens, H. A.,Van de Ven, L. J. M., Cramers, C. A., Kinkel, J. N., Anal.Chem. 1990, 62, 2288–2296.

[23] Glajch, J. L., Kirkland, J. J., US Patent ] 4,705,725 1987.

[24] Kirkland, J. J., Glajch, J. L., Farlee, R. D., Anal. Chem.1989, 61, 2–11.

[25] Wirth, M. J., Fairbank, R. W. P., Fatunmbi, H. O., Science1997, 275, 44–47.

[26] Wirth, M. J., Fatunmbi, H. O., Anal. Chem. 1992, 64,2783–2786.

[27] Wirth, M. J., Fatunmbi, H. O., Anal. Chem. 1993, 65,822–826.

[28] Wirth, M. J., Fatunmbi, H. O., US Patent ] 5,599,6251997.

[29] Wirth, M. J., Fatunmbi, H. O., US Patent ] 5,716,7051998.

[30] Neue, U. D., in: Meyers, R. A. (Ed.), Encyclopedia ofAnalytical Chemistry, Wiley, New York 2000.

[31] Trammell, B. C., Ma, L., Luo, H., Jin, D., Hillmyer, M. A,Carr, P. W., Anal. Chem. 2002, 74, 4634–4639.

[32] Fatunmbi, H. O., Bruch, M. D., Wirth, M. J., Anal. Chem.1993, 65, 2048–2054.

[33] Pesek, J. J., Rashet, W., Chromatographia 1990, 30,442–446.

[34] Buchmeiser, M. R., J. Chromatogr. A 2001, 918,233–266.

[35] Henry, R. A., Am. Lab. 2002, 34, 18, 20–22, 24–25.

[36] Pesek, J. J., Matyska, M. T., J. Chromatogr. A 2002, 952,1–11.

[37] Kirkland, J. J., LC-GC 1996, 14, 486–500.

[38] Majors, R. E., LC-GC 1986, 4, 872, 874, 876.

[39] Nawrocki, J., Dunlap, C., McCormick, A., Carr, P. W.,J. Chromatogr. A 2004, 1028, 1–30.

[40] Snyder, L. R., Glajch, J. L., Kirkland, J. J., Practical HPLCMethod Development, Wiley-Interscience, New York1996.

[41] Unger, K. K., Packings and Stationary Phases in Chro-matographic Techniques, M. Dekker, New York 1989.

[42] Luo, H., Carr, P. W., Anal. Bioanal. Chem. 2008, 391,919–923.

[43] Luo, H., Ma, L. J., Zhang, Y., Carr, P. W., J. Chromatogr.A 2008, 1182, 41–55.

[44] Ma, L. J., Luo, H., Dai, J., Carr, P. W., J. Chromatogr. A2006, 1114, 21–28.

[45] Trammell, B. C., Ma, L. J., Luo, H., Hillmyer, M. A., Carr,P. W., J. Chromatogr. A 2004, 1060, 61–76.

[46] Davankov, V. A., Sychov, C. S., Ilyin, M. M., Sochilina, K.O., J. Chromatogr. A 2003, 987, 67–75.

[47] Penner, N. A., Nesterenko, P. N., Rybalko, M. A., J. Anal.Chem. 2001, 56, 934–939.

[48] Rosenberg, G. I., Shabaeva, A. S., Moryakov, V. S.,Musin, T. G., Tsyurupa, M. P., Davankov, V. A., ReactivePolymers, Ion Exchangers, Sorbents 1983, 1, 175–182.

[49] Ma, L., Acid Stable Hyper Crosslinked Stationary Phasesfor the Reversed Phase High Performance Liquid Chro-matographic Separation of Basic and BiologicalAnalytes. Ph.D. Thesis, University of MinnesotaUniversity of Minnesota, Minneapolis, 2005.

[50] Luo, H., A Silica-Based Hydrophobic Cation ExchangePhase for Water Soluble Pharmaceuticals and BioactiveAnalytes. Ph.D. Thesis, University of Minnesota 2006.

[51] Halasz, I., Martin, K., Ber. Bunsen-Ges. Phys. Chem.1975, 79, 731–732.

[52] Olah, G. A., Kobayashi, S., Tashiro, M., J. Am. Chem.Soc. 1972, 94, 7448–7461.

[53] Olah, G. A., Kuhn, S. J., J. Org. Chem. 1964, 29,2317–2320.

[54] Hanson, M., Eray, B., Unger, K., Neimark, A. V., Schmid,J., Albert, K., Bayer, E., Chromatographia 1993, 35,403–409.

[55] Hanson, M., Unger, K. K., Trac. Trend Anal. Chem. 1992,11, 368–373.

[56] Li, J., Carr, P. W., Anal. Chem. 1997, 69, 2193–2201.

[57] Li, J., Reeder, D. H., McCormick, A. V., Carr, P. W.,J. Chromatogr. A 1997, 791, 45–52.

[58] Dai, J., Carr, P. W., McCalley, D. V., J. Chromatogr. A2009, 1216, 2474–2482.



[59] McCalley, D. V., Anal. Chem. 2003, 75, 3404–3410.

[60] Dai, J., Carr, P. W., J. Chromatogr. A 2005, 1072,169–184.

[61] Dai, J., Mendonsa, S. D., Bowser, M. T., Lucy, C. A.,Carr, P. W., J. Chromatogr. A 2005, 1069, 225–234.

[62] Snyder, L. R., Dolan, J. W., Carr, P. W., J. Chromatogr. A2004, 1060, 77–116.

[63] Gilroy, J. J., Dolan, J. W., Snyder, L. R., J. Chromatogr.A 2003, 1000, 757–778.

[64] Gilroy, J. J., Dolan, J. W., Carr, P. W., Snyder, L. R.,J. Chromatogr. A 2004, 1026, 77–89.

[65] Wilson, N. S., Gilroy, J., Dolan, J. W., Snyder, L. R.,J. Chromatogr. A 2004, 1026, 91–100.

[66] Marchand, D. H., Croes, K., Dolan, J. W., Snyder, L. R.,J. Chromatogr. A 2005, 1062, 57–64.

[67] Marchand, D. H., Croes, K., Dolan, J. W., Snyder, L. R. ,Henry, R. A., Kallury, K. M. R., Waite, S., Carr, P. W.,J. Chromatogr. A 2005, 1062, 65–78.

[68] Wilson, N. S., Nelson, M. D., Dolan, J. W., Snyder, L. R.,Wolcott, R. G., Carr, P. W., J. Chromatogr. A 2002, 961,171–193.

[69] Wilson, N. S., Nelson, M. D., Dolan, J. W., Snyder, L. R.,Carr, P. W., J. Chromatogr. A 2002, 961, 195–215.

[70] Wilson, N. S., Dolan, J. W., Snyder, L. R., Carr, P. W.,Sander, L. C., J. Chromatogr. A 2002, 961, 217–236.

[71] Zhang, Y., Carr, P. W., J. Chromatogr. A 2011, 1218,763–777.

[72] Melander, W., Stoveken, J., Horvath, C., J. Chromatogr.1980, 199, 35–56.

[73] Hanai, T., Hubert, J., J. Chromatogr. 1984, 291, 81–89.

[74] Tanaka, N., Tokuda, Y., Iwaguchi, K., Araki, M.,J. Chromatogr. 1982, 239, 761–772.

[75] Yang, X., Ma, L., Carr, P. W., J. Chromatogr. A 2005,1079, 213–220.

[76] Stoll, D. R., Li, X., Wang, X., Carr, P. W., Porter, S. E. G.,Rutan, S. C., J. Chromatogr. A 2007, 1168, 3–43.



Documents

Optimization of the synthesis of a hyper-crosslinked stationary phases: A new generation of highly efficient, acid-stable hyper-crosslinked materials for HPLC