Introduction to Modern Liquid Chromatography, Third Edition part 42 pot

Consequently, for a solute molecule to be retained in NPC, one or more previously adsorbed solvent molecules must be displaced from leave the silica surface in order to make room for the

hydrocarbons—according to the number of double bonds in the molecule, but with little effect of differences in alkyl substitution or solute molecular weight Similarly lipid samples can be resolved into mono-, di-, and tri-glycerides (as well as other compound classes)

Isomeric solutes are usually much better separated by NPC than by RPC, as

seen in Figure 8.2b (C1or methyl-substituted anilines), Figure 8.2c (C2or

dimethyl-plus ethyl-substituted anilines), and Figure 8.2d (C3 or trimethyl-, methylethyl-,

and n-propyl-substituted anilines) The RPC retention of isomers (e.g., C1 anilines)

is generally similar (with marginal separation), while the reverse is true for NPC

(more varied values of k for different isomers, and better separation) The greater isomer-selectivity of NPC versus RPC is also shown by the range in values of k for each set of isomers; that is, the ratio of k-values for the most retained and least retained isomers (‘‘range in k,’’ Fig 8.2e) The spread in k values for a group of

isomers (C1, C2, or C3) ranges from 1.1- to 1.2-fold for RPC, versus 2.0- to 3.4-fold for NPC, which is many times larger for NPC

When two isomeric compounds prove difficult to separate by RPC (Section 6.3.5), NPC will usually prove more effective Thus, for preparative separations (Chapter 15), where the largest possible values ofα are desirable, NPC with a silica

column is strongly recommended for the separation of achiral isomers (for chiral isomers, see Chapter 14) For a review of the RPC separation of isomers, see Section 6.3.5; for a further discussion of isomer separation by NPC, see Section 8.3.5

If in doubt as to whether a separation is based on NPC or RPC retention,

a simple test is to vary the polarity of the mobile phase If retention increases with increased mobile-phase polarity, RPC is involved; if retention decreases with increased mobile-phase polarity, NPC can be assumed This test holds whether the stationary phase is unbonded or bonded silica, and whether the mobile phase is aqueous or nonaqueous (note that water is the most polar of all solvents).

8.2.1 Theory

Retention in NPC is best described by a displacement process, based on the fact that

the silica surface is covered by a monolayer of solvent molecules that are adsorbed from the mobile phase [1, 5, 6] Consequently, for a solute molecule to be retained

in NPC, one or more previously adsorbed solvent molecules must be displaced from (leave) the silica surface in order to make room for the adsorbing solute

Displacement in NPC is illustrated in Figure 8.3a, b for a relatively nonpolar solute

(chlorobenzene) and a weaker, less-polar mobile-phase solvent methylene chloride

When a molecule of chlorobenzene moves from the mobile phase in Figure 8.3a to the stationary phase in Figure 8.3b, one or more pre-adsorbed solvent molecules

CH2Cl2must be displaced from the stationary phase and return to the mobile phase

In this example the adsorbed solute molecule is assumed to lie flat on the surface

of the silica, and to occupy an area that is indicated in Figure 8.3b by the dotted

rectangle that surrounds the molecule of retained chlorobenzene By reference to

Figure 8.3a, it is seen that this same area was originally occupied by (approximately)

two retained molecules of CH2Cl2 Consequently the resulting retention equilibrium can be written as

where n is the number of solvent molecules M displaced by a retained solute molecule

Y (n = 2 in the example of Fig 8.3a,b) Y (m) and Y (s) refer to a molecule of solute

Y in the mobile and stationary phases, respectively, while M (m) and M (s) refer to a

molecule of solvent M in the mobile and stationary phases, respectively The quantity

n in Equation (8.1) is thus the ratio of molecular areas for the solute with relation

to the mobile phase

Retention differs for a more-polar mobile-phase solvent such as

tetrahydrofu-ran (THF) and a more polar solute such as phenol (Fig.8.3c,d) Here the interaction

of solvent and solute molecules with surface silanols will be stronger, as indicated

by the arrows that connect the two interacting species—in contrast to the weaker

HO

H O

Mobile phase

Stationary phase

Mobile phase

Stationary phase

H

O

H O

H O

H O

Cl

H O

H O

H O

H O

H O

Cl

CH2Cl2 CH2Cl2

CH2Cl2 CH2Cl2

CH2Cl2

CH2Cl2

CH2Cl2

CH2Cl2

CH2Cl2

CH2Cl2

CH2Cl2

CH2Cl2

CH2Cl2

CH2Cl2

CH2Cl2

CH2Cl2 CH2Cl2

CH2Cl2

HO

Figure8.3 Hypothetical examples of solute retention on silica for chlorobenzene (a,b non-localized) and phenol (c,d non-localized) Mobile phase in (a,b) is a less-polar solvent (CH2Cl2);

mobile phase in (c,d) is a more-polar solvent (tetrahydrofuran, THF).

and less specific interactions shown in Figure 8.3a,b As a result there is a ratio 1 :

1 interaction of a surface silanol with a polar group in a molecule of either solute

or mobile phase—called localized adsorption Under these conditions adsorbed

molecules can assume a vertical rather than flat configuration, as illustrated by the

retention of phenol in Figure 8.3d In the example of Figure 8.3a,b, the entire solute

molecule (chlorobenzene) is attracted more strongly to the silica surface than are molecules of the less-polar solvent CH2Cl2 In Figure 8.3c,d, the very polar –OH

group of the phenol solute interacts strongly with a silanol (–OH) group on the silica surface, while the less-polar phenyl group is attracted less strongly than are molecules of the strong (very polar) solvent THF As a result the phenyl group cannot compete with molecules of THF for a place on the silica surface; the phenyl group therefore dangles out into the mobile phase, tethered to the surface by the interaction of a silanol with solute-hydroxyl groups For solutes with some number

n of more strongly retained (very polar) substituent groups (e.g., –OH, –NH2),

Equation (8.1) will still apply, with n referring now to the number of polar groups

in the solute molecule that can simultaneously interact with silica silanols However,

for less-polar molecules of the mobile phase as in Figure 8.3a,b, n refers to the ratio

of areas required by molecules of the solute and mobile-phase, respectively

The competition of solute and solvent molecules for a place on the silica surface will be affected by the interactions of each molecule with the mobile and stationary phases Because polar interactions predominate in NPC, polar molecules of solute

or solvent will interact much more strongly with the more-polar silica surface than with the less-polar mobile phase As a first approximation we can ignore interactions between solute and solvent molecules and consider only interactions of the solute and solvent with the stationary phase This allows the derivation of a simple equation

for k as a function of the concentration of the B-solvent in a binary mobile phase

A–B [1, 6]:

where k A is the value of k for a nonpolar A-solvent (for which ε is zero), A s refers

to the molecular area of the solute molecule, and the mobile-phase solvent strength

ε can be calculated as a function of (1) the solvent strengths εA andεBof the pure A- and B-solvents, respectively, (2) the molecular area of the B-solvent, and (3) the concentration (%B) of the B-solvent in the mobile phase (see Eq 8.5 in Section 8.2 following)

The remainder of this section provides a quantitative, somewhat detailed, discussion of retention as a function of the solute and mobile phase The reader may prefer to skip to Section 8.2.2 (which provides a practical summary of the discussion below) and return to this section as appropriate.

Equation (8.2) applies approximately for all solvents and solutes For more

polar solutes and B-solvents, as in Figure 8.3c,d, polar solute groups can attach to individual silanols So Equation (8.1) now applies with n equal to the number of

polar groups in the solute molecule (rather than the area of the solute molecule) If

%B is large enough, and the B-solvent is strong enough, the silica surface will be covered almost entirely by the B-solvent (to the exclusion of the A-solvent); under

these conditions the concentration of adsorbed B-solvent will not vary much when

%B is changed For this case the Soczewinski equation can be derived from Equation

(8.1) [7, 8] Thus the equilibrium for Equation (8.1) can be written as

K eq= X Y,s X M,m n

X Y,m X M,s n

(8.3)

where X Y,s and X Y,m refer to the concentrations (mole-fractions) of solute Y in the stationary (s) and mobile (m) phases, respectively Similarly X M,s and X M,mare the

mole-fractions of the B-solvent M in the stationary (s) and mobile (m) phases The retention factor k can be written as

k = ψ



X Y,s

X Y,m



(8.3a)

where ψ is the phase ratio: the volume-ratio of stationary phase to mobile phase

within the column If the silica surface is covered almost entirely by the B-solvent,

X M,s ≈ 1, Equations (8.3) and (8.3a) then yield

log k = log k B − n log X B (Soczewinski equation) (8.4)

where n is the number of B-solvent molecules displaced by the solute (approximately equal to the number of polar substituent groups in the solute molecule); k B refers

to the value of k for pure B-solvent as the mobile phase (100% B), and X B is the mole-fraction of the B-solvent in the mobile phase Equation (8.4) can be regarded

as a special case of Equation (8.2), whenever X M,s≈ 1 (i.e., for larger concentrations

of the B-solvent)

Equation (8.4) is more conveniently (and approximately) expressed as

whereφ is the volume-fraction of the (polar) B-solvent in the mobile phase (= 0.01×

%B) An example of the application of Equation (8.4a) is shown in Figure 8.4 for two different (polar) solutes, and a mobile phase composed of hexane (A) plus tetrahydrofuran (B), with %B varying between 8 and 27% In each case an

approximately linear plot of log k against φ is observed Values of n from Equation

(8.4a) for the two plots of Figure 8.4 are each equal to about 1.4, which is somewhat

greater than the expected value of n = 1.0 (as there is just one polar –OH group in

each solute molecule); this is a consequence of the approximate nature of Equation

(8.4a) Equations (8.4) and (8.4a) are reliable for more-polar B-solvents ( ε0

B> 0.4)

and higher concentrations of the B-solvent (>10% B)—conditions that assure that

the surface of the silica will be covered almost entirely by molecules of the B-solvent

(instead of the A-solvent) Typically values of n in Equation (8.4a) fall between one

and two for representative small-molecule solutes (molecular weights<500 Da), so

that on average a change in %B by a factor of two (e.g., a change in mobile phase

from 100% B to 50% B) will change values of k by a factor of 2 to 4.

Similar plots of log k against log %B as in Figure 8.4 are shown in Figure 8.5a

for two less-polar solute molecules, using the weaker B-solvent chloroform (ε0

B= 0.26 with lower values of %B Because these conditions fail to meet the above

1.2

1.0

0.8

0.6

0.4

0.2

log k

benzyl alcohol 3-phenyl-1-propanol

y = x

–1.0

log φ

10 15 20 25 30 %-THF

Figure8.4 Dependence of log k on log %B for solutions of a polar (localizing) B-solvent.

Sample: benzyl alcohol and 3-phenyl-1-propanol Conditions: silica column; tetrahydrofuran (THF)-hexane mobile phases; 25◦C Data from [11]

requirements for the application of Equation (8.4a), the resulting plots are curved rather than straight (as required by the theory upon which Eqn 8.4a is based), and

the best-fit values of n are

example produces a change in k by much less than 2- to 4-fold However, the use of the more accurate Equation (8.2) for these data (Fig 8.5b) yields a near-linear fit of values of log k versus ε0as expected

8.2.2 Solvent Strength as a Function of the B-Solvent and %B

Solvent strength in NPC depends on the polarity of the solvent; more polar solvents

are stronger, resulting in smaller values of k for a sample The strength of a

pure solvent for unbonded silica as column packing can be expressed by the solvent-strength parameterε of Equation (8.2); for pure solvents, ε will be referred

to asε0 As the value ofε increases, the solvent becomes stronger, and solute k-values

decrease Values ofε0for some commonly used NPC solvents are listed in Table 8.1; for additional values ofε0for other pure B-solvents, see Appendix I The value ofε

for a mixture of A- and B-solvents is given by [1, 6]

ε = ε O

A +log[N B10n B(ε0

B −ε0

A)+ 1 − N B]

Here,ε0

Aandε0

Brefer to theε0values of pure solvents A and B, N Bis the mole-fraction

of solvent B in the mobile phase, and n Brepresents the relative size (area) of solvent

B (relative to a value of n B= 6 for benzene as B-solvent) Equation (8.5) assumes a fully active (non–water-deactivated) adsorbent such as silica

When a weaker A-solvent (e.g., hexane) is mixed with a stronger B-solvent (e.g., CH2Cl2), the resulting mobile phase will have an intermediate strength and

value of ε that is given by Equation (8.5) Figure 8.6 is a solvent nomograph for

NPC with a silica column (similar to that for RPC in Fig 6.11), which compares the strengths of different mobile-phase mixtures in terms of their values of ε (see

values at top of Fig 8.6, calculated from Eq 8.5) A change inε by 0.05 units will change values of k by roughly a factor of 2 For example, a mobile phase of 50%

methylene chloride-hexane hasε = 0.24 (dotted vertical line of Fig 8.6) If a change

nitrobenzene phenanthrene

(a)

(b)

2.5

2.0

1.5

1.0

0.5

0.0 log k

0.00 0.02 0.04 0.06 0.08

ε 0

%B = 0.5% 1% 2% 3% 4% 5%

–2.5 –2.0 –1.5

log %B

2.0

1.5

1.0

0.5

log k

y = x

nitrobenzene phenanthrene

%B = 0.5% 1% 2% 5%

Figure8.5 Dependence of log k on log %B for dilute solutions of a lesspolar (nonlocalizing)

B-solvent Sample: nitrobenzene and phenanthrene Conditions: silica column; CHCl3-hexane mobile phases; 25◦C (a) unsatisfactory correlations of log k with log φ (Eq 8.4); (b) improved correlations of log k with ε (Eq 8.2) Data from [11].

Table 8.1

Solvent Strength (ε 0 ), Molecular Area (n B), and UV Absorbance as a Function of Wavelength for Normal-Phase Solvents

200 210 220 230 240 250 260 Acetonitrilea,b 0.52 3.1 0.03 0.02 0.00 0.00 0.00 0.00 0.00 Ethoxynonafluorobutane 0.01 h Presumed similar to ethyl ether

Chloroformc 0.26 5.0 >1.0 >1.0 >1.0 >1.0 >1.0 0.25 0.08 Ethyl acetateb 0.48 5.2 >1.0 >1.0 >1.0 >1.0 >1.0 >1.0 0.10 Ethyl etherd,e 0.43 4.4 >1.0 >1.0 0.46 0.27 0.18 0.10 0.05 Hexane, heptane 0.00 h 0.35 0.20 0.07 0.03 0.02 0.01 0.00 Methanola 0.70 3.7 >1.0 0.53 0.23 0.10 0.04 0.02 0.01 Methylene chloridec 0.30 4.1 >1.0 >1.0 >1.0 >1.0 0.09 0.00 0.00

Methyl-t-butyl ether d 0.48 4.1 >1.0 0.70 0.54 0.45 0.28 0.10 0.05

n-Propanol f 0.60 4.4 >1.0 0.65 0.35 0.15 0.07 0.03 0.01

i- Propanol f 0.60 4.4 >1.0 0.44 0.20 0.11 0.05 0.03 0.02 Tetrahydrofurand,e 0.53 5.0 >1.0 >1.0 0.60 0.40 0.21 0.18 0.09 Sources: Data from [6, 13].

aimmiscible with hexane.

bNonbasic localizing.

cNonlocalizing.

dBasic localizing.

eEasily oxidized and therefore less useful in practice.

fVery strong (localizing), proton-donor solvent; classification as ‘‘basic’’ or ‘‘nonbasic’’ may not be relevant.

gValues from [6], derived as described in [1].

h Values of n Bfor A-solvents are not required in Equation (8.5).

in values of k by 2-fold is needed, a mobile phase of 30% methylene chloride-hexane

(ε = 0.19) will increase k by about 2-fold, while 95% methylene chloride-hexane

(ε = 0.29) will decrease k by about 2-fold.

Mobile phases with the same values ofε in Figure 8.6 should provide similar values of k for a given sample; for example, suppose that 50% CH2Cl2-hexane (ε = 0.24) has been found to provide 1 ≤ k ≤ 10 for given sample and a silica

column From Figure 8.6 we can predict that 3.5% MTBE-hexane, 6% THF-hexane,

or 4% ethyl acetate-hexane will each have a similar solvent strength (ε = 0.24); each

of these four mobile phases should therefore provide a retention range of about

1≤ k ≤ 10 for the same sample Figure 8.6 is thus useful for selecting a different

B-solvent in order to change selectivity (Section 8.3.2), while maintaining the same

solvent strength and similar values of k Because of potentially large changes in

solvent-type selectivity for NPC (Section 8.3.2), the NPC solvent-nomograph of Figure 8.6 is somewhat more approximate than the corresponding nomograph of Figure 6.11 for RPC

An example of NPC separation as a function of %B is shown in Figure 8.7 for

an arbitrary mixture of organic compounds, using a silica column with mixtures of

2 4 6 10 20 50 100

1 2 4 6 10 50 100

1 2 4 6 10 50 100

1 2 4 6 10 50 100

0.5 1 2 3 5 10 20 50 100

e

%v CH2Cl2-hexane

%v MTBE-hexane

%v THF-hexane

%v Ethyl acetate-hexane

%v propanol-hexane

10 20 50 100

10

10 20

%v MTBE-CH2Cl2

%v THF-CH2Cl2

%v Ethyl acetate-CH2Cl2

%v

propanol-CH2Cl2

100 50

Adapted from [12]

ethyl acetate (B) and cyclohexane (A) as mobile phase An increase in %B by a factor

of two (40% B in Fig 8.7b vs 20% B in Fig 8.7a) leads to a decrease in values

of k by a factor of 2 to 3 in this example Changes in relative retention with %B

are also seen in Figure 8.7, as discussed in following Section 8.3.1 (solvent-strength selectivity) Because of these changes in relative retention with %B, an intermediate

mobile-phase composition (Fig 8.7c) provides the best resolution for this sample.

8.2.3 Use of TLC Data for Predicting NPC Retention

Thin-layer chromatography (TLC) can be a useful complement to the use of HPLC

with a silica column Corresponding separations by TLC and column

chromatogra-phy (involving the same sample, mobile phase, temperature, and especially the same

silica as stationary phase) should yield similar values of k for each compound in the

sample As illustrated in Figure 8.8, the positions of separated bands (spots) on a

TLC plate can be used to estimate values of k for a corresponding column separa-tion The R Fvalue of a solute in TLC is defined as its fractional migration from the original sample spot (point at which the sample is applied) toward the solvent front (end of solvent migration during TLC) Thus a sample band that migrates half as

far as the solvent front (e.g., spot D in Fig 8.8) would have R F = 0.5; the R Fvalues

of solutes A through D in Figure 8.8 are 0.15, 0.25, 0.38, and 0.50, respectively

Time (min)

Time (min)

Time (min)

20% B, e = 0.31

3 ≤ k ≤ 5, Rs = 0.1

40% B, e = 0.38

1 ≤ k ≤ 3, Rs = 1.3

25% B, e = 0.33

2 ≤ k ≤ 4, Rs = 1.6

1 2

3 + 4 5

6

1

4 2

3 5

6

1

4 2

6

(a)

(b)

(c)

Figure8.7 Solvent-strength selectivity in normal-phase chromatography Sample: 1, 2-aminonaphthalene; 2, 2,6-dimethylquinoline; 3, 2,4-dimethylquinoline; 4, 4-nitrophenol;

5, quinoline; 6, isoquinoline Conditions: 150 × 4.6-mm silica column (5-μm particles);

ethy-lacetate (B)-cyclohexane (A) mixtures as mobile phase; ambient temperature; 2.0 mL/min Peaks 1 and 4 are shaded to emphasize their change in relative retention as %B is varied Chro-matograms recreated from data of [14]

Corresponding values of k can be obtained from the relationship

k= 1− R F

In the example of Figure 8.8, the R F values of the spots vary from 0.15 to 0.50, corresponding to 1≤ k ≤ 5.7 Similar k-values are expected for a corresponding

HPLC separation (use of the same mobile phase with a silica column), hence providing an acceptable retention range (1≤ k ≤ 10) for this sample.

TLC is sometimes used prior to NPC separation with a column, as a convenient way of anticipating possible problems and/or exploring different mobile phases

Sample components (bands) that do not move (R F ≈ 0.0) will be visible in TLC,

whereas any strongly retained solutes in column NPC will remain in the column and hence be undetected and perhaps missed A related advantage of a preliminary TLC separation is that samples that might otherwise contaminate a column—and therefore require pretreatment for column chromatography (Chapter 16)—can usually be applied directly in TLC because a TLC plate is used only once and then discarded By trial-and-error changes in %B (e.g., for methylene chloride-hexane mixtures) a value of %B that can provide values of 1≤ k ≤ 10 for HPLC separation

solvent front

original sample spot

1.0 0.0 0.9 0.1 0.8 0.2 0.7 0.3 0.6 0.7 0.5 1.0 0.4 1.5 0.3 2.3 0.2 4.0 0.1 9.0

D C B A

Figure8.8 Use of thin-layer chromatography (TLC) for selecting a mobile phase to be used for HPLC with a silica column (hypothetical sample)

can be obtained in a short time by means of TLC—provided that isocratic elution is possible If 1≤ k ≤ 10 is the goal for all peaks in isocratic column chromatography,

R Fvalues in a corresponding TLC separation should fall between 0.1 and 0.5 TLC experiments may show that the sample cannot be separated isocratically by column NPC (no value of %B results in sample bands with 0.1 ≤ R F ≤ 0.5) Changes

in relative retention as a result of solvent-strength selectivity (Section 8.3.1) or solvent-type selectivity (Section 8.3.2) can also be explored by TLC—often more conveniently than by HPLC (see the example of Section 8.4.3)

While the use of TLC as described above is reasonably reliable, it should be

noted that solvent demixing (Section 8.5.2) can lead to misleading results in some

cases When dilute solutions (<10% B) of strong B-solvents (ε0> 0.4) are used

as the mobile phase in TLC, the B-solvent can be strongly retained by the silica, leading to its removal from the mobile phase during separation As a result the concentration of B-solvent in the mobile phase will be lowered, the mobile-phase

strength will be reduced, and observed values of R Fwill be too small; thus values of

k that are estimated from R Fvalues may be too large when solvent demixing occurs This problem can be minimized by first equilibrating the TLC plate with the mobile phase The plate should be placed in the developing chamber for 15 minutes—but without allowing the mobile phase to touch the plate and initiate sample migration [9] This way the vapor above the solvent will equilibrate the plate and minimize any solvent demixing, after which the plate is lowered to contact the mobile phase and begin the separation

TLC can be especially useful for monitoring preparative separations Thus several fractions from a separation by column chromatography can be conveniently analyzed at one time, by spotting a single (suitably large) plate with multiple samples—each in a different lane of the plate Fractions that are seen to contain the desired product, uncontaminated by adjacent impurity peaks, can be combined for further processing