Correlations of surface measurements with the chromatographic performance of chemically modified glass capillary columns

Journal of Chronmratrnphr, 94 (1974) 35-51

33 Elsevier Scientific Publishing Company, Amsterdam - Printed in The Netherlands CHROM_ 7432

CORRELATIONS OF SURFACE MEASUREMENTS WITH THE CHROMATOGRAPHIC PERFORMANCE OF CHEMICALLY MODIFIED GLASS CAPILLARY COLUMNS K, D. BARTLE' and M . NOVOTNY Deparuucnr of Clzr,nivirr, Indiana Unirersitr . Bluonrintto.r, 1nd_ 47401 (U-SA_)

(Received February

26th . 1974)

SUMMARY

Wetting phenomena for dillerent liquids were studied on various stir/aces -- tailored" by the permanent bonding of selective monolavers of silanes to the inner wall of glass capillaries . Surfaces were characterized by graphing the cosine of advancing contact angle on differently modified _glass against liquid surface tension_ The thicknesses of coated films of stationary phases, measured chromatographically . were found to he in good a greement with thOSe calculated from the Fairbrother-Stuhbs equation . However, the presence of the monolayers si gnificantly alters the surface characteristics of the glas,_ and the chromatographic performance of capillary columns the, reflects the film homogeneity, which was found to be a sensitive function of the chemical similarities between the surface structure and the phase- The mechanism of compatibility proposed earlier for the wettability of glass capillary walls is discussed in terms of both the surface phenomena described and a molecular model of the modified surface . INTRODUCTION Surface wetting phenomena are among the most important subjects of study in chromatography_ To achieve high separation efficiencies with gas chromatographic columns, a uniform and homogeneous stationary-phase layer on the supporting solid must be obtained . Wetting phenomena are particularly critical for glass capillary columns, and these problems have been discussed in a review article by Novotnv and Zlatkis' . In the same way, improved vv-ettability would also be beneficial for ordinary chromatographic materials (diatomaceous earths_ silica gels, glass beads, etc .) in terms of better efficiency and deactivation . According to the theoretical model of Giddings, a solid support is first coated with a thin layer of the stationary phase, if this process is allowed by the surface wettability, before any further accumulation of the liquid into the pores takes place . On leave from Department of Physical Chemistry, University of

Leeds . Leeds,

Great Britain .



36

K . D . BARTLE, -Ni- NOVOTNT

While wettability problems with glass capillary columns can be solved to a certain degree by the general approach of increasing the roughness of the glass surface'` with corresponding decrease of liquid contact angles on thermodynamic grounds, the molecular basis of wetting cannot be overlooked- Based on our previous experimental et-idence'- 5 a _general surface model for modified Mass has been suggested' in which a permanently bonded monolayer should be attached to the silica framework, while the "exposed" portions of the attached molecules should be available to comply with the polarity or other property of a given stationary phase . While hydrophobic monolayers prepared by simple trialkylsilylation can be used with non-polar phases', only limited success was achieved with more polar substrates because of the unavailability of suitable silanes_ Recently . however, a universal experimental realization or this above model has been described", in which dintethvlchloro[4(4-chloronmethylphenyl)butyl]slatte (I) is reacted with the surface hydroxyl groups : the benzylic chlorine atom is then replaced, in subsequent reactions, by suitable groups compatible with the stationary phase- Only qualitative proof of the validity of this approach was presented" as reflected in the improved efficiency of glass capillary columns coated with several polar stationary phases_ Consequently, the main objective of the present work has been to verify the compatibility concept' in a more quantitative fashion . If it indeed implies that chemi cal similarities between the wall of capillary columns and the stationary phase are conducive to an efficient spreading of the latter, such phenomena should be reflected in the surface characteristics_ Therefore_ the contact angles of various liquids on differently -- tailored" surfaces were measured together with respective surface tensions where necessary, and these were then correlated with chromatographic measurements . It is found that both chromatographic efficiencies and measurements of the average film thickness' in _glass capillary columns strongly support the validity of the compatibility concept. A discussion of the molecular basis of wetting is presented . and a model of the surface is used to explore relevant phenomena at the molecular level . EXPERIMENTAL

Soda-,-lass capillaries . 14-25 m -- 0.2 mm I .D ., were filled with a solution of compound I in dry benzene and the surface groups were allowed to react before washing with benzene and acetone. Further reactions or thee surface-bonded -CH .C1 groups" were then carried out bye (a) hydrolysis with potassium carbonate in tetrahydrofuran-water before successive washings with a series of solvents of decreasing polarity: (b) treatment as in (a) followed by esterification of the hydroxyl groups with benzoyl chloride and subsequent washing as in (a) : (c) reaction with potassium cyanide before washing as in (a) : and (d) reaction with a weak solution of ammonium hydroxide before washing as in (a) . ftw-+-Other capillaries were filled with benzene solutions of : (c) diethoxyntetltyl(l0-ethoxycarbonyldecyl)silane (l1) : (f) 3 .3,4,4,5 .5.5-Iteptafluoropentyldimethylchlorosilane (Ill) : (g) N-(2-aminoethyl)-3-aminopropyltrimethoxysilane (IV) : and (h) a 1 :1 mixture of I,3-di(3-heptafluoroisopropoxypropyl)tetramethyldisilazane (V) and (3-heptafluoroisopropoxypropyl)dimethylchlorosilane (VI) ; after reaction, the capillaries were washed with benzene and acetone . In a capillary, treated (i) as in (h), hydrolysis was carried out by means of an HCI solution in methanol (in order to yield



37

CHEMICALLY MODIFIED GLASS CAPILLARY COLUMNS

CH, CI -C

U

I

-CHp-CHp-CHp-CH2-si -CI I CH3 OC2Hg I

C2HS000-(CH 2 h o Si-CH3 f

OC2Hy CH3 1

CF3 CF2-CFZ CH2 CHZ Si-CI I

CH3 OCH3 I

!!H2-c"2-CHZ NH-CHZ CHZ CH 2 5~-OCH3 I

OCH3 CH,

CF3\ y

CF-O C

CHZ CH2-Si-

CF3

I

CH3 3 CH I

CF3 \

CF-O-CH2 -CH2-CH2 - i i-CI CF3

CH3

dillerent types of surface hydroxyl groups) . and followed by washing with methanol and acetone . All capillaries were finally dried by passing dry nitrogen gas at 50 . Procedures described elsewhere were used to prepare (j) acetone-washed" : (k) chromic acid-washed' : and (I) silanized' ?0 m : 01 mm I .D . capillaries .

Plt_rsieczi measurements The surface tensions of stationary phases and stationary phase solutions were measured by the capillary-rise method using capillaries (k) above : calibrating liquids were analytical grade toluene and chloroform . Contact angles were measured by the capillary-rise method on segments of the columnss later used in coating experiments . The meniscus height was measured with a cathetometer with the tangent to the centre of the meniscus normal to the perpendicular in both surface-tension and contactanele measurements . Densities and viscosities were measured with a pyknometer and a Ubbelohde suspended-level viscometer, respectively' .

Coating of capillary colunzzzs The dynamic method was used, as described previously' . with the coils hanging vertically . The required plug velocity for a given film thickness was calculated from the Fairbrother-Stubbs equation" using values of the physical properties listed in Table I .

Gas chromatographic measurements Capacity-ratio and column efficiency measurements were made on a Varian Model 1400 gas chromatograph with modified injector and detector manifold . Injec-



38

K . D_ BARTLE, M_ NOVOTNY

TABLE I PROPERTIES OF lo-' vJv STATIONARY-PHASE SOLUTIONS AT 25 For details of the surface treatments, see Experimental_ Statinneu_rphase

Solrext

Densitr Surface Ikg;'nr') tension (to-2 nice)

Ucon oil 50-HB-2000 2 Y-Oxydipropionitrile

Toluene Toluene Dinonyl phthalate Toluene Triethanolamine Methanol Silicone oil OV-21O Acetone 1,2,3-Tris(cyanoethoxy)propane Methanol

883 890

873 821 8,24 822

2.80 2.79 2-SO 2_32 212 2.25

Viscositr (g.'nrsee)

Chenuraltreaarrent of fhegdasN surface

where zero contact angles ore obtained 1319 070 0 .66 0 .54

0 .80 0 .75

(a), (c), (i)

(a), (c) (b), (e) (f) (c)

lions of r pl of benzene headspace were made with nitrogen carrier gas- a split ratio of 1 : 2200 . and injector and detector temperatures of 130 =_ Specific retention volumes, I"' , for 2 .2'-oxydipropionitrile . Ucon oil 50-HB-2000, and OV-210 silicone oil were determined with previously described equipment'_ and film thicknesses . d,, . were derived by combining values of the capacity ratio . k . and 1 y measured at the same temperature. RESULTS AND DISCUSSION Contact angles Characterization of the chemically modified capillary surfaces was attempted by plotting cos 0 (where 0 is advancing contact angle- Tables I and Il) against - • (surface tension) for various stationary phases, stationary-phase solutions, and miscellaneous liquids '° •" _ Untreated (i_e- merely acetone washed) glass shows effects expected of a heterogeneous surface (Fig . 1) . A slightly greater degree of wetting (C-) -= 0, cos ) = I ) is evident with this glass as compared to the literature tattle 13 of the critical surface tension . yr (the maximum value of ,,, at which H - 0 (ref. 10)), obtained for Pyrex glass, but there is wide scatter in this graph . Autophobic eftectsi 3 -'= apparently operate for the esters di(2-ethylhexyl) sebacate and dinonyl phthalate . On the other hand, surfaces treated by silanization with trimethvlchlorosilane_ compound Ill : a mixture of V and VI : V and V I followed by hydrolysis with HCI : I treated first to yield hydroxy structures and then followed by reaction with benzoyl chloride ; and I followed by reaction with ammonia, all gave more regular cos (9 rs . -behaviour (Figs . 2-6, and Tables II and III) . The autophobic behaviour of esters is generally suppressed, and straight lines with good correlation coefficients (Table Ill) may be drawn through the points for which y (liquid) is greater than y c (surface)_ The intercept on the cos 0 = I axis gives the value of yc (Table I11) : for our surfaces covered With surface bonded -Si(CH,) 3 , -Si(CH,),CH,CH,CF_CF,CF, and -Si(CH ;),CH,CH,CH_CH_ C6 H jCH,000-C6 H ; groups, these values are very close to yr for corresponding monolayers and crystal structures reported in the literature (Table IlI)so that a fairly complete coverage of the surface is suggested . The gradients of the rectilinear plots shown in Figs . 2-6 are all similar .



CHEMICALLY MODIFIED GLASS CAPILLARY COLUMNS

39

TABLE It CONTACT ANGLES ON CHEMICALLY TREATED SODA-GLASS SURFACES AT 25 . For details of the surface treatments, see ExperimentalStationary phase or standard liquid

Silicone oil SF-96 Ucon oil 50-H82000 r-Oxydipropionitrile Di(2-ethylhexvl) sebacate Dinonylphthalate Triethanolamine Silicone oil OV-210 I? 3-Tris(cyanoethoxv)propanc Formamide Water Glycerol Toluene Methanol

Contact angle (_ 10") ear rarioaslr treated soda glass ' (a) (b) (e) (d) (e) (f) (g) (lt) (i) (f) (k) (I)

Density Surface (kg/n') tension (10 - ' N,%at)

954

2.06

12

0

19 23 21

27

20 25

1051

3 .57

1045

4 .81

20 28 27 59

912 964' . 1119 1190

332--2 .88 4.90 2 .19

22 20 35 27 - -- - 31 17 15 24 16 23 39 0 43 0 30 20 26 36 50 66 64 38 73 56 21 10 0 0 16 0 38 0 28 39 0

22 15 0 0

36 35 61 0

1114 1129" 997" 1258" 863" 787"

5.19= 79" 7_20 6 .29 -2 .74" 2 .22"

37 41 30 50 78

30 ---0 -

67 59 68 71 -

0 0 - 0 27 59 46 33

31 37 41 63 33 . 45 0 0 -- -

27 45 41 0

61 81 68 0 -

60 65

16

0

0 0

52 25

0

0

59 73

-

--

42

44

74 65 50 78 . 85 62 77 85 59 0 40 0 -- -

77 67 92 56 88 67 -10 0 0 --

0 33 14 0 -

38

0 72

Value from ref- 3 3_ Values front refs . 34 and 35. "' Value from ref. 36 . . 37). s Based on (-) -- 30` on chromic acid-washed glass (ref

1 7 10r

12

2

9 .11

6 F

5

4

06 '1

o 0

10 6

1

2

3

4

5

6

;p (10 -2 Nlm) Fig . l . Graph of cos H against ; for acetone-washed glass (treatment (j))- 1 = Silicone oil SF-96 2 - Ucon oil 50-HB-2000 : 3 = 22'-oxydipropionitrile ; 4 -= di(2-ethvlhexyl) sebacate : 5 = dinonyl phthalate : 6 = triethanolamine ; 7 -= silicone oil OV-210 ; ti -1 12,3-tris(evanoethoxy)propane_ 9 -__

formamide : 10 = water; I I = glycerol : 12 = toluene : 13 = methanol- 14 = 10% v!v Ucon oil 50HB-2000 in toluene . 15 = 10% vlv 2 .2'-oxvdipropionitrile in toluene : 16 = 10°;; vjv dinonvl 2 phthalate in toluene ; 17 = 10 % :v/v silicone oil OV-210 in acetone : i8 = 10". ; v v-1,2,3-tris (cvanoethoxy) propane in methanol ; 19 = t0°;; viv silicone oil in toluene .



40

K . D . BARTLE, M . NOVOTNY

.9 10

a

5

7(10_ 2 Nlm) Fig . 2. Graph of cos (d against ; • for silanized glass (treatment (U) . For explanation of the numbers see legend to FU_ I-

y (10"Nlm) Fig . 3_ Graph of cos H against- , for glass with bonded -Si(CH,),CH :CH :CF:CF,CF, groups (treatment (f)). For explanation of the numbers, see legend to Fig . I . 13

125

021 4 6 7 .10 j(1O-2 NIm) Fig . 4 . Graph ofcas (-l against 7 for glass with (e) bonded -Si(CH,),CH,CH_CH,OCF(CF,). groups (treatment (h)), and for glass similarly treated and then hydrolyzed (treatment (i)) For explanation of the numbers, see legend to Fig . 1 . -01



CHEMICALLY MODIFIED GLASS CAPILLARY COLUMNS

2

4

y('o -2 NIm)

5

41 .

7

Fib . 5 . Graph of cos H against 7 for glass with bonded -Si(CH,),(CH,),-C„H,-CH_OOC-C„H, groups (treatment (h)). For explanation of the numbers . see legend to Fig. L On the other hand, surfaces covered with -Si(CH,),CH,CH_CH_CH,--C,H,CH,OH and -Si(CH,),CH,CH,CH,CH_ C,H,--CH,CN groups show rectilinear cos 49 rs- y plots (Figs . 7 and S). but with smaller negative gradients than the earlier graphs (Table 111) . 1c for both is approximately 35-10 -2 N/m . Esters show sonic degree or autophobia on both surfaces, and 3 2'-oxvdipropionitrile still shows an appreciable 0 on the surface containing cyano groups_ Anomalous cos 0 rs. relationships were observed for surf tces'rreated with compounds II and IV . It should be noted that a nudtilayer surface coverage is expected with these compounds due to the formation of polymers, and complicated solvation

4

6

7

7 (SO'2 Nlm) Fig . 6. Graph or cos H against for glass with bonded -Si(CH,),(CH,),-C6H, CH,NH' groups (treatment (d)) : For explanation of the numbers, see legend to Fig . 1 .



42

K . D . BARTLE, M . NOVOTNY

TABLE III DATA DERIVED FROM GRAPHS OF cos N AGAINST :- FOR CHEMICALLY TREATED SODA-GLASS SURFACES

Treatment (a) Cmdient mN-r 1W;' dNm_ , Number of points Correlation coefficient

(c)

(b)

-7 -16 36 4.1' 7 6

(d)

(f)

Ref

(1)

(1)

-7 -18 -17 -20 --19 --18 33 28 1_S' -29 21"' 1 .7" 7 9 11 11 6 9

--090 -0.91

0.89 --0 .98 --0.97

--0.99

-0_99 -088

of ;c = 4 .3-10 - ' Nlm (ref. 10) for polyethylene terephthalatet 4,0 - 0 .2-10 - ' N/nl (ref. 12) for polyethyleneglycal terephthalate_ cj. ; " -- 1_7-i0 N .m for surface composed of -CF --- and CF, groups (ref. 10) . "' rf ;, =?2-2 .4-I0 - ' N,'m for monolavercomposed of-CH, groups : 2-0-10-' N .m for-CH, crystal surface (ref. 10) .

lo t

o a

7

121415

os± t I 051

-

f 04 1 O

4

5

y(10_

2

N1m)

Fig, 7- Graph of cos f-1 against for glass with bonded -Sit CH,),(CH,I,-CJ H,-CH .OH groups (treatment (a))_ For explanation of the numbers, see legend to Fig . 1_

041 5

(10_ 2 Nfm) Fig. S . Graph of cos (d against for glass with bonded (treatment (c))_ For explanation of the numbers, see legend to Fig . 1 .

groups



CHEMICALLY MODIFIED GLASS CAPILLARY COLUMNS

43

phenomena 15 may complicate the measurements_ Surface hydrolysis of the ester groups with 11 may also be implicatedHydrolysis of the ester groups in capillaries with bonded -Si(CH 3),CH,CH,CH,CH_ C6H,-CH_OOC-C 6 H; groups was observed and studied in detail (Fig_ 9) ; the rise of the water level in capillary-rise experiments did not follow the Washburn equation 16 as in the case for capillaries in which no surface reaction occurred (e-,r- with bonded -Si(CH,),(:H,CH,CH,CH= C6H, CH,OH groups -Fig . 9) . and continued to increase as the ester groupings were hydrolysed to hydroxyl groups for which the contact angle is smaller and the meniscus height greater . A more rapid increase in level was observed for 0-1 111 hydrochloric acid in the ester-bonded capillary_ consistent with a greater reaction rate for the hydrolysis_ Such observations are . of course . strong evidence for the presence of the ester groups after the sequence of reactions outlined in Experimental10H2O

9

H2O OW HCI

720

(50

iSO

Time (mm)

Fig. 9- Graph of height of liquid in capillary tube against tine forglass with bonded -S1CH,)_ICH,),Ce HrCH.000-C,H s (S). and -Si(CH,)_[CH_L-Cj-L-CH,OH groups ( 1 .

Gam' chrotnatograplne- properties q/- treated capillaries

The properties(Table IV) of columns coated with plugs of stationary-phase solutions at coating rates appropriate to give film thicknesses near 1-i-f0 - ' m demonstrate clearly the importance of column pretreatment to give a surface compatible with the coated phase . For example, 3,Y-oxydipropionitrile coated on untreated glass results in a poor column, but much improved column efficiency is achieved on a~surface with bonded cyano groups_ In both cases . the measured' film thickness is very near

(i)

(j)

(c)

(a)

(j)

(1)

(h)

(e) (d)

(g)

Ic)

17

I5

16

15

15

16

17

19 IS

20

14

"

(c)

14

Srarirmni ;r phase

Si-C'II ;CII ;C'I-I ;NIIC'II ;CII ;NIi ;

-Si(ClI,)(Cll),,,C000 ;H, --Si(CH .d=(CI'I,h-C h I1 4 -CH :NII,

1,2 . .3-'frislcynno0thoxplpri' I1)c

Trielhanolamine

Dinonylphthalale Trieihanolantine

1 .5

1 .6

1 .4 1 .4

1 .4

1,3

1,4

.2 1

1 .3

1 .4

1 .5

0,3403 ')

0,24(47)

1,511(22) 1),22 (47')

1.611281

0,30 (53')

0,311 (53' )

(1,41 (53 )

(1.411 (53')

0.51 (53)

0.86(53`)

1,14(53')

1,15 (53')

1 .5

(I,1

(nrn :cue)

1 .15 (53')

a

111)

dleasurrr( k-ralar

1.9

1 .8

(10 - '

Colculolerl ihirknes's

.0 1

1 .1

.6 1

1 .6

1 .1

1,1

1.4

1 .6

2,1

2 .1

2,1 :1; (1,1

00`111)

lixperiuu •u Hd th rkiIess

260

ISO

360 60

220

1400

230()

300

1101)

170

1100

90()

2100

140

(plurexlm)

fin. hoir:('na

IiJ]iriem;r

0,2 mm) WITH VARIOUS TRWVTMP-NTS AND COATINGS

Uean oil 50 .1-IB • 20011' Won uil 50 .1-113-SI(CFI,),ICII r )4 -C1 N,-•C Ii ;OII 2(100 Si(Cli,) ;ICI-I ;).,-G,II r CIi ;CN Won oil 50 .11132000 Uconoil50-H1i-Si(CHs)1CH ;Cl1 ;C11 ;011 2001) None 2,2'-Oxydipropinnilrile'' Si1CII,I ;(CI-h) 1 -C„ 14-Cli h CN 2,2'-Oxydipropionilrile SilCli,)4CII ;) 4 -C„H,-Cllpll 2 .2'-O .xydipropionilrilc None Silicone oil OV-210''' -SiHCI1,1 ;CH ;CI I ;CI';CF ;Ch,, Silicone oil OV-210 -Si(CIi .J;(Cli ;l n -C' M1 li .,-CH ;-000' •C „II, DinonylphtlmhOC4

None

Bonded sm ;fare 'ernups

11 .7 , 10 - ' m'' :kg. 14 .8' 10-' mykg . 6.7-10 -2 nr'lkg . 69,2 .10 - ' nr' ;kg (ref, 33) .

(a)

II

V„ V, 11„ . VV •

(j)

Stir flirt' nrnnnenr

14

(rrr)

ivu rh

Cahumr

For details nithe surll,ce arenlmeals, see Experimental,

PROPIiRTII:S OF SDDA-C1LASS CAPILLARY COLUMNS (I.D .

TARl-I? IV



Slight tailing Tailing

Visible droplets

Visible droplets

Visible droplets

Reurar'hs

L

0

7C 0

i

m m

m D

x

t



CHEMICALLY MODIFIED GLASS CAPILLARY COLUMNS

47

that calculated from the Fairbrother-Stubbs relation, but compatibility of the surface and the stationary phase induces film homogeneity . Similar results were observed for columns coated with Ucon oil 50-HB-2000 (a polypropylene glycol)_ Poor efficiency on untreated glass was observed- while considerably improved results were obtained on the -Si(CH,),CH,CH,CH,CH,-C e H,CH,OH bonded surface_ On the other hand_ attempts to make such columns with d, : near 2 .5-10-' m . by changing the parameters of the Fairbrother-Stubbs equation_ failed . The range of stabilizing action of the hydroxyl group-containing surface is evidently comparatively limited . Similar eliects have been observed' . for trimethylsilanized columns coated with SF-96 methyl silicone oil and Apiezon L_ Thicker films may be prepared by increasing the surface area (e ;,, by etching) before a selective treatment . The compatibility effect is remarkably specific_ and complements . and may even over-ride. the requirement that contact angle be small for the stationary phase on the _glass surface . For example_ 2 2 •- oxvdipropionitrile has non-zero Cl (27") on the compatible surface which itcoatcsuceessfully :(-tier this phase on -Si(CH 3 ),CH,CH,CH .CH_ C„H, CH,OH bonded glass is similar (20 )_ butamchlowerunfi-v ciencv is observed . Conversely. Ucon oil has (1 - 0 on both hvdroxv and cyano surfaces . but a lower column eflicienev is observed in this case for the column with the surface containing cvano groupsThe interplay of contact angle and compatibility eliects is further demon- strated by columns coated with the fluorinated silicone polymer OV-210 . This phase has low ; •( 2 .19-10 _ N,'m) and coats untreated glass quite well- so that a column of good efficiency is obtained . However_ a surface treated with -Si(CH ;),CH,CH,CF,CF,CI= ; _groups has (1 .5 . 10 - ' N m) below even the -, value for OV-210 . and contact angle (I --- 3S for [his . liquid . The increased compatibility of the phase and tluorin :ated surface, however, results in only a relatively small decrease in column etliciencv . Similarly_ another surface compatible witli Ucon oil is the one containin g_ hydroxy groups prepared through the hydrolysis of -Si(CH,) .CH .CH_CH_ O-CF(CF;)_ with dilute HCI : although (9 is 25'_ a fairly efficient column Can he prepared . Regrettably . -compatibility is not always ensured by merely matching functional groups of the stationary phase with those of the surface : 12.3-tris(eyanoethoxy) propane hasstructuresomewhatanalo_goustothat or22 oxydipropionitrile.h :ssimilar (-1 on the cyano-treated glass (30 ), but gives poor columns . Iii the same wav, attempts

CH, to find Surfaces compatible with dinonyl phthalate through -Si-(CH,),,,-COOC .H_ and -Si(CH,) 2CH,CH_CH,CH,-C aH,-CH,000-C,H ; were unsuccessful In any case_ the observed surface hydrolysis off the ester linkage makes use of these compounds for practical chromatography rather limited . The treatment previously described by us'- in which an allvlsilane layer is bonded to the glass and oxidized in situ appears to provide a suitable Surface for - dinonyl phthalate No surface compatible with triethanolantine was found in this study_ However . gas-phase etching which was previously shown to aid the spreading of triethanolamine' may be combined with a selective treatment in orderr to bring about better efficiency .



46

K . I7. 13ARTLE, M . NOVOTNY

The molecular basis o/ compatihilit_t' A solid surface niay alter the structure of an adjacent liquid l ver . over itt distance of up to about 10 - ' ill (refs_ 18 _ 19)- This phenomenon is thought to be caused by the induction by the Solid Surlace of a molecular long-range orientation ill the vice. cal liquid . This condition alters the viscosity"'-" . thermal conductivity', and dielectric constant of liquid surface zones . These eirects cannot he explained on the basis of long-range dispersion force, extended from the solid surfaces . which are calculated to have much smaller eileets"-"- A more forceful explanation involve, molecular rotation restriction in oriented surface zones-'- For example . the rotational movement of an adsorbed or surface-bonded molecule is restricted about at least one axis . and the intermolecular distance to the next layer ill the hint is decreased compared ttttli condition, in the bulk liquid in which rotation is f ice . A reduction in chemical potential is expected . and the decreased distance gives rise to an excess dispersion interaction- In this taaV . nsuty molecular layers can be built up- but with increasing distance from the solid surface the rotations about other molecular axes become important and stahilizin _e excess interaction, are diminished \ :oreoter, the stability of certain dispersions ha ; been explained by the existence of an oriented later, of the liquid crystal type- of alk%lhenzenc molecule ; lying parallel to it graphite surface . and extending some distance from it'"- Boundary layers with -specific~properties may also he formed flu nonpol r compound, under the influence of adsorbed monoiavers of polar compounds of situiltu- structure' . Boundary layers arc thus 'tell-documented for low-molecular-height materials- and their existence would explain our obser%ttions of them stability in c :tpillare columns- - I he necessary similarity -of functional groups in a surface modifying nionolacer and the stationary phase may then hee explained li nu+lerulau - packing- . Thi, proposal was tested by model huilding-

A 'no alel uh flit• rlx •n ?irec/!r atodillezl ,lass sttrlare • It i, tsttally supposed that the structure of silic to glasses can he represented by the assumptions that the basic units are silica groupings in tt hich ;alkali and other basic oxides arc distributed- sold that the tyhole i, bonded together through oxygen . The mean spacing between the ,lightly irregular% arranged SiO, tetralledra i, atoms rather tt ider than the regular spacing in the silica allotrope tridvmile tt hich obtains in A glass surface etched by NaOH 'tax reported by Lihcrti' to have a mash silica structure- Accordingly_ a model planar silicon oxygen network was chosen to represent a glass surface with standard bond distance,`" (see p 491 . i.rIll= Such a structure could aecomoda to right hvdroxvl groups per unit area f l( 10tI A=)- but in practice (Table V) : smaller numbers sure available for bonding . and it figure of four (of tthich half %ere a>sttnled centime) per unit area was taken as best in keeping with other work . A possible model fi,r a chemically nn,dilied surface was then constructed by bonding --Si(CH :),(CI-l •) 4_C„H, - CH .C N groups through all non _eniinal and half of' the gentian -O(H) groups of the network (I-ig- 101 lower lacer . Bonding to geniinal groups is unlikely on steric __grounds ; reduced numbers of silaune groups with bulky constituents were found liv Unger" to bond to gees, surltccs (Table V) . Fig- Ill shuns that this procedure leads to a surface ellictivily comprised of Curly widespread cvano groups . Stationary phase molecules pith cvano groups such

: .red,siIicult :)eIlute . ,.unlcd Si1CII,l,I('I1 2 ), C„I I ., ('II :('\grttIllscrnncduills 2 .2 .oi)dipl'nplnnltrile ('nluureude I'Iv . I 11 Modelnf'lasssurl'aceteithh I ; bloc, nilrngen, Ln~cer : ~chite, hpdrngen : block, carbon :uud grrngtx nzypen in h ;lsic silica I'r ;uneevrk : rednr ;wge, nx)'gen : invnge-yepun, osggcn in I'It'e sil . • la)cr n( :,'_'-m)diprupinniu'ile lirsl nwlecultu nunlilied surface ; tipper la)cr l:n'er



J

A-



CHEMICALLY MODIFIED GLASS CAPILLARY COLUMNS

si

~N

/

0

0

0

49

si \/

/\0\

si

si-

0

0 0

I 0

0 I

0 si

0 si-

\0\5 / /

as 2,2'-oxydipropionitrile (upper layer in Fig_ 10) adsorbed on such a surface are thus constrained to locations specified by the packing of like groups_ Thus, the oriented surface zones necessary for the formation of stable layers of a compatible phase on chemically modified glass may arise. TABLE V

r

ESTIMATES OF NUMBERS OF SURFACE HYDROXYL GROUPS PER UNIT AREA (10- " m= = 100 A) ON SILICA SURFACE Method

Number per IO -1° tit' Isolated

Crystal structure Model 'H i0 exchange Gravimetry Reaction with SiCl,(CH,) ., T iCI 4 and BCI3 Reaction with SiCI(CH 3)3 Reaction with SiCIAC,H5)

Vicinal

2 .5

2S

1 .4

3.2

Reference

Total 8

38

8 5 6 .8 4.6 3-4 2.5

This work 39 3 (Etched glass) 40 41 31

CONCLUSIONS The previously suggested compatibility model for rationalizing wetting phenomena in glass capillary columns has now been tested with a series of monolayers of



50

K . D. BARTLE, MI . NOVOTN'

different structures and with liquids of different polarities . Although contact angles may be generally decreased through suitable surface modifications, zero values (which are the thermodynamic indication of an ideal wetting) are difficult to achieve_ This study underlines the great importance of chemical similarities between the surface and the wetting liquids for the preparation of homogeneous films inside cylindrical glass_ tubes_ It appears that the contact angle and the chemical compatibility interplay in the wetting effect_ The chromatographic efficiencies obtained with stationary-phase films on suitably modified monolayers give strong support to the high selectivity of "tailored -- surfaces . Consequently . stationary phases of different polarities can be effectively coated inside glass capillaries . thus creating homogeneous films of the order of 10 - ' m thick . "Compatibility' of the surface and wetting liquid must still be understood in a broader sense : the molecular mechanism of wetting may be attributed to quite complex phenomena among which boundary-induced long-range order is particularly important . and each matching of the - 'tailored - surface and the wetting liquid should be treated separately- A combination of surface roughness and selective chemical modi(ication appears necessary for inducing better adhesion of thicker liquid fibres` . . Temperature changes (e .,-. temperature programming in gas chromatography) should theoretically effect the film homogeneity'' less for a -- tailored-- surface than for ordinary Ihetero _geneous glass_ However_ the validity of this prediction has yet to he verified experimentally_ ACKNOWLEDGEMENTS This work was supported by the Grant No . GM-19232-02 of the National Institute of General Medical Sciences- U_S_ Public Health Service, We thank Ms . Leora :Medsker for technical assistance_ Several silanes used in this study were kindly supplied by Dr_ K . Grohniann, Division of Biology. California Institute of Technofogy, Pasadena . Calif. . U .S .A ., and Dr_ W_ Parr- Department of Chemistry_ University of Houston . Texas . U .S .A . REFERENCES . I M_ Novotne and-A. Zlatkis, Chrumar,gr. Rev., 14 (1971) 12 J . C_ Giddings, .Anal. Cheat . ; 34 11962) 458_ 3 A_ Liberti, in A . B . Littlewood (Editor) . Geis Chrontarogruphr 1966, Institute of Petroleum . London, 1967, p . 96 . 4 M . Novotny and K_ Trsatik, Chromumerapltia. I (1968) 331 5 M . Novotn}' .and K . D . Bartle, C/erantatugraphia, 3 (1970) 272_ 6 \i . Novotny and K . Grohmann.l. Chrunuauyr. . 84 (1973) 167. 7 K_ D . Ban le. Anal, Chem ., 45 (1973) 1831 S 4I . Nocotny, L. Blomberg and K_ D_ Bartle . J. C17rantrucn!r. Sri- S (1970) 390 9 F_ Fairbrother and A. E . Stubbs, J. Chem . Sue . (London), (1935) 527 10 W_ A . Zisman, Ltd. Eug . Cheat., 55 (1963) ISI I W_ A . Zisman, Advent . Chem . Ser., 43 (1964) 112 F_ Farrd-Rius, J . Henniker and G_ Guioclmn, Nature (Loaduui, 196 (1962) 63 . 1 3 E . F_ Hare and W. A . Zisman, J. P/i s_ Chent_, 59 (1955) 335 . 14 H_ W . Fox. E_ F_ Hare and W_ A. Zisman. J. P/ns. Cheat_, 59 (1955) 1097 15 J_ J_ Kirkland, J. Chranturugr . Sri„ 9 (1971) 206. 16 E_ W . Washburn, Plits . Rer_ . 17 (1921) 374 .



CHEMICALLY MODIFIED GLASS CAPILLARY COLUMNS

51

17 M . Novotny and IC . D. Bartle, J. Chrotaatoi+: 93 (1973) 405 . IS J . C. Henniker, Rev. Mud. P/ns., 21 (1949) 322 19 B . V. Der aguin, Discuss. Fareu/err Sac, 42 (1966) 109 . 20 C . Peschel and K . H . Adiflnger, Z Natmforseh. A- 24 (1969) 1113. 21 G- Peschel and K . H . Adllinger, Bee. BWLveu,es_ PI{vs. Chem ., 74 (1970) 35L 22 M . S . Metsik and O . S . Aidanova . in B . V. Dervagun (Editor) . Research in Surface Filnm- Consultants Bureau, New York. 1965. 23 G . Peschel, Z. Plus. Cheese. (Frankfurt ma Main) . 59 (1968) 27 . 24 J . Frenkel, Kinetic Theories of Liquids, Dover, New York, 195i25 K . H . :Adllinger and G . Peschcl, Spec- Discuss. Faradur Sue., 1 (1970) 89 . 26 C. D . Partite and E . Willis, J. Cullaid interface Sri., 22 (1966) 1011 . 27 B. V. Dervaguin. Diseuus.v. F,reulue• Sar_ 42 (1966) 138 . 2S J . R . Partington. .4n Ad ran red Trearise on P/psical Che mivtrr, Vol . III, Lon__nians, London . 1952. 29 J . H . Konnert, J- Karle and C . A . Fergttson, Science' . 179 (1973) 178 . 30 Chemical Society Special Publications II and IS . London . 1959 and 1965 . +I K . Unger . Am, ,r. Chem . Get- F-d. 6e.,t, II (1972) 267. . 32 F . C . Shairin and W . A_ Zisman . J. P/tnt Chem .- 76 11972) 3529. 3 E . R . Adlard . %t . A . Khan and B . T- Whitman, in R . P . WV. Scott (Editor), Gus Cf+rumcnq;rapfir 1960, Butterwortits, London . 1960, p . 251 . 34 J_ Timmermans, P/n:cica.chemicu! Ccucsca+ns ed Purr Or,unio Ceunpuane/ .v, Elsevicr, Amsterdam, 1950 . 35 /(aodhuok of C/tentistrr and Phracs. Chemical Rubber Publishin g-, Co ., Cleveland, Ohio . 36 M . Neti tso%d and K . Tesaiik . J. Chromani_v-.- 79 (1973) 15 . 37 fl . Cassiot-flatus- J . O, Pa_scual-Catrer s and A_ Serra-\taeia . C/u-emeam,rttphia. 5 ( 1972) 328 . 38 R . K . Iler, The Cu/fold Cftemicu_r of Silica s mud Sifieattec-Cornell University Press . Ithaca. X .Y. . 195539 V. Ya. Davydov. L . T- Zhuravtev and A . V . Kiselev- Zh- his . Ahint ., 37 (1963) 14114 . 40 C. G . Armistead . A . J . Tyler, F . H . Hamble[on, S . A . Mitchell and J . A . Hockev .I. P/i s . Cheat, 73 11969) 3947 . 41 R . F . Gilpin and M . F. Burke. :icmf. Chenc_ 45 (1973) 1381