Note : Les descriptions sont présentées dans la langue officielle dans laquelle elles ont été soumises.
BOND~D O~G~NO-PhLI!ICULPR PACKINGS
FOR C~ROMATOGF~PHIC COL~S
'llhis invention relates to impro~ed silica
packin~s for use in chroma-to~raphic analysis and to
a process f or makina them. It relates particularly
to $inel~ divi.ded silica suppor ~s havillg a very thln
pol-ylneric o~-~anic ~ chPmicaily 'oon~e~ to theiL
surf ac:e .
The ~roblem of optimizing chromatGgraphic
perlormance is one that has ~ersisted throu~hout the
history of chro natographic separations. A1-though
si~lificant advances have beell ~.ade in the apP11-
cation of g~s-liguid chromatog--aphy to analytical
lr~ problems, most of the~.e have r~sulte~1 from improve-
ments in apparatus rather tharl in th.~ column packlng
itself.
In order t.o avoi~ or minimize problems
caused b~ active sites Oll tke surface 0$ a silica
chroniato~raphic colum~ packing, ccat~d packings
have b2en prepar~d ~ eatir.g an acti~at~d sllica
'
,
? 5,5'36~
.,.... , . :
~-2--
with an ~pproprlc-te alcohol under conditions which
allow continuous removal of water, thereby causing
ethel-ification of the alcohol with silanol groups
on the si ica surface, see ~alasz et al., J. Chrom.
Sci., 12, 161 (197~ owe~rer, the coating obtained
is often no~ miform and the method gives poor
reproduci~ ty. Stehl, U.S. 3,654,967 describes
a method wh2reby a silica or alumina gel is reacted
with an organohalosilane, the halosilane groups
thus at-tached io the surface are reac-ted with. an
alcohol arid the produc-t is halogenated to provide
a haloorganic coa~ing ~onded to the support sur-
face. Ho~ever, the im~roved chromatographic packings
thus obtained are still not entirely satisfactory.
1~ lt has now been found that novel chron~ato-
grap~ic packings of uniformly high quality are ~ro~
duced ~y a process which cornprises ~1) contacting an
activated silica sup~ort havins a sign1ficant propor-
tion of hydroxyl ~roups bond~d to silicon atoms at
2~ the ~upport surface with silicon tetrachloride to
react essentially all of said h~rdroxyl g-oups, ihereby
- pLO~ucing a chlorosilylated surface, (2~ reactin~
by coiltacting the chlorosilylated surface ~ith an
inert solvent soluiion of a pol~ro' hav~Ilg an avera~
molecular weight o fror~l 3,000 -to 100,000 at a tempera
ture of from 100 to 250C, (~ ccntaztillg the polyol-
-chlorosilylated surface reaction product ~ith a
lower alkanol in sufIicient quantity to neutralize
residual chlorosilyi groups, and (4) separating
3~ t~.e neu-traLized product from the reaction mixture
as an esselltially ~ule and dry solid. The reaciion
prod~c-~ has 2 ~mifol~, chemicall~ ~onded crganic
surface c~atiily ~.r~lich is essenti~-~11y monomolecular
~5,596-F
.., ~ : ` '
.
~ 23 ~ . 3
_3.~
in -thickness. This bonded coating provides sub-
s~antially complete sulface coverage and reduces
~ surface activity ~o a minimum.
; The chromatog~aphic packings of this inven-
tion offer the advantages o:E high thermal stability,
increased selectivity, controll~le functionality,
reduced analysis time, low reactivity to sample
components~ and sharp separation of solute species
at lo~er ten~peratures than those required for elutior
on conventional coated packings.
These advantages are obtained by followin~
the above-described process steps and -they are
maxim ~ed by following those steps in their pre-
ferred modGs of operation. For example, the sur~
face of a silica support is preferably specially
ac'ci~ated to provide a laryer number of hydroxyl
groups bonde~ to the su;-face silicon atoms by treat-
ing a cle2n~d silica -~ith vaporized conce~tr~-ted
hydrochloric acid at 100 to 300C for 0.5 to S
hours. The vaporized aqueous ~Cl is most conveniently
- applied as a mixture with all inert gas, or example,
nitro~en, argon or helium.
The hydroxy (or silanol) groups on the
silica surfa~e are then reacted with silicon tetra-
~5 cnloride in either a slurry r~action with the liquidxeagen~ or, preferably, by a gas-solid reaction in
which SiCl4 vapors are contacted with a bed of the
silica particles. In either case, the Si~14 reaction
is Garried Olit at a temperatur~ Irom 50C to 300C,
preferably from 53~ to ~50~ for -he gas phase
reaction, at a somewhat lower tem~erature for the
25,59r~F
,
~3~23~3
reaction with li ~lid silicon tetrachloride in order
to avo:id excessi~e reactor pressure. This chlo1-o
silylation reaction is preferably ~arried to the
extent of 0.002 to 0.01 ~ram atoms of silicon-bound
chlorine per s~uare meter of silica surface.
The reaction of the chlorosilylate~ produ~t
wi-th the polyol or polyester pol~ol is carrie~ out
at a temper~ture from 100C to 25~C by contacting
the li~uid polyol reactant with the solid chloro-
silylated si'ica in a slurry reaction, pref'erablyin the presence of an inert solvent for the polyol.
Suitable solvents are those boiling at or above
lOO~C and inert to bot,h reactants under the reaction
conditiolls. ~romatic hydroc~rbons su~h as xylene,
diethyl~enæene, and durene are examples.
Lo~7er alkylene polyglycols of at least
3,0C0 average molecular weight arc preferred poliol
reactants. These include polyethylene glycol, poly-
propylene glycol, polybutyle~e glycol, ~lock co-
polymers of tWG Ol' more Or these oxyalkylene units,and physical mixtl~res of any of these. The minimum
molecular weight is a measure of the mini.mum length
o molecule required to give effQctive surace
~overage and conse~uent surface deactivatio~.
For poly~-thylen~ glyccl, the minimum molecular
weight indicates a chain of 65-70 oxyethylene units
in the average molecule. For pol,vpropylene and poly~
butylene ~lycols, molecular ~eights of about 4,000
and 5,000, respectively, correspond to mole~ules of
3i~ similar lengt'l. Polyglycols having an a~erage molec~
ular weight of about 100, noo represent a pr~ctical
maximu~ mol~cular size limit.
..
25,596-F
,.. ~ -, i
: ~ -
~3~..3
Polyester polyols are another class of
polyol reactant. Polymers n~ade by esterifying an
alkylene diol of 2-16 carbon atoms with rl dicar-
boxylic acid of 3-10 carbon atoms are preferred
5 examples. All~ylene diols include ethylene glycol,
propylene ~lycol, butylene glycol, diethylene
glycol, triethy].ene glycol, dibutylene glycol,
trimethylene glycol, 1,4~butanediol, and 1,12-
-dodecanediol, and mixtures of these. Alipha-tlc
dicarboxylic ac.ids such as malonic acid, suc-
eini- acid, and sebacic acid are prefexred although
aromatic diacids such as terephth~alic acid an~ iso-
ph$halic carl also be used, alone or in mixtur~
witll the acids deIined above. The polyesterifi
eation reac-tion is normally earried out for con-
venience witn tlle diacid chloride. Other reactive
dihalides can be mixed in minor proportion ~ith
the diaci~ chloride reactant in the polyesterifi
eation reaction -~o vary the properties of the
~0 resulting poly.~er, for example, organic silico~
dichlorides, or disulfonyl dichlorides. A minimum
molecular wei~ht of about 3,000 is also appro-
priate for po:Lyester polyol reaetants. Preferabl~,
the polyester polyol is prepared 1 _ itu, in the
presence of the chlorosilyla~ed silica so ~hat
the polyesterification reaction and the reaetion
o~ the polyol molecules with tke chlorosilyl grou~s
take place more or less simultaneously.
When the reaction o the polyol or poly-
ester polyol reactant -~litll the chlorosilylated
si~iea has essf.-nti~lly ceased, a small proportion
of unreacted chlorGsil~l srowps renlaiils o~ the sili.ca
surface. In crcler to eliminate these highly unde-
sirable active ~roups they are ne~tLalized by
~,
~5, 596~F
.:
~,
adding a lo~er al~ nol such as methanol, ethanol,
or isopropvl alcoi~ol to the reaction mixture and
heating a5 before. Preferably, an intermediate
neutrali~ativn with a lower molecular weight and
con~e~len-tly more reactive polyol is carried out,
most pref~rably with a series of such polyols of
progressiv~l-y decreasing molecuLar weight. In this
way, the s.ilica surface is blan}ceted to the gr~.atest
extent po.ssible with bonded molecules of ma~imum
leng~h. For example~ chlorosilylated silica Cail
be reacted with polyethylene glycol of about
20,000 molecular weight and remaining si.licon~bound
chlorine atoms then neutraliæ~d by successive re-
actions with polyethylene glycols of a~out 5,000
and abGut l,Q00 molecular weight, then with tri-
ethylene gl-rcol, and inally with methanol to
ensure the neutralization of all possible residual
chlorosil~l groups.
~3~L~ 9~5~ Surface
2C About 100 g portions of 80-100 mesh Chromo-
sorb ~Y, a flux-calcined celite diatomaceous silica
specially processed for chromatographic use by
Jo~s-M~nville Co~p., were e.~tracted for ~4~72 hours
in a Soxhle~ extraction apparatus with constant
25 boiling hydrochloric acid. The extracted silica l;
was pui in a washing column and washed for i~-24
hours with deionized water at room temperature using
a fluid bed back~flushing procedure to remove acid
and fines. The washed silica was rinsed thoroughly
with metr~an31 and then dried by passing filt2red
air ~rougn th~ colunln or about two hours. The
dried sili.ca was stcred in closed glass bottles
uniil sub~ec';ed to su^f;lce reac~ion.
!`
~,5~6-~
.. . .
, ~ :
_ --7--
r.~he si].i.ca surface was acti.vated by packing
about 30 g of the aci.d-extracted sili~z in a g~ass
reactor tllbe hected ~y a clamshell electric furnace and
passing about ~5 ml/min nitrogen t:llrou~h the bed while
its temperature was raised .n 40C steps to 200GC' over
a pexiod of about 40 minutes, t~len the .incoming nitrogen
was switched throug~l a conc. HCl bubb].er so that the
nitrogen ~.assin~ through the bed was essential`ly saturated
with HCl an~ water ~apor. The H~l-satura~ed nitrogen
stream was _ontinued at ~OO~C ar. tne s,~ne xate .~r
three ho-urs, -then the bubbler WdS bypassed and the bed
was flushed witn pur~ nitroger. for one hour, also at
200C.
Reaction with SiCl
A~ this point, a bubbler charged with SiCl4
~as connec-ted into the nitrogen supply line aDcl the
silica bed was contacted with SiCl4 vapor in nitrvgen
lor 90 mlnutes, the tempe.rature and nitrogen flow rate
remainin~ constant at t~e prior lev~e:Ls. 'rhe bed Q~
~0 chlorosilylated silica ~12S th?n flushed with nitrogen
2S before for 15 minutes and allowed to cool to room
temperature ~fter removal of the furnace with continuPd
'low of nitrogen.
~:
The followirlg procedure was employe~ wi~h ~i
2~ modifications as noted fGr th~ reaction of the chloro--
silylated silica with a polygiycol. A somewhat lnodified
procedure was used for the corresponding reaction o~ a
polyester polyol ~s described in those exan-.p].e~.
Re~ction Proced.ure
~ glass reac-~or flask ~ ipped with reflux
sondenser ald nitr~en ~ .et wa., c~iargefl .~it:h about
25,596 E
;,
i
--8--
10 g of polyglycol reactant and 300 ml of o-xylene
and the contents refluxed for about an hour with dry
nitrogen flush to remove small amounts of water,
then the contents were cooled to 110C and about
30 g of the chlorosilylated silica were added under
nitrogen. Nitrogen flow through the flask was
reestablished and the reaction mixture was heated
at reflux temperature for 2-24 hours. The reaction
was then quenched by successive addition of poly-
ethylene glycols as described in the examples while
maintaining reflux temperature. The reactor was
then cooled to 100C, the heat source was removed,
and 50 ml of anhydrous methanol were added slowly
to neutralize any residual chlorosilane groups.
After the addition was completed and the reaction
mixture had cooled to 55C to 60C, the liquid in
the reactor flask, consisting essentially of xylene
and unreacted polyglycols, was decanted and the
reacted silica was washed by decantation with three
portions of methanol followed by three portions of
chloroform. The washed silica was then carefully
transferred to a washing column where it was
throughly washed by gravity flow with successive
300 ml portions of methanol, chloroform, and methy- ~
lene chloride. The washed silica was then dried ~`
by drawing filtered air through the column for about
an hour. The finished bonded silica packing was
stored in sealed glass bottles until used. The
product was a free-flowing fine white powder.
Example 1
Chlorosilylated 100-200 mesh Chromosorb~ W
(a series of screened calcined and flux calcined
diatomite aggregates) was reacted with polyethylene
Trademark
25,596-F
., . '
g
glycol of 2~,0~ average molecular weight ~E 20,000)
by the procedure descri~ed above. T~e reaction was
~uenched bv adding about 2 g of melt,ed polyethylene
ylycol OI 6,0~0 average molecular weight ~E~,000),
refluxing the reaction rllixture for about 2C ~linutes
and repe~ Li ng this proceclure with successi.ve
2 g portions of polyethylene glycols E ~,OOo,
E-1,000, and E-400 and, finally, diethylene glycol.
The reacted silica was then trea-ted with methanol
and washed and dried by the previously described
procedure.
Fvr purposes of comp~rison, "bonded" or
coated siliGa packings were prepared by coating
100-120 n~esh Chromosorb~ W-HP Witil polyetnylene
glycol of 20,000 molecular wei~ht in quanti.ties
sufficient to produce loadings of 5 percent and
3 percent by weight using the con~entiona' slurr~
method (described by Halasz et al., J. Chrom. Sci.,
12, 161 (1~74). These packings were preconditioned
for 12 hours at 220C and 60 ml/min helillm flow.
These packings were compared in ~ G . 1 ~m
x 160 cm co]umn maintained at 75C and-using 300
~g/ml n-dodecane i.n me~hy~.ene chlori.~ as t~e test
solute and helium as the carrier gas. Column
efficiency for each packing at optimum carrier
flow was calculated from the plot o~ test results .~,
and is li.sted in Table 1.
.
~5,596~F
. ~ . , , .:
. . . . :
! , ' :: . ; - .
- : ' .' ~ ~ :,
PY~13
--10--
TABLE 1
Optimum
Column Plate Height, Carrier Flow
Packin~ mm/Theoretical Plateml/min _
5 bonded 0.34 5s
coated, 5% 0.67 27
coated, 3% 0.62 32
It is apparent that the bonded packing provided
substantially greater ~fficiency in terms of plat~
height and also offers faster analysis ti~.es sin~e
the o~timwn carrier flow was about twice that fo~
the coate~ packings prepared by a ~reviously known
method.
-Example 2 -
A bonded packing was prepaxed as .in Example
1 using 80 lG0 r.~esh Chromosorb~ W and about 8 ~ of
polyethylene glycol of 4,000 average molecular weight
(E-4,000). ~olyglycols used in the quench cycle of
the process were polyethylene glycols of 1,000 and
600 molecular weight, tetraethylene glycol and dieth-
ylene glycol, respecti~ely. This pc.cking was conl-
pared with a conventionally prepaxed polyester-coated
silica packing by the method of Example 1 in the
analysis of impurities in 1~2-dibromo-3-chloroplopane,
a commercial soil fumigant. Both packings showe~
the presence of allyl chloride and 1,2,3-tribromo-
propane in the product but the bonded packing Or this
inv~ntion also showed the presence o 1,2~5,6-
-tetrabromoheYane w~lich was not previously observed
30 USillg the col~vention^~l packing. Ad~itionally, us~
of thc bonded packing cut the analysis time in h~lf.
25,596-F
... . ,~ ,.
~2
Exa~ple 3
Bonded packings were also prepared ky the
method of Examples 1 and 2 using polyethylene glycols
with average molecular weights o a~out 6,000 and
abou-t 1,50G. Chromatographic testing showed excellerlt
results for the filst of these, eomparable with results
o~tained with the prodllcts of Examples 1 and 2.
However, t~e bonded pac~ing made with E-1,500 showed
severe peak tailing, charaeter:istie of high surfaee
activity. Evidently, for packings having a bonded
polyethvlene glycol layer, the minimum average
molecular size that provides adequa-tP surface cover-
age is in the molecular weight range of about ~,000,
eorresponding to polymers having about 65 to 70
oxyethylene units in the polyglycol moleeule.
This eonelusion was supported by ~he
properties o~ a bonded polyglyeol packing descri~ed
in Example 4 where the polyglycol reaetant was a
pol~ropylene glycol o 4,000 average molecular
weight, corresponding to about 65 to 70 oxypropylene
~mits per molecule. The bonded packing was evidently
at about the lower moleeular size limit for -the
bonded molecules covering the silica surface, for
it showed some peak tailing although suecessful
chromatographie separations were obtained.
Example 4
A bor.ded ehromatographie eolumn paeking
was prepared with polypropylene glyeol of 4000
average molecular weight (P-4000) and subjeeted
to the same eva'uation as descri~ed in Example 1.
Wherl examinecl using n-tetradec~lle at 100C as in
Example 1, theoretical plate he:ights of 0.88 mm
25,596-F
3~
-12-
were observed -For the P-4000 product as compared
to 0.45 ~-n for the bonded E-4000 packing. These
hydrocarbons are eluted faster at the same tempe.ra-
tures from the bonded polypropylene glycol columns.
~s a result, the number of components that ca~ be
separa-ted in a given leng-th o time is virtually
identical on both bonded columns and both are signi-
ficantly better than a conventionally coated column
packing as noted in Table I.
1~ ln addition to the improved efficiency,
the lonyer molec~les represented by the polypropylene
glycols afford this high efficiency at lower flow
rates than the corresponding bonded pol,yethylene
glycols, allowi~lg observations and detection of more
volatile components.
Example 5
The proc~dure described above for the prepa-
ration of the bol-lded polyglycol packings wa modified
to make a corresponding silica having surface-bonded
diethylene glycol succinate polymex. Equimolar quan~
- tities OI diethylene glycoi and succinyl chloride
(0.0472 g mole each) in o-xylene solutio~ were added
from separate dropping funnels to a flask reactor
containing about 30 g of chlorosilylated Chromosorh~
W-AW in refluxing o-,~ylene under a nitrogen atmo-
sphere. The resulting reaction mixture was quenched
by adding about 3 g of diethylene glycol and refluxing
an additisnal half hour. The mixture was the~ cooied
to about 100C and 50 ml of methanol were add?d drop~
wise with ~radual cooling to abvut 55C. Liqui.d was
decanted of~ and the c~oated silica was washed and
dried as before.
; 25,59~-F
;, . , ,, :
.: - .. . :
- , : . ..
2~
-~3-
The p~lye~ter-coated silica w~s packed
into a colun~n similar to that of E~ample 1 and the
column was used to separate a mixture of close:Ly
related phenols and cresols (~00 ~g in ether). The
com~onent~ of the ~ixture ~phenol, o-cresol, o-
-chlorophenol, p-chlorophenol, 6-chloro-o-cresol,
4-chloro-o-cresol, 2,4-dichlorophenol, and 4,6~
-dichloro-o-cresol) were all separated efficiently
and sharply.
- lG Example 6
The proce~ure of Example 5 was followed
in the reaction of chlorosilylated Chromosorb~ W-AW
with 1,12-dodecanediol sebacate polymer, the polymer
being formed ln situ by reaction of the diol and
the acid chloride as before. The reaction mixture
~as c~enched by the addition of 2 g of molten
dodecanediol with a one-}lour reflux followed l~
addition OI 2 g of diethylene glycol and another
20 minutes of reflux.
The bonded packing thereby produced was
found -to be particularly useful for analyzing mix ;'
tures of nonpolar compounds. For example, it was
highly effective for the isothermal chromatographic
analysis of thr~e alkanes (C14, C15, and C16) in
hexane. It also provided efficient separation of
polychlorinat~d ~ibenzo-p-dioxin isomers.
Example 7
A bcnded silica packing was made by the
pxocedure d~scribe~ in Exam~le 5 except th~t the
succinyl chloride ~as replaced by a mixture of 6.6 g
(0.042~ mol~) succillyl chloride and 1.2 g (0.0048
mole) y-c~anopl^~p~l phenol dichlorosilane. The
25,595-~
.
I
¢~3
bonded coating thereby obtained was polymeric di-
ethylene glycol succinate wherein a ten-th of the
succ:inyl groups were replaced by y-cyanopropyl
pheneyl silyl moieties. The r~action product was
quenched by reacting with diethylene glycol and
then with methanol as in Example 5.
The highly polar nature of the bonded
silica packing thereby obtained permitted efficient
gas chroma-tographic separation of brominated penta-
erythritols, the dibromo and t:ribromo compolmdsboth eluting in sharp peaks wi-th minimal tai.ling.
This pac~ing also proYidec improved separating
po~er and considerably reduced analysis time as
eompared to a conventional coated silicone packing
in the separation of components present in crude
pentabromochlorocyclohexane.
In all of the bonded polyester packing
products described in Examples 5-7 and in other
sueh bonded polyester packings prepared similarly
from other diol and dibasic acid reactants, t~e
polyester moieties had a eomparatively broad dis-
tribution of molecular weights .in the approximate
rangP of 1,000 to 20,000 based on examination of
the nonbonded polyester byproduet of the reaction.
~ ~5,59~-~
. . , . ~ -
, . : ........ :
