Professional Documents
Culture Documents
(
1
R
1
+
1
R
2
-
2
r
0
)
2
+ j
(
1
R
1
R
2
)
(1)
f ) 2K
(
1
R
-
1
R
0
)
2
2K ) 2 + j (2)
R
0
)
2 + j
2
r
0
X
N
N
)
{
X
M
M
exp
[
M(
M
0
-
N
0
)
k
B
T
]}
N/M
(3)
(
N
0
-
M
0
) )
4R
2
f
N
)
8K
(
1 -
R
R
0
)
2
N
(4)
C
N
)
{
C
M
exp
[
-8K
k
B
T
(
1 -
R
0
R
)
2
]}
R
2
/R
0
2
(5)
VesicleSizeDistributions via Electron Microscopy Langmuir, Vol. 19, No. 14, 2003 5635
determi ne that the opti mi zed K ) 0.15 ( 0.03 k
B
T, and
R
0
) 55 ( 10 nm, wi th a reduced Chi test ) 2.3. The
uncertai nty i n R
0
i s qui te hi gh when worki ng wi th such
broad di stri buti ons because the model i s onl y weakl y
sensi ti ve to R
0
. The rel ati vel y poor fi t qual i ty al so can be
attri buted to the mi smatch between theory and experi -
ment for R < 10 nm. The si mpl e Hel fri ch theory does not
account for hi gher order curvature terms that are i m-
portant when the bi l ayer radi us i s onl y a few mul ti pl es
of the bi l ayer thi ckness. On the basi s of raw numeri cal
data, R ) 42.6 nm and /R ) 0.59. For these broad
di stri buti ons, R
0
> R, as entropy favors smal l er ag-
gregates.
4,5
Comparison of Mean Vesicle Sizes Measured by
QLSand Cryo-TEM. Al though the data are di ffi cul t to
i nterpret quanti tati vel y, QLS i s a conveni ent tool for
qui ckl y compari ng vesi cl e si zes when deal i ng wi th a l arge
number of sampl es. I n addi ti on, QLS does not requi re any
speci al sampl e preparati on except for fi l teri ng dust and
other parti cl es. The vesi cl e si zes we observed by QLS are
general l y i n good agreement wi th vesi cl e sampl es of
si mi l ar composi ti ons publ i shed previ ousl y.
16
For both
CTAT-ri ch and SDBS-ri ch vesi cl es wi thi n each vesi cl e
l obe, there i s novari ati on i n vesi cl e si ze wi th SDBS/CTAT
rati o. For CTAT-ri ch vesi cl es, we observe a hydrodynami c
radi us R
QLS
) 57 ( 2 nm, and a pol ydi spersi ty i ndex (PI )
of 0.25 ( 0.03. For SDBS-ri ch vesi cl es, R
QLS
) 49 ( 3 nm
and PI ) 0.2 ( 0.03. The l ack of si ze dependence wi th
composi ti on i s consi stent wi th the i dea that the actual
vesi cl e bi l ayer composi ti on i s rel ati vel y fi xed and that
excess ani oni c or cati oni c surfactant i s expel l ed to coexi st-
i ng mi cel l es or sol uti on.
30
I t al so suggests that the bi l ayer
el asti c parameters and spontaneous curvature do not
depend on composi ti on wi thi n a vesi cl e l obe. The pol y-
di spersi ty i ndex i s a rough measure of the wi dth of the
si ze di stri buti on, normal i zed to the mean hydrodynami c
radi us, but shoul d not be confused wi th the standard
devi ati on of the di stri buti on.
27
Qual i tati vel y, the QLS data
agree wi th both the cryo-TEMresul ts that show the SDBS-
ri ch vesi cl es to be smal l er and more monodi sperse than
the CTAT-ri ch vesi cl es. However, the hydrodynami c
radi us measured by QLS i s si gni fi cantl y l arger than the
number-average radi us measured by both cryo-TEM and
FF-TEM.
Thi s i s because QLS does not gi ve a number-average
mean radi us; rather, the hydrodynami c radi us measured
by QLS i s skewed toward l arger si zes due to the depen-
dence of the scatteri ng i ntensi ty on the si ze of the object.
The i ntensi ty of the scattered l i ght from a vesi cl e wi th a
mass M
i
i s proporti onal to M
i
2
. QLS measures the
z-averaged di ffusi on coeffi ci ent, D
z
, as
41
where x
i
i s the number fracti on of vesi cl es of radi us i, M
i
i s the mass of si ze i, and D
i
i s the di ffusi vi ty of si ze i. For
a vesi cl e, M
i
i s proporti onal toR
i
2
as the vesi cl e i s a hol l ow
shel l . Hence,
and i nserti ng the Stokes-Ei nstei n rel ati on between si ze
and di ffusi vi ty, D
i
) k
B
T/(6R
i
), we have
The mean hydrodynami c radi us, R
DLS
, can be extracted
(41) Berne, B. J.; Pecora, R. DynamicLight Scattering; J. Wi l ey and
Sons: New York, 1976.
Figure3. (A) Cryo-TEM i mage of 7:3 w/w CTAT/SDBS, 1 wt
%total surfactant i n water at 25 C. Onl y spheri cal vesi cl es are
observed, and they are l arger and more broadl y si ze di sperse
than the SDBS-ri ch vesi cl es i n Fi gure 1. Smal l bl ack spots are
frost contami nati on. (B) Hi stogram of vesi cl e si zes from cryo-
TEM i mages of CTAT/SDBS, w/w, at 2 wt % total surfactant
i n water. The fi t of the bendi ng ri gi di ty model was opti mi zed
vi a the reduced Chi test, gi vi ng R
0 ) 55 nm and K ) 0.15 kBT.
On the basi s of raw numeri cal data, R ) 43 nm and /R )
0.59, whi ch are agai n i n excel l ent agreement wi th the corre-
spondi ng FF-TEM data.
D
z
)
i
x
i
M
i
2
D
i
i
x
i
M
i
2
(6)
D
z
)
i
x
i
R
i
4
D
i
i
x
i
R
i
4
(7)
D
z
)
k
B
T
6
i
x
i
R
i
3
i
x
i
R
i
4
(8)
5636 Langmuir, Vol. 19, No. 14, 2003 Coldren et al.
from thi s expressi on for D
z
by agai n usi ng the Stokes-
Ei nstei n rel ati on, R
hyd
) k
B
T/(6D
z
), yi el di ng
From eq 9, the hydrodynami c radi us can be ap-
proxi mated as the fourth moment of the vesi cl e si ze
di stri buti on di vi ded by the thi rd moment. Whi l e i t i s
i mpossi bl e to determi ne the ful l si ze di stri buti on from
QLS to compare to the cryo-TEM si ze di stri buti ons, the
hydrodynami c radi us can be cal cul ated from the si ze
di stri buti on measured by cryo-TEM.
For the CTAT-ri ch vesi cl es, the number-average radi us
determi ned from the cryo-TEM i mages i s R ) 43 nm.
From the si ze di stri buti on i n Fi gure 3B, we can use eq 9
to cal cul ate an equi val ent hydrodynami c radi us, R
hyd
)
i
x
i
R
i
4
/
i
x
i
R
i
3
) 63 nm, whi ch i s cl ose to the R
QLS
) 57 (
2 nm measured by QLS. The cal cul ated hydrodynami c
radi us i s very sensi ti ve to the fracti on of the l argest
vesi cl es; i f the l argest 1% of vesi cl es i n the cryo-data i s
l eft out of the cal cul ati on, R
hyd
drops to 59 nm. Hence,
whi l e QLS can cl earl y gi ve a consi stent measure of the
average radi us, thi s average shoul d al ways be greater
than that determi ned by TEM and i s domi nated by any
l arge parti cl es that may be i n the sol uti on. For the SDBS-
ri ch sol uti on, the number-average radi us of the spheri cal
vesi cl es i s R ) 30 nm. From Fi gure 1B and eq 9, R
hyd
)
i
x
i
R
i
4
/x
i
R
i
3
) 37 nm, whi ch i s qui te a bi t smal l er than
the 49 (3 nm measured by QLS. Thi s di screpancy poi nts
out a l i mi tati on of QLS rel ati ve to di rect i magi ng. I n the
cryo-TEM i mages, Fi gure 1A, of the SDBS-ri ch sol uti ons,
there are l ong, cyl i ndri cal bi l ayer aggregates i n addi ti on
to the spheri cal vesi cl es. QLS gi ves an average hydro-
dynami c radi us for al l of the structures i n sol uti on; the
cyl i ndri cal vesi cl es l i kel y skew the mean si ze and the
pol ydi spersi ty toward l arger val ues.
Freeze-Fracture Results. Si ze di stri buti ons from
cryo-TEM sampl es can be compl i cated by vesi cl e segrega-
ti on duri ng sampl e preparati on. The water fi l m i n a cryo-
TEM sampl e vari es i n thi ckness from a few mi crons to a
few hundred nanometers over the TEM gri d, thi cker near
the supporti ng hol ey pol ymer fi l ms of the gri d and thi nner
i n the gaps between the pol ymer fi l ms. Thi s shoul d not
si gni fi cantl y affect the si ze di stri buti ons shown i n Fi gures
1A and 3A as the vesi cl es are smal l compared to al l the
sampl e thi ckness. However, toconfi rm the cryo-TEM si ze
di stri buti ons, i t i s useful to compare the mean radi us and
standard devi ati on between sampl es that have the same
mean radi us and pol ydi spersi ty when measured by QLS
by a techni que that does not requi re such thi n water fi l ms.
Hence, we used freeze-fracture to measure the si ze
di stri buti on i n sampl es wi th sl i ghtl y di fferent composi ti on,
but that had the same mean radi us and pol ydi spersi ty as
measured by QLS.
However, eval uati ng a mean vesi cl e si ze from a freeze-
fracture el ectron mi croscopy i mage i s compl i cated by the
fracture process. I n freeze-fracture, vesi cl es appear as
hemi spheri cal domes or craters wi th di sti nct shadows
(absence of pl ati num). The fracture pl ane can pass through
any part of the vesi cl e, skewi ng the apparent si ze
di stri buti on toward smal l er si zes. By accounti ng for the
probabi l i ty densi ty of cross fracture pl anes, Hal l ett et al .
42
determi ned si mpl e correcti ons to gi ve better esti mates of
the mean and standard devi ati on of a spheri cal vesi cl e
di stri buti on. I n a freeze-fracture repl i ca, fractured vesi cl es
can be i nterpreted as the i ntersecti on of a random fracture
pl ane wi th a sphere of radi us R. The i mage then appears
i n the mi crograph as a ci rcl e of radi us r eR. I f the cl eavage
pl ane i s equal l y probabl e at al l posi ti ons wi thi n the vesi cl e,
the probabi l i ty that the apparent radi us i n the i mage i s
between r and r + dr i s G(r) dr, whi ch, for a vesi cl e of
radi us R, i s
42
I f H(R) i s the actual vesi cl e si ze di stri buti on, then the
apparent di stri buti on of radi i determi ned from a freeze-
fracture i mage i s I (r):
The i ntegrati on goes from r to because the true vesi cl e
radi us, R, i s never smal l er than the apparent radi us on
the freeze-fracture i mage, r. As can be seen from eq 11,
eval uati ng the si ze di stri buti on requi res a pri ori knowl -
edge of the functi onal form of the di stri buti on. Even i f we
assume the functi onal form gi ven by our spontaneous
curvature theory, eq 5, a compl ete deconvol uti on of the
freeze-fracture si ze di stri buti on i s a practi cal i mpossi bi l i ty.
Fortunatel y, the fi rst twomoments of the si ze di stri buti on
can be determi ned from FF-TEM data independent of any
functi onal form of the actual si ze di stri buti on. Si nce
eq 12 rel ates the mean radi us measured from the freeze-
fracture i mages, r, wi th the true mean radi us of the
di stri buti on, R: R ) 4r/. By determi nati on of the
average apparent radi us of the fractured vesi cl es, the
average vesi cl e radi us can be cal cul ated by a si mpl e
mul ti pl i cati ve factor.
The standard devi ati on of the vesi cl e radi i can al so be
determi ned from the freeze-fracture i mages i n a si mi l ar
way. The second moment of I (r) yi el ds the mean square
val ue, r
2
:
From eqs 12 and 13, the standard devi ati on of the vesi cl e
si ze di stri buti on can be easi l y cal cul ated:
Al though freeze-fracture i mages cannot easi l y gi ve a
compl ete representati on of the vesi cl e si ze di stri buti on,
the mean and standard devi ati ons of the si ze di stri buti on
can be compared between the cryo-TEM and freeze-
fracture sampl e preparati ons. As freeze-fracture sampl es
(42) Hal l et, F. R.; Ni ckel , B.; Samuel s, C.; Krygsman, P. H. J . Electron
Microsc. Tech. 1991, 17, 459.
R
hyd
)
i
x
i
R
i
4
i
x
i
R
i
3
(9)
G(r) )
r
R(R
2
- r
2
)
1/2
(10)
I (r) )
r
r
R(R
2
- r
2
)
1/2
H(R) dR (11)
r )
rI (r) dr )
H(R)
[
0
R r
2
R(R
2
- r
2
)
1/2
dr
]
dR
(12)
r )
RH(R) dR )
4
R
r
2
)
r
2
I (r) dr )
H(R)
[
0
R r
3
R(R
2
- r
2
)
1/2
dr
]
dR
(13)
r
2
)
2
3
R
2
H(R) dR )
2
3
R
2
2
) R
2
- R
2
)
3
2
r
2
-
16
2
r
2
(14)
VesicleSizeDistributions via Electron Microscopy Langmuir, Vol. 19, No. 14, 2003 5637
are typi cal l y 10-100 mi crons thi ck, probl ems of si ze
segregati on, compressi on, and ori entati on that troubl e
cryo-TEM are not present i n freeze-fracture. Whi l e both
techni ques may suffer from freezi ng arti facts (whi ch are
mi ni mi zed i n cryo-TEM sampl es due to thei r smal l er
thi ckness and can be eval uated by di ffracti on todetermi ne
i f there has been i ce crystal formati on, see refs 25 and 26),
freeze-fracture sampl es are not damaged by el ectron beam
i rradi ati on, and the contrast i n the i mage i s generated by
di fferent processes. Hence, agreement between the two
methods i s good evi dence that the si ze di stri buti ons
determi ned by both methods are accurate.
Fi gure 4 shows vesi cl e si ze di stri buti on hi stograms
determi ned from freeze-fracture TEM i mages (not shown)
for CTAT/SDBS, 2:8 w/w at 2 wt % (Fi gure 4A) and 8:2
w/w at 1 wt % total surfactant i n water at 25 C (Fi gure
4B). The si ze di stri buti on shows the apparent vesi cl e
radi us, uncorrected for the effect of random fracture
pl anes, and thus i s skewed to smal l er si zes. Fi gure 4A
shows CTAT/SDBS 2:8 w/w at 2 wt % total surfactant i n
water at 25 C. The corrected R )30 nm and /R )0.34
are i n excel l ent agreement wi th cryo-TEM of CTAT/SDBS
3:7 w/w at 2% total surfactant of R ) 30 nm and /R
) 0.32. For the 8:2 w/w CTAT/SDBS at 1 wt % total
surfactant i n water at 25 C, the apparent si ze di stri buti on
i s shown i n Fi gure 4B. As expected from the random
fracture pl anes, the di stri buti on i s bi ased toward smal l er
radi i than the cryo-TEM di stri buti ons. After deconvol u-
ti on, R ) 37 nm and /R ) 0.59, i n good agreement
wi th the cryo-TEM whi ch gave R ) 43 nm and /R )
0.59. As di scussed previ ousl y, QLS shows that the
hydrodynami c radi us and pol ydi spersi ty do not change
si gni fi cantl y wi th composi ti on over thi s range, sothi s good
agreement speaks to the consi stency between al l three
methods of measuri ng the si ze di stri buti on. Apparentl y,
nei ther the smal l di fference i n vesi cl e composi ti on nor
the FF-TEM deconvol uti on i ntroduced si gni fi cant errors.
Furthermore, the twosets of i mages were generated years
apart from di fferent surfactant stocks, i ndi cati ng the
repeatabl e properti es of these thermodynami cal l y con-
trol l ed vesi cl e systems.
For both vesi cl e composi ti ons i n the CTAT/SDBS/water
system, cryo-TEM and FF-TEM si ze di stri buti ons are i n
excel l ent agreement once R and /R are properl y scal ed.
However, cryo-TEM remai ns the onl y techni que for
unambi guousl y determi ni ng the ful l si ze di stri buti on of
pol ydi sperse fl uctuati ng vesi cl es and, hence, esti mati ng
el asti c properti es. Neverthel ess, FF-TEM i s cl earl y a
rel i abl e tool for determi ni ng R and /R as wel l as
veri fyi ng a uni modal vesi cl e si ze di stri buti on. Further-
more, FF-TEM shares cryo-TEMs advantages over scat-
teri ng i n the abi l i ty to i gnore dust and generate number-
average data rather than hi gher moment averages. There
are even advantages of FF-TEM over cryo-TEM: a much
l ower probabi l i ty of shear-i nduced arti facts and arti facts
due toevaporati on/condensati on. Thus, FF-TEMand cryo-
TEM shoul d be vi ewed as compl ementary techni ques that
shoul d be used i n paral l el to generate unambi guous
col l oi dal si ze di stri buti ons.
Conclusions
Cryo-TEM, freeze-fracture TEM, and quasi -el asti cl i ght
scatteri ng are compl ementary methods todetermi ne mean
radi i , pol ydi spersi ty, and even the compl ete si ze di stri bu-
ti on of submi cron col l oi dal parti cl es i ncl udi ng spontaneous
vesi cl es. Cryo-TEM and FF-TEM show surpri si ng agree-
ment i n both the mean radi i and the standard devi ati on
of both CTAT-ri ch and SDBS-ri ch spontaneous vesi cl e
sampl es, once the FF-TEM data are corrected for the
fracture process. Moreover, both are number-average
methods so they general l y are more sui tabl e for com-
pari son to theoreti cal si ze di stri buti ons. QLS shows a
systemati cal l y l arger mean vesi cl e si ze, but compari sons
of the rel ati ve si zes and pol ydi spersi ti es between the
CTAT-ri ch and the SDBS-ri ch sampl es are consi stent wi th
both mi croscopy methods. For the CTAT-ri ch sampl es that
onl y form spheri cal vesi cl es, the cryo-mi croscopy data,
when wei ghted appropri atel y for the scatteri ng i ntensi ty,
gi ve a mean radi us i n good agreement wi th the hydro-
dynami c radi us measured by QLS. However, the agree-
ment i s l ess good for the SDBS-ri ch sampl e. Thi s poi nts
out a di sadvantage of QLS that i s not often appreci ated
i n compl ex surfactant systems. Surfactant aggregates can
take on a vari ety of shapes i ncl udi ng di sks, cyl i nders, and
spheres. QLS averages over al l such aggregates i n sol uti on,
Figure4. (A) Hi stogram of vesi cl e si zes from freeze-fracture
TEM i mages (not shown) of 2:8 w/w CTAT/SDBS, 2 wt %total
surfactant i n water at 25 C, uncorrected for the effect of random
fracture pl anes i n the FF-TEM i mage. After correcti on, R )
30 nm and /R ) 0.34, i n excel l ent agreement wi th the
correspondi ng cryo-TEM si ze di stri buti ons of the SDBS-ri ch
vesi cl es (Fi gure 1A). (B) Vesi cl e si ze di stri buti on hi stogram
from FF-TEM i mages (not shown) of 8:2 w/w CTAT/SDBS at
1 wt % total surfactant at 25 C. Si zes are raw FF-TEM data,
uncorrected for the effect of random fracture pl anes, and thus
are skewed to smal l er si zes. After correcti on, R ) 37 nm and
/R )0.59, whi ch are i n good agreement wi th cryo-TEM data
above. Apparentl y, nei ther the smal l di fference i n vesi cl e
composi ti on nor the FF-TEM deconvol uti on i ntroduces si g-
ni fi cant errors. The two sets of i mages were generated years
apart from di fferent surfactant stocks.
5638 Langmuir, Vol. 19, No. 14, 2003 Coldren et al.
and both the mean radi us and pol ydi spersi ty of each
structure mi ght be qui te di fferent from the average
hydrodynami c radi us. The mai n advantage of QLS i s the
mi ni mum amount of sampl e preparati on and the qui ck
data accumul ati on, whi ch makes QLS i deal for i ni ti al
qual i tati ve compar i sons between l ar ge number s of
sampl es. However, the combi nati on of QLS, freeze-fracture
TEM, and cryo-TEMcan provi de rel i abl e si ze di stri buti ons
of equi l i bri um vesi cl es that can then be fi t to theoreti cal
expressi ons to extract materi al parameters.
Wi th the excepti on of systems contai ni ng l onger fl u-
ori nated chai ns (>7 carbons),
1
equi l i bri um vesi cl es formed
from catani oni c surfactant mi xtures of unequal chai n
l ength are typi cal l y pol ydi sperse. Wi thi n the confi nes of
our thermodynami c model i ncorporati ng spontaneous
curvature, thi s woul d i ndi cate that these catani oni c
surfactant mi xtures form bi l ayers of very l ow ri gi di ty (of
order k
B
T) regardl ess of di fferences i n headgroup chem-
i stry (sul fate, sul fonate, carboxyl ate, etc.), counteri on
(bromi de, chl ori de, tosyl ate, etc.), chai n l ength (6-16
carbons), and degree of branchi ng (SDBS). The tosyl ate
counteri on i n the CTAT/SDBS system al so l eads to a
decrease i n the effecti ve bi l ayer ri gi di ty. For SDBS-ri ch
bi l ayers, the effecti ve el asti cconstant K )0.54 k
B
T, whi ch
i s si mi l ar to that of l i near-chai n bi l ayers i n cetyl tri m-
ethyl ammoni um bromi de (CTAB)/sodi um octyl sul fate
(SOS) and CTAB/FC5, whi ch al so have K 0.5 k
B
T.
1
Hence, i t appears that hydrocarbon branchi ng does not
reduce membrane ri gi di ty si gni fi cantl y over si mpl e chai n
mi xi ng. However, CTAT-ri ch vesi cl es have an effecti ve
el asti c constant K ) 0.15 k
B
T, si gni fi cantl y l ower than
those of CTAB/SOS, CTAB/FC5, or SDBS-ri ch bi l ayers.
I n fact, i t i s si mi l ar to that of equi mol ar dodecyl tri -
methyl ammoni um chl ori de (DTAC)/SOSo bi l ayers (K )
0.12 k
B
T, tobe publ i shed). Si nce the DTAC/SOSobi l ayers
compri se ful l y di ssoci ati ng hydrophi l i c counteri ons and
much shorter unbranched chai ns, we may concl ude that
the tosyl ate counteri ons i n CTAT-ri ch bi l ayers probabl y
do remai n associ ated wi th the bi l ayer, i ncrease the area
per surfactant head, and dramati cal l y reduce the mea-
sured ri gi di ty. I n that case, the rol e of tosyl ate i n the
bi l ayer i s anal ogous to that of short-chai n fatty al cohol s
that al so promote the reducti on i n measured membrane
ri gi di ty.
43
Hence, i t appears that much of the reducti on
i n bi l ayer el asti ci ty common to spontaneous vesi cl e
bi l ayers i s due to si mpl e chai n mi xi ng, regardl ess of the
type of chai n bei ng mi xed. However, i ncreasi ng the area
of the headgroup can reduce the el asti ci ty even more,
l eadi ng to more pol ydi sperse vesi cl e popul ati ons.
Acknowledgment. The authors thank J. I srael achvi l i
and E. Kal er for ongoi ng col l aborati ons and di scussi ons
and Dr. D. Dani no for trai ni ng R.V.Z. i n cryo-mi croscopy
techni ques. The authors grateful l y acknowl edge support
by Nati onal I nsti tutes of Heal th Grant EB-00380 and
Nati onal Sci ence Foundati on Grant CTS-9814399 and the
use of equi pment supported by the NSF MSERC Program.
LA034311+
(43) Chi ruvol u, S.; Warri ner, H. E.; Naranjo, E.; I dzi ak, S.; Radl er,
J. O.; Pl ano, R. J.; Zasadzi nski , J. A.; Safi nya, C. R. Science1994, 266,
1222.
VesicleSizeDistributions via Electron Microscopy Langmuir, Vol. 19, No. 14, 2003 5639