Physics of Compaction of Fine Cohesive Particles
Abstract
Fluidized fractal clusters of fine particles display critical-like dynamics at the jamming transition, characterized by a power law relating consolidation stress with volume fraction increment. At a critical stress clusters are disrupted and there is a crossover to a logarithmic law resembling the phenomenology of soils. We measure _ _ _@__1=__=@ log_^ c / Bo0:2 g , where Bog is the ratio of interparticle attractive force (in the fluidlike regime) to particle weight. This law suggests that compaction is ruled by the internal packing structure of the jammed clusters at nearly zero consolidation.
Full text
Physics o Compac ion o Fine Cohesi e Pa icles
A. Cas ellanos, J. M. Val e de, and M. A. S. Quin anilla
Depa amen o de Elec onica y Elec omagne ismo, Uni e sidad de Se illa, A enida Reina Me cedes s/n, 41012 Se illa, Spain
(Recei ed 9 No embe 2004; published 22 Feb ua y 2005)
Fluidized ac al clus e s o ine pa icles display c i ical-like dynamics a he jamming ansi ion,
cha ac e ized by a powe law ela ing consolida ion s ess wi h olume ac ion inc emen [ ^c/].
A a c i ical s ess clus e s a e dis up ed and he e is a c osso e o a loga i hmic law (log ^c)
esembling he phenomenology o soils. We measu e @1==@ log ^c/Bo0:2
g, whe e Bogis he
a io o in e pa icle a ac i e o ce (in he luidlike egime) o pa icle weigh . This law sugges s ha
compac ion is uled by he in e nal packing s uc u e o he jammed clus e s a nea ly ze o consolida ion.
DOI: 10.1103/PhysRe Le .94.075501 PACS numbe s: 61.43.G , 45.70.Cc, 61.43.H , 81.20.E
Empi ical s udies on he compac ion o soils da e back o
he beginning o he las cen u y. Walke [1] i ed his da a
by he loga i hmic law 1=logc=c0, whe e
is he pa icle olume ac ion, c he applied consoli-
da ion s ess, and (comp ession index) and c0a e em-
pi ical pa ame e s. This equa ion applies well in loose
samples, whe e compac ion is d i en by ea angemen
o pa icles, and has been adi ionally used in ci il engi-
nee ing [2,3]. An essen ial ing edien in mos g anula sys-
ems is cohesion. Tes s on cohesi e powde s show ha
dec eases wi h he pa icle olume ac ion o he ini ial
s a e [4,5], indica ing ha in e pa icle a ac i e o ces,
which a o he o ma ion o po ous s uc u es, play a ele-
an ole in he compac ion p ocess. Ye he ini ial s a e in
ypical enginee ing expe imen s in ol es consolida ion
s esses c0>10 kPa [5]. Many indus y applica ions de-
mand esea ch on smalle consolida ions as hese co e-
spond o condi ions o powde low. Fo example, in he
handling o xe og aphic one s, ypical consolida ions ange
om a ew pascals o a ew hund ed pascals. Mo eo e ,
expe imen s a low consolida ions ha e a undamen al in-
e es in o de o cha ac e ize he ansi ion om he luid-
like o he solidlike s a e (jamming) [6] since he s uc u al
p ope ies o he unconsolida ed jammed s a e (c’0),
which is he uly ini ial s a e in any compac ion p ocess,
a e de e minan on he ea angemen o he u he loaded
pa icles. We s udy he compac ion o ine pa icles wi h
con olled a ac i e o ce, ini ially luidized and la e sub-
jec ed o loads om jus a ew pascals up o 10 kPa. Ou
no el expe imen al s udy is aimed o shed ligh on he ole
o he ini ial s a e, i.e., he unconsolida ed jammed s a e, on
compac ion. The powde s es ed a e xe og aphic one s
based on polyme (pa icle densi y p’1g=cm3). They
a e p oduced by an a i ion p ocess, hus ha ing an i egu-
la shape, and size classi ied in a ange o pa icle sizes
(dp) om 19.1 o 7mby ae odynamic classi ica ion,
showing a na ow pa icle size dis ibu ion (see Fig. 1).
Addi ionally, he powde s a e blended wi h umed silica
nanopa icles (ei he 8 o 40 nm nominal diame e s) o coa
uni o mly he polyme pa icle su ace in concen a ions
om 10% o 100% o su ace a ea co e age (SAC).
In he luidized egime he e is an a ac i e o ce F0
be ween he d y and uncha ged pa icles mainly a ising
om he an de Waals in e ac ion F0F dW ’Ada=
24z2
0, whe e z0’4
Ais he dis ance o closes app oach
be ween wo molecules, Ais he Hamake cons an , and da
is he ypical size o he su ace aspe i ies ( ypically A
1019 Jand da0:2m) [7]. An es ima ion o elec o-
s a ic o ces om cha ge spec og aph measu emen s
shows ha hey a e much smalle han an de Waals o ces
as commonly accep ed in he li e a u e o elec oneu al
ine powde s [7,8]. A d y ni ogen a mosphe e minimizes
also capilla y o ces. Because o he s ong in e pa icle
a ac i e o ce as compa ed o pa icle weigh , one pa -
icles a e clus e ed in he luidlike egime [9]. Acco ding o
ou p e ious expe imen al esul s he ypical numbe o
pa icles pe clus e Nand ypical a io o clus e size o
pa icle size depend on he a io o a ac i e o ce o
pa icle weigh F0=mpgBog(g anula Bond num-
be ). In pa icula we ound NBo
g(’0:7) and D
lnN=ln’2:5 o he ac al dimension, in ag eemen
FIG. 1. Typical pic u e om he op ical mic oscope o one
pa icles clus e ed in a nonaqueous liquid suspension ha
eminds one o a di usion-limi ed agg ega e. Fo his one
(12:7mpa icle size and 10% SAC) he ac al dimension
ob ained om se ling expe imen s in gas luidiza ion [9] is
D’2:53.
PRL 94, 075501 (2005) PHYSICAL REVIEW LETTERS week ending
25 FEBRUARY 2005
0031-9007=05=94(7)=075501(4)$23.00 075501-1 2005 The Ame ican Physical Socie y
wi h he di usion-limi ed-agg ega ion (DLA) model p e-
dic ion. The ypical size o su ace aspe i ies a con ac can
be dec eased down o he size o silica agglome a es co e -
ing he pa icle su ace o la ge enough SAC, hus educ-
ing F dW. (F om scanning elec on misc osopy mic o-
g aphs we obse e silica agglome a es o ypical size da’
50 nm o 8 nm silica nanopa icles and da’200 nm o
50 nm silica nanopa icles.) Since ou ine pa icles a e
a he mal, he limi s o clus e g ow h in he ini ial luidized
s a e esul om he in e play o g a i a ional and low
shea e ec s [10]. In o de o es ima e he limi size o ou
DLA clus e s we will adop a simila app oach o ha used
in Re . [11], whe e he limi s o gela ion in colloidal
agg ega ion we e s udied. In he luidized bed he weigh
o he clus e is balanced by he hyd odynamic d ag om
he su ounding gas. F ac al clus e s sc een ex e nal ields
e y e ec i ely and he luid low inside he clus e is
negligible compa ed o he low ou side, hus he d ag
ac s mainly a he su ace o he clus e whe eas g a i y
is a body o ce ac ing uni o mly h ough he clus e . This
esul s in shea o ces dis ibu ed ac oss he clus e limi ing
i s size. Using a simple sp ing model o he clus e , i has
been shown [10,11] ha he ypical s ain on he clus e is
Nmpg=KcRc, whe e Kcis he clus e sp ing cons an
and Rcdp=2is he clus e adius. Kcis gi en by
k0=, whe e k0is he in e pa icle o ce cons an , and
he elas ici y exponen is 3in he 3D case [12]. Thus
he local shea o ce inside he clus e is Fsk0dp=2
mpgD2. Manley e al. [11] use a c i ical alue, mea-
su ed independen ly, o he maximum s ain sus ainable o
calcula e he maximum size o hei agg ega es. Mo e
gene ally, we may es ima e ha he c i ical shea o ce
mus be o o de o he in e pa icle a ac i e o ce Fmax
s
F0, which leads o BogD2, hus he maximum numbe
o pa icles pe clus e should be NDBoD=D2
g.
Fo DLA clus e s (D2:5) we ob ain NBo0:6
g, in close
ag eemen wi h ou p e ious expe imen al esul s [9]. Now
we can explain why, o a cons an F0(cons an SAC), he
size o ou clus e s measu ed in Re . [9] was weakly
dependen on pa icle size since he c i e ion p edic s
kdp/d0:3
p. Mo eo e , o ou ypical clus e s (<10)
he in e clus e Bond numbe is Bo
gBog=N 2<
100, i.e., in e clus e cohesi eness is small.
In he expe imen al pa o ou wo k we use he luidized
bed es e o measu e as a unc ion o c. A de ailed
epo abou he unc ioning o his appa a us can be ound
in Re . [13]. A d y ni ogen se es o con ol c, being
pumped upwa d o downwa d h ough he powde bed
while he gas p essu e d op pac oss he bed is ead
om a di e en ial p essu e ansduce . is de i ed om
he heigh o he bed, which is measu ed by means o an
ul asonic senso . In o de o subjec he powde o e y
low s esses, like in mic og a i y, he bed is allowed o
se le unde a small upwa ds di ec ed gas low. In his way
cis lowe ed down o cWp, whe e Wis he
powde weigh pe uni a ea and pinc eases as he alue
o he decomp essing gas low used is la ge . In Fig. 2 we
ha e plo ed nea he jamming ansi ion as a unc ion o
c o one Canon CLC700 (100% SAC, in his comme -
cial one he addi i e is TiO2) and o an expe imen al
one wi h simila pa icle size (7:8m) bu only 32%
SAC. The jamming ansi ion is discussed in de ail in a
p e ious wo k [14]. As seen in he da a depic ed in Fig. 2, i
was gene ally obse ed ha in a ange o e y small
s esses he s c ela ionship ollows a c i ical-like
unc ional o m c/J, p edic ed by simula ions
[15] and eminiscen o an equilib ium c i ical phenomena.
In ou new expe imen al s udy consolida ion s esses
la ge han Wa e applied by a downwa ds di ec ed gas
low. cis hus inc eased up o cWp. We see in
Fig. 2 ha he inc emen o as cis inc eased de ia es
om he c i ical-like powe law, and a a c i ical s ess
c010 Pa c osses o e o a loga i hmic law ’J
log ^c(^cc=c0) ema kably simila o he empi i-
cal equa ion usually employed o desc ibe he compac ion
o g anula ma e ials such as soils [2] in he ea angemen
egime. Likely a ^c’1clus e s ha e eached hei closes
andom packing (
RCP). The c osso e o he loga i hmic
law occu s o olume ac ions o clus e s smalle han
0.64 ( andom close packing o noncohesi e ha d sphe es)
as i migh be expec ed om he exis ence, al hough small,
o in e clus e cohesi eness. Fo example,
RCP ’0:53
FIG. 2. Pa icle olume ac ion as a unc ion o he consoli-
da ion s ess. Da a ob ained by consolida ing he powde allow-
ing i o se le unde upwa d di ec ed gas lows (solid sym-
bols) a e join ly plo ed wi h da a ob ained by consolida ing
he powde by means o a downwa d di ec ed gas low (open
symbols). Da a co espond o comme cial one Canon CLC700
(100% SAC) and o an expe imen al one wi h educed su ace
addi i e co e age (32% SAC) and simila pa icle size. The lines
co espond o he powe law c/J ha i s o he da a
in a ange c&c010 Pa and o a loga i hmic i J’
logc=c0 ha i s o he da a in he ange c*c0. Inse :
da a in he loga i hmic egion o one s wi h he same SAC
(32%) bu di e en pa icle size (indica ed).
PRL 94, 075501 (2005) PHYSICAL REVIEW LETTERS week ending
25 FEBRUARY 2005
075501-2
0:02 o clus e s o p ima y pa icle size om 7.8 o
19:1mand 32% SAC (see Fig. 4 in Re . [14]), ma ching
he epo ed alue [16] o sphe es o equi alen size o
hese clus e s (kdp’50 m). The inse o Fig. 2 shows
da a o hese one s wi h he same SAC (32%) bu di e -
en pa icle size. I is obse ed ha ’0:04 is almos
independen o pa icle size, sligh ly dec easing when he
SAC is inc eased (see main g aph). Thus he co ela ion
be ween and Bogseems o be a leas a second o de
e ec (no e ha a dec ease o dp om 19.1 o 7:8m
co e s a wide ange o Bog). Howe e , he da a epo ed in
he enginee ing li e a u e e eal clea co ela ions be ween
@1==@ log ^cand he pa icle olume ac ion
o he lowes consolida ion s a e. We plo in he inse o
Fig. 3 he da a o he same one s (32% SAC) o 1= s
^c, which a e also well i ed o a loga i hmic law 1= ’
1=Jlog ^c. The main g aph shows as a unc ion
o Bog(calcula ed assuming F0F dW). A powe law
’0:1Bo0:21
gis clea ly seen. The ex apola ed alue o
Bog1(’0:1) ma ches he ypical alue epo ed
o noncohesi e g anula ma e ials (Bog&1) such as
sand [3]. Wha is he physical o igin o his law? In a
i s o de app oach we may app oxima e 1=’
@1==@ log ^cJlog ^c’1=J2log ^c, whe e
’@=@ log ^cJ. Thus ’1=J2. Le us w i e
J’
RCPc, whe e cN=3D3is he pa -
icle olume ac ion wi hin each clus e . Then ’
1=
RCP2Bo62D=D2
g1=
RCP2Bo0:22
g, whe e
we ha e used he clus e size limi c i e ion (Bog
D2), and D2:5. Clus e s beha e as low cohesi e
e ec i e sphe es and hus
RCP will be almos independen
on Bog. Using ’0:04 and
RCP ’0:53 o he one s
wi h 32% SAC, we would p edic ’0:14Bo0:22
g, in good
ag eemen wi h he expe imen al esul . We plo in Fig. 4
he da a o s Bog o o he one s wi h SAC >30% o
which we admi F0F dW. The new da a also scale wi h
Bogin acco dance wi h he p edic ed law. In summa y, his
law emphasizes he undamen al ole o size and ac al
s uc u e o he jammed clus e s on he dis ibu ion o
oids o be illed in he compac ion p ocess. The case o
highly cohesi e powde s (SAC <30%) needs, howe e ,
addi ional discussion. In he inse o Fig. 4 we include da a
o highly cohesi e one s wi h only 20% SAC using F0
F dW o calcula e Bog. The pa icle size o hese one s is
’7mand he base polyme is polyes e ea ed wi h
di e en amoun s o a c oss-linking agen (gel) ha p o-
duces a sligh inc ease o he polyme ha dness (in any case
small compa ed wi h he e ec o silica). The da a de ia e
clea ly om he scaling law, showing unexpec edly la ge
alues o ha mus indica e he exis ence o la ge clus-
e s. Mo eo e , in spi e o he simila alues o Bog(same
pa icle weigh and same an de Waals o ce: A’1019 J,
FIG. 3. Comp ession index e sus he g anula Bond numbe
o one s wi h he same su ace addi i e co e age (32% SAC)
and a ying pa icle size (indica ed). The con inuous line is a
powe law i o he da a (’0:1Bo0:21
g). Inse : in e se o
pa icle olume ac ion e sus he consolida ion s ess, whe e
lines a e loga i hmic i s o he da a (c010 Pa).
FIG. 4. Comp ession index e sus g anula Bond numbe o
one s wi h di e en pa icle size and % SAC (indica ed). In he
main g aph Bogis calcula ed assuming ha he in e pa icle
a ac i e o ce in luidiza ion is he an de Waals o ce o
SAC >30%, while con ac memo y (see ex ) is conside ed o
one s wi h SAC <30%. In he inse he an de Waals o ce is
used o all he one s. Fo one s wi h 20% and 100% SAC, he
e ec o inc easing he size o silica nanopa icle addi i es om
8 o 50 nm has been es ed. Addi ionally, ou di e en amoun s
o c oss-linking agen in he pa en polyme ( om 0% o 45%)
ha e been used in hese one s. The con inuous lines a e powe
law i s o he da a, only o one s wi h SAC >30% in he inse
(/Bo0:24
g) and o he whole se o one s in he main g aph
(/Bo0:25
g).
PRL 94, 075501 (2005) PHYSICAL REVIEW LETTERS week ending
25 FEBRUARY 2005
075501-3
da’0:2m), he e is a clea di e ence be ween he
comp ession indexes o one s wi h di e en gel con en .
Silica addi i es imp o e powde lowabili y mainly by
inc easing he local ha dness Hon he con ac as i is
p o en by he be e lowabili y o one s blended wi h
200 nm silica agglome a es, o simila size o he ypical
polyme aspe i y size. The s ong a ac i e o ces be ween
loaded ine pa icles cause plas ic de o ma ion o con ac s,
leading o a ele an inc ease o he adhesi e o ce wi h he
applied load [17]. Tone s ei he wi hou o wi h low pe -
cen age o he ha d silica addi i e ha e a e y poo low-
abili y because i is di icul o b eak in e pa icle con ac s
be ween p e iously loaded pa icles, i.e., he in e pa icle
con ac s in highly cohesi e powde will p ese e he mem-
o y o he ini ial loaded s a e. I is likely ha agmen s ha
ea lie exis ed as agg ega es in he loaded powde pe sis in
luidiza ion gi ing ise o la ge clus e s o s ongly adhe ed
pa icles (a simila phenomenon has been ecognized in
agmen a ion o colloidal suspensions o s ongly cohe-
si e pa icles [18]). Thus he a ac i e o ce be ween
clus e ed pa icles in luidiza ion o hese powde s mus
be much la ge han he an de Waals o ce. The inc eased
con ac ha dness by silica addi i e educes he a e o
inc ease o he adhesi e o ce wi h load [17], hus allowing
o an easy b eaking o in e pa icle con ac s by an ex e nal
ene gy sou ce (such as gas luidiza ion), and he e o e
imp o ing lowabili y. Fo hese low cohesi e one s he
an de Waals o ce was indeed a good app oxima ion o
he in e pa icle a ac i e o ce in luidiza ion. Fo one s
wi h only 20% SAC many con ac s a e be ween polyme
su aces. The a e o inc ease o he adhesi e o ce wi h
load inc eases as His inc eased [17], which means ha
con ac s wi h smalle ha dness will gi e ise o la ge
clus e s and hus o la ge alues o . (This explains he
e ec o gel seen on .) The p oblem is, How can we
es ima e Bog o hese highly cohesi e one s wi h ha d-
ness dependen memo ies? One possibili y is o use he
c i e ion o clus e limi size in luidiza ion, BogD2,
whe e D’2:5acco ding o sedimen a ion es s [9], and
may be ob ained om he in o ma ion on he ini ial
jammed s a e: D=3J=
J. In his equa ion he
mos impo an and a iable pa ame e is J, which is
accu a ely ob ained om he loga i hmic i equa ion J
log ^c, while
Jcan be expec ed o show a small
a ia ion be ween 0:5(10% SAC) and 0:55 (abo e
60% SAC). Fo low cohesi e one s (SAC *30%), he
es ima ed Bogin his way is simila o he p e iously
calcula ed one assuming F0F dW, bu , o he highly
cohesi e ones, Bogis signi ican ly la ge as we an icipa ed.
The main g aph o Fig. 4 includes he da a o 20% and
10% SAC one s wi h he new es ima ion o Bog, showing
a good i o he p edic ed powe law. We can also dis-
c imina e now be ween he comp ession indices o 20%
SAC one s in he base o he memo y e asing e ec o he
c oss-linking agen added o he pa en polyme .
To conclude, we ha e in es iga ed he compac ion be-
ha io o cohesi e pa icles which a e clus e ed in he
luidlike egime. The numbe o pa icles in ou DLA
clus e s, and hus he clus e packing ac ion (c), is
con olled by he a io o he in e pa icle a ac i e o ce
o pa icle weigh Bog(c’Bo0:1
g). In he close icini y
o jamming and abo e a c i ical s ess c010 Pa, he e
is a c osso e o he loga i hmic law 1= ’1=J
logc=c0. Expe imen al s udies in he enginee ing
li e a u e ha e usually shown his beha io , and he com-
p ession index has been co ela ed o he minimum
pa icle olume ac ion (co esponding o he smalle
s ess applicable, ypically 10 kPa); he la ge i is, he
la ge he . Ou luidiza ion echnique allows o a s udy
o compac ion beha io om jus a ew pascals, hus we
a e able o ela e o he uly ini ial s a e o ea ange-
men . The ini ial dis ibu ion o oids o be illed, which
ules he compac ion p ocess, is mainly de e mined by he
in e nal packing s uc u e o clus e s ha a e jammed a he
ini ial unconsolida ed s a e. In a i s o de analysis we
es ima e /1=c
J2’Bo0:2
g, in ag eemen wi h ou mea-
su emen s. Ou analysis implies also ha , due o he high
plas ici y o in e pa icle con ac s in highly cohesi e pow-
de s, hese powde s mus e ain memo y in luidiza ion o
p e ious loaded s a es; when jammed, e y la ge clus e s
p oduce e y po ous ini ial s a es and as a consequence
la ge alues o he comp ession index.
We acknowledge he Xe ox Founda ion and he
Spanish Minis e io de Ciencia y Tecnologia (Con ac
No. BMF2003-01739).
[1] E. E. Walke , T ans. Fa aday Soc. 19, 73 (1923).
[2] J. A kinson, The Mechanics o Soils and Founda ions
(McG aw-Hill, London, 1993).
[3] P. E esque, Poud es and G ains 10, 6 (1999).
[4] D. Poquillon e al., Powde Technol. 126, 65 (2002).
[5] J. H. Pa k and T. Koumo o, J. Geo ech. Geoen i on. Eng.
130, 223 (2004).
[6] J. M. Val e de e al., Phys. Re . Le . 86, 3020 (2001).
[7] K. Rie ema, The Dynamics o Fine Powde s (Else ie ,
London, 1991).
[8] H. K upp, Ad . Colloid In e ace Sci. 1, 111 (1967).
[9] A. Cas ellanos e al., Phys. Re . E 64, 041304 (2001).
[10] Y. Kan o and T. A. Wi en, J. Phys. Le . 45, L675 (1984).
[11] S. Manley e al., Phys. Re . Le . 93, 108302 (2004).
[12] Y. Kan o and I. Webman, Phys. Re . Le . 52, 1891
(1984).
[13] J. M. Val e de e al., Re . Sci. Ins um. 71, 2791 (2000).
[14] J. M. Val e de e al., Phys. Re . Le . 92, 258303 (2004).
[15] C. S. O’He n e al., Phys. Re . E 68, 011306 (2003).
[16] R. Y. Yang e al., Phys. Re . E 62, 3900 (2000).
[17] M. A. S. Quin anilla e al., Phys. Re . E 64, 031301
(2001).
[18] Y. Ta ek e al., Powde Technol. 143–144, 117 (2004).
PRL 94, 075501 (2005) PHYSICAL REVIEW LETTERS week ending
25 FEBRUARY 2005
075501-4