Вы находитесь на странице: 1из 7

Apparent viscosity of suspensions of rods using falling ball rheometry

Robert L. Powell, Thomas G. Morrison, and William J. Milliken


Citation: Phys. Fluids 13, 588 (2001); doi: 10.1063/1.1345717
View online: http://dx.doi.org/10.1063/1.1345717
View Table of Contents: http://pof.aip.org/resource/1/PHFLE6/v13/i3
Published by the American Institute of Physics.

Additional information on Phys. Fluids


Journal Homepage: http://pof.aip.org/
Journal Information: http://pof.aip.org/about/about_the_journal
Top downloads: http://pof.aip.org/features/most_downloaded
Information for Authors: http://pof.aip.org/authors

Downloaded 11 May 2013 to 131.211.208.19. This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://pof.aip.org/about/rights_and_permissions

PHYSICS OF FLUIDS

VOLUME 13, NUMBER 3

MARCH 2001

Apparent viscosity of suspensions of rods using falling ball rheometry


Robert L. Powella) and Thomas G. Morrisonb)
Department of Chemical Engineering and Materials Science, University of California, Davis,
California 95616

William J. Milliken
Chevron Petroleum Technology Company, P.O. Box 6019, San Ramon, California 94583

Received 19 April 1999; accepted 4 December 2000


Falling ball rheometry measurements of the apparent relative viscosity of suspensions of randomly
oriented rodlike particles are reported for rods having a wide range of aspect ratios and
concentrations. These data, combined with those from earlier studies Milliken et al., J. Fluid Mech.
202, 217 1989; Powell et al., J. Rheology 33, 1173 1989, provide a master curve describing the
apparent viscosity in terms of a single parameter, nL 3 , where L is the length of the particle and n
is the number density. 2001 American Institute of Physics. DOI: 10.1063/1.1345717

INTRODUCTION

EXPERIMENTAL

The rheology of suspensions of rodlike particles receives


considerable attention in the literature due to such diverse
applications as the manufacture of short fiber composites and
the processing of pulp used in making paper see the reviews
of Powell1 and Milliken and Powell2. In these systems, the
particles are generally large enough that Brownian forces are
negligible and viscous forces are assumed to dominate. In a
series of studies, falling ball rheometry has been used to
determine the viscosity of suspensions of rods37 by observing the trajectory of a sedimenting, test sphere. The terminal
velocity of the sphere, , is used to calculate the drag and to
infer an apparent or effective viscosity, , which is calculated using Stokess equation with the appropriate corrections for wall effects4

The apparatus used for these experiments is described in


detail in previous papers.3,4 The suspensions were composed
of individually prepared rods and a density matched suspending fluid. Five different populations of rods were used.
The rods were made of either polymethyl methacrylate
PMMA Thermo Plastics Inc., Cleveland, OH or polystyrene Dow Chemical, Walnut Creek, CA and all of the rods
were individually cut to the desired lengths. A summary of
the dimensions of the rods and the statistical properties of
randomly chosen samples is provided in Table I.
Two different suspending fluids were used: one each to
match the densities of the PMMA and polystyrene rods. For
the PMMA rods, the fluid was composed of a mixture of
polyalkylene glycol UCON-HB-9500, Union Carbide, Danbury, CT, 1,1,2,2-tetrabromoethane Eastman Kodak Co.,
Rochester, NY, and a small quantity of Tinuvin 328 CibaGeigy, Ashley, NY. The mixture was density matched to
that of the PMMA rods, 1,180 kg/m3, and had a viscosity of
11.42 Pa s at 20.0 C. For the polystyrene rods, an 84%/16%
by weight mixture of two polyalkylene glycols, UCON 75H-90000 and UCON 50-HB-5100 was used to match the
polystyrene rod density of 1080 kg/m3. Initially after preparation, this mixture had a viscosity of 18.2 Pa s at 20.6 C.
Over the 18-month period during which these experiments
were performed, the viscosity decreased to 16.2 Pa s. However, since the viscosity of the suspending fluid was regularly monitored and our results are all in terms of the relative
viscosity, this change results in no errors in the reported
values.
The apparatus consisted of a cylindrical glass column
540 mm in height and 142 mm in diameter placed in an
insulated water tank, the temperature of which was controlled by a constant temperature circulator Model 2067,
Forma Scientific, Marietta, OH. The temperature of the suspensions was measured to within 0.05 C at the beginning
of an experiment and after each set of ten measurements.
Experiments were performed only when the measured temperature was within 0.20 C of the desired set point.

2
gd ball
b
d ball
d ball
12.104
2.09
18
D col
D col

0.95

d ball
D col

d ball 5
.
D col

Here, g is the gravitational acceleration, b and d ball are the


density and diameter of the falling ball, respectively, and
D col is the diameter of the cylindrical container containing
the suspension.
Two key assumptions inherent in falling ball rheometry
are that the suspension exhibits a linear stressstrain relationship and that the sphere disturbs the microstructure of the
suspension by only a small amount. These have been validated experimentally,4 although computational studies have
suggested otherwise Claeys and Brady8.
In this paper, experimental results using suspensions of
fibers having a number of different aspect ratios and over a
wide range of concentrations are reported. These data augment those presented previously.3,4
a

Author to whom correspondence should be addressed.


Present address: Arco Products Co., 1990 West Crescent Ave., Anaheim,
CA 92801.

1070-6631/2001/13(3)/588/6/$18.00

588

2001 American Institute of Physics

Downloaded 11 May 2013 to 131.211.208.19. This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://pof.aip.org/about/rights_and_permissions

Phys. Fluids, Vol. 13, No. 3, March 2001

Viscosity of rod suspensions by falling ball rheometry

589

TABLE I. Physical and geometrical properties of the rods.

Material
Polymethyl
Methacrylate

Polystyrene

Elastic
modulus N/m2

Density
g/cm3

(2.4 3.4)109 b

1.1818

3.1109 c

1.08

Aspect
ratio
Stan. dev.a

Nominal
length cm
Stan. dev.

Nominal
diameter cm
Stan. dev.

5.00

1.59

0.318

0.04
30.7
1.00
30.7
1.23
47.8
2.54
48.9
2.92

0.004
4.76
0.004
3.01
0.02
4.47
0.02
3.55
0.09

0.002
0.158
0.006
0.098
0.004
0.098
0.005
0.074
0.006

Standard deviation, based on sample of 25 to 40 rods.


From Weast 1986 Ref. 14.
c
From Perry 1984 Ref. 15.
b

An experiment consisted of having a spherical brass ball


bearing Anti-Friction Bearing Manufacturers grade 200,
Hoover Universal Co., Ann Arbor, MI fall through the suspension and measuring its sedimentation velocity. The balls
were placed in the water bath and thermally equilibrated to
the temperature of the suspension prior to the experiments.
The diameter of each ball was measured to an accuracy of
0.003 mm and the weight of each was determined to
0.0002 g for calculation of the viscosity. An eddy current
system4,9 was used to measure the velocity of the falling
balls. The suspensions were mixed before each individual
experiment.
RESULTS

found previously,3 the data exhibit two distinct regimes. The


first is a lower concentration region in which apparent viscosity varies linearly with the volume fraction. The second is
a higher concentration region in which the apparent viscosity
is proportional to the cube of the volume fraction. A relatively sharp change between the two regimes is also seen.
This behavior is observed for all aspect ratios studied which
suggests that a single relationship exists between viscosity
and concentration. If the apparent viscosity is graphed as a
function of nL 3 Fig. 3, all of the data, except for those
representing the smallest aspect ratios, collapse onto a single
curve. The data describing this master curve cover two decades of concentration: 1.25nL 3 131.
A least squares fit of the data in Fig. 3 to equations of the

The apparent viscosities of the suspensions with particles having a r 48.9 are shown in Fig. 1 as a function of
d (d /L) for 0.00250.0276. As expected, Fig. 1
ball
shows that the rods augment the apparent viscosity of the
suspension by a substantially larger amount than do spheres
at equivalent volume fractions.1 For d greater than about
0.35, the apparent viscosity at each volume fraction is approximately constant and an average for each concentration
can be determined. For smaller test sphere diameters, there is
a decrease in the measured relative viscosity small ball
effect.5 Although this behavior is seen for all concentrations and aspect ratios examined, it is most evident at volume
fractions of 0.0135 and 0.009. In Tables IIV, the average
values of the viscosity are reported for all the suspensions
used in this study. Values were determined by averaging the
apparent relative viscosities for those in which there is no
variation of r with d .
DISCUSSION

The data presented here, along with those of Milliken


et al.3 and Powell et al.,4 allow the dependence of the apparent viscosity on both the aspect ratio and the concentration to
be determined. Indeed, that is the principal goal of this study.
Figure 2 shows the dependence of the apparent specific viscosity, sp , on the volume fraction, , for a r 30.7. As

FIG. 1. The relative viscosity as a function of dimensionless ball diameter


for suspensions of fibers with an aspect ratio of 48.9. The for a volume
fraction of 0.0276, the for a volume fraction of 0.0245, the for a
volume fraction of 0.0218, the for a volume fraction of 0.0135, the
for a volume fraction of 0.009, the for a volume fraction of 0.005, and
the for a volume fraction of 0.0025. One representative set of error bars
95% confidence limits is shown for each concentration.

Downloaded 11 May 2013 to 131.211.208.19. This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://pof.aip.org/about/rights_and_permissions

590

Phys. Fluids, Vol. 13, No. 3, March 2001

Powell, Morrison, and Milliken

TABLE II. Apparent relative viscosities and 95% confidence intervals for a r 30.7, L3.01 cm. The number in parentheses under each viscosity value is the
number of experiments used in calculating the average viscosity.

d/L0.623
d/L0.519
d/L0.415
d/L0.363
d/L0.311
d/L0.259
d/L0.208
d/L0.156
Total average

0.009
10.8

0.0218
26.2

0.0325
39.0

1.5500.095
16
1.6240.085
14
1.6010.108
10
1.6100.091
16
1.5460.090
10
1.5390.078
10
1.4420.108
8
1.2520.092
7
1.5940.042
56

2.3190.125
10
2.3320.149
20
2.2300.149
23
2.3440.162
18
2.2600.160
19
2.1110.093
10
2.0220.118
11

2.8610.221
15
2.9050.207
15
2.8380.268
14
2.7790.219
14

2.2920.065
90

2.8460.103
58

nL 3
0.050
60.0

0.070
84.0

0.080
96.0

0.090
108.0

4.0100.326
14
4.0830.211
13
4.0600.303
15
3.9010.358
19
3.9390.357
16
4.0530.446
17
3.5470.325
22
3.0260.404
7
3.9940.133
94

7.1550.815
19
7.5120.887
19
6.5920.783
15
6.3280.844
20
7.1730.939
14
6.3500.799
19
5.6340.752
18
4.5790.698
9
6.9490.361
87

12.272.49
11
9.5831.709
8
9.3521.093
10
9.8082.842
9

12.272.49
11
11.851.67
9

9.5730.980
27

12.081.42
20

form sp c(nL 3 ) b where c and b are constants, finds that at


the lower concentrations, nL 3 50, if the data from the two
smallest aspect ratios are excluded, there is a linear relationship between sp and nL 3 ,

sp 0.050 nL 3 1.00.

At higher concentrations there is a roughly cubic relationship

sp 0.000 045 nL 3 2.7.

The slight difference from the strictly cubic dependence between sp and nL 3 found for individual aspect ratios and Eq.
3 is due to differences in the changeover point from the

0.1094
131.3

27.155.77
11
23.036.28
15
24.717.44
15

24.753.49
41

linear to non-linear regime for suspensions of the different


aspect ratio rods. Equations 2 and 3 describe quite simply
the dependence of the apparent viscosity on the aspect ratio
and concentration through one parameter, nL 3 . Even the data
for a r 5 and 10 suspensions deviate from these curves by
less than 15%. To our knowledge, this is the first experimental verification of the fact that the apparent viscosity of nonBrownian rod suspensions can described through a single
dimensionless parameter, nL 3 .
Although the experimental data for low concentrations
exhibit a linear dependence on nL 3 the slope of the vs
line does not agree with theoretical predictions. The dilute

TABLE III. Apparent relative viscosities and 95% confidence intervals for a r 48.9. The number in parentheses under each viscosity value is the number of
experiments used in calculating the average viscosity.

d/L0.534
d/L0.445
d/L0.356
d/L0.311

nL 3
0.0135
41.1

0.0218
66.4

2.4340.175
15
2.4250.234
15
2.4400.219
15
2.3770.220
15
2.4980.241
14
2.2810.261
16
2.0520.222
12

3.0810.233
14
3.0350.277
15
3.1680.284
10
3.1320.365
14
3.0860.354
14
2.8840.289
14
2.5030.262
14
2.1380.252
12

4.7980.535
21
4.8410.500
18
4.9860.564
17
4.6120.553
18
4.7710.641
15
4.5540.646
18
4.0100.598
14
3.5290.475
10

2.4340.088
74

3.0950.122
67

4.8010.227
89

0.0025
7.6

0.0050
15.2

0.0090
27.4

1.3380.048
9
1.3820.066
10
1.3860.114
10
1.3820.102
10

1.6670.118
12
1.6790.149
10
1.7490.101
16
1.7410.138
13

1.3730.038
39

1.7140.055
51

d/L0.267
d/L0.222
d/L0.178
d/L0.133
Total average

0.0245
74.6

0.0276
84.0

5.9720.702
15
6.2280.692
15
5.5630.608
15

6.9801.158
11
8.6161.410
15
8.4081.048
15

5.9210.355
45

8.0860.675
39

Downloaded 11 May 2013 to 131.211.208.19. This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://pof.aip.org/about/rights_and_permissions

Phys. Fluids, Vol. 13, No. 3, March 2001

Viscosity of rod suspensions by falling ball rheometry

591

TABLE IV. Apparent relative viscosity values and 95% confidence intervals for rods of aspect ratio 5.0 and 30.7. The number in parentheses under
each viscosity value is the number of experiments used in calculating the
average viscosity. The volume fraction for both is 0.05. The suspensions
with aspect ratio 30.7 reported in this table have a length of 4.76 cm and a
ratio D cof /L that is smaller than considered necessary for measuring the
viscosity this issue is addressed in the Discussion.
a r 30.7
(L4.76 cm)

a r 5.0
d/L0.598
d/L0.500
d/L0.397
d/L0.348

1.2410.020
20
1.2310.026
19
1.2700.027
30
1.2470.025
19

d/L0.200
d/L0.167
d/L0.133
d/L0.117
d/L0.100
d/L0.084

Total average

1.2500.013
88

3.1120.475
13
2.9040.324
14
2.9700.445
10
3.1530.750
9
3.1500.311
10
2.9030.382
10

FIG. 2. The specific viscosity as a function of volume fraction for rods of


aspect ratio 30.7. The solid line in the figure has a slope of 1.0; the dashed
line has a slope of 3.0.

3.0260.146
66

theory of Batchelor10 predicts a coefficient of /900.035


whereas the best fit line shown in Fig. 3 has a value of 0.2,
over five times greater. The cause of this is not at all clear.
Harlen et al.11 suggest that this difference is in part due to
the small size of the falling ball relative to the particles and
to particleparticle contacts. Claeys and Brady8 suggest that
this discrepancy is due to changes in the fiber distribution as
a consequence of the flow induced by the falling ball. It is
also possible that the explanation is simply due to the fact

that the suspensions in the experiments are not dilute. The


predictions of Batchelor are for dilute suspensions with
nL 3 1 whereas, the low concentration data in this study fall
in the range 1nL 3 50.
Earlier studies,3,4 identified a significant decrease in the
apparent viscosity if the diameter of the falling ball was
smaller than some critical value the small ball effect.
Figure 1 shows a corresponding decrease in the apparent
viscosities reported here. Following Milliken et al.,4 the data
are normalized using

TABLE V. Apparent relative viscosities and 95% confidence intervals for suspensions with rods of aspect ratio 47.8. The number in parentheses under each
viscosity value is the number of experiments used in calculating the average viscosity. The data from this table are not included in Fig. 3 because the ratio
D col /L is smaller than considered necessary for measuring the viscosity this issue is addressed in the Discussion.

d/L0.400
d/L0.334
d/L0.267
d/L0.233
d/L0.200
d/L0.167
d/L0.133

0.0218
63.3

0.0302
87.8

0.035
101.7

2.8510.267
20
2.6540.305
20
2.9290.350
23
2.5030.251
19
2.5580.391
20
2.4270.282
15
2.0190.192
9
1.7430.179
9
1.6770.135
10

4.1890.435
18
3.9570.544
20
4.0730.450
15
3.8730.477
15
3.5170.504
19
3.3640.515
20
2.9690.372
9
2.4130.222
10
1.9620.308
10

4.9322.291
5
5.9562.532
5
4.6160.940
10
5.2600.958
10

1.4170.066
15
1.4220.096
10
1.3140.070
10
1.2240.093
10
1.1580.043
10

2.0790.160
25
2.2220.204
25
2.1780.144
20
2.0750.168
10
1.8950.145
27
2.0050.175
17
1.7370.228
15
1.7170.133
19
1.3890.057
10

1.4330.021
85

2.1480.101
80

2.8170.171
63

4.0260.224
68

5.1070.549
30

0.0050
14.5

1.2490.030
19
1.2200.056
14
1.2110.035
14
1.2380.077
11
1.2000.059
13
1.1730.040
14
1.1470.052
14

1.4380.076
20
1.4610.128
20
1.4150.081
20

1.2310.021
58

d/L0.100
d/L0.067
Total average

nL 3
0.0135
39.2

0.0025
7.26

4.0961.452
9

Downloaded 11 May 2013 to 131.211.208.19. This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://pof.aip.org/about/rights_and_permissions

592

Phys. Fluids, Vol. 13, No. 3, March 2001

Powell, Morrison, and Milliken

FIG. 3. The specific viscosity as a function of nL 3 . The are for aspect


ratio 5.0, the are for aspect ratio 10, the are for aspect ratio 48.9,
the are for aspect ratio 30.7, and the are for aspect ratio 19.8. The
two solid lines in the figure have slopes of 1.0 and 3.0; the dotted line is
drawn at the intersection of the two solid lines.

A S
,
sus S

where A is the apparent viscosity for a given falling ball


size and sus is the suspension viscosity as calculated from
the data for the larger test ball sizes. Figure 4 plots the
normalized viscosity as a function of d ball /L. For the suspensions at all concentrations and aspect ratios, the normalized
viscosity decreases when the falling ball diameter is less than
about 0.3. This decrease is also seen in the numerical results
of Harlen et al.11 However, in that study the apparent viscos-

ity peaks at d ball /L1.5 and reaches a constant value for


d ball /L1. In these experiments a constant viscosity is found
for d ball /L0.3.
In Milliken et al.,3 the suspension viscosity was found to
be independent of the size of the container column holding
the suspension if the ratio D col /L3.2. That conclusion was
drawn by varying the diameter of the column without changing the dimensions of the rods and is in general agreement
with results obtained in shearing flows.12,13 Here, the effect
of changing the dimensions of the rods while holding the
column diameter is considered. Two sets of suspensions with
virtually equal aspect ratios but different particle dimensions
were studied.
In the first case, two sets of rods with aspect ratio of
about 48 were examined. For the first set, a r 47.8 and the
mean length was 44.7 mm, making D col /L3.2. The second
set, shown in Fig. 1, had an a r 48.9 but a length of 35.5
mm making D col /L4.0. For the four concentrations tested,
the suspensions with rods having a r 48.9 the shorter rods
had a higher relative viscosity by an average of almost 25%
cf. Tables III and V. This difference is larger than can be
explained based solely on the difference in aspect ratio. In
addition, in the high concentration regime, the viscosity of
the suspensions of a r 47.8 rods the longer rods increases
as the square of the volume fraction rather than as the cube
as is found in all of the other suspensions studied both here
and previously.3,4
In the second case, we examined two sets of rods with
aspect ratio 30.7. One had a length of 30.1 mm the other had
a length of 47.6 mm. Here, a decrease in the viscosity of
about one-third is found as D col /L is decreased from 4.7 to
3.0 cf. Tables II and IV. Hence, the effect of container
walls is significant at larger values of D col /L in this study
than that found previously; Milliken et al. found that a ratio
of D col /L3.2 was sufficient to insure that the container
boundary was not affecting measurements. It is important to
emphasize here that the wall effects found for these long
particles (D col /L3.2) results from the container boundaries
influencing the motion of particles and not due to Faxen type
effects for which we have adequately accounted.4
CONCLUSIONS

The data presented here augment earlier studies using


falling ball rheometry to examine the rheology of suspensions of rods. These data represent many experimental repetitions and are consistent with previous findings. These results now represent a comprehensive set of carefully
performed experiments of fiber suspensions spanning a wide
range of concentrations and aspect ratios and provide a data
set for comparisons to theoretical predictions and numerical
simulations. The results of this study provide a benchmark
against which theories for the rheology of randomly oriented
suspensions can be tested.
ACKNOWLEDGMENTS
FIG. 4. The normalized viscosity as a function of d ball /L. The data are those
taken from the two aspect ratios reported in this study, plus the data of
Milliken et al. Ref. 3 and Powell et al. Ref. 4.

This work was supported by a grant from the National


Science Foundation CBT-8718261. Equipment used for

Downloaded 11 May 2013 to 131.211.208.19. This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://pof.aip.org/about/rights_and_permissions

Phys. Fluids, Vol. 13, No. 3, March 2001

these studies was loaned to the Rheology Laboratory at the


University of California at Davis by Sandia National
LaboratoriesAlbuquerque through the efforts of Lisa A.
Mondy.
1

R. L. Powell, Rheology of suspensions of rodlike particles, J. Stat.


Phys. 62, 1073 1991.
2
W. J. Milliken and R. L. Powell, Short fiber suspensions, in Flow and
Rheology in Polymer Composites Manufacturing, edited by S. G. Advani
Elsevier, Amsterdam, 1994.
3
W. J. Milliken, M. Gottlieb, A. L. Graham, L. A. Mondy, and R. L.
Powell, The viscosity volume fraction relation for suspensions of randomly oriented rods by falling ball rheometry, J. Fluid Mech. 202, 217
1989.
4
R. L. Powell, L. A. Mondy, G. G. Stoker, W. J. Milliken, and A. L.
Graham, Development of a falling ball rheometer with applications to
opaque systems: Measurements of the rheology of suspensions of rods, J.
Rheol. 33, 1173 1989.
5
W. J. Milliken, L. A. Mondy, M. Gottlieb, A. L. Graham, and R. L.
Powell, The effect of the diameter of falling balls on the apparent viscosity of suspensions of spheres and rods, PCH, PhysicoChem. Hydrodyn. 11, 341 1986.
6
A. L. Graham, L. A. Mondy, M. Gottlieb, and R. L. Powell, Rheological

Viscosity of rod suspensions by falling ball rheometry

593

behavior of a randomly oriented suspension of rods, Appl. Phys. Lett. 50,


127 1987.
7
L. A. Mondy, T. G. Morrison, A. L. Graham, and R. L. Powell, Measurements of the viscosities of suspensions of oriented rods using falling
ball rheometry, Int. J. Multiphase Flow 16, 651 1990.
8
I. L. Claeys and J. F. Brady, The dynamics of prolate spheroids in Stokes
flow. Part 2: Statistically homogeneous dispersions, J. Fluid Mech. 251,
443 1993.
9
T. G. Morrison, Experimental investigation of the mechanics of suspensions of elongated particles using falling ball rheometry, M.S. thesis,
University of California, Davis, 1989.
10
G. K. Batchelor, The stress generated in a non-dilute suspension of elongated particles, J. Fluid Mech. 46, 813 1971.
11
O. G. Harlen, R. R. Sundararajakumar, and D. L. Koch, Numerical simulations of a ball falling through a suspension of neutrally buoyant fibers,
J. Fluid Mech. submitted.
12
M. A. Nawab and S. G. Mason, The viscosity of dilute suspensions of
thread-like particles, J. Phys. Chem. 62, 1248 1958.
13
W. R. Blakney, Viscosity of suspensions of straight, rigid rods, J. Colloid Sci. 22, 324 1966.
14
R. C. Weast, ed., Handbook of Chemistry and Physics CRC Press, Boca
Raton, FL, 1986.
15
R. H. Perry and D. Green, ed., Perrys Chemical Engineers Handbook,
6th ed. McGraw-Hill, New York, 1984.

Downloaded 11 May 2013 to 131.211.208.19. This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://pof.aip.org/about/rights_and_permissions

Вам также может понравиться