148

Part. Part. Syst. Charact. 12 (1995) 148-157

Comparative Test of Methods to Determine Particle Size and Particle Size Distribution in the Submicron Range
Heinz Lange *
Dedicated to Professor Klaus Elgeti on the occasion of his 60th birthday
(Received: 7 February 1994; resubmitted: 2 January 1995)

Abstract
Among the most important characteristic properties of disperse systems such as latices, pigments, ceramic materials or drug formulations are the particle size and the particle size distribution. To measure these quantities, several methods and measuring instruments based on different physical principles are available. These include turbidimetry, dynamic and static light scattering, electron microscopy with image analysis, ultra- and disc centrifugation, light diffraction and the electrical sensing zone method. All these measuring techniques are doubtless necessary because of the large product variety and the broad particle size range. However, some problems arise if different techniques are used and the results are compared uncritically without considering to the application range and the resolution of the methods. An extensive comparative test was therefore carried out using seven latices in the submicron range with defined monomodal, bimodal and hexamodal particle size distributions. The most important methods of determining average particle size values and particle size distributions were tested and compared. Of the methods to determine only average particle sizes, turbidimetry is the most efficient, followed by dynamic light scattering with cumulants evaluation. Static light scattering only yields accurate results for small particles with narrow particle size distributions. Of the methods to determine particle size distributions, ultracentrifugation and, somewhat less, disc centrifugation and electron microscopy with image analysis are the most efficient. Dynamic light scattering only yields reliable results in the case of small particles with narrow distribution curves. Light diffraction and the electrical sensing zone method are less suitable for the submicron range.

1 Introduction and Methods

Among the most important characteristic data for disperse systems such as latices, pigments, ceramic materials or drug formulations are the particle size and the particle size distribution. In order to measure these quantities, several methods and measuring instruments based on different physical principles are available. These include turbidity measurements, dynamic and static light scattering, electron microscopy with image analysis, ultra- and disc centrifugation, light diffraction and the electrical sensing zone method. All these measuring techniques are doubtless necessary because of the large product variety and the broad particle size range of about 10' to lo6 nm. However, some problems arise if different techniques are used and the results are compared uncritically without considering the application range and resolution of the methods. In order to achieve some improvements in this field, an extensive comparative test was carried out in which the most important methods to determine average particle size values and particle size distributions were tested and compared with regard to their efficiency in the submicron range').

This test involved 17 laboratories within the Bayer Group worldwide, using all the methods mentioned above. Each laboratory was sent seven test samples (latices) with defined particle size distributions, unknown to the participants. They were asked to measure the average particle size values and particle size distributions routinely without special effort.

2 Test Latices
The test latices were chosen in such a way that the efficiency of the methods could be examined and evaluated for both small and relatively large particle diameters of about 100 and 800 nm and for a large distribution range between about 100 and 1600 nm. The efficiency of the methods was also tested for latices of different chemical composition. The test latex set therefore consisted of two almost monodisperse polystyrene latices, two bimodal polystyrene latex blends, one hexamodal polystyrene latex blend and two latex blends of polystyrene and polybutadiene and of poly(styrene-co-butadiene), polystyrene and polychloroprene. The detailed data (composition, modality, particle diameter and distribution range) of the test latices are given in Table 1.

* Dr. H. Lange, Bayer AG, ZF-TPF 2, E

[l-41.

41, 51368 Leverkusen (Ger-

many). 1) For comparative tests with particle diameters larger than 1 pm, see

0 VCH Verlagsgesellschaft mbH, D-69469 Weinheim, 1995

0934-0866/95/0306-0148 $5.00 + .25/0

Part. Part. Syst. Charact. I2 (1995) 148-157 Table 1: Test latices. test substance latex polystyrene polystyrene polystyrene polystyrene polystyrene polystyrene polystyrene polystyrene polystyrene polystyrene polystyrene polystyrene polystyrene polybutadiene poly(styreneco-butadiene) polystyrene polychloroprene composition
(070) 100 100 50 50 10 90 16.67 16.67 16.67 16.67 16.67 16.67 33.33 66.67 33.33 33.33 33.33 3 1 1 2 2 213 815 137 213 605 815 137 213 327 552 815
1,550

149

3 Results
modality
d,
U

dPfd

(nm)

(nm)
7.9 7.8

(nm)
213 815 175 794

3.1 Determination of Average Particle Size Values

7'0 7.9
8.9 7.8
7.0 7.9

The first part of this paper discusses the results of those methods which are able to determine only average particle size values and not particle size distributions. These methods are turbidimetry [5-71, dynamic light scattering with cumulants evaluation [8] and static light scattering with evaluation of the angle dependence of the scattered intensity [9]. The results of these methods are shown in Table 2 and Figures 1-6.
3.1.1 Test Latex A

9*0 10.4 7.8

599

213 2 363 80 113 200

7.9

313

Test latex A with a particle diameter of 213 nm and a very narrow monomodal particle size distribution (see thick full line in Figure 1) is characterized very well by turbidimetry and dynamic light scattering with cumulants evaluation. The values obtained by static light scattering are slightly too high (see Table 2 and Figure 1). 3.1.2 Test Latex B

132

The particle size (diameter d,) and the standard deviation ( 0 ) of the polystyrene test latices except the last constituent of latex E and the middle constituent of latex G are data from the supplier Serva, Heidelberg, based on electron microscopic measurements. The particle size of the last constituent of latex E was measured in the laboratory 11 (see Table 4) by electron microscopy and the particle size of the second constituent (polybutadiene) of latex F and of all constituents of latex G were determined in the laboratory 2 (see the same table) by ultracentrifugation [lo]. All test latices were sent to the participating laboratories as 10 ml samples with 10% solids content together with a small amount of the emulsifier K30 (sodium alkanesulfonate) for further dilution with a defined dispersion medium (0.2 g/l K30 in distilled water). dppFal is the weight-average particle diameter calculated from the d, values and the compositions.

Test latex B with a particle diameter of 815 n m and a very narrow monomodal particle size distribution (see thick full line in Figure 2) is also characterized very well by turbidimetry in laboratories 2 and 3 and by dynamic light scattering with cumulants evaluation in laboratories 4, 8 and 9. Only turbidimetry in laboratory 1 and dynamic light scattering in laboratories 5, 6 and 7 yield values which are too low (see Table 2 and Figure 2). The static light scattering measurements could not be evaluated because of the curvature of the resulting Zimm plots.
3.1.3 Test Latex C Test latex C is a blend of two latices with small particle diameters of 137 and 213 nm and a mass ratio of l : l. From this, cal was calculated as the weight-average particle diameter 175 nm (see thick full lines in Figure 3) and the Z-average as 183 nm. These average values are characterized very well by tur-

4,

Table 2: Results of the comparative test: particle size average values. method laboratory equipment test latex d,,, (nm) turbidimetry dynamic light scattering (cumulants evaluation)
1 2 3 4 5 6 7 8 9 A 213 210 220 206 211 227 226 220 217 213 214 224 246 246
~p,meas

(nm)
D 794 700 800 748 769 766 533 698 744 773 785 820

B 815
700 800 796 840 616 451 692 814 769 779 875 -

C 175
180 180 179 187 198 204 197 192 197 190 208 217 224

E 599
700 700 449 266 283 286 279 323 408 291 395

F 313

static light scattering (evaluation of the angle dependence)

10

Photometer 1 Photometer 1 Photometer 2 Goniometer 1 + Correlator Measuring instrument D1 Measuring instrument D1 Measuring instrument D1 Measuring instrument D1 Measuring instrument D1 Measuring instrument D2 Measuring instrument D3 Measuring instrument S1 Measuring instrument S2

290 252 249 262 261 265 256 236 256 341 136

150

Part. Part, Syst. Charact. 12 (1995) 148-157

-5
0
QIQ

12

E *-

8
4
0 0 400
800

-IE0

d p Inml

1200 d lnml

1600

2000

Fig. 1: Particle size distribution (mass distribution) and the measured average particle diameters dp,meas(s. Table 2) of test latex A (polystyrene; monomodal; d, = 213 nm, o = 7.9 nm).

Fig. 5: Particle size distribution (mass distribution) and the measured average particle diameters dp,meas (s. Table 2) of test latex E (polystyrene; hexamodal; d , ] = 137 nm, o, = 7.0 nm; dp,*= 213 nm, o2 = 7.9 nm; dp,3 = 327 nm, o3 = 9.0 nm; d , , = 552 nm, o4 = 10.4 nm; dp,5= 815 nm, a5= 7.8 nm; d , , = 1,550 nm, EM; mass ratio
1: 1: 1 : 1: 1: 1).

500

600

700

800 d s [ml

900

1000

Fig. 2: Particle size distribution (mass distribution) and the measured average particle diameters dp,me,,(s. Table 2) of test latex 5 (polystyrene; monomodal; d, = 815 nm, o = 7.8 nm). Fig. 6: Particle size distribution (mass distribution) and the measured average particle size diameters dP,,,,, (s. Table 2) of test latex F (polystyrene, polybutadiene; bimodal; d,,] = 213 nm, 6 ) = 7.9 nm; d , , = 363 nm, UC; mass ratio 1 : 2).

80

0,
x
I

70 60 50
40

$la-I E0

30

2o
10

bidimetry and well by dynamic light scattering with cumulants evaluation. The values obtained by static light scattering are slightly too high (see Table 2 and Figure 3).
3.1.4 Test Latex D

100

200 300 d p [nml

400

500

Fig. 3: Particle size distribution (mass distribution) and the measured average particle diameters (s. Table 2) of test latex C (polystyrene; bimodal; d , ] = 137 nm, o1 = 7.0 nm; dp,z= 213 nm, o2 = 7.9 nm; mass ratio 1 : 1).

E
-

I .

t
9W
1000

Test latex D is a blend of two latices with relatively large particle diameters of 605 and 815 nm and a mass ratio of 1 :9. From this, the weight-average particle diameter dp,cal was calculated as 794 nm (see thick full lines in Figure 4) and the Z-average as 799 nm. These average values are characterized very well by turbidimetry in laboratories 2 and 3 and by dynamic light scattering with cumulants evaluation in laboratories 4, 5 , 8 and 9. Only turbidimetry in laboratory 1 and dynamic light scattering with cumulants evaluation in laboratory 7 yield values which are slightly too low and in laboratory 6 values which are too low (see Table 2 and Figure 4). See Section 3.1.2 for results of static light scattering. 3.1.5 Test Latex E Test latex E is a blend of six latices with particle diameters of 137, 213, 327, 552, 815 and 1550 nm and a mass ratio of 1 : 1 : 1 : 1 : 1 : 1. From this, the weight-average particle diamewas calculated as 599 nm (see thick full lines in Figter dp,cal ure 5 ) and the Z-average as 986 nm. These average values cannot be characterized well. Tho of the turbidimetric results are between the two average values and one is too low. The results

500

600

700

dp"nl

800

Fig. 4: Particle size distribution (mass distribution) and the measured average particle diameters d,,,,,, (s. Table 2) of test latex D (polystyrene; bimodal; d , , = 605 nm, o, = 8.9 nm; d,,, = 815 nm, ( 5 , = 7.8 nm; mass ratio 1 :9).

Part. Part. Syst. Charact. 12 (1995) 148-157

Table 3: Efficiency evaluation of the methods to determine average particle sizes.

151

method

small particles
d , s 500 nm

large particles
d,
d, 2 500 nm 5 1,000 nm

narrow distributions monomodal bimodal turbidimetry

narrow distributions monomodal bimodal

~~~~

large particle particles of different size region chemical composition dp z 100 nm d, s 500 nm d, 5 2,000 nm narrow distributions hexamodal bimodal

advantages

disadvantages

++

++

(+I

refractive index of small, inexpensive the particles and instrumentation at low particle sizes additionally the concentration and the density of the particles needed, sphere equivalent particle diameter
no parameters

dynamic
light

++

scattering (cumulants evaluation) static light scattering (evaluation of the angle

dependence)

of the particles needed, very fast method

sphere equivalent particle diameter

(+I

--

--

no parameters
of the particles

needed

method more appropriate for polymers in solution

+ + very good, + good, (+) less good, - bad, - - very bad appropriate for the determination of average particle sizes. This efficiency evaluation is exclusively related to the results of the present comparative test.

of dynamic light scattering with cumulants evaluation are too low overall (see Table 2 and Figure 5). See Section 3.1.2 for results of static light scattering.
3.1.6 Test Latex F Test latex F is a blend of two latices with chemically different particles of polystyrene and polybutadiene. The particle diameters are 213 and 363 nm and the mass ratio is l :2. From this, is calculated as the weight-average particle diameter 313 nm (see thick full lines in Figure 6) and the Z-average as 329 nm. These average values are characterized well. Turbidimetry and dynamic light scattering with cumulants evaluation yield values which are only slightly too low. The exception is the result of the light scattering measurement with the measuring instrument D3. The value obtained by static light scattering is much too low (see Table 2 and Figure 6).

Similar good results are obtained using the very fast method of dynamic light scattering with cumulants evaluation [8]. Only in the case of the latex with a very wide particle size distribution are the measured values too low. The static light scattering method [9], which is more suitable for characterizing macromolecules in solution, yields useful results only in the case of latices with very small particles and narrow particle size distributions.

ap,cal

3.2 Determination of Particle Size Distributions The second part of this paper discusses the results of those methods which are able to determine particle size distributions. These methods include electron microscopy with image analysis, ultra- and disc centrifugation [lo, 111, dynamic ligth scattering with non-negative least-squares fit (NNLS) evaluation [8], light diffraction [12, 131 and the electrical sensing zone method [14]. The results of these methods are shown in Tables 4 and 5 and Figures 7-26. In all these figures the plotting scales are linear in order to illustrate the different efficiency of the methods clearly. Because of the very broad distribution curves obtained by the methods of light diffraction and electrical sensing zone in this small particle size range, no plots of the results of these methods are shown in the figures. 3.2.1 Test Latices A and B The particle size distribution of test latex A (see thick full lines in Figures 7-9) is characterized very well by electron micro-

3.1.7 Efficiency Evaluation of the Methods to Determine Average Particle Sizes Summarizing the results given in Table 2 and Figures 1-6, the efficiency evaluation of the methods to determine average particle sizes is as follows (see Table 3). Turbidimetry [5-71 with its small and inexpensive instrumentation yields nearly the exact average particle diameters over the whole submicron range. Only the results for the latices with a very wide particle size distribution vary considerably between the laboratories and the mesured values for the latices with chemically different particles are slightly too low.

152
Table 4: Results of the comparative test: particle size distribution of test latices A, B, C, and D. method laboratory equipment

Part. Part. Syst. Charact. I2 (1995) 148-157

results latex A latex B latex C latex D d, ( m a ) mass ratio d, (max) mass ratio d, ( m u ) mass ratio d, ( m a ) mass ratio (nm) W) (nm) (Oh) (nm) (Yo) (nm) (70)
230 218 204 212 208 208

electron microscopy

11 12 13

ultracentrifugation disc centrifugation dynamic light scattering (NNLS*) evaluation)

2 14 4

Electron microscope I + image analysis 1 Electron microscope 2 + image analysis 2 Electron microscope 2 + image analysis 3 Ultracentrifuge + in-house method [lo] Disc centrifuge Goniometer 1 + Correlator Measuring instrument D1 Measuring Measuring Measuring Measuring Measuring Measuring instrument instrument instrument instrument instrument instrument
D1 D3 L1 L2 L3 El

836
815

98 2

148 214 139 210

810 815 819

= 950
820

145 215 153 220 195

610 836 555 805 545 814 580 843 567 815

8 92 20 80 8 92 11 89 7 93

= 800
= 830

215

189

= 70

light diffraction

6 9 9
9

= 950 5 630 5 720

15 electrical sensing 8 zone method 9 16

= 595 5 720 = 620 = 850 = 750

= 28 = 72 = 50 = 50

= 1800
165 165 150 = 160 = 145 -

= 30

= 650 = 630 5 670 = 650

= 37
= 63 -51 = 49 -

Measuring instrument E2 Measuring instrument E3

= 750
= 630 = 850 = 690

*) NNLS = non-negative least-squares fit.

60

50

50

40

40-

laboratory 4 laboratory 5 labralory 6

----- * -. -

30

20
10

20 10 30
0
100
200 300 d p "4

0 400

500

100

200

-% f2. ; , , 300 d p [=I

_-. , ,
400

f
500

Fig. 7: Particle size distribution (mass distribution) of test latex A, given distribution (-) and the results of electron microscopy.

Fig. 9: Particle size distribution (mass distribution) of test latex A, given distribution (-) and the results of dynamic light scattering.

50

40

centrifugation laboralmy 2 laboralmy 14 ----

I:
.
500

20 30 lo 0

1
0

.
300
400

100

' O 0 d

"nl

Fig. 8: Particle size distribution (mass distribution) of test latex A, given distribution (-) and the results of ultra- (-) and disc (---) centrifugation.

Fig. 10: Particle size distribution (mass distribution) of test latex B, given distribution (-) and the results of elecctron microscopy.

a scopy with image analysis, by ultra- and disc centrifugation and by dynamic light scattering with NNLS evaluation in laboratory 4. The distribution curves produced by the dynamic light scattering method in laboratories 5 and 6 are too broad. The latter is also true for the results of the light diffraction method (see Table 4 and Figures 7-9). The particle size distribution of latex B (see thick full lines in Figures 10-12) is characterized well by electron microscopy with image analysis and by ultra- and disc centrifugation and less well by dynamic light scattering in laboratory 4. The distribution curves obtained using dynamic light scattering in laboratory 9, light diffraction and the electrical sensing zone method are partially bimodal and much too broad (see Table 4 and Figures 10- 12).

Fig. 11: Particle size distribution (mass distribution) of test latex B, given distribution (-) and the results of ultra- (-) and disc (---) centrifugation.

3.2.2 Test Latices C and D

604

'
laboratory 4

'

'

'

'

'

'

40

20 30
10

,
I

,.~,. ,
600

I I

3-

-. .

O r 500

1 - 7_7 _ . _ _ _ _ _
1 7

600

700

+
1

_ _ k.,' - - - - - 900

d [nml

1000

Fig. 12: Particle size distribution (mass distribution) of test latex B, given distribution (-) and the results of dynamic light scattering.

60 50
40 30

The particle size distribution of test latex C (see thick full lines in Figures 13-15) is characterized very well by electron microscopy with image analysis and by ultra- and disc centrifugation. The resolution of the two peaks of the distribution curve is almost perfect using these methods. The correlation with the given particle sizes and the mass ratio of 1 : 1 is also very good. Dynamic light scattering and light diffraction are not able to resolve the two peaks and the distribution curves obtained are monomodal and very broad (see Table 4 and Figures 13-15). The particle size distribution of test latex D (see thick full lines in Figures 16-18) is also characterized well by ultra- and disc centrifugation. These methods reproduce the given particle sizes and the extreme mass ratio of 1 :9 very well. Dynamic light scattering, light diffraction and the electrical sensing zone method are not able to resolve the two peaks of the given particle size distribution. The curves obtained are partly monomodal and much too broad (see %ble 4 and Figures 16-18). 3.2.3 Test Latices E and F The particle size distribution of test latex E with its six peaks (see thick full lines in Figures 19-21) is reproduced with only limited success by electron microscopy with image analysis. Only four (laboratory 11) or five (laboratory 12) constituents are found. The largest particles are missed in the electron microscopic image because of their extremely small number. This and the small overall number of particles (about 500) for all the constituents together are also the reason for the lack of correlation between the measured and given mass ratios, since the numbers of particles of the constituents on the pictures taken are more or less random and not always representative (see Table 5 and Figure 19). However, the ultra- and disc centrifugation methods record all six constituents, not only with the right particle diameters but also with the right mass ratios (see Table 5 and Figure 20). Dynamic light scattering and light diffraction are not able to resolve the six constituents of latex E. The distribution curves obtained are partially monomodal and partially bimodal and very broad (see Table 5 and Figure 21). The electrical sensing zone method cannot resolve the constituents of latex E. This method yields a very broad monomodal distribution curve with particle diameters larger than about 1 vm. The particle sizes of test latex F with its two chemically different constituents (see thick full lines in Figurs 22-24) are characterized very well by electron microscopy with image analysis. However, before the image is taken, the polybutadiene particles have to be contrasted, e.g, with OsO,. The correlation with the mass ratios is not so good, probably because of the relatively

20
10

0 0

100

200 300 dp [nml

400

500

Fig. 13: Particle size distribution (mass distribution) of test latex C, given distribution (-) and the results of electron microscopy.

Fig. 14: Particle size distribution (mass distribution) of test latex C, given distribution (-) and the results of ultra- (-) and disc (---) centrifugation.

154
~

Part. Part. Syst. Charact. 12 (1995) 148-157

Table 5: Results of the comparative test: particle size distribution of test latices E, F, and G.
~ ~ ~~~~~~~

method

laboratory

equipment latex E d, ( m a ) mass ratio (nm) (070)

results latex F d, ( m a ) mass ratio (nm) (@fo)

220 375 25 75

latex G d, ( m a ) mass ratio (nm) (410)

82 = 250 110

electron microscopy

11

Electron microscope 1 +image analysis 1

12

Electron microscope 2 + image analysis 2

155 215 345 565 140 205 330 550 790 150 220 345 585 825 = 1,750 155 213 330 565 805 1,470 = 1,760 180 570 = 480 = 820 400 610 = 525 = 1,900 = 180 = 550 295 ~2,500

22 44 22 12 12 17 4 7 60 18 17 14 16 18 17 12 17 19 17 15 18

= 39

213 376

48 52

ultracentrifugation

Ultracentrifuge + in-house method [lo]

21 1 377

35 65

83 198 114

35 31 34

disc centrifugation

14

Disc centrifuge

218 263

97 3

dynamic light scattering (NNLS evaluation)

Goniometer 1 + Correlator Measuring instrument D1 Measuring instrument D1 Measuring instrument D3 Measuring instrument L1 Measuring instrument L2 Measuring instrument L3 Measuring instrument E3

light diffraction

6 9 9 9

38 62 = 74 = 26 -

252

= 70
= 30

212 360 = 320 = 280

15 electrical sensing 17 zone method

= 30 = 70 -

= 250

--

220

20

I
10

eIccIron microscopy

laboratory 5 laboratory6

-----.-.-

l a h a t o n , 12

-----

0
x
I

%-I En
I-

301
20
10

100

200

300
d p" n l

400

500

500

600

700
dBb

800

900

1000

Fig. 15: Particle size distribution (mass distribution) of test latex C , given distribution (-) and the results of dynamic light scattering.

Fig. 16: Particle size distribution (mass distribution) of test latex D, given distribution (-) and the results of electron microscopy.

Part. Part. Syst. Charact. 12 (1995) 148-157

20
16

155

12
X
I

5I s
-I

ED

500

600

700 800 d p [nml

900

1000

400

800

1200

16W

2000

d p [nml

Fig. 17: Particle size distribution (mass distribution) of test latex D, given distribution (-) and the results of ultra- (-) and disc (---) centrifugation.

Fig. 21 : Particle size distribution (mass distribution) of test latex E, given distribution (-) and the results of dynamic light scattering.

60-

'

'

'

'

'

'

5
0

50

40-

.x. 30 2 la" 20 -lEo

10

IOO

10o
il Inml

300

400

mi

Fig. 22: Particle size distribution (mass distribution) of test latex F, given distribution (-) and the results of electron microscopy.

20

z
"?

0
x

12

I
0
400

21a-I'
4

0
800
1200 1600

2WO

100

200
d,, lnml

300

400

500

d p "4

Fig. 19: Particle size distribution (mass distribution) of test latex E, given distribution (-) and the results of electron microscopy.

Fig. 23: Particle size distribution (mass distribution) of test latex F, given distribution (-) and the results of ultra- (-) and disc (---) centrifugation.

20

16 12 8 4

0
x
I

213' -I-

0
0
400

800
~

1200

1600

2000

100

mi

d [nml

3on <I lnrnj

4oo

sno

Fig. 20: Particle size distribution (mass distribution) of test latex E, given distribution (-) and the results of ultra- (-) and disc (---) centrifugation.

Fig. 24: Particle size distribution (mass distribution) of test latex F, given distribution (-) and the results of dynamic light scattering.

156

Part. Part. Syst. Charact. 12 (1995) 148-157

3.2.5 Efficiency Evaluation of the Methods to Determine Particle Size Distributions

I
101

ik;-,
n

~ , ,,

n
0

.--. .+. .. .... .: -... L...

*r

I
I
500

100

200

300
d,,lnml

300

Fig. 25: Particle size distribution (mass distribution) of test latex G, given distributions (-, ---, - .. . -) and the results of electron microscopy.

30
centrifugation

It

c
I

Fig. 26: Particle size distribution (mass distribution) of test latex G, given distributions (-, ---, - ... -) and the results of ultracentrifugation.

Summarizing the results given in Tables 4 and 5 and Figures 7-26, the efficiency evaluation of the methods to determine particle size distributions is as follows (see Table 6). Electron microscopy with quantitative image analysis yields correct particle size distributions with very high resolution in the small and large particle size ranges and also in a very wide particle size range. The condition for this, especially in the case of a wide particle size distribution, is the investigation of an adequate large number of particles which accurately represents the whole sample. However, this is not always achieved. Because of the restricted possibilities to contrast specifically, some problems also arise with the identification of particles of different chemical composition. With ultracentrifugation, very accurate particle size distributions with very high resolution in particle size and accurate mass ratios are obtained, not only in the small and in the large particle size range but also for very broad particle size distributions. Similar results are yielded by disc centrifugation, except in the case of floating (not sedimenting) particles, which cannot measured by this method. Dynamic light scattering with NNLS evaluation yields different results. The particle size distributions are reproduced fairly well for small particle sizes and relatively narrow distribution curves. However, in the large and wide particle size ranges, the distribution curves are too broad and partly bimodal and the resolution is poor. Even worse results are obtained using light diffraction and the electrical sensing zone method. These methods are more appropriate for the characterization of particles with diameters larger than 1 pm but not of those in the submicron range.

small number of particles investigated (see Table 5 and Figure 22). The results of the centrifugation methods have to be considered separately. The particle size distribution is characterized very well by ultracentrifugation. The disc centrifugation method records only the polystyrene particles because only sedimenting particles can be measured by this method (see Table 5 and Figure 23). Dynamic light scattering and light diffraction are not able to resolve the two peaks. The resulting distribution curves are mostly monomodal and too broad (see Table 5 and Figure 24).

4 Acknowledgements
The author thanks Drs. A . Becker, A . Biirkholz, b K Jacobsen, G. Junkers, A. Karbach, G. Klug, H. Krtimer, H. G. Miiller, T Miinzmay, E-M. Rateike, A . Schmidt, S. Storp, K. Siimmermann, H. G. Vogt and K. Wieser, Dip1.-Phys. E.-R Kops, Ing. R. Sneyders, Ch. van Roost, D. Pfiitzenreuther and their co-workers for participating in the comparative tests and for useful discussions.

3.2.4 Test Latex G

Test latex G with its three constituents of different chemical composition of poly(styrene-co-butadiene)(SBR), polystyrene and polychloroprene (see thick full and dotted lines in Figures 25-26) was investigated using only electron microscopy and ultracentrifugation. After contrasting the particles with OsO,, electron microscopy finds only two kinds of particles. On the one hand SBR and polychloroprene and on the other polystyrene. The first kind cannot be differentiated because both constituents contain double bonds that add OsO,. However, the particle size distributions of all three constituents are reproduced relatively well (see Table 5 and Figure 25). In contrast, ultracentrifugation yields, after preseparation according to the different particle densities by preparative ultracentrifugation, excellent results for particle size distribution and chemical composition (see Table 5 and Figure 26).

5 References
[l] I: Allen, R. Davies: Evaluation of Instruments for Particle Size Analysis. Report of Du Pont Engineering Services, 1987. IA . Davies, D. L. Colhs: Comparison of the Size Distribution [2] . of Boron Powders as Measured by Malvern Diffractometer and Coulter Counter. Part. Part. Syst. Charact. 5 (1988) 116-121. [3] A. BllrkholG R. Boeck: KompatibilitSlt unterschiedlicher Korngr(iBenme8verfahren. Report of Bayer AG, 1989. [4] l ? Jokela, l? D. I . Fletcher, R. Aveyard, J-R. Lu: The Use of Computerized Microscopic Image Analysis to Determine Emulsion Droplets Size Distributions. J. Colloid Interface Sci. 134 (1990) 417. IB. Bateman, E. J Weneck, D. C. Eshler: Determination of Parti[5] . cle Size and Concentration from Spectraphotometric Ransmission. J. Colloid Sci. 14 (1959) 308. K Heller, H. L. Bhatnagar, M. Nakagaki: Theoretical Investiga[6] F tions on the Light Scattering of Spheres XIII. The Wavelength Exponent of Differential nrbidity Spectra. J. Chem. Phys. 36 (1962) 1163.

Part. Part. Syst. Charact. 12 (1995) 148-157

~~ ~~

157

Table 6: Efficiency evaluation of the methods to determine particle size distributions. method small particles d,
5

large particles d, 5 500 nm d, s 1,000 nm narrow distributions monomodal bimodal

500 nm

narrow distributions monomodal bimodal electron microscopy

large particle particles of different size region chemical composition d, 2 100 nm d, s 500 nm d, s 2,000 nm narrow distributions hexamodal bi- and trimodal

advantages

disadvantages

++

++

++

++

4 -

no parameters of the particles needed, information about the particle shape high statistical security

low statistical security, limited contrastability of particles with different chemical composition refractive index and density of the particles and for particles with different chemical Composition preseparation according to particle density needed ditto, only sedimenting particles can be measured refractive index of the particles needed, sphere equivalent particle diameter in the submicron region refractive index of the particles needed, sphere equivalent particle diameter low statistical security

ultracentrifugation

++

++

++

++

++

disc centrifugation

++

++
_

++
_

+
_-

++

ditto

dynamic light scattering (NNLS evaluation) light diffraction

(+)

high statistical security, very fast method

_-

_-

__

_-

high statistical security, very fast method

electrical sensing zone method

no parameters of the particles needed

+ + very good, + good, (+)less good, - bad, - - very bad appropriate for the determination of particle size distributions. The efficiency evaluation is exclusively related to the results of the present comparative test.

[7] H. Lunge: Bestimmung von TeilchengrdDen aus Triibung und Brechungsinkrement. Kolloid-Z. Z. Polym. 223 (1968) 34-30. [8]B. J Berne, R. Pecoru: Dynamic Light Scattering. Wiley, New York 1976. [9] H. A. Stuart: Die Physik der Hochpolymeren, Bd. 11. SpringerVerlag, Berlin, Gbttingen, Heidelberg 1953. K Scholtun, H. Lunge: Bestimmung der TeilchengrbDenverteilung [lo] F von Latices mit der Ultrazentrifuge. Kolloid-Z. Z. Polym. 2.50 (1972) 782-796.

1111 T Allen: Photocentrifuges. Powder Technol. 50 (1987) 193. [12] K. Leschonski, F K Alex, B. Koglzn: TeilchengrdI3enanalyse. Chem.Ing. Tech. 47 (1975) Sonderdruck Berufspraxis. [13] R. Weichert: Untersuchungen zur photometrischen Charakterisierung von Partikelkollektiven. Habilitationsschrift, Univ. Karlsruhe 1983. [14] T Allen: Particle Size Measurement. Chapman and Hall, London, New York 1981.