This leads to the distinguished history, but for smaller colloidal danger that artefacts may creep into the image particles its resolution is frequently inadequate. Donald et al. Aspects 37—53 the imaging of colloidal dispersions in their native nition by Danilatos that by utilising a column state, yet still with the high resolution of an with a system of differential pumping zones, it electron microscope.
The introduction of the envi- was possible to satisfy both the requirement that ronmental scanning electron microscope ESEM the electron gun was kept at the necessary high in the last decade, and its various cousins at the vacuum, and that the sample was kept wet for a low vacuum end known variously as leaky useful early review see Danilatos [2].
However, different pressure zones identified. The vacuum comparatively little work has yet been done utilis- considerations in the vicinity of the gun are suffi- ing the full power of the instrument for colloidal cient to permit a LaB6 filament to be used, with systems, not least because of a failure of an its high brightness, as well as the more common appreciation of the potential of the instrument.
Below the gun are a series of This article serves to highlight the power of the pressure-limiting apertures typically three across technique, in the hopes that other workers in the which there is differential pumping to allow a field will feel encouraged to apply it to their pressure difference to be maintained.
As we field of particle aggregation and coalescence. A will see below, this is sufficient to ensure that a brief overview of the limited applications of wet sample can be maintained in its wet state. ESEM to colloids in the literature can be found in However maintaining this pressure differential Donald [1]. There are two further require- electron microscope and still form a high resolu- ments: firstly to ensure that the scattering which tion image?
The key development lay in the recog- the electron beam undergoes as it passes through Fig. Schematic of the ESEM column, identifying the different pressure regimes. Aspects 37—53 39 that is capable of operating outside a vacuum. The requirement that the scattering is not exces- sive is most easily met by ensuring that the dis- tance from the final pressure-limiting aperture to the sample is only short, the so-called working distance.
This skirt effectively acts as a DC background, but does not significantly degrade the resolution. To ensure that this so-called oligoscattering regime is not exceeded, the combi- nation crudely this can be represented by the product of working distance and chamber pres- Fig. Scattering of the incident beam by the gas molecules sure needs to be kept sufficiently low.
However, as which forms as a result. As regards the detector, Danilatos back in the early s designed a simple and effective detec- tor, quite distinct from the Everhardt—Thornley detector used in conventional SEMs, known as the environmental secondary electron detector.
Since then detectors have continued to evolve, and are still doing so as our understanding of how to optimise signal detection continues to improve. For all the detectors the basic principles are the same.
Electrons emitted from the sample, be they secondary electrons or the higher energy backscat- tered, have to travel back through the gas-con- taining chamber to the detector. In traversing the gas, these electrons will undergo collisions with the gas molecules. The collisions may be of vari- ous types both inelastic and elastic. Of crucial Fig. Schematic representation of the cascade amplification process, whereby daughter electrons are created en route to the importance are the ionising collisions which lead detector.
In this way a cascade amplification process occurs as shown in Fig. The degree of amplification has been it is no longer better than an optical microscope studied in detail, Fletcher et al. Aspects 37—53 pressure and working distance. A highly desirable problem is to cool the sample slightly to a few side effect of the cascade amplification process is degrees centigrade.
There they can neu- SVP and hence prevent loss of water from the tralise the usual build-up of charge associated sample. Therefore by using a Peltier cooling stage with electron microscopy of insulators.
This to control the temperature slightly below ambient, means that insulating samples can be viewed with- and by appropriate control of the chamber pres- out the need to apply any conducting coating sure, not only can the sample be maintained wet first. A corollary of this is that fine surface detail but it is also possible systematically to permit can be seen without the risk of it being obscured either hydration or dehydration experiments to be by a thin conductive layer.
All these instrumental developments will not Examples of the utility of these procedures will be avail the study of wet dispersions unless control given below. This article aims to illustrate by suitable exam- There are two necessary problems to overcome ples from the work at the Cavendish the great here: maintaining the sample in its wet state as the potential power of the technique of ESEM for whole instrument is pumped down from ambient colloids.
Three illustrative examples are presented to a few torr in the chamber, and thereafter here in some depth in order to achieve this goal. The former condition can Electroscan E3 model, operating with a LaB6 be met by suitable control of the way evacuation filament, or a model with a tungsten is carried out, performing this in steps to maintain filament.
In both cases a Peltier cooling stage was as high a percentage of water in the chamber as used, as indicated above, to ensure correct control possible at each step, Cameron and Donald [5]. Al- vapour pressure curve SVP for water as a func- though in principle gases other than water can be tion of temperature, as shown in Fig. From this used in the chamber Fletcher et al. The solution to this 2. Aggregation of non-film forming PMMA latices 2.
Experimental The majority of the experiments described here were performed on a latex comprising core-shell rubber particles supplied by ICI plc: each particle had a poly methyl methacrylate PMMA core, a rubber middle layer and a PMMA outer layer, with an overall dimension of 0.
Particles were dispersed in water containing MgSO4 at a concentration ranging from 0 to 0. The temperature of the stage was maintained at Fig. Saturated vapour pressure curve SVP curve for water.
Aspects 37—53 41 2. Results and discussion Fig. It must be remembered that what this image shows is the surface only of the sample. Thus what can be seen is where individual parti- cles are protruding through the surface or covered Fig. Acrylic latex in water. Thus all the grey regions in between the individual particles corre- spond to water where there are no particles close enough to the surface to be visible. At this early stage the particles remain distinct, and there are no aggregates visible.
As evaporation proceeds, aggregation begins Fig. The net result is the slow formation of a well-ordered colloidal crystal. In the surface plane the majority of the packing is Fig.
As water is allowed to evaporate in the ESEM, a found to have hexagonal symmetry, although oc- colloidal crystal starts to form from the acrylic latex. After tation of the planes changes. Systematic changes imaging was continued. In this way the evolving can be seen as the molarity of the MgSO4 is structure during aggregation could be followed in raised, He and Donald [6].
The addition of salt real time. A few experiments were also carried out has the effect of reducing the long-range repulsion on a similar particle, but lacking the outside layer so that the barrier to the deep primary minimum of PMMA. When two as the outer layer has a Tg of below zero. In particles come together the probability of sticking principle videos can be taken in real time of the therefore increases and this can be directly viewed whole process, so that each stage can be examined in the ESEM with the particles coming together in in detail later.
In practice the danger of beam the primary minimum. This aggregation will be damage affecting the response of the system, if a irreversible Jeffrey and Ottewill [7].
The net ef- given area is viewed for extended periods, means fect is therefore to lead to a much more disor- videoing must be treated with caution. Aspects 37—53 Fig. It can be seen that a much more the fractal structure has formed.
Colloidal aggre- compact structure is formed for the latter case gation is usually described by one of three limiting than for the former, and in neither case is there case models: fast diffusion limited cluster aggrega- any regularity of packing of the type seen in the tion DLCA , slow reaction limited cluster aggre- colloidal crystals when no salt is present.
In the RLCA case there are more detail to provide a novel way of characteris- significant repulsive forces acting between neigh- ing the fractal dimension of the flocs that form. Diffusion limited aggre- gation DLA idealises the irreversible aggrega- tion of colloidal particle aggregates, in which individual particles join pre-existing clusters.
This is in contrast to the DLCA regime, in which clusters diffuse to join other, pre-existing clusters. According to theory and simulation, the fractal dimension of two dimensional flocs should be 1. The fractal dimensions were calculated for the flocs in Fig. From such an analysis it should be possible to deduce, for differ- ent conditions of particle and salt concentration, what regime of aggregation the system is in.
The flocs that form are of course three dimensional, but the way the image is formed means that only the top quasi-two-dimensional plane is sampled. This means that the situation is more complicated than where the floc is constrained to lie at an interface and is therefore necessarily two-dimen- sional, as in many of the earlier works, Hurd and Schaefer [12], Stankiewicz et al.
It is more akin to that of fat crystal networks recently Fig. Flocs formed during the evaporation of water from a analysed via optical microscopy Narine and latex containing a 0.
Marangoni [15]. Aspects 37—53 43 When the salt concentration is raised to 0. Because of the increased screening due to the salt, essentially every collision leads to sticking. Be- cause of this a large number of small aggregates form. These clusters then move around and aggre- gate further, exactly the situation envisaged in the cluster-cluster aggregation model. The fractal di- mensions ranged from 1. This is similar to the exper- imental results obtained by Stankiewicz et al.
Fractal dimensions calculated from the flocs shown in 1. If the sticking probability is high, then a analysis for the two flocs of Fig. Although it is new shape will form, whose fractal dimension no dangerous to draw too many conclusions from longer reflects the mechanism by which the origi- such a limited data set, it can be seen that the two nal structure formed. In our work we saw evi- slopes are somewhat different for the two differ- dence for this in the flocs formed from rather ent salt levels.
The biggest aggregate in Fig. This can stabilised. Since the outer layer was rubber and be compared with a value for d of 1. It sometimes small clusters may also approach and can be seen that the floc contains loops of parti- stick. The salt concentration employed, 0. Thus the observations suggest ture subsequently restructuring as the two arms that the mechanism operating is DLA, but not in stick together. Of course, it is additionally possi- the ideal sense of every encounter leading to ble that the particles themselves are so soft that sticking, but with a lower probability of aggrega- they deform individually; this will also have the tion than that.
This may explain why the fractal effect of altering the fractal dimension deter- dimension determined from Fig. Aspects 37—53 One caveat should be mentioned at this point. In the preceding paragraphs it has been implied that real time observations can be made as evapo- ration proceeds.
This is of course true, but the observer must always be aware of the possibility of beam damage becoming a limiting factor.
Beam damage is a ubiquitous problem for elec- tron microscopists working on organic and many inorganic systems but not in general metals. This is for two reasons. Firstly there is no conductive coating required for the surface of an insulator, as discussed in the introduction.
Secondly, and probably more importantly, the various processes that occur when electrons interact with water Fig. These The ESEM used in this way therefore opens up free radicals can be very deleterious for organic the possibility of a high resolution way of study- materials. If the exposure is continued for suffi- ing fractal structure evolution.
Whereas the num- ciently long, the damage is easily visible. First, ESEM has the resolution, which during hydration of some substituted lyocell light microscopy does not, to image individual fibres. In the presence of water the fibres should latex particles. Thus the details of the encounters swell, but this real effect may be obscured by which occur — does a particle-particle collision beam-induced damage that can look broadly simi- lead to sticking or not?
In Fig. Secondly, whereas previous work using beam plus water, there is substantial bubbling and electron microscopy to study fractal structures, swelling of most of the fibres viewed end on. The Lin et al. Consequently, in its native state. What is observed is, however, right at the edge of the figure it seems as if all the clearly only the surface of a three-dimensional originally distinct fibres have fused together.
This structure. Thus, unlike in many of the optical effect appears to be general for cellulosic materi- studies where a true 2 dimensional structure is als, and additional information can be found in formed by constraining the aggregation to occur refs [18—20]. FTIR spec- obtained from the images. From this summary it troscopy has been used to quantify the extent of can be seen how ESEM studies may contribute damage under different conditions, in this case for further to studies of the fractal nature of colloidal samples of polypropylene, [21].
This demonstrated aggregation. Aspects 37—53 45 Fig. Micrographs illustrating the visible changes to carboxymethyl-derivatised lyocell fibres in the ESEM under damaging conditions of the incident beam. The images show a undamaged fibres; b the start of bubbling; c continuing damage, with the effects of the beam raster pause evident down the left hand side of the image; and d the fibres obscured 15 min after the initiation of damage in b. Scale bar represents 50 mm. This of the flocs comprising the soft particles; that is effect is presumed to be due to the ready mobility exactly the sort of situation where long term of radicals through the water layer, increasing the observations would need to be made.
Neverthe- rate of polymer hydrolysis. This necessarily means less, as the work on damage to cellulose fibres has that damage is likely to be an issue when imaging shown elsewhere, as long as one knows how long wet dispersions of PMMA latices, PMMA being one can safely image a sample without significant notoriously susceptible to beam damage, and in- beam damage occurring, then imaging over com- deed used as an electron resist in e-beam lithogra- paratively long periods can be made, [20,22].
Whereas this does not mean that any given image is necessarily artefactual, it does mean that caution must be exercised when sitting with the 3. Structure and aggregation of vinyl latices beam on any given area of a sample and watching how the structure evolves.
Thus although one 3. Experimental may have confidence that a conclusion such as that concerning the probability of sticking is Both standard vinyl latices supplied by ICI valid, it does not mean that a video taken of a Paints and those with starch incorporated were single floc is necessarily free of artefacts.
It is for studied, to examine the effect of starch upon this reason that we have refrained from making aggregation and coalescence. Aspects 37—53 ther in the form of unmodified material, or by interdiffusion across the initial particle derivatised potato starch. The effect of the addi- boundaries occur to yield a film with good me- tion of surfactant was also explored. The latices chanical properties. The required stages in film were imaged after spin coating onto glass stubs formation are described in detail elsewhere, [23].
First the individual particles have a very narrow size distri- 3. Results and discussion bution compared with the standard commercial formulation shown in Fig. How- Vinyl latices are very often used in paint formu- ever, although this might seem like an attractive lations. As such it is desirable that the individual advantage the particles show no tendency to coa- particles rapidly aggregate and coalesce followed lesce even after 2 weeks of drying.
This means Fig. Aspects 37—53 47 Fig. Ex- particle size is significantly smaller, but the main actly what role the native starch is playing cannot consequence is the apparently very rapid coales- be assessed from this information alone, but it is cence of the particles. After 1 h all particle clear that it is radically affecting the surfaces of boundaries have disappeared. However at much the particles.
This altered surface chemistry will longer times a new effect is seen: regions of bright then affect both the polydispersity and the contrast slowly begin to appear on the surface. This effect is attributed to the slow exudation of If on the other hand the starch is chemically excess surfactant to the surface to form pools modified, so that the original granular structure is upon it.
This interpretation is confirmed by alter- totally lost before addition, the response of the ing the level of surfactant addition, which reduces latex the same base vinyl formulation is rather the amount of exudate which appears on the different.
Although changing structure of latex films as water evapo- in this case both at early and late stages the films ration occurs. Because water loss does not occur are fairly flat and particle deformation can be seen due simply to the high vacuum conditions of even at early times, after 3 weeks there are numer- observation, as in a conventional SEM, we can ous imperfections still present in the film, al- have confidence that the changes that are ob- though some particle coalescence has occurred.
In served during drying are real and not artefactual. The particle polydis- gree of particle coalescence can readily be persity is greater than in the case of the addition obtained. In terms of identifying whether stan- of unmodified starch.
The particle polydis- affects both particle polydispersity during poly- persity is now even greater although the mean merisation, and ability to film form. ESEM means that these factors can be unambigu- that the presence of an increasing concentration ously determined, rather than simply relying on of non-film forming particles led to an increase in parameters such as film formation temperature or the number of voids.
This had the effect of in- gloss as a quality determinant. However the ability to image the particles directly in the ESEM enabled a distinction to be made between the situation where a void was associated 4. Film formation in the presence of hard inclusions with only one individual hard particle, from the situation in which a void is formed interstitially Finally, let us consider the situation when hard between several.
In the latter case void closure particles are also present. These might either be was much harder. Consequently, when hard parti- particles of a higher glass transition temperature cles are present it is important to ensure a uni- than the remainder, and hence unable to partici- form distribution to minimise the risk of hard pate in film formation, or an inorganic inclusion particle clusters. In an example of In this section we will consider what happens the former situation, in which ESEM was used in when silica particles are added to an acrylic latex conjunction with ellipsometry [24], it was found as a matting agent.
Aspects 37—53 49 ing an optimised pumpdown procedure, Cameron and Donald [5] to prevent premature dehydration. Series of drying images were obtained by a gradual increase of the sample temperature. Once silica particles could be distinguished form the aqueous back- ground, images were recorded every few minutes.
Thus all the grey regions in between the individual particles corre- spond to water where there are no particles close enough to the surface to be visible. At this early stage the particles remain distinct, and there are no aggregates visible. As evaporation proceeds, aggregation begins Fig. The net result is the slow formation of a well-ordered colloidal crystal.
In the surface plane the majority of the packing is Fig. As water is allowed to evaporate in the ESEM, a found to have hexagonal symmetry, although oc- colloidal crystal starts to form from the acrylic latex. After tation of the planes changes. Systematic changes imaging was continued. In this way the evolving can be seen as the molarity of the MgSO4 is structure during aggregation could be followed in raised, He and Donald [6]. The addition of salt real time.
A few experiments were also carried out has the effect of reducing the long-range repulsion on a similar particle, but lacking the outside layer so that the barrier to the deep primary minimum of PMMA.
When two as the outer layer has a Tg of below zero. In particles come together the probability of sticking principle videos can be taken in real time of the therefore increases and this can be directly viewed whole process, so that each stage can be examined in the ESEM with the particles coming together in in detail later.
In practice the danger of beam the primary minimum. This aggregation will be damage affecting the response of the system, if a irreversible Jeffrey and Ottewill [7].
The net ef- given area is viewed for extended periods, means fect is therefore to lead to a much more disor- videoing must be treated with caution. Aspects 37—53 Fig.
It can be seen that a much more the fractal structure has formed. Colloidal aggre- compact structure is formed for the latter case gation is usually described by one of three limiting than for the former, and in neither case is there case models: fast diffusion limited cluster aggrega- any regularity of packing of the type seen in the tion DLCA , slow reaction limited cluster aggre- colloidal crystals when no salt is present. In the RLCA case there are more detail to provide a novel way of characteris- significant repulsive forces acting between neigh- ing the fractal dimension of the flocs that form.
Diffusion limited aggre- gation DLA idealises the irreversible aggrega- tion of colloidal particle aggregates, in which individual particles join pre-existing clusters. This is in contrast to the DLCA regime, in which clusters diffuse to join other, pre-existing clusters. According to theory and simulation, the fractal dimension of two dimensional flocs should be 1. The fractal dimensions were calculated for the flocs in Fig. From such an analysis it should be possible to deduce, for differ- ent conditions of particle and salt concentration, what regime of aggregation the system is in.
The flocs that form are of course three dimensional, but the way the image is formed means that only the top quasi-two-dimensional plane is sampled.
This means that the situation is more complicated than where the floc is constrained to lie at an interface and is therefore necessarily two-dimen- sional, as in many of the earlier works, Hurd and Schaefer [12], Stankiewicz et al. It is more akin to that of fat crystal networks recently Fig.
Flocs formed during the evaporation of water from a analysed via optical microscopy Narine and latex containing a 0. Marangoni [15]. Aspects 37—53 43 When the salt concentration is raised to 0. Because of the increased screening due to the salt, essentially every collision leads to sticking. Be- cause of this a large number of small aggregates form. These clusters then move around and aggre- gate further, exactly the situation envisaged in the cluster-cluster aggregation model.
The fractal di- mensions ranged from 1. This is similar to the exper- imental results obtained by Stankiewicz et al. Fractal dimensions calculated from the flocs shown in 1. If the sticking probability is high, then a analysis for the two flocs of Fig. Although it is new shape will form, whose fractal dimension no dangerous to draw too many conclusions from longer reflects the mechanism by which the origi- such a limited data set, it can be seen that the two nal structure formed.
In our work we saw evi- slopes are somewhat different for the two differ- dence for this in the flocs formed from rather ent salt levels. The biggest aggregate in Fig. This can stabilised. Since the outer layer was rubber and be compared with a value for d of 1. It sometimes small clusters may also approach and can be seen that the floc contains loops of parti- stick. The salt concentration employed, 0. Thus the observations suggest ture subsequently restructuring as the two arms that the mechanism operating is DLA, but not in stick together.
Of course, it is additionally possi- the ideal sense of every encounter leading to ble that the particles themselves are so soft that sticking, but with a lower probability of aggrega- they deform individually; this will also have the tion than that. This may explain why the fractal effect of altering the fractal dimension deter- dimension determined from Fig.
Aspects 37—53 One caveat should be mentioned at this point. In the preceding paragraphs it has been implied that real time observations can be made as evapo- ration proceeds. This is of course true, but the observer must always be aware of the possibility of beam damage becoming a limiting factor.
Beam damage is a ubiquitous problem for elec- tron microscopists working on organic and many inorganic systems but not in general metals.
This is for two reasons. Firstly there is no conductive coating required for the surface of an insulator, as discussed in the introduction. Secondly, and probably more importantly, the various processes that occur when electrons interact with water Fig.
These The ESEM used in this way therefore opens up free radicals can be very deleterious for organic the possibility of a high resolution way of study- materials. If the exposure is continued for suffi- ing fractal structure evolution. Whereas the num- ciently long, the damage is easily visible.
First, ESEM has the resolution, which during hydration of some substituted lyocell light microscopy does not, to image individual fibres. In the presence of water the fibres should latex particles. Thus the details of the encounters swell, but this real effect may be obscured by which occur — does a particle-particle collision beam-induced damage that can look broadly simi- lead to sticking or not?
In Fig. Secondly, whereas previous work using beam plus water, there is substantial bubbling and electron microscopy to study fractal structures, swelling of most of the fibres viewed end on. The Lin et al. Consequently, in its native state. What is observed is, however, right at the edge of the figure it seems as if all the clearly only the surface of a three-dimensional originally distinct fibres have fused together. This structure. Thus, unlike in many of the optical effect appears to be general for cellulosic materi- studies where a true 2 dimensional structure is als, and additional information can be found in formed by constraining the aggregation to occur refs [18—20].
FTIR spec- obtained from the images. From this summary it troscopy has been used to quantify the extent of can be seen how ESEM studies may contribute damage under different conditions, in this case for further to studies of the fractal nature of colloidal samples of polypropylene, [21]. This demonstrated aggregation.
Aspects 37—53 45 Fig. Micrographs illustrating the visible changes to carboxymethyl-derivatised lyocell fibres in the ESEM under damaging conditions of the incident beam. The images show a undamaged fibres; b the start of bubbling; c continuing damage, with the effects of the beam raster pause evident down the left hand side of the image; and d the fibres obscured 15 min after the initiation of damage in b.
Scale bar represents 50 mm. This of the flocs comprising the soft particles; that is effect is presumed to be due to the ready mobility exactly the sort of situation where long term of radicals through the water layer, increasing the observations would need to be made. Neverthe- rate of polymer hydrolysis. This necessarily means less, as the work on damage to cellulose fibres has that damage is likely to be an issue when imaging shown elsewhere, as long as one knows how long wet dispersions of PMMA latices, PMMA being one can safely image a sample without significant notoriously susceptible to beam damage, and in- beam damage occurring, then imaging over com- deed used as an electron resist in e-beam lithogra- paratively long periods can be made, [20,22].
Whereas this does not mean that any given image is necessarily artefactual, it does mean that caution must be exercised when sitting with the 3.
Structure and aggregation of vinyl latices beam on any given area of a sample and watching how the structure evolves. Thus although one 3. Experimental may have confidence that a conclusion such as that concerning the probability of sticking is Both standard vinyl latices supplied by ICI valid, it does not mean that a video taken of a Paints and those with starch incorporated were single floc is necessarily free of artefacts.
It is for studied, to examine the effect of starch upon this reason that we have refrained from making aggregation and coalescence. Aspects 37—53 ther in the form of unmodified material, or by interdiffusion across the initial particle derivatised potato starch.
The effect of the addi- boundaries occur to yield a film with good me- tion of surfactant was also explored. The latices chanical properties. The required stages in film were imaged after spin coating onto glass stubs formation are described in detail elsewhere, [23]. First the individual particles have a very narrow size distri- 3. Results and discussion bution compared with the standard commercial formulation shown in Fig.
How- Vinyl latices are very often used in paint formu- ever, although this might seem like an attractive lations. As such it is desirable that the individual advantage the particles show no tendency to coa- particles rapidly aggregate and coalesce followed lesce even after 2 weeks of drying. This means Fig. Aspects 37—53 47 Fig.
Ex- particle size is significantly smaller, but the main actly what role the native starch is playing cannot consequence is the apparently very rapid coales- be assessed from this information alone, but it is cence of the particles. After 1 h all particle clear that it is radically affecting the surfaces of boundaries have disappeared. However at much the particles. This altered surface chemistry will longer times a new effect is seen: regions of bright then affect both the polydispersity and the contrast slowly begin to appear on the surface.
This effect is attributed to the slow exudation of If on the other hand the starch is chemically excess surfactant to the surface to form pools modified, so that the original granular structure is upon it. This interpretation is confirmed by alter- totally lost before addition, the response of the ing the level of surfactant addition, which reduces latex the same base vinyl formulation is rather the amount of exudate which appears on the different. Although changing structure of latex films as water evapo- in this case both at early and late stages the films ration occurs.
Because water loss does not occur are fairly flat and particle deformation can be seen due simply to the high vacuum conditions of even at early times, after 3 weeks there are numer- observation, as in a conventional SEM, we can ous imperfections still present in the film, al- have confidence that the changes that are ob- though some particle coalescence has occurred.
In served during drying are real and not artefactual. The particle polydis- gree of particle coalescence can readily be persity is greater than in the case of the addition obtained. In terms of identifying whether stan- of unmodified starch.
The particle polydis- affects both particle polydispersity during poly- persity is now even greater although the mean merisation, and ability to film form.
ESEM means that these factors can be unambigu- that the presence of an increasing concentration ously determined, rather than simply relying on of non-film forming particles led to an increase in parameters such as film formation temperature or the number of voids. This had the effect of in- gloss as a quality determinant. However the ability to image the particles directly in the ESEM enabled a distinction to be made between the situation where a void was associated 4.
Film formation in the presence of hard inclusions with only one individual hard particle, from the situation in which a void is formed interstitially Finally, let us consider the situation when hard between several. In the latter case void closure particles are also present.
These might either be was much harder. Consequently, when hard parti- particles of a higher glass transition temperature cles are present it is important to ensure a uni- than the remainder, and hence unable to partici- form distribution to minimise the risk of hard pate in film formation, or an inorganic inclusion particle clusters.
In an example of In this section we will consider what happens the former situation, in which ESEM was used in when silica particles are added to an acrylic latex conjunction with ellipsometry [24], it was found as a matting agent. Aspects 37—53 49 ing an optimised pumpdown procedure, Cameron and Donald [5] to prevent premature dehydration.
Series of drying images were obtained by a gradual increase of the sample temperature. Once silica particles could be distinguished form the aqueous back- ground, images were recorded every few minutes.
A continuous transformation from a uniform- aqueous latex to a dried film in which the silica is clearly distinct was observed. This system proved quite susceptible to beam damage, as shown in Fig. Primary beam damage of a latex containing silica, Fig. This figure shows a central rectangle imaged at 12 keV accelerating voltage. The sample was viewed at higher magnification and then the magnification reduced so which has been heavily irradiated at high magnifi- that the fingerprint of the original raster can clearly be seen.
It can be seen as in Fig. To reduce beam damage to an acceptable tive they are as matting agents will obviously level, the accelerating voltage was reduced to 8 depend both on how they interact with the film- keV, and a short working distance used. Results and discussion 4. Experimental Fig. The first trices containing 2. Firstly it is clear that the silica particles and the second was a fumed silica obtained from appear bright.
This is not surprising since they are Degussa. These in a conventional SEM, their backscattered emis- aggregate to give a loose, porous structure with a sion is therefore higher.
At the magnifications broad distribution of particle sizes peaking at employed, the individual primary particles cannot around 7 mm. The 10 mm present. The uniform structure of the concentration of silica particles was varied from fumed silica aggregates is thought to be due to the 0. Experiments were carried out way they agglomerate through hydrogen bonding. Drying was carried out in situ follow- arate particles.
Aspects 37—53 51 Thus at a simple level, the contrast in the more deeply the effect of water upon image for- images of Fig. However, mation. It was first noticed several years ago that aside from the relative brightness of the silica a water layer surrounding a latex particle gave an particles there is a more subtle type of contrast anomalously high signal, which could not be ex- visible in these images.
As drying proceeds the plained by either of the standard SEM contrast background gets systematically darker. This is mechanisms, namely topographic contrast which seen for both series of images in Fig. In order arises primarily from the secondary electron sig- to understand this we have to consider a little nal and atomic number contrast coming from the higher energy backscattered electrons [25]. Imaging of oil-in-water emulsions showed that the water phase looked bright, and this was maintained when the emulsion was inverted to form a water- in-oil emulsion, again demonstrating that the brightness was not a subtle topographic effect.
Qualitatively the origin of the contrast can be attributed to the way in which the secondary electrons which are by far the major constituent of the signal lose energy as they travel through different materials.
Water can be considered as a wide band gap semiconductor, in the sense that there is a signifi- cant range of electron energies over which no inelastic collisions can occur to cause loss of energy to the secondary electrons as they travel through the material.
Hence a greater number of them will have sufficient energy to overcome the surface affinity and escape to be subsequently detected. It is for this reason that water looks bright compared with oils, or indeed a polymer matrix. For the images in Fig. A second set of experiments was concerned with looking at the morphology of the dried films as a function of silica content. It appears that there is more silica in the Fig. A comparison of the surface of dried films containing surface of the sample with the precipitated silica 1.
Aspects 37—53 titative approach permits a correlation between type of silica added including its density to the surface content. Further studies on surface rough- ness at the micron level can then rationalise the performance of different silicas.
However, it must be borne in mind that this technique can only reveal the top layer of a sample. Depth of field is good, as with conventional SEM, so that there is little difficulty associated with imaging a rough surface, but it is not possible to see through a significant layer of water, [27,28]. Thus, when wet colloidal dispersions are imaged there is a problem if none of the particles are sufficiently close to the surface for imaging. Image analysis as a function of silica content compar- face is not typical of the bulk, this cannot be ing fumed and precipitated silica.
For this reason it is highly desirable, whenever possible, to correlate information ob- quantitative by carrying out image analysis on tained from ESEM imaging with other techniques. The analysis was carried out by we are in the process of obtaining quantitative data imposing a threshold grey level: pixels brighter than on the distribution of the silica throughout the this threshold were presumed to be part of a silica drying film via confocal microscopy, which permits particle, and all other pixels were treated as matrix.
The results of but imaging below the surface is straightforward. It confirms that The two techniques can therefore be seen to be at low silica concentrations the surface occupancy highly complementary, and work is ongoing to of the samples with precipitated silica is higher than correlate surface and sub-surface packing of the for the samples with fumed silica. However at silicas. Beyond this value for the bulk 5. Conclusions concentration it seems that the surface structure does not change any further.
This article has aimed to identify the many Clearly this technique offers new possibilities of strengths of ESEM for the field of colloidal disper- studying in detail the way the silica is distributed sions. Recognition of the fact that it is possible to at the surface of a drying film. And since it is the image a wet sample in its natural state with nature of the surface which imparts the matting resolution comparable to conventional high vac- properties to the film, this may provide additional uum SEM, by dropping the temperature of the insight into how to optimise the silica type and stage to slightly below ambient opens up a wide properties.
This brief example of the utilisation of range of possibilities. Moreover, it is also possible image analysis also demonstrates how at least to change the state of the sample, to move it semi-quantitative information can be obtained between hydrating and dehydrating conditions, by from such images of the surface.
Aspects 37—53 53 niently, chamber pressure. Thus dynamic experi- —
0コメント