Volume 1 Number 6 Dennis R. Dinger 1 April 2003
An Update
Please forward any issues of the E-zine, or my web address, to any ceramists or materials engineers who might be interested. If this is the first issue you've seen, you can add your name to the mailing list by clicking HERE. All back issues can be accessed from the Publications page at the web site. Questions, suggestions, and/or requests for topics to be covered in future issues of this e-zine can be sent to QuestionsandComments@DingerCeramics.com .
The Short Courses are ON! In response to my questionnaire, sufficient interest was expressed for me to firm up the arrangements at the Ramada Inn. So we are all set. You should all have received a copy of the short course announcement. Click on this link if you didn't receive the announcement, or if your e-mail program didn't display it properly: Short Course Brochure . Courses will be held during the week of 16-20 June 2003 at the Clemson Ramada Inn. June is a very beautiful time of year in Clemson -- lots of green trees, green grass, and colorful flowers. Registration, hotel, and payment information and forms are all on the web site. Don't expect Uncle Sam to be delivering brochures by snail mail. Everything is on the web.
If you are planning to attend any of the courses, please send me a completed registration form at your earliest convenience. Thanks.
According to the printer, the Rheology for Ceramists book should be completed by mid-April. As soon as I receive them, they will be available at the web-site. Downloadable copies in PDF format are available now at the Books and Downloads page, but the paperbacks will be available soon.
I'm still looking for suggested topics. Please send questions and suggestions. Although I'm not a glaze expert, the first article is in response to a question, and the second article is a subject I think many should find interesting.
Pitting in Glazes
Why does pitting and pin holes occur in glazes? There are several possible reasons (and probably more that I am failing to mention): impurities in the glazes, bodies, and/or at the interface, reactions between the glaze and body, decomposition reactions that have not completed before the glaze begins to melt, non-uniformity of temperature in the kiln, kiln settings, and overfiring. I will try to suggest lots of questions and bring out points to consider in this article. Jim Funk always used to say that if you were going to be wishy-washy, at least be firm about it! So I can say with certainty, beyond a shadow of a doubt, that I might be able to help with this question (or maybe not).
Impurities
All raw materials contain impurities of one sort or another. The important issue is not whether impurities are present (because they always are), but whether such impurities cause processing problems. This applies to all bodies, raw glazes, and fritted glazes. Fritted glazes may be somewhat less problematic than raw glazes because the frits have had time to react, melt, and homogenize at temperature before they are added to the final glaze compositions. During the fritting process, minor constituents (impurities?) have time to react, homogenize, and be incorporated into the glass structure where they may then be less harmful during later firings than if they remained free, crystalline (unmelted) particles.
Some raw materials are less pure than others. In my opinion (and it's certainly a biased opinion), American raw materials (generally speaking) are nice, clean materials. But in some parts of the world, you too can open your own raw materials business if your land contains mineral deposits and you have access to a back hoe. Some such materials are by no means nice, or clean. Others are highly beneficiated and they, too, are nice and clean. The particular impurities associated with each raw material differ according to the geographic location of the deposit and according to mining and beneficiation procedures. So there are wide variabilities that are possible with each raw material depending on its source. This also applies to deposits of the same raw material mined from different geographic locations by a single company. And this applies to chemically prepared (supposedly pure) materials as well. (Several years ago, I was told that there were only three sources in the US of a particular chemically prepared oxide powder, and the three were distinguishable by their impurities.)
Some impurities are no problem in the body alone, but they cause problems when glaze is applied. This makes it difficult to determine which ingredient is causing the problem because bisque firings may proceed without incident. Yet when glazes are applied, problems occur. When clays are added to fritted glazes to improve their application properties, the raw clays may cause glaze problems for this reason.
Another issue which applies to this problem concerns organic additives. What is the nature of the additives in a body or a glaze? At what temperatures do they burn out? Do they burn out cleanly? Do they burn out completely? Is their oxidation finished well before the glaze begins to melt? With regard to combustion, if a gas or fuel oil is not oxidized completely in the flame, soot can form. Once formed, soot particles are very difficult to oxidize completely. This same phenomenon should be expected to apply to remnants of organic additives. If they are not oxidized completely, but have been changed to mostly carbon, will they oxidize completely (and easily) at higher temperatures?
How is the ware handled before glaze is applied? Is it sponged well immediately before glaze is applied? If the ware is sponged before glazing, is it done carefully and completely? Is it done too carefully and too completely? In the one case, if parts of the ware are cleaned and parts are not, the glaze may not adhere properly in the dusty areas and it may adhere well in the clean areas. In the other case, too much sponging can clean away fine surface clays and expose larger non-clay particles to the glaze.
How is ware stored before and after glazing? Does it remain clean? Can dust settle on it? Are the kiln cars clean and free of dust? Are the atmosphere and roof support structures above the ware clean and free of dust? Dust won't necessarily cause pitting, but the natures of the problems it can cause depend when and where it is deposited on the ware.
Body/Glaze Reactions
It is possible that interactions take place at the body/glaze interface. If the ware is bisque fired and then glaze fired, this problem depends on the bisque firing temperature vs the glaze firing temperature. If it is a single fired product, glaze ingredients can react with surface body ingredients on the way to the final firing temperature. Some reactions don't have enough time or temperature to run to completion by the end of bisque firing cycles. Those remnant intermediate products may then interact with the glaze ingredients. By their very nature, glazes contain large amounts of feldspathic and other low melting materials that can interact with the body at the interface during firing (and especially near the final firing temperatures).
Incomplete Decomposition Reactions
If sulfates and carbonates, for example, are present in a body, are those decomposition reactions completed by the time the glaze begins to melt and seal? If carbon remnants from organic binders and additives remain to higher temperatures, will they finally oxidize if and when they come in contact with molten glaze materials? One would think that this is not possible -- that the organics should be completely oxidized by the time the bodies reach glazing temperatures. But some binders like PVA generally don't work well with sodium dispersants. Such polymers can roll up into balls in the presence of the wrong dispersants. When binders don't appear to be performing their intended functions, and strengths are lower than desired, the decision may be made to increase binder concentrations in the body composition. But binder additions raise slip viscosities, so more dispersants may then be added, too. If the binders and deflocculants are not compatible, the new, higher concentration of dispersants may adversely affect the higher binder concentrations, and cause the new polymers to roll up as well. This cycle can then continue: more binder (higher viscosity), more dispersant (lower viscosity), more binder (higher viscosity), more dispersant (etc.), etc. If this happens and binders roll up due to incompatibilities with dispersants, in today's quest for the lowest possible firing temperatures and the lowest possible firing durations, can such concentrated zones of organics burn out properly? It depends on the process and the firing cycle, but it is certainly questionable and it can be a problem. If any of this happens and the reactions are delayed and/or are only partially complete when the glazes begin to melt and seal over, bubbles and pitting can occur.
Non-Uniformity of Kiln Temperature
Today's kilns frequently are large in cross-section and it is difficult to guarantee perfect uniformity of firing temperature and kiln atmosphere throughout the whole kiln. Dead spots in and around the ware can produce zones that are slightly lower in temperature, lower in oxygen content, and/or lower in convective scrubbing velocities which can cause non-uniform glaze properties on the finished wares. This also depends on the narrowness of the firing range of a particular product.
Kiln Settings
This is closely related to the question of kiln temperature uniformity. Hot gases will take the shortest (easiest) paths from the burners, around the ware, to the flue. When the ware is set properly, the short-cuts will be blocked, and the gases will be forced to travel through the settings as desired. But subtle changes in setting can affect kiln flows. How many times has the rotation of a single piece of ware by 90o or 180o at its location in the kiln solved (or caused) a firing problem? I think this happens frequently.
I know of one example where large square floor tiles with lengthwise grooves on the back were stacked in cubes with the directions of grooves of each tile in the stack alternated 90o from the direction of the grooves in the tile below it. Set in this way, gases could pass through the stacks from all four sides, and they fired without problem. But when rectangular tiles half the size were fired, all grooves were parallel, and two adjacent rectangular stacks formed the same size cube of tiles to be placed in the kiln. In this case gases could only pass through the stacks from two of the four sides, and it depended on the exactness of the registration of the stacks at the central plane whether gases could easily pass through the stacks at all. These were smaller tiles, so it was expected that they would be easier to fire, but there were black coring problems in the centers of these stacks.
Small, random problem areas on glazes can easily be the result of temperature non-uniformity and/or setting arrangements. This can be checked by placing lots of pyrometric cones or Buller's rings at various locations around the ware. Such problems can also be the result of kiln temperatures that are close, but not quite at, the desired temperatures (e.g., 98% of the kiln sees the desired firing temperature; 2% does not).
Overfiring
Pitting frequently occurs when wares are overfired. But in this case, the problem doesn't usually disappear when the ware are refired. This too can be a kiln uniformity problem and/or a setting problem. Minor compositional variations from batch to batch can cause the firing temperature requirements to vary slightly from batch to batch. So there is a double problem here. Kiln temperatures and kiln atmospheres can vary slightly from firing to firing, and body requirements can also vary slightly from firing to firing. On the days when the kiln temperatures are on the low side of the acceptable range, and the body requirements are on the high side of the acceptable range, firing problems can occur. On the days when the kiln temperatures are on the high side of the acceptable range, and the body requirements are on the low side of the acceptable range, firing problems can also occur.
Lots of process problems will come and go for no apparent reason (maybe the moon was in the right phase, or something like that). The best we can do is continue to strive for tight controls of bodies and properties, and to look for tests that correlate to the apparently random problems. No problems are random. They all happen for very specific reasons. But some of those reasons and conditions only occur occasionally within any process. We just need to keep our eyes open and pay close attention to all of the possibilities.
Log-Log PSD Plots and Similarity of Packing
This is an interesting topic regarding particle size distributions, packing, and log scales. My son (a college freshman) was complaining to me yesterday that he doesn't like logs. I showed him the title of this topic and he rolled his eyes and walked away. His dislike for logs is typical of many college students and engineers, but logs, and especially log axes, are very frequently useful to ceramists. They are especially useful for particle size distributions and rheograms -- which are two areas of great interest to me. Hopefully, you won't be turned off by the word Log in the title, and you'll continue reading.
Similarity of Packing - Bimodal Systems
If a process requires the packing of a bimodal system of particles, such as packing baseballs within the pores of basketballs, then the size ratio Dbaseballs/Dbasketballs (where D is diameter) will define the size of the next smaller size balls (maybe relatively large marbles) that will pack similarly within the pores of the baseballs. If the size ratios of baseballs to basketballs and of marbles to baseballs are the same, the packing in each bimodal system will be similar.
Similarity of packing occurs when the ratios between particle diameters of adjacent size classes are identical. Any two other particle sizes that have the same diameter ratio as that of baseballs to basketballs will also pack similarly. This would apply, for example, to a fine powder and a coarse powder with the same diameter ratio.
Similarity of Packing - Continuous Distributions
How does this translate to continuous distributions? If you have a distribution comprised of basketballs, soccer balls, softballs, baseballs, and golf balls, what other distribution will have similarity of packing to this distribution? If you measure the diameters of each of these balls (in millimeters) and then you put together a series of spherical powders with identical numeric sizes but with units of micrometers, the two distributions, each with 5 particle sizes, one with coarse sizes (the balls) and one with fine sizes (the powders), will be similar. In fact, in this example, the small particles all have diameters that are exactly 1/1000th the size of the balls. If the volume percentage of coarse particles is the same as the volume percentage of basketballs, and the volume percentage of the next largest particles is the same as the volume percentage of soccer balls, and so on for all of the five size classes in each distribution, the two distributions will be geometrically similar, and they will pack similarly as well. Geometric similarity defines similarity of packing.
Using Log-Log Axes for Particle Size Distributions
Because particle sizes can cover many orders of magnitude, log axes work well when charting particle sizes. Similarly, cumulative percent finer than (CPFT) percentages and histogram (frequency) percentages can each also cover several orders of magnitude. They, too, are both well suited for log axes.
Log axes are also appropriate from the point of view of similarity. If the size of each class in a distribution must be similar to the size of its nearest neighbor in a uniform way, then each size class (from coarse to fine) must be a constant percentage of the size of its nearest neighbor. For example, if the diameter of each size class in a distribution is 95% of the diameter of its neighbor, this will form a geometric series: 100mm, 95mm, 90.25mm, 85.74mm, 81.45mm, etc. Another way to express this relationship is to say that the ratio between any two adjacent sizes is the same (as in the example above.) When plotted on a log axis, each size will be a uniform distance from the next. Multiplying each size by a constant factor is the requirement to achieve uniform increments on a log axis. By comparison, adding (or subtracting) a constant factor to each size would be the way to achieve uniform increments on a linear axis. For example, if the diameter of each size class in a distribution is 5mm less than its neighbor, this produces the series: 100mm, 95mm, 90mm, 85mm, 80mm, etc. When plotted on a linear axis (standard graph paper), each size will be a uniform distance from the next.
If the volume percentage of particles in each size class is a constant factor times the amount in its neighbor's class, this too will form a geometric series. For example, if the amount of particles in each size class is 50% of the amount in the next larger size class, these amounts will form a geometric series: 10%, 5%, 2.5%, 1.25%, 0.625%, etc. When these percentages are marked on a log axis, they too will produce equidistant tick marks.
Perfect Packing
Perfect packing occurs in a broad distribution when the coarsest two size classes pack perfectly with each other, and all other size classes have diameter ratios and percentage in class (histogram) ratios that are geometrically similar to these first two. The combination of the geometric similarity of behaviors of both size and histogram % produce linear behavior on log-log axes. The slopes of the histogram lines vary depending on the natures of the two geometric similarities. Perfect packing in which each size class packs similar to every other size class occurs when histograms are linear (plotted on log-log axes) and when the slope of the line is ~0.37. (The 0.37 value comes from our particle packing research.)
Figure 1 shows the histogram of a perfect packing distribution with a slope of 0.37. Notice that the size classes are all identical widths (the X increments) and the percentages contained in each size class are all uniform step sizes as well (the Y increments). This figure also shows the corresponding cumulative percent finer than (CPFT) distribution plot (at the top). It is not linear. Only the histogram is linear.
Figure 1 Histogram and CPFT of A Perfect Packing
Distribution
Similarity of Packing in Particle Size Distribution Charts
All processes don't need perfect particle size distributions, as shown in Figure 1, and most processes don't use such distributions. But how can similarity of packing be recognized on particle size distribution charts? Similarity of packing occurs when histograms and/or CPFT plots are parallel on log-log axes. Figure 2 shows three non-perfect histograms that are similar from a packing point of view. Rather than drawing these distributions using bars as was done for the histogram in Figure 1, one point represents the top of each bar (the amount in each size class) and each of the points are connected by straight lines.
These distributions don't pack perfectly, but they are typical of ingredient powders. The three distributions in Figure 2 are geometrically similar and they should be expected to exhibit similarity of packing. Slight shifts of the whole distribution to coarser or finer sizes (producing distributions that are parallel on log-log axes) produces distributions that are similar and that can pack similarly. These three distributions are exactly similar, and they should be expected to pack similarly. This can only be seen on log-log axes. Linear axes won't show it.
Figure 2 Three Histograms Exhibiting Similarity of Packing
Rheological Properties May Differ in Geometrically Similar Distributions
Just because distributions are geometrically similar, however, doesn't mean suspensions of those distributions will behave similarly. Geometrically similar distributions should be expected to pack similarly. But suspensions of geometrically similar distributions can have different rheological properties because the fines and colloidal contents differ. The green distribution in Figure 2 has more fines and higher surface area than the two blue distributions, so it will respond slightly differently to the interparticle chemicals and fluids. The interparticle friction due to differences in surface areas, the interaction intensities between particles during collisions, and the interactions between particles and interparticle fluids can all cause rheological differences. This can and should be expected. When the size differences between similar distributions are small (such as between the two blue distributions in Figure 2), rheological differences will be small if they are noticeable at all.
Conclusion
Log-log axes (rather than linear or even semi-log axes) should be used when charting particle size distributions. Fine distribution details at both coarse and fine particle sizes will be readily apparent on log-log histograms. When comparing distributions, similar distributions are easy to spot using log-log axes. And the packing capability of a distribution is also relatively easy to spot. A broad, relatively linear histogram distribution with a gentle positive slope is indicative of good packing. Bodies comprised of distributions that pack well are also usually capable of excellent rheological properties. Histogram distributions with lots of peaks, modes, gaps, and/or spikes are signs that the powders will not pack well, and that the bodies will probably also have poor (troublesome) rheological properties.
Miscellany
Thank you to those of you who answered my questionnaire. I appreciate the help.
I apologize to those of you who received multiple copies of the short course announcement. My old modem didn't like a nearby lightning strike the other night. The first replacement modem didn't work at all. The second replacement worked, but it disconnected frequently from the line. With this modem, I made four tries until I was able to successfully distribute the announcement. I'm now on my third replacement modem, and this one seems to be working OK. It's possible, therefore, that some of you received four copies of the short course announcement, and it's possible some of you don't know what I'm talking about because you received only one copy. Anyway, I'm sorry to those of you who received multiple copies.
See you next time.
Thanks for signing up to receive Ceramic
Processing E-zine
To unsubscribe, click the "Remove Me" link below.
Copyright © 2003 Dennis R Dinger
103 Augusta Rd, Clemson, SC 29631 (864) 654-3155
All Rights Reserved.
All Rights Reserved.