By Wendell Dubberke
The following linked SEM images and corresponding element maps are from faulty pcc box beams fabricated in Texas. The SEM work was done in 1997. For those using dial-up modems, it will take a while to upload each image. The image may upload as an image too large for your screen, but should automatically rescale to fit your screen. We are still looking for the original images that were stored on disk. These images were uploaded with the use of a scanner and hard copy images, consequently they will not be as sharp as the originals. The links will retrieve the images stored at the MARL facility at Iowa State University in Ames, Iowa.
Click here or click on image1 (bottom of page) to view an environmental scanning electron microscope (ESEM) in Backscattered electron (BSE) mode taken at 30X for a 4" x5" polaroid negative. Viewing full screen images will correspond to a higher magnification. These pcc samples were ground and polished next door to the ESEM facility and were immediately moved from the polishing room to the ESEM sample chamber so that only a minimum of time was allowed for oxidation and carbonation to occur. The ESEM obtains its information from the sample surface and BSE images are enhanced by the atomic weight of the material observed. Consequently, fresh, clean surfaces will give the best images. The lighter gray levels relate to heavier atomic weights when viewing samples in the BSE mode.
Most of the aggregate particles appear durable. Most of the ettringite (gel?) filled microcracks go around the aggregate particles. However, after some searching, a cracked aggregate particle, containing silica gel, was found and imaged at higher magnifications.
Click here or click on image2 (bottom of page) to view element map for previous BSE image. The upper left image is a repeat of the BSE image. The next 11 images are element maps for carbon (C), oxygen (O), sodium (Na), Magnesium (Mg), aluminum (Al), silicon (Si), sulfur (S), chlorine (Cl), Potassium (K), calcium (Ca) and iron (Fe). Black or dark gray indicates areas of low counts for a particular element and light gray or white indicates areas of high counts for a particular element. Ettringite is hydrous compound containing mainly sulfur, aluminum and calcium. Silica gel is a hydrous compound containing silicon, potassium and/or calcium and/or sodium. Both compounds also contain oxygen and hydrogen.
Click here or click on image3 below to view an enlargement of the silica particle (chert?) containing silica gel in the fracture. Most of the small aggregate particles appear to be non-porous, igneous quartz, but a few appear to be finely porous sedimentary chert. Both chert and quartz are nearly 100% silicon dioxide and appear to be very similar in ESEM observations. Some brown cherts in Iowa gravels can be very fine-grained with no pore system. If the cracked silica particle is alkali/silica reactive (ASR), as the silica gel in the fracture would seem to indicate, why is there no silica gel in the aggregate-paste bond area where it would be expected to form in abundance? This area is filled with ettringite as this image and the next element map clearly show. One pcc investigator, who attended our pcc durability meetings, believed that silica gel was in the cracks first, then disappeared, and was replaced by ettringite. It is quite a stretch, but hard to disprove.
The easiest explanation (but not necessarily correct) for the deterioration seen in these images is suggested in the following sequence: (1) initial microcracking due to a mix using too much finely ground, type 3 cement containing excessive amounts of potassium sulfate, (2) the use of water reducers, plasticizers and grinding aids requiring reduced water in the mix, which in turn, will not allow all of the ettringite to form up front while the concrete is still plastic, (3) water more easily able to enter the hardened concrete through the microcracked pore system, (4) delayed ettringite forming (DEF) in the microcracks and air voids, (5) silica gel forming in a few particles and (6) extensive freeze/thaw damage if moisture and freezing temperatures are present.
Cement company representatives argued that the extra surface area of finely ground cement particles required more sulfur to control set. If the extra sulfur is not from highly soluble potassium sulfate and if the mix contains enough water to allow all of the ettringite to form very early, the cement may perform satisfactorily. Type 3 cement may require more grinding aid to promote flowability in storage silos. Some grinding aids also work as water reducers and therefore may not allow for more water in the mix.
Perhaps the cement and water should be premixed as a slurry prior to being introduced to the rest of the mix. This would make it easier for all of the sulfur to convert. Another mix method that might work would be to withhold the fine-aggregate (sand)for 15 or 30 seconds. Coarse aggregate in a water/cement slurry would stir the mix and break up any cement clumps. In one Duggan sample pcc mix, using a lesser quality cement, sand was inadvertently added late to a sloppy mix. That sample passed the Duggan test, while all of the other samples, from that cement source, did not.
As can be seen on this ESEM BSE image and the following element map, the silica gel in the fractured silica particle and the ettringite in the fractured matrix on each side of the fractured particle have not intruded into each others space indicating equal pressure if indeed the silica gel and the ettringite are expansive.
Click here or click on image4 below to view the element map for the previous BSE ESEM image. This element map shows the silica gel in the fractured aggregate particle to contain potassium and calcium, but no sodium. A small feldspar particle, at the lower right, contains potassium, aluminum and silicon. A few small limestone (calcium carbonate) particles can also be seen. The amount of ettringite compared to the amount of silica gel is readily apparent in this image even though extra effort was expended finding a particle containing silica gel.
Click here or click on Image5 below to view a higher magnification of the fractured particle image at 300X. The pore system in the aggregate particle is becoming apparent and looks more like chert than quartz.
Click here or click on image6 below to view the element map for the previous ESEM BSE image. At 300X, the sulfur map now exposes a fracture in the matrix (upper right corner) that is nearly invisible in the previous enlarged BSE image. Can optical microscopy, using thin sections, identify fractures like these? Grinding and polishing a sample for ESEM analysis takes less than 10 minutes. How long does it take and how much does it cost to make a thin section for optical microscopy? Do all of the minerals and other compounds remain in place while making a thin section? Some thin section fabrication techniques use heat and vacuum that may alter or remove certain compounds.
Click here or click on image7 below to observe ESEM BSE deteriorated pcc,at 350X, from another location on the box beam sample.
Click here or click on image8 below to view element map for the previous 350X BSE image. This element map was improperly scanned and consequently is a little fuzzy. Even at this magnification, the microcracks would be difficult to identify on the BSE image if it was not for the aid of the sulfur map. How are shrinkage microcracks identified in faulty pcc when a shrinkage prone cement without excessive sulfur is used in the concrete mix? The pcc sample can not be vacuum saturated with a tracer because excessive vacuum can cause cracking as we discovered when using a conventional high vacumm SEM early in these investigations. When using the ESEM, only a very weak vacuum is applied to the sample chamber and an inert gas is leaked into the chamber to enhance the BSE and element map images. Most likely ASR and freeze/thaw damage would also be enhanced by an initially microcracked pcc matrix. Shrinkage prone cements can be identified by using the ring test method or the Duggan test method. Potassium sulfate particles in cement can easily be identified by generating an SEM element map. Where potassium and sulfur overlay each other, arcanite particles are present in the cement.
Click here or click on image9 below to view what may be the best ESEM BSE image in this box beam series. The fiberous appearance of this type of ettringite is apparent in this image. A 400 micron diameter air void full of ettringite is located at the right center of this image. Even though nearly all pcc investigators refer to this ettringite material as crystalline, diffraction tests, on this material removed from air voids, showed it to be amorphous. Also, this material has the consistency of a thick paste when removed with a sharp needle. In some older faulty concrete, the ettringite balls, removed from a fractured surface, were hard and broke like glass. These too, were amorphous.
The ettringite fibers (bundles of tiny filaments), growing from the base (being fed by capillary action) or possibly being extruded, are able to lift flyash spheres off of air void surfaces. An ESEM image showing tiny filaments (in the process of forming a bundle) with a flyash sphere atop is available and may show up on another site page when time permits.
Some pcc investigators have argued that ettringite has no expansive force and merely grows into existing space. Other investigators think ettringite's capacity to take on moisture (32 waters per molecule)is the expansive force. If ettringite grows from the base, the force could be similar to that of ice when when liquid water is fed to the base by capillary action. Under these circumstances, the growth of the ice is related to the conversion of water to ice, but in ettringite, the expansion would relate to the uptake of moisture.
Click here or click on image10 below to view the ESEM element map for the previous BSE image. The crack in the feldspar particle (lower left) contains parent material. The microcracks in the matrix show a general trend and indicate the direction of the shrinkage force. A small microcrack cutting through the four larger microcracks, in the center of the image, indicate that a sequence of events apparently occured.
Click here or click on image11 below to view an ESEM pcc fractured surface instead of a flat, polished surface. Note the clarity of the image even though the sample has substantial topography. Having a large depth-of-field is another advantage of the SEM over optical microscopy. Also, notice the ettringite "egg shells" around the aggregate particles. A considerable amount of ettringite is also present on the matrix fractures. Ettringite lined aggregate sockets can be seen in the upper left area of this image. The fiberous nature of this type of ettringite is readily evident in these sockets. This image was taken at 80X with BSE detectors near the electron beam at the top of the sample chamber.
Click here or click on image12 below to view the element map for the previous BSE image. Normally element maps are not generated from fractured surfaces because the topographic high areas will shadow out information needed by the element detector which is located towards the side of the sample chamber. On the element map, areas that are black for all elements have been shadowed out. The amount of sulfur on the matrix fractures, aggregate sockets and aggregate/paste interface is significant.
When evaluating faulty pcc, the ESEM should be the primary instrument of choice. Either there are too many optical microscopists unable to differentiate ettringite and silica gel or the thin section fabrication techniques are altering the pcc sample.
(This paragraph added on 4-1-2005) Recently I had an email exchange with another pcc investigator who also analyzed the faulty Texas pcc box beams. He forwarded high quality SEM images relating mainly to faulty aggregate particles used in the mix. If he gives permission, links to these images will be made available later. Based on crack propagation analysis and the generation of silica gel, his conclusion was that the pcc deterioration was primarily due to ASR. After looking at his SEM images, I would counter that the reason these cracks appear around reactive aggregate particles is because the matrix was already microcracked prior to the generation of silica gel. The previously microcracked matrix allowed moisture to easily enter the concrete and also makes it easy for expansive silica gel to add cracks to the matrix in the area around the reactive particles, similar to what happens when a reactive particle causes a popout when located near the surface of pcc. A few, more deeply buried, reactive particles will not crack the matrix. The tensile strength of a good matrix can easily handle minor amounts of silica gel or ettringite. Excessive amounts of silica gel or delayed ettringite can cause the matrix to crack. A tight, non-microcracked matrix should restrict moisture availability to the potential generation of silica gel or delayed ettringite. Personnel at the TxDOT said that the aggregates used in the faulty box beam mixes, was used previously in other pcc mixes, and performed adequately.
(The next 2 paragraphs added on 02/26/2006) As indicated above, ESEM images of early forming ettringite would be added to this site as time permits. These ESEM images were obtained in the mid 1990s when the new Hitachi ESEM was delivered. The pcc sample was subjected to a variety of vacuum strengths. Consequently, the matrix of the pcc sample shows vacuum induced cracking due to excessive vacuum applied to the sample. After some experimentation, a proper level of vacuum was found that would not damage the pcc sample and yet give good image results. Also, the images were only downloaded to a polaroid camera. The images were recently uploaded from hardcopy and therefore have lost some of their sharpness. Click here to see an ettringite related report by Stark & Bollmann showing ettringite images similar to ours, but with greater detail.
The Iowa DOT ettringite-in-pcc images were from a slice taken off the third beam of a set fabricated for ASTM C666-b durability (freeze/thaw) testing. This set of beams had an extra 1.0% K2SO4 (arcanite) added to the mix. The Iowa DOT cures the beams for 90 days in a moist room prior to testing. The slices were periodically taken off of the third beam while it was in the moist room. These images show the ettringite growing up from the base where it is in contact with the surface of the air void. Is water in the air void combining with a water-starved calcium sulfoaluminate to form ettringite? Is this taking place at the pore openings in the concrete matrix surface? The newly formed ettringite filaments bunch together to form hexagonal protocrystals which are soft and pliable like balsa wood. XRD test results show this newly formed type of ettringite to be amorphous even though it is hexagonal. Click on images 13 through 18 for more information concerning ettringite growth in young pcc.
(The next 3 paragraphs added on 03/20/2006) The remaining two beams from the above three beam set, along with other three beam sets were freeze/thaw tested, using the ASTM C666-B test method. All four sets in this grouping easily passed the test, which indicates that the aggregate fraction was durable and also that the matrix was durable and was able to handle moderate amounts of ettringite. Click here to see a graph of the ASTM C666-B test results for this grouping of beams.
The Iowa DOT 90 day moist room curing, of pcc beams destined for ASTM C666-B testing, probably precludes any autogenous shrinkage taking place. The 90 day moist room cure was adopted because beams intentionally fabricated with faulty coarse aggregates would not consistently fail when cures of 7, 14 and 45 days were used. Initially, the test was only used to evaluate coarse aggregate for use in pcc.
If the curing conditions were modified to more closely duplicate field conditions, the test might be able to differentiate shrinkage prone cements and/or mix designs. The modification should also include a 1/2 inch wire mesh inserted when the mold is half full. If autogenous shrinkage takes place when a shrinkage prone cement is used or if an excessive amount of fine grained cement is used in the mix, microcracking should occur in the area of the wire mesh. Beams should fail when microcracking preceeds freeze/thaw cycling. The Iowa DOT fabricated beams, for ASTM C66-B testing, measure 4" X 4" X 16". Without some kind of restraint, the whole beam may shrink and generate only a minor amount of microcracking. Perhaps something better than wire mesh should be used, as the strands could move during autogenous shrinking. On failing pcc pavements, quite often the deterioration is first noticed at the transverse joints where metal baskets are located.
(Next 2 paragraphs added on 3/29/2006) From the images, there is evidence that the ettringite filaments are growing or extruding at their base. Image13 shows a flyash sphere being lifted off of the air void wall by ettringite filaments. Image15 shows ettringite filaments being scrunched towards the center of a smaller air void. Ettringite filaments are bending and breaking in this small air void. Some of the bending and breaking may have been caused by sample preparation. The gap between the ettringite filaments and small air void wall was caused by excessive ESEM vacuum being applied to the sample.
Click here to see an EDS spectrum of the ettringite filaments in the large air void.
(The next 2 paragraphs added 04-23-2006) According to information in Paula Raivio's report, pcc mix designs and/or pcc mixes, made with cements that are prone to generating excessive ettringite, can be identified by immersing crushed pcc particles in water and then, using an electron microscope, look for ettringite filaments on the particle surface. This is more evidence that the ettringite filaments are growing from their base rather than growing on their surfaces as most researchers believe. The easiest explanation for the ettringite filament growth on the surface of the submerged pcc particles is that water starved calcium sulfoaluminate is lurking in the micro pores of the pcc particles prior to immersion.
Richard Burrows, who has done considerable work involving pcc microcracking related to autogenous shrinking, told me by phone that high alkali in a shrinkage prone cement or mix design could make a bad situation worse. Richard did not define the type of high alkali, but if it is the soluble type (potassium or sodium sulfate) then there could be an explanation for the correlation. In this type of faulty pcc, if autogenous shrinkage induced microcracking occurs first, followed by water intrusion and then calcium sulfoaluminate extrusion from the non-microcracked areas in the matrix, wouldn't it be easier for these non-microcracked areas to shrink even more if the calcium sulfoaluminate is allowed to escape from the micro pores. It is possible that the generation of ettringite in the microcracks is also enhancing the shrinkage potential in the surrounding matrix. Shrinkage first, then expansion if enough of the ettringite building components are available.
Home Page
Image1
Image2
Image3
Image4
Image5
Image6
Image7
Image8
Image9
Image10
Image11
Image12
Image13
Image14
Image15
Image16
Image17
Image18
