A CASE STUDY OF -SILICA REACTION IN (INVOLVING VERY REACTIVE SILICEOUS GLASS) USING PETROGRAPHIC TECHNIQUES

KEN SPRING, DIRECTOR AND SENIOR PETROLOGIST HEATHER SPRING AND THOMAS SPRING

Geochempet Services, Unit 5/14 Redcliffe Gardens Drive, Clontarf, Queensland Phone: +617 32840020 Website: www.geochemept.com Email: [email protected]

ABSTRACT

The forms of potentially reactive silica most likely to result in a deleterious degree of alkali- silica reactivity (ASR) in concrete are finely microcrystalline and moderately to heavily- strained quartz. Other forms of potentially more reactive silica of significance comprise , , tridymite, cristobalite and siliceous glass. Determination of the abundance and form of the potentially reactive free silica is best achieved by petrographic examination, based mainly on microscopic analysis of the concrete.

Free silica refers to in a disordered state (such as glass) or within various crystalline (of which quartz is only one example). It is different to combined silica which refers to silicon dioxide which is chemically combined with other elements or oxides to create chemically more complex silicate minerals (of which feldspar will serve as a common example). The total silica content of an aggregate or sand which can be determined by bulk chemical analysis (such as X-ray fluorescence - XRF) is not an effective guide to its potential for deleterious alkali-silica reactivity: it reports the total of combined silica and free silica. Determination of the abundance and form of the free silica is best achieved by petrographic examination, but supplemented where necessary by X-ray diffraction (XRD) or electron microscopy methods such as SEM/EDS.

The most reactive forms of silica are richly siliceous opal, chalcedony and glass. The substances are essentially disordered, non-crystalline to cryptocrystalline (individual not resolvable by optical microscopy) and/or unstable minerals. Natural glass in rocks carries variable amounts of other oxides in addition to silica. The glass is generally produced by quenching of volcanic igneous or tuffaceous rocks. Although all types of glass in rocks carry silica in a non-combined or "free" condition, it seems that only those compositions which would have been capable of crystallizing free silica minerals (if not quenched) give rise to deleterious amounts of silica in glass for reaction with in concrete. Thus, glass of acid and intermediate igneous compositions is usually deleterious and amounts of richly siliceous glass as low as several percent can be quite deleterious within aggregate or sand.

A case study of concrete which contained acid and/or intermediate lavas and tuffs of Tertiary or younger age is presented. Very ASR reactive rock types (broken fragments of pumice and glassy acid and intermediate volcanic rocks) were identified in the petrographic examination of the supplied samples including liberated high temperature quartz phenocrysts (tridymite and cristobalite) with glassy rims. These highly reactive forms of alkaline silica appear to cause accelerated ASR effects in the investigated concrete.

Forensic petrographic investigations of ASR damaged are important in characterising ASR cracking as well as determining the significance of cracking or degree of damage. The petrographic examination can also determine if ASR-type gels are present as well as assessing the potential risk for future ASR impacts from potentially reactive aggregates. Determination of the abundance and form of the potentially reactive free silica and related cracking is best achieved by petrographic examination, based mainly on optical microscopic analysis of the concrete.

If enough ASR products have formed, further analysis using XRD and SEM/EDS techniques can confirm or otherwise, the presence of alkali-silica gels as the primary cause of the damage to the concrete.

Keywords: ASR, Concrete, Siliceous Glass, Petrographic, Case Study

INTRODUCTION

Alkali-silica reactivity (ASR) is widely accepted as a chemical reaction between alkalis in Portland and particular forms of reactive silica that occur naturally in aggregates used in concrete production. It results in the formation of an alkali-silica gel which is expansive. The expansion generates stresses causing cracking in the affected concrete structure over a period of time.

ASR in concrete involves a chemical reaction between alkalis in pore forming an alkaline solution (contributed mainly but not exclusively from the cement) and reactive forms of silica (within the coarse and/or fine aggregate). The result of the alkali-silica reaction is an alkali silicate gel which in the presence of moisture causes expansion in the concrete. The expansion results in pressures in the concrete producing cracking in the structure (Farny and Kerkhoff, 2007). It can be over a long time interval where mild and slow reactivity is occurring or accelerated to shorter time duration where reactivity is substantial.

Conditions that favour ASR development in concrete is moist and hot conditions (i.e. tropical conditions), but local factors around the in situ concrete may contribute to favourable conditions promoting ASR effects. In NZ, the main factor appears to be exposure to moisture, and in most cases, ASR evidence occurs within 5-10 years of pouring the concrete indicating the presence of very reactive silica. Once ASR damage commences in concrete there is no way to stop the reaction until the alkalis are exhausted.

The key ingredients for deleterious alkali-silica reaction in concrete are elevated sodium and/or potassium (mainly from the cement, but supplemented from other sources such as saline external in some environments) combined with enough but not too much reactive silica and moisture. Some other factors contribute to the rate of reaction such as high temperatures combined with high humidity which promotes alkali-silica reactions.

Research into bridge structures in South Island demonstrated the presence of ASR impacts in concrete piles subjected to intertidal marine conditions (Freitag, SA; Bruce, SM and Shayan, A – 2011, Bruce et al – 2008 and Freitag et al – 2009, 2010).

Prior to this research, it was generally considered that ASR was not a significant issue in South Island concretes. Accelerated mortar bar and concrete prism tests showed that the material sourced from beach sands and associated river gravel alluvials are potentially alkali reactive in some localities. The reactive rock types are interpreted to be acid and intermediate volcanics and metamorphosed sedimentary rocks (greywacke) containing some of these volcanic rock fragments. Similar assemblages are commonly found throughout NZ. The occurrence of ASR in concrete appears to be readily accepted where these potentially reactive aggregates have been recognised (Freitag et al – 2000).

Steps can be taken in the design of concrete mixes and in the design of concrete structures to control and possibly eliminate or minimise a deleterious degree of reaction. Guidelines and recommended practices for minimizing ASR risks in concrete are explained in CCANZ (2003). Preventative measures include rejecting potentially reactive aggregates, limiting alkali content of cement and using sufficient amounts of to effectively reduce ASR effects.

PETROGRAPHIC EXAMINATION

The petrographic analyses were performed in 2013 on supplied concrete samples taking account of ASTM C856 “Standard Recommended Practice for PETROGRAPHIC EXAMINATION OF HARDENED CONCRETE”. Information provided on the General Purpose used to produce the concrete showed, on average, 2.7% SO3 and 0.57% Na equivalent which appear to be within acceptable limits for cement products as are the compressive strength tests on mortar and concrete generated using the cement. Basically, the relatively recent concrete examples contained empty carbonated cracks that displayed ASR characteristics as well as cracked quartz and feldspar grains; the presence of potentially very reactive silica is also noted.

Very ASR reactive siliceous glassy fragments (broken fragments of pumice are shown in Plates 1-3) were identified in the petrographic examination of concrete containing aggregate from a North Island sand source along with liberated high temperature quartz phenocrysts (tridymite and cristobalite) with glassy rims (Plates 4 and 5). These highly reactive forms of silica appear to be the reason for accelerated ASR effects in the concrete.

Plate 1. Medium magnification, plane transmitted light image of the sand sample showing a cuspate broken pumice fragments (circled in red). Image width is 0.59 mm.

Plate 2. Medium magnification, crossed polarised, transmitted light image of part of concrete showing a cuspate fragment of pumice consisting of isotropic glass adjacent to an air bubble. Image width is 0.59 mm.

Plate 3. Medium magnification, crossed polarised, transmitted light image of part of concrete showing relatively common cuspate fragments of pumice consisting of isotropic glass. Image width is 0.59 mm.

Plate 4. Medium magnification, crossed polarised, transmitted light image of part of concrete showing wedge-shaped high temperature beta-quartz. Image width is 0.59.

Plate 5. Medium magnification, crossed polarised, transmitted light images of the concrete showing a wedge-shaped beta quartz with a glassy rim (the hardened cement paste is carbonated). Image width is 0.59 mm.

SEM Analysis

SEM/EDS analysis was undertaken to further investigate the cracking in concrete and a conspicuous crack was targeted for probing for evidence of ASR gel in the crack. A line scan was conducted across the crack and elemental results for 15 points recorded in Table 1.

ELEMENTAL ABUNDANCES (MASS PERCENT) Data Point C O Na Al Si S Ca Fe 1 5.21 66.48 0 0.09 27.97 0.06 0.19 0 2 3.45 55.68 0 0 40.69 0.05 0.14 0 3 12.85 49.67 0.03 0.05 36.67 0.09 0.64 0 4 80.05 13.7 0.09 0.14 0.91 0.29 4.17 0.65 5 24.03 9.38 0 0.15 3.21 0.59 61.37 1.27 6 17.6 16.02 0.24 0.08 1.55 0.35 64.16 0 7 69.78 14.13 0.2 1.11 1.73 1.88 7.48 3.71 8 36.86 35.48 0.46 2.83 13.83 0.24 9.24 1.05 9 27.92 36.53 0.58 1.51 25.8 0.29 4.65 2.73 10 16.69 37.24 0.04 0.79 6.01 0.16 38.73 0.35 11 16.07 21.95 0.44 0.87 5.17 0.12 54.27 1.11 12 4.27 47.12 0.44 16.81 2.5 0.39 15.01 13.46 13 3.48 42.58 7.82 11.15 33.3 0.05 1.27 0.35 14 5.21 44.83 6.46 10.47 32.7 0.01 0.26 0.05 15 2.41 44.19 7.61 11.25 33.74 0 0.67 0.13

Table 1: Data matrix of elemental abundances in line scan across an open crack in the concrete.

The first three points are in a quartz grain (reflected by high silicon and oxygen values). Point 4 shows very high carbon content indicating presence of carbon in an air space or void. Point 5 and 6 are interpreted to be carbonated cement paste (elevated carbon, oxygen and calcium values) but is depleted in alkalis. Point 7 has high carbon again and is interpreted to represent the presence of an open crack. Points 8 to 12 are located in the cement matrix and are elevated in carbon, oxygen and calcium but show a degree of variability in elemental distribution. The last three points are suspected to cover either a lithic fragment or feldspar grain (based on relatively higher sodium and aluminium results combined with elevated silicon and oxygen).

The sodium results for the cement matrix show some depleted areas (Points 5, 6 and 10 on each side of the empty crack) but sodium is still available (up to 0.58%) in the hardened cement paste. Points 8 and 9 adjacent to the open crack show slightly elevated silicon combined with the highest sodium values in cement paste which seems to suggest migration of these elements towards the open crack to form sites for the development of expansive alkali-silica gels.

Concrete samples (polished blocks) were then examined by SEM/EDS to search for evidence of ASR gels or indirect evidence of ASR impacts in micro-cracks. Effects that characterise ASR activity include cracking that passes into and through aggregate fragments. Cracking although mainly confined to cement paste (Plates 6 and 7) follow the edges of aggregate fragments (where weaknesses can be concentrated) and occasionally pass through lithic clasts and pumice fragments (Plate 8).

Plate 6. SEM images of micro-cracks confined to cement matrix. Note that the fine cracks run alongside aggregate and pumice fragments.

Plate 7. Two SEM images of the same sample, showing an open crack confined to cement matrix and running alongside aggregate fragments and intersecting void spaces.

Plate 8. Two SEM images of open cracks passing through a lithic clast (upper image) and pumice fragment (lower image) which are interpreted to be indirect evidence of ASR impacts.

The crack linings were then analysed by EDS and SEM (Plate 9 and Figures 1 and 2) and shown to be essentially composed of calcium from effects. It is considered that ASR cracking in the cement paste allows carbonation to penetrate deeper into the concrete. The effect of flushing ASR gels from cracks, when air and moisture are introduced into opening cracks, lines these cracks with calcium .

Plate 9. SEM image of the open crack showing location of probed areas.

Figure 1. EDS Elemental analysis of material lining crack (dominated by Calcium and Oxygen).

Figure 2. EDS Elemental analysis of material lining crack (dominated by Calcium and Oxygen).

Plate 10. SEM image of feldspar showing internal cracking of probable ASR origin.

Plate 11. SEM image of quartz crystal showing internal cracking of probable ASR origin.

In plates 10 and 11 above, further indications of ASR activity is shown by internal cracking of feldspar and quartz grains which are regarded as sites where alkalis and silica can combine to form expansive gel which in advance case would shatter the sand grain.

Partial infill around an air bubble were analysed (Plate 12 and figures 3, 4 and 5) and interpreted to be essentially composed of ettringite, a hydrous calcium aluminium sulphate – Ca6Al2(SO4)3(OH)12-26H2O. It is considered to be benign, early form of ettringite filling void spaces in the cement paste.

Plate 12. SEM image of an air bubble showing location of probed areas in thin coatings.

Figure 3. EDS Elemental analysis of material lining air bubble (dominated by Calcium and Oxygen along with Sulphur and Aluminium) – ettringite, a hydrous calcium aluminium sulphate has this composition.

Figure 4. EDS Elemental analysis of material lining air bubble (dominated by Calcium and Oxygen along with Sulphur and Aluminium) – ettringite, a hydrous calcium aluminium sulphate has this composition.

Figure 5. EDS Elemental analysis of material lining air bubble (dominated by Calcium and Oxygen along with Sulphur and Aluminium) – ettringite, a hydrous calcium aluminium sulphate has this composition.

Plate 13. SEM image of a finely vesicular and siliceous pumice fragment showing location of probed area.

Figure 6. EDS Elemental analysis of the pumice fragment (dominated by Silica and Oxygen along with subordinate abundances of alkalis).

Some pumice fragments were analysis for chemical composition (Plates 13 and figure 6). The EDS analysis indicated that the pumice fragments contain an alkaline and siliceous volcanic glass and is a possible source for both very reactive silica and alkalis.

XRD Analysis

The XRD analysis of the sand provided support for the presence of glassy volcanics (possibly containing comingled glass and microlites in the lithic fragments) and pumice fragments via a high amorphous content (Table 2 and Figure 7). The presence of cristobalite was interpreted from powder XRD patterns but tridymite was not confirmed.

XAF 8614 8614R non-diffracting/ 34.7 33.9 unidentified Plagioclase (Anorthite) 46.5 48.4 Quartz 16.5 14.8 1.4 1.7 Biotite 0.3 0.3 Amphibole 0.7 0.9 (Hornblende)

Table 2. Quantitative XRD results in wt% (with duplicated sample).

Figure 7. Powder and clay XRD patterns – note amorphous mineral signature between 0-9 degrees (two-theta) In thin section examination, some of wedged-shaped quartz (probably tridymite) is rimmed by volcanic glass which may mask its presence. The presence of volcanic glass and cristobalite and tridymite, high temperature beta quartz confirm that this sand is potentially very reactive in relation to alkali-silica reactivity in concrete. These highly reactive forms of silica combining with alkalis in the cement and siliceous glass appear to be the reason for what seems to be advance ASR effects in the concrete samples. XRD analysis confirms petrographic examination that the sand sourced from North Island, NZ consist of quartzo-feldspathic and volcanic lithic sand. It is composed of water-worn rounded volcanic and pumice clasts and free mineral grains liberated from these volcanic rocks. The volcanics appear to be derived from quenched acid and intermediate volcanic rock with a component of frothy pumice fragments.

Accelerated Mortar Bar Testing

The sample of the North Island sand and associated cement was despatched to Material Technical Services, Boral Construction Services to test the sand for Alkali-Aggregate Reactivity (AAR). A report was using Test Method – Accelerated Mortar Bar Test for AAR Assessment. (RMS T363). The results are tabulated below:

Table 4. Data matrix of age of concrete and percentage expansion determined on an average of 3 specimens.

Age (in days) Expansion%

3 0.100 7 0.312 10 0.389 14 0.483 17 0.534 21 0.585

The results were compared to a classification table shown below to assess the reactivity of the sand sample:

Table 5. Classification of Alkali-Aggregate Reactivity of Concrete soaked in 1 molar NaOH solution at 80C (based on Test Method WA 624.1 – 2012).

Percentage expansion at 21 days Expansion%

Fine Aggregate AAR classification <0.15 Non-reactive >0.15, <0.45 Having potential for slow/mild ASR >0.45 Having potential for substantial ASR

The result for the sand is 0.585 after 21 days which is classified as having potential for substantial ASR. This result confirms the petrographic assessment of potential for alkali-silica reactivity. The report commented “Specific components in the sand which are perceived to have potential for alkali-silica reaction in concrete comprise tridymite/cristobalite and siliceous glass in pumice, acid and intermediate volcanic fragments (around 9-10%), a potentially very reactive form of amorphous silica. The sand as a whole is predicted to have potential for substantial deleterious alkali-silica reactivity in concrete.”

The presence of reactive components was also indirectly supported by XRD results with high amorphous content interpreted to be related to the glassy volcanic clasts as well as glassy pumice fragments and the likelihood of the presence of cristobalite and inferred presence of tridymite in the supplied sand sample.

DISCUSSION

The concretes showed the following characteristics:

1. After 3 years, the concrete showed more cracking than usually observed with associated loss of strength 2. The cracks appears to be rectilinear in pattern 3. Cracks are generally associated with a white stain marking

The sodium results from EDS/SEM for the cement matrix show some areas depleted in alkalis demonstrating mobility of alkalis in the concrete; the depletion may be caused by generation of ASR gel in the probed crack that has been flushed and replaced during carbonation of the concrete. Sodium is still available (up to 0.58% in other areas) for further ASR development in the hardened cement paste. Points 8 and 9 adjacent to the open crack show slightly elevated silicon combined with the highest sodium values in cement paste. This may be indicative of early stages of ASR gel formation in areas sub-parallel to the open crack related to migration of these elements to sites of potential ASR development.

The accelerated mortar bar result on mortar mixed with the sand used in the concrete is classified as having potential for substantial ASR. This result confirms the petrographic assessment of substantial potential for alkali-silica reactivity. The presence of potentially reactive components was also indirectly supported by XRD results with high amorphous content interpreted to be related to the glassy volcanic clasts as well as glassy pumice fragments and the likelihood of the presence of cristobalite and inferred presence of tridymite in the supplied sand sample.

In our opinion, petrographic investigations of ASR damaged concretes are important in characterising ASR cracking as well as determining the significance of cracking or degree of damage. The petrographic examination can also determine if ASR-type gels are present as well as assessing the potential risk for future ASR impacts from potentially reactive aggregates. Determination of the abundance and form of the potentially reactive free silica and related cracking is best achieved by petrographic examination, based mainly on microscopic analysis of the concrete.

If enough ASR products have formed, further analysis using XRD and SEM/EDS techniques can confirm or otherwise, the presence of alkali-silica gels as the primary cause of the damage to the concrete.

CONCLUSIONS

The Portland cement from which the concretes are produced are not blended with a pozzolan and it contained on average 0.57% Na equivalent. Recommended practices in NZ suggests that cement have an alkali content of less than 0.6% with replacement of at least 15% of the cement by pozzolan when used with reactive aggregates.

The potentially reactive nature of the aggregate used in the manufacture of these concretes has been demonstrated by petrographic, SEM/EDS and XRD assessment of reactivity in sands within concrete and substantial reactivity of sand in mortar bar testing. ASR impacts can occur over a long time interval where mild and slow reactivity is operating in the concrete or accelerated to a shorter time frame where reactivity is substantial. The fine aggregates used are very reactive (due to the presence of pumice fragments, volcanic glass in acid to intermediate volcanic fragments and high temperature quartz) and likely to show ASR cracking at an earlier stage than the typical ASR reactive in concretes.

Thus, sand of the type investigated may be used in concrete provided that appropriate precautions are taken in mix and engineering design to take account of its perceived potential for substantial deleterious alkali-silica reactivity. Care needs to be exercised in understanding and, if necessary, controlling the perceived quite substantial potential for alkali-silica reactivity which may vary in particular mixes from benign (if the content of siliceous glass is high enough) to deleterious (if the content of amorphous silica decreases towards a pessimum value which is around 2%).

While preventative measures such as use of low alkali cement and may minimise the risk of ASR, lithium treatments by impregnating damaged concrete with lithium salts solutions can assist in alleviating future expansion in the concrete. The limitation of this type of treatment is that it requires the concrete to be fully impregnated which can be difficult to achieve and practical experience in the use of the treatment appears to be limited.

Since, it is considered that ASR damage will continue while alkalis and reactive silica is present in the concrete and conditions of high moisture and humidity persist. Temperatures above 38C can favour ASR expansion but the availability of moisture is considered to be more significant in NZ but both contribute to humidity in the concrete. Treatments based on limiting moisture in the concrete will have limited value if they cannot restrict water ingress from all sources and permit water already in the concrete to evaporate.

ACKNOWLEDGEMENT

The XRD and SEM/EDS analyses were undertaken using QUT facilities. Accelerated Mortar Bar Test was undertaken by Material Technical Services, Boral Construction Services.

REFERENCES

Bruce et al – 2008, Identifying the cause of concrete degradation in prestressed bridge piles, and Prevention 2008. CCANZ (2003), Alkali silica reaction. Minimizing the risk of damage to concrete – guidance notes and recommended practices. CCANZ technical report TR3 (second edition). Freitag et al – 2000, ASTM C1260 and the alkali reactivity of NZ greywackes, 11th International Conference on alkali aggregate reaction. Farny, J.A and Kerkhoff, B.. 2007, ‘Diagnosis and Control of Alkali-aggregate reactions in concrete’, Concrete Technology, Portland Cement Association 1-25. Freitag et al – 2009, Concrete pile durability in South Island bridges, Coast and Ports 2009. Freitag et al – 2010, Concrete pile durability in South Island bridges, The NZ Concrete Industry Conference. Freitag, SA; Bruce, SM and Shayan, A – 2011, Concrete pile durability in South Island bridges, NZ Transport Agency Research Report 454.