BRE's involvement in several cases of failure of lime and cement stabilised clay soils from sulfate attack in the past three years (Figure 1) has prompted a review of methods by which sulfatebearing ground is assessed prior to stabilisation.This article describes the reasons for failure and suggests improvements to site assessment.
Sulfate attack in stabilised ground
The addition of lime and cement to cohesive (clay) soils to improve their engineering characteristics is well established and generally problem free. However, in a small number of cases each year the stabilised layer fails to meet specification owing to sulfate attack.
In such attack sulfates and aluminium silicate minerals in the ground react with the lime (calcium oxide or calcium hydroxide) or Portland cement and water to form the highly expansive sulfate hydrates ettringite and thaumasite.
This has affected structures including capping layers for roads and hardstanding for vehicles, and stabilised materials under ground-bearing floors of warehouses and retail units.
The formation of ettringite (3CaO.Al 2O3.3CaSO 4.32H 2O) is favoured by wet conditions at a wide range of ambient temperatures. BRE found this reaction product (Figure 2) on the intended site of a retail store where Lower Lias Clay was lime/cement stabilised in May and in excess of 100mm of ground heave had occurred by September .
In contrast, thaumasite (CaCO 3.CaSO 4.CaSiO 3.15H 2O ) particularly likes cool (generally less than 15infinityC) wet conditions for formation, an environment typical of UK ground conditions for much of the year.Thaumasite also requires a source of carbonates, often supplied from calcium carbonate present as shell fragments and limestone particles in clay soils, although it can also come from bicarbonate in groundwater. BRE found thaumasite to be the principal expansive mineral involved in heave of the site of a building that occurred over the winter following cement stabilisation of Kimmeridge Clay.
In a landmark case that occurred during construction of the Banbury section of M40 motorway in the late 1980s [2, 3], both ettringite and thaumasite were reported by investigators following lime stabilisation of Lower Lias Clay. The initially 250mm thick stabilised layer heaved by up to 150mm, equivalent to a 60% expansion. Investigation indicated that ettringite, initially formed during summer months, was converted to thaumasite by exposure to cold winter conditions.
In both these Lower Lias Clay cases, some of the sulfate involved in the expansive reaction was originally present in the clay as iron sulfide (FeS 2). This commonly occurs in clays (up to 5% by mass in Lower Lias Clay) in the form of disseminated minute grains of the mineral pyrite (Figure 3).
In the presence of air and water, the pyrite oxidises, initially leading to the formation of sulfuric acid.This reacts with any calcium carbonate in the ground to form gypsum (CaSO 4.2H 2O).This process is itself significantly expansive, but its main significance is that pyrite oxidation can raise sulfate levels in stabilised material to harmful levels.
While in nature the process of clay oxidation is relatively slow, possibly taking many years, it is important to understand that oxidation is greatly speeded up by the 'maceration' of the clay needed to mix in lime and/or cement. This exposes the pyrite to water and air, and also possibly renders the pyrite less stable owing to an increase in pH to values in excess of 12.
Guidance BS 1924:1990 , the British standard for testing stabilised materials for civil engineering purposes, does not warn of possible failure arising from sulfate reactions, or include tests for sulfates or sulfides in the section devoted to chemical tests. Also, while it includes a swelling test procedure in Clause 4.5 'Laboratory determination of California Bearing Ratio' this is not ideally designed for the determination of swelling resulting from sulfide/sulfate reactions.
Because of these shortfalls, designers of soil stabilisation for built development have generally looked to Highways Agency publication HA 74/00  for guidance in respect of assessment of materials to avoid sulfate attack.
This key document, written in the light of the M40 case, gives extensive guidance on the assessment of sulfates and sulfides in materials stabilised for use in highway construction. However, while it mentions 'detrimental' swelling of lime stabilised clay soils resulting from reaction with sulfates, it does not point out that a similar problem can occur with cement stabilised clay soils. The latter phenomenon has occurred on UK sites and, indeed, is to be equally expected since this reaction is akin to the long recognised sulfate attack on concrete that has been extensively researched in the UK and elsewhere [6, 7, 8].
New European standards are being introduced for testing stabilised materials as BS EN 13286 Parts 1 to 53. They will replace BS 1924:1990 by December 2004. Still in draft for future publication are general specifications for stabilised materials.These will be issued as Parts 1-14 of a future BS EN 14227 for 'Unbound and hydraulically bound mixtures' Test protocols for sulfates and sulfides A background factor in this review is the author's responsibility in recent years for BRE's research and preparation of guidance on sulfates and sulfides in natural ground, fills, and hardcore.
Assessment of ground for sulfates and sulfides in respect of sulfate attack is an area where BRE has long taken the UK lead, the earliest guidance dating back to 1938 when concrete design was first specified in terms of sulfate classes . Guidance has been published in a long series of BRE Digests and supporting technical reports, the most recent ones being Special Digest 1 (SD1)  and BR 279 . BSI standards for sulfate and sulfide testing, both for concrete specification  and more general civil engineering application , have generally followed the BRE precedent guidance.
When BRE SD1 was much revised relative to predecessor digests to cater better for the thaumasite form of sulfate attack (TSA), it also overhauled the protocol for sulfate and sulfide assessment.
Key among the changes was the introduction of a formal procedure for taking into account the sulfide (pyrite) content of the ground for cases in which pyrite might oxidise owing to disturbance by construction processes.
Failure to take this process into account is thought to be a key factor in over 30 cases of TSA found in recent years in motorway and highway bridges founded on pyritic Lower Lias Clay . Another change was the recommendation of modern instrument-based chemical analytical methods to replace manual chemical procedures advocated by the British standard for soils testing, BS 1377:1990 .
The recommended new procedures were harmonised with a new TRL protocol (TRL 447 ) for assessment of sulfate and sulfide for application to structural backfills for highway use.The TRL work was, coincidentally, commissioned in response to corrosion of buried steel structures owing to failure to take into account pyrite in fill materials.
In its recent studies of soil stabilisation problems BRE has re-evaluated the recommendations of HA 74 for sulfate and sulfide testing in the light of the new BRE SD1 protocols. The following sections of this paper suggest some improvements to the HA 74 recommendations and, where appropriate, offer alternative procedures for evaluation of material prior to soil stabilisation.
Because of the often highly variable distribution of sulfate and sulfide in the ground, it is important to test a sizeable number of samples of the material to be stabilised. Preference is therefore given to methods which are quick and economical to carry out, rather than those which deliver highest accuracy.
Tests for total sulfate
The total sulfate content (also referred to as the acid soluble content, AS) is of key importance to soil stabilisation as it gives a measure of the 'reservoir' of sulfates available to react with lime and/or cement, and therefore (in the absence of pyrite) gives an indication of the likely magnitude of ground expansion.
In this respect, assessment of the ground for soil stabilisation is different to assessment of the ground for concrete durability. For the latter, the key parameter is the likely concentration of sulfates that may be leached from the soil and transported to the concrete by groundwater - this is appropriately assessed by an index test, the 2:1 water/soil extract test, the result of which depends more on the relative proportions of calcium, magnesium, and sodium sulfate than the total amount of sulfates present.
HA 74 recommends the procedures given in Part 3 of BS 1377.While these are technically acceptable, they are slow to carry out and relatively expensive. SD1 and TRL 447 additionally recommend analysis of the acid extract by inductively-coupled plasma atomic emission spectroscopy (ICP-AES).
This modern instrumental method is widely available in the major test houses as it is extensively used for ground contamination analytical work.The technique is fast and economical for large numbers of samples, a typical acid-soluble sulfate test costing under £10. Results should be requested in terms of %SO 4.Tests for total sulfur The total sulfur (TS) content is important as it gives a measure of the total amount of sulfate that could potentially accumulate in the ground, if that resulting from oxidation of sulfides (particularly pyrite) is added to the existing sulfates.
HA 74 follows BS 1377:1990 in recommending that total sulfur be determined by the procedure detailed in BS 1047, the standard for specification of blastfurnace slags for use as aggregates .
However, this test is a multi-stage wet chemical method that is laborious, time-consuming and expensive to carry out, and is therefore far from optimal for site assessment work.
BRE has long recommended , as a preferred alternative, an instrumental method based on high temperature combustion in an oxygen environment.The sulfur present is evolved as sulfur dioxide which can be detected by passing through an infra-red cell. The instruments most commonly used by UK test houses are sulfur/carbon determinators from the Leco Corporation and Eltra.
Instruments such as the Leco SC-144 heat a small (0.5g) sample together with an appropriate catalyst in a ceramic crucible to about 1,400infinityC. An analysis takes about five minutes. Provided the instrument is appropriately set up and calibrated the valid measurement range for sulfur is from about 0.01% to over 5%. For large batches, the cost per analysis can be as little as £5.
A further practicable method for sulfur determination for site assessment, described in TRL 447, Test 4A, is the microwave digestion method. For this, the sulfur is first extracted by addition of aqua regia (3 part nitric acid:1 part hydrochloric acid) while the sample is being irradiated with microwaves in a pressure vessel. The amount of sulfur is then determined by ICP-AES. The procedure is said to be relatively simple, but does take two hours per sample.
In respect of detection and measurement of pyrite content, it is recommended that the procedure given in BRE SD1 be followed:
(i) Calculate the Total Potential Sulfate (TPS), defined as the sulfate equivalent of the total sulfur (TS).
By calculation based on atomic mass, TPS %SO 4= 3 x TS %S.
(ii) Note that TPS may be considered equivalent to the sum of the acid-soluble sulfate (AS) pre-existing in the material, plus that which might be produced by oxidation of any pyrite.
(iii) Calculate the oxidisable sulfides (OS) of the sample defined as the difference between TPS and AS, ie OS %SO 4= TPS - AS.
(iv) Note that OS is a measure of the pyrite content, though expressed as SO 4.The value will be conservative if the sample initially contained sulfur in organic inclusions or in minerals such as barite which are not acid-soluble.
The presence of any significant level of OS should trigger a detailed review as to whether the material is suitable for lime/cement stabilisation. Specialist advice should be sought if necessary.
Tests for sulfate in 2:1 water/soil extract As explained above, water-soluble sulfate by the 2:1 water/soil extract method will not usually be a primary parameter for assessment of the material for soil stabilisation. However, its determination in the material to be stabilised and in adjacent materials will aid judgement of suitability for soil stabilisation and the need for design details such as installation of site drainage.
HA 74 recommends the determination of water-soluble sulfate in materials present within 500mm of stabilised capping 'as sulfates in solution can migrate into the capping' For the built environment, where permissible risks perhaps need to be set at lower levels than for roads, the extent of the possible sulfate source zone to be investigated seems to be set too small at 500mm.
Judgement should be exercised on distances well in excess of this when mass soil permeability is moderate to high and there is a possibility of sulfate transfer by water flow, including rain infiltration.
The methods for analysis of water-soluble sulfate recommended by HA 74 are those of BS 1377: Part 3, namely the gravimetric method and the ion-exchange method.The first is a tedious wet chemical method, which is relatively slow and expensive to carry out. It is suggested that the new recommendations of BRE SD1 and TRL 447 be additionally adopted, allowing determination of the sulfate (as elemental sulfur) by ICP-AES.The results should be requested in terms of g/litre SO 4rather than the obsolescent convention of g/litre SO 3used in BS 1377.
In respect of interpretation of the results, it is relevant to note that the maximum solubility of calcium sulfate (gypsum) in solution is 1.44g/litre SO 4.In UK clays, values of water-soluble sulfate in excess of this limiting value generally indicate the presence of the more highly soluble and mobile magnesium and sodium sulfates which demand caution.The magnesium cation is particularly implicated as increasing the risk of severe sulfate attack of Portland cement binders .
Tests for sulfate in groundwater
Determination of sulfate in groundwater is important because it may, under conditions of adverse drainage, result in sulfate ingress into the stabilised material. HA 74 recommends the same BS 1377: Part 3 sulfate determination methods for this as for water-soluble sulfate. Again, it is suggested that the new recommendations of BRE SD1 be additionally adopted, allowing determination of the sulfate (as elemental sulfur) by ICP-AES.
Swelling tests on candidate stabilisation mixes are an important laboratory determinator as to suitability for use whenever chemical assessment of the material shows a significant presence of sulfate/sulfide. HA 74 refers practitioners to the BS 1924: Part 2:1990 for a basic test procedure, though some modifications to test duration are recommended. The relevant test in this BS is being replaced (by December 2004) by a new European Standard, BS EN 13286-47: 2004 Test method for the determination of California bearing ratio, immediate bearing index and linear swelling .
Both the former and new tests specify measurement of swell of a sample in a CBR mould that is laterally constrained and exposed top and bottom to water at a temperature of 20 ±2 infinityC.While BRE has not had the opportunity to carry out research on this swelling test procedure, its experience with sulfate reactions in cementitious materials suggests this temperature is too high to facilitate a thaumasite type of sulfate reaction, as thaumasite forms much more readily at temperatures in the range 5degreesC-15degreesC.
In its view, a procedure is called for that includes testing at temperatures that are more typical of UK ground conditions, which for much of the year are less than 10infinityC.
The BS EN test also specifies a minimum swelling test duration of four days, after which it may be terminated if 'flattening of the curve indicates swelling is substantially complete' It should be noted that this relatively short duration may be insufficient to detect swelling caused by sulfates where the soils to be stabilised initially contain pyrite rather than sulfate.
In the context of discussion of appropriate test temperatures, it should be noted that there is also a new European standard BS EN 13286-49: 2004: Accelerated swelling test for soil treated by lime and/or hydraulic binder .
BRE does not consider this a generally appropriate test for expansion caused by sulfates, since the test temperature in the swelling phase is 40 ±2infinityC. This is well beyond the temperature range where thaumasite forms and is also rather high for ettringite formation.However, the test procedure does have some merits as compared to the BS EN 1328647 test: it allows the sample freer access to water via a circumferential permeable jacket and permits lateral expansion which can be important in some construction applications.
Of these documents, only HA 74 gives guidance as to what magnitude of swelling is acceptable.
For capping materials for highway construction, the recommendation is that the average degree of swelling should be less than 5mm (3.9%) as measured on a 127mm high CBR mould sample, with no individual sample swelling more than 10mm (7.8%).
It is important for designers of built development to note that, while these limits may have been found acceptable in practice for road construction, they may well be too lax for other applications such as under floor slabs of buildings, where movement tolerances are much less and the engineering and financial consequences of ground heave are much greater.
Strategy for clay site and material appraisal Experience of sulfate-bearing ground suggests that a primary consideration in any assessment of the ground or of imported material for soil stabilisation is the variability of sulfate concentration in the ground. The following outline strategy is suggested for evaluation of clay soils for possible stabilisation for built development:
(i) Carry out a site investigation to determine the geological and manmade materials on site and the hydrology (water regime).
l Note that sulfates and/or sulfides are found in many UK sedimentary strata, including ancient marine clays such as Carboniferous shales, Lias Clay, Oxford Clay, Kimmeridge Clay, Gault and London Clay, in Glacial Till derived from these, and in recent estuarine silts and clays.
l Note that in most geological cases, the top metre or so of undisturbed ground is normally low in sulfate, due to leaching by rainfall. High levels of sulfate are often found at the base of the (former) tree root zone and at the base of the weathered zone.
l Note that sulfates and/or sulfides are also found extensively in made ground comprised of reworked geological strata, industrial wastes and building debris. Distribution of sulfate and sulfide in such material is likely to be much more variable and zonation with depth due to leaching may not be well developed.
(ii) Carry out systematic or targeted sampling of the materials to be stabilised and any adjacent material using trial pits or boreholes, as seems appropriate to the identified materials.
l Systematic sampling would typically be on a specified grid pattern, say at 10m intervals laterally and every 200mm vertically.
l Target sampling would be for visually identified 'interesting'materials.
(iii) Carry out total sulfur tests on 'quartered down' fractions of the samples using the high temperature combustion method: 100 tests would probably cost less than £500.
(iv) Review the total sulfur test results together with other site investigation data. If any total potential sulfate (TPS) equivalents to the total sulfur are in excess of 0.2% SO 4, then (a) consider taking more samples on a closer spaced grid to delineate hotspots, (b) proceed to (v).
(v) Carry out acid-soluble sulfate tests on materials with TPS >0.2% SO 4and determine the present concentration of sulfates (AS) and by comparison with TPS, any presence of pyrite.
(vi) For material with AS confirmed as >0.2% SO 4or concluded to contain pyrite, carry out swelling tests on candidate stabilisation mixes to determine the magnitude and rate of expansion. Consider, modification of current test procedure in the light of foregoing comments.
(vii) Evaluate the risks to the proposed development from any observed swelling, noting that the above quoted HA 74 swelling test limits of 3.9% average and 7.8% max may be too high as an indicator of allowable swelling for built development.
Conclusions In recent years there have been numerous cases where lime and cement stabilised clay soils intended for built development have failed to meet performance specification owing to sulfate attack that has resulted in ground heave and softening. Unlike a current case on a highway under construction [18, GE April], most of such cases have not received publicity, yet some have been costly to rectify and have been the subject of behind the scenes dispute.
From long experience of appraisal of sulfates and sulfides in the ground, expansive sulfate reactions in cementitious materials and from recent involvement in some cases of failure, BRE has become aware of deficiencies in practice for assessing soil stabilisation for built development. Reasons include:
the failure to employ appropriately qualified practitioners for appraisal;
the failure to test enough soil samples of material where sulfate levels vary;
the use of inappropriate guidance on sulfate, sulfide and swelling testing;
not selecting appropriate stabilisers when sulfate levels are substantial, eg granulated ground blastfurnace slag (ggbs) may have greater tolerance for sulfates than lime or Portland cements .
the use of sulfate and swelling tolerance limits that are designed for roads and are inappropriate for buildings.
In part, this scenario is similar to a position we have been in before in respect of built development: about a decade ago it was realised that fills used as support for building construction purposes were often being under-specified because the guidance being used was that prepared for road design. A study of the problem culminated in 'a model specification for engineered fill for building purposes' as reported by Trenter and Charles (1996) .
It is our view that an authoritative guide should likewise be produced for application of soil stabilisation to the built environment. This paper based on BRE's experience is offered as the first step towards new guidance, but with the awareness that firmly based authoritative guidance will require a pooling of experience on soil stabilisation problems, and research, particularly on swelling tests, to underpin recommendations.
1 BRE (2002). BRE Report 441. Avoiding deterioration of cement-based building materials and components. Lessons from case studies: 4. CRC, London, 2002.
2 Snedker EA (1996). M40 Lime stabilisation experiences. Lime Stabilisation: Proceedings of Seminar at Loughborough University.
Thomas Telford. London. pp 142-158.
3 Snedker & Temporal J (1990). M40 Motorway Banbury IV Contract - Lime stabilisation. Highways & Transportation. December 1990.
4 British Standards Institution (1990). BS 1924-2: 1990. Stabilized materials for civil engineering purposes. Part 2: Methods of test for cement-stabilized and lime-stabilized materials.
5 Highways Agency (2000). Report HA 74/00. Treatment of fill and capping materials using either lime or cement or both. Design Manual for Roads and Bridges.
6 Lea F and Desch C (1935). The chemistry of cement and concrete, Edward Arnold, London.
7 Department of Environment, Transport and the Regions. Report of the Thaumasite Expert Group: The thaumasite form of sulfate attack: Risks, diagnosis, remedial works and guidance on new construction. Department of Environment, Transport and the Regions, January 1999.
8 Skalny J, Marchant J and Older I (2002). Sulfate attack on concrete. Spon Press, London.
9 Longworth TI (2004). Development of guidance on classification of sulfate-bearing ground for concrete. Concrete.Vol 38, No 2, pp.2526.
10 BRE (2003). Special Digest 1: Concrete in aggressive ground. CRC, London.
11 Bowley MJ (1995). Sulphate and acid attack on concrete in the ground: recommended procedures for soil analysis. Report BR 279.
12 British Standards Institution (2002). BS 8500: 2002. Concrete - complementary British standard to BS EN 206-1: Part 1: Method of specification and guidance for specifiers: Part 2: Specification for constituent materials and concrete.
13 British Standards Institution (1990). BS 1377: 1990. Methods of testing soils for civil engineering purposes. Part 3: Chemical and electro-chemical tests.
14 Reid JM, Czerewko MA and Cripps JC (2001). Sulfate specification for structural backfills. TRL Report 447. Transport Research Laboratory.
15 British Standards Institution (1983). BS 1047: 1983. Specification for air-cooled blastfurnace slag aggregate for use in construction.
16 British Standards Institution (2004). BS EN 13286-47: 2004. Unbound and hydraulically bound mixtures for roads - Part 47 Test method for the determination of California bearing ratio, immediate bearing index and linear swelling.
17 British Standards Institution (2004). BS EN 13286-49: 2004 Unbound and hydraulically bound mixtures for roads - Part 49:
Accelerated swelling test for soil treated by lime and/or hydraulic binder.
18 Parker D (2004). Hertfordshire bypass heaves under sulphate attack. New Civil Engineer. 11 March 2004.
19 Higgins DD (1998). The use of lime/GGBS in soil stabilisation. Concrete. Vol 32, No 5, 1998, pp15-18.
20 Trenter NA and Charles JA (1996). A model specification for engineered fills for building purposes. Proceedings Institution of Civil Engineers, Geotechnical Engineering, Vol 119, October 1996, pp219-230.