GLOBALISATION OF ANALYSIS OF SMELTER GRADE ALUMINA

Frank M. Kimmerle1 and Gregory J. Yaxley2

1 Convenor ISO/TC 47/ SC 7 Working Group 3

Alcan International Limited, Arvida Research and Development Centre

P.Q. Box 1250, Jonquire, Qubec, Canada G7S 4K8

2 Chairman, Standards Australia MN/9 Committee

Yaxley Scientific, 38 Maroong Drive

Research, Melbourne, Victoria, Australia, 3095

ABSTRACT

Only a quarter of the annual world smelter-grade alumina production is used within vertically integrated aluminium companies, with the rest being exchanged or traded on national and international markets. Although common analytical techniques within the industry are a pre-requisite to assure conformity to contractual specifications, most ISO methods for the characterisation of alumina have not been revised since the mid-seventies. The more recent Standards Australia MN/9 Committee methods correspond well to the modern techniques widely used within the industry but international consensus had not been sought during their development. The major alumina producers met informally in 1995 and annually since then, as an "International Smelter Grade Alumina International Standards Group" to identify the critical parameters and prioritise efforts towards establishing common analytical protocols.

This communication reports on the efforts we have made since as an ISO Working Group to organise international round-robins, to consider direct adoption or modification of MN/9 Committee standards for recommendation as ISO methods and our progress towards the production of common certified reference materials. It seeks contributions from a wider community to accelerate the standardisation of methods for the physical and chemical characterisation of alumina.

KEY WORDS:

smelter grade alumina, analysis, characterisation, ISO, standards

GLOBALISATION OF ANALYSIS OF SMELTER GRADE ALUMINA

Frank M. Kimmerle and Gregory J. Yaxley

1.0 INTRODUCTION

Only a quarter of the annual world smelter-grade alumina production is used within vertically integrated aluminium companies, with the rest being exchanged or traded on national and international markets. Many of the Bayer plants commissioned in the last few years are owned by consortia involving the major aluminium companies and yet no common analytical methodology has been adopted throughout the industry. Few of the ISO methods for the characterisation of alumina have been revised since the mid-seventies and most of the ISO methods lack the now mandatory (for new and revised standards) precision data. The more recent Standards Australia MN/9 Committee methods correspond well to the modern techniques widely used within the industry but were developed and accredited with almost no participation from the larger international community. Following the presentation of Ledru and Roach (Ledru, 1993) at the 3rd Alumina Quality Workshop, management of Pechiney and Alcoa invited representatives from Alcan, Alcoa of Australia, Alumax, Billiton, Comalco, Kaiser, Norsk Hydro, Pechiney and Reynolds Metal Co. to meet during the 1995 Light Metals Conference (TMS) in Las Vegas. The objectives of this initial informal meeting were:

  • to determine whether a need exists to establish standard protocols, reference materials and instrumentation for the characterisation of smelter grade alumina (SGA),
  • to establish the priorities of subjects to be tackled, and
  • to establish a working context.

It was agreed that the discussions be public, be published and, in order to tackle the technical issues, a very loose structure was adopted. While the participants were reluctant to simply adopt the Standards Australia methods, it was agreed that where possible they should serve as a starting point for any development or be validated through round-robins if judged appropriate in their present form. Discussions during the three following Light Metals conferences and the 1996 Alumina Quality Workshop have led to consensus on several analytical issues and this SGA International Standards Group (SGA-ISG) was subsequently restructured as ISO working committee ISO/TC47/SC7/WG3 in 1997.

2.0 EVOLUTION of ISO/TC47/SC7 Working Group #3

The chairman of ISO/TC47/SC7, (W. Schmidt-Hatting), attended the 1997 SGA-ISG meeting and recommended that the group accept independent Working Group status of the ISO SC7 committee which deals with Standards for Alumina, Fluoride and Carbonaceous Materials for the Production of Aluminium. This recommendation was made at the behest of the SC7 committee in order to facilitate the upgrading of ISO alumina standards since the expertise of the majority of SC7 members was in carbonaceous materials which had been the focus of SC7 since the mid 1970's. Members agreed to the recommendation and elected F. M. Kimmerle as Convenor and G. Yaxley agreed to act as Secretary. The SGA-ISG now carries the ISO designation ISO/TC47/SC7/WG3 as ratified at the Seattle meeting of SC7 in September 1997.

This mechanism allows WG3 to present draft standards for consideration by SC7 with subsequent "fast tracking" to DIS and subsequently ISO status. It also permits the use of precision data from MN/9 standards for WG3 submissions, as long as the analytical methodology has not changed in a significant manner.

3.0 ROLE of MN/9 and CURRENT ACTIVITIES

The Standards Australia MN/9 Committee has been active in the development of alumina standards since its formation at the request of the combined Australian alumina refining and smelting industry in the early 1980's. It has the same role within Australia as the SC7 committee has within ISO but has dealt only with alumina standards and these have had the benefit of continuing development over the last two decades. These standards generally correspond to current industry methods and all incorporate precision data determined by round-robins within the Australian industry. Because of these benefits, the MN/9 standards were suggested by the convenors as a basis for the SGA-ISG's work prior to the inaugural meeting.

Although direct adoption of the MN/9 standards existing in 1995 was rejected at the inaugural meeting, WG3 has collaborated with MN/9 in order to have its views considered in new and revised MN/9 standards to facilitate their subsequent adoption. Several of these are now close to adoption as detailed in Section 5. Additionally, one of the two most important existing standards, AS2879.6, "Particle Sizing by Electroformed Sieves", has with minor modifications been passed by WG3 to SC7 for approval as an ISO standard, see Section 5.1

Because WG3 will continue to rely on previous work by MN/9 and is likely to participate in future in some of their analytical developments, it is pertinent to summarise their current program. Table 1 lists MN/9's current (as of Sept. 98) program including development of new standards and "guides" and revisions to existing standards. It also notes certain points of interest in this work and gives a summary of major changes proposed to standards under revision.

Table 1

Current Status of MN/9 Activities

  Standard

Likely Changes, Points of Interest and Status

AS 2879-1

Under revision

Loss of mass at 300oC & 1000oC

Choice of manual or automatic instrument method.

Use of vacuum desiccation, air inlet through moisture trap.

More detailed description of sample transfer and handling.

  • Revision close to finalisation.
AS 2879-2

Under revision

Particles passing a 20m aperture sieve

Removal of calibration by NIST glass beads.

Mandatory use of round-holed electroformed sieves.

Retain use of full sample, i.e., no scalping.

  • Round-robin in progress.
AS 2879-3

Under revision

Alpha alumina content by X-ray diffraction

Precision limitations on flash-calcined alumina noted; in-house variations and improvements are solicited.

  • Revision in early stages.
AS 2879-4

Under revision

Specific surface area by N2 adsorption

Normalisation to carbon black CRM to be dropped.

  • Alumina reference material being developed with WG3.
AS 2879-5 Angle of flow
  • No current work.
AS 2879-6 Mass distribution of particle sizes using electroformed sieves

No objection to WG3’s extension of sample mass to 50g.

  • Nominated for ISO DIS status.
AS 2879-7 Trace elements - XRF method

Interest by some MN/9 members to extend to include determination of S and F.

  • WG3 working on mods. for flux choice and S addition.
New Standard Attrition Index

Original approach of tight parameter and design standardization abandoned.

Convergence of results finally achieved using Alcan reference alumina method. Working draft now based on this method.

  • Alumina reference material being developed with WG3.
New Standard Flow Time

Based on Alcoa flow funnel.

  • Second round-robin in progress.
New Standard Bulk Density- Loose and Packed

Draft based on QAL modification of ISO 901 (Davies, 1993).

Moisture correction included – improves precision twofold.

  • Public review imminent.
New "Guide" Guide to sampling

Generic ore sampling guide by MN/10 adapted for sampling of alumina - mainly intended for belt sampling.

  • Public review done. Publication likely first half 1999.
New "Guide" Guide to sample preparation
  • Public draft likely to be available early 1999.

4.0 ANALYTICAL PRIORITIES and INITIAL SGA-ISG ROUND-ROBIN RESULTS

Replies to the questionnaire distributed at the inaugural 1995 SGA-ISG meeting indicated a much higher urgency for agreement on methods for physical characterisation than for chemical analysis. The order of the priority expressed was as given in Table 2, column 1. To determine the level of agreement available at the time for a range of parameters, a first round-robin using a carefully homogenised, typical smelter grade alumina was organised with eight participating laboratories analysing in triplicate using their own in-house methods. The results of the round-robin were reported at the October 1995 Montreal meeting and are discussed in Sections 4.1 and 4.2 below.

Since it is not possible within the scope of this communication to describe in detail the various methodologies employed by the participating laboratories, we will largely limit the discussions to the differences observed. Analysis of the data using ASTM E691-92 statistical treatment is summarised in Table 2 where "r" indicates the average intralaboratory repeatability and "R" the interlaboratory reproducibility expressed as one standard deviation in the same units as the average value. Parameters reported by only one laboratory are not included. The results largely confirmed the perceptions expressed and lay the groundwork for further development work.

Table 2

Initial Survey - Expressed Priority of Analytical Needs and Round-Robin Results

Test Method

1st Round-Robin Results

 

Average

Sr*

SR

Comment

Physical Characterisation

       

Granulometry by sieving (Ro-Tap)

50.9 %

2.9

10.4

> 75 m

Granulometry of superfines

1.07 %

0.09

0.56

< 20m

BET surface area

63.7 m2/g

0.7

3.4

1 outlier

Attrition Index

16.1

1.0

5.0

7 labs

Flowability

different methods

   

2 labs only

Moisture RT- 300oC

2.62 %

0.03

0.23

7 labs

LOM 300 – 1000oC,

0.77 %

0.022

0.070

 

Alpha alumina

4.2 %

0.26

0.9

7 labs

Bulk Density (loose)

0.980 g/cm3

0.008

0.028

6 labs

Bulk Density (tapped)

1.173 g/cm3

0.011

0.040

6 labs

Angle of repose

32.9

0.6

0.8

4 labs

Angle of flow

55.5 o

0.5

16.7

2 labs only

Chemical Analysis

       

Na2O

0.373 %

0.009

0.020

 

Fe2O3

0.0082 %

0.0005

0.0009

 

SiO2

0.0144 %

0.0013

0.0024

 

CaO

0.0245 %

0.0010

0.0023

 

P2O5

<0.0003 %

   

LOD

SO3

0.014 %

   

Sensitivity

Other minor oxides

       

* Sr, SR standard deviation of repeatability, reproducibility

4.1 Physical Characterisation

A large variety of methodologies, a number of in-laboratory protocols and both commercial and home-made equipment were used by the various laboratories, making exact comparisons more difficult but underlining the need for standardisation. The greatest problem lay with interlaboratory precision of granulometry methods (both for shipment and fines samples), followed by surface area and then attrition index.

4.1.1 Particle size analysis

Granulometry by Ro-Tap: Electroformed sieves showed better repeatability (0.3%) than woven sieves (1.5%) and higher particle size for the coarser fractions. Contrary opinions were expressed whether sieves should be certified using glass beads or whether manufacturer’s certification should be relied upon. The numerical results showed unacceptable scatter between laboratories for this important parameter (See item 5.1).

Granulometry of Fines: One laboratory reported results for the <45 m fraction by wet sieving using electroformed sieves, one laboratory reported single orifice electric sensing (ELZONE or Coulter Counter equipment); the others used various versions and models of laser diffraction instruments that use different software treatments. For the <20 m fraction, three laboratories first scalped the sample on a dry 45 m sieve, whereas three others used the entire sample.

(At the present stage of development (1998) it is unlikely that a laser diffraction industry standard will be developed soon. Recent perceptions are that for contractual purposes dry sieving is likely to remain the preferred technique for the > 45m fraction and wet sieving for <45m fractions.)

4.1.2 Specific surface area

When one outlier result (also using conventional single point BET) was discarded, the agreement among the remainder was adequate and no immediate action by the group was deemed necessary. Precision but not accuracy of the analysis seemed to be inversely proportional to the aliquot weight employed (indicative of sampling problems).

4.1.3 Attrition index

In-house equipment and different operating parameters led to considerable scatter of results. Using calibration techniques within its own laboratories however, Alcan reported interlaboratory standard deviations half of those found during this initial international round-robin. (See item 5.3).

4.1.4 Moisture and loss of mass

Almost all laboratories are switching or have switched to automatic equipment with a major improvement in repeatability over the manual techniques. Within the reproducibility reported, variations in the end temperature from 1000oC to 1200oC did not affect the LOM, although standardisation would be desirable. Sample pre-treatment constitutes the main issue.

4.1.5 Alpha alumina, gibbsite

All laboratories used an XRD calibration for alpha alumina, with the differences arising from the different 100% alpha alumina reference materials used. For the analysis of gibbsite it was generally agreed that differential scanning calorimetry (DSC) is the preferred method, but in any case this parameter is too low in fluid flashed alumina to be of great interest.

4.1.6 Bulk density (loose and tapped), angle of repose, angle of flow

Different definitions and different in-house equipment were used for bulk densities (e.g., free flowing or tapping a graduated cylinder with a rubber mallet or spoon, commercial or semi-commercial instrumentation). Angle of repose can distinguish between sandy and floury alumina but is too insensitive to distinguish between grades of modern SGA. Angle of flow has greater sensitivity but has not gained wide acceptance. These five parameters have historically received low priority, however for various reasons still find their way into contract specifications.

4.1.7 Sampling and sample handling

Opinions were mixed whether samples should be tested "as received", dried at 105oC or dried at 300oC. Exposure to atmosphere prior to analysis was seen to be more important for freshly produced alumina and considered to have a direct influence on some of the physical characteristics measured.

4.2 Impurity (Elemental) Analysis

Six of the eight participants used XRF (by briquetting or fusion), one laboratory used ICP and one employed classical chemical analysis. For Na2O, a relative standard deviation (RSD) of 5% was considered acceptable, while for Fe2O3, SiO2, and CaO, RSD’s of up to 17% were observed. Phosphorus analysis sufficed at the 50 ppm level but gave little information below 10 ppm; zinc analysis was poor but did allow distinguishing low zinc from high zinc alumina. Few laboratories reported the XRF sensitivity necessary for a sulphur mass balance involving gas-fired alumina.

5.0 ANALYTICAL DEVELOPMENTS

Table 3 indicates the current ISO and Standards Australia protocols for the characterisation of SGA. The dates of revision confirm that very little ISO activity has addressed the needs of the alumina industry in the past two decades. The status of current work on the Australian standards is given in Table 1. The Standards Australia protocols complemented by in-house methods used by the principal alumina producers are thus a more likely starting point towards achieving consensus on modern analytical techniques.

 

Table 3

ISO and Standards Australia Methods for the Characterisation of Alumina

#: date of issue

ISO International Standard

#: date of issue

Standards Australia Standard

AS 2879 (closest equivalent)

802:1976 Alumina, preparation & storage of test samples New 'Guide' Series Guide to the sampling of alumina:

Part 2: Sample preparation

803:1976 Alumina, loss of mass, 300oC 1:1986 Loss of mass at 300oC & 1000oC
806:1976 Alumina, loss of mass, 1000 & 1200oC As above As above
901:1976 Alumina, absolute density   No equivalent
902:1976 Alumina, angle of repose 5:1994 Determination of angle of flow
903:1976 Alumina, untapped density 8:1998 Determination of bulk density
2926:1974 Alumina, particle size analysis, sieving 6:1995 Determination of the mass distribution of particle sizes using electroformed sieves
2927:1973 Alumina, sampling New 'Guide' Series Guide to the sampling of alumina:

Part 1: Sampling procedures

2961:1974 Alumina, adsorption index   No equivalent
8008:1986 Alumina, specific surface area 4:1991 Determination of specific surface area by nitrogen adsorption
8220:1986 Alumina, <60 m particle size distribution, electroformed sieves 2:1991 Determination of particles passing a 20 m aperture sieve
  No equivalent 3:1991 Determination of alpha alumina content by X-ray diffraction
804:1976

2073:1976

Alumina, alkaline fusion

Alumina, HCl attack

(precursors to wet chemical elemental determinations below)

7:1997 Determination of trace elements -Wavelength dispersive XRF method.

Elements included: Na,Si,Fe,Ca,Ti,P,V,Zn,Mn,Ga,K,Cu,Cr,Ni

805:1976 Alumina, Fe content   As above
900:1977 Alumina, Ti content   As above
1232:1976 Alumina, Si content   As above
1617:1976 Alumina, Na content   As above
1618:1976 Alumina, V content   As above
2069:1976 Alumina, Ca content, AAS   As above
2070:1981 Alumina, Ca content, s'photom.   As above
2071:1981 Alumina, Zn content, AAS   As above
2072:1981 Alumina, Zn, PAN method   As above
2828:1973 Alumina, F content   As above (but F not included)
2829:1973 Alumina, P content   As above
2865:1973 Alumina, B content   As above (but B not included)
3390:1976 Alumina, Mn content, AAS   As above

At the October 1995 Montreal meeting, project leaders were identified to tackle the analytical priorities identified from the initial survey and round-robin results discussed in Section 4. Preliminary results were reported during informal meetings at the Darwin Alumina Quality Workshop and at subsequent meetings. The subsequent evaluations and actions taken are reported below.

5.1 Particle Size Distribution by Sieving

C. Franz (Reynolds) undertook an evaluation of particle size analysis by dry sieving, comparing results from wire-woven (ASTM E-11) sieves with those from electroformed (ASTM E-161) sieves involving seven laboratories. Differences in results seemed to be largely caused by sample splitting, but a number of other extraneous factors were identified in this and follow-up work. The influence of different models of Ro-Tap shaker used was eliminated by following the manufacturer’s instructions (e.g. of hammer drop settings) and choice of sample weight between 30 and 50g was not deemed critical. However, for both types of sieves, brushing to weigh the individual alumina fractions led to larger dispersion than weighing the entire sieve, particularly for the >75 m fraction.

Figure 1 illustrates work done to compare the accuracy of effective openings of woven wire sieves (calibrated using NBS glass beads SRM 1004A) to that of electroformed sieves. For both types of sieves, the microscopically measured values of the screen openings (diagonal stripes) typically correspond closer to the manufacturers’ nominal values than to the "calibrated" effective openings (hatched bars). Moreover, attempts to correct the values of the openings using the older glass beads, NBS 1004, instead of 1004A led to different results again. In all cases the microscopically measured openings of the electroformed sieves were much closer to the nominal values than the wire sieves and indicated a much narrower spread of the openings (average 0.6 to 1.2 m) than for the wire sieves (average 1.4 to 2.8 m).

Along with other compelling practical reasons reported to WG3 by MN/9, this study convinced us in 1997 to adopt the Australian electroformed sieve standard AS 2879.6 (with minor modifications) as revised method ISO 2926 for dry sieve SGA analysis. This adoption is now at the ISO "DIS" stage.

 

Figure 1

Certification of Sieves using NBS SRM 1004A or image analysis

5.2 Particle Size Analysis <20m

Because of the concern over handling dustiness, several smelter organisations would like to introduce alumina specifications for the < 20 m fraction. K. Hamberg (Norsk Hydro) examined a number of techniques including image analysis and sedimentation methods. Comparative characterisation of the < 20 m SGA fraction showed increasing values in the order: Elzone/Coulter Counter, Dry Sieving + Laser diffraction on < 45 m fraction, laser diffraction on entire sample and finally, wet sieving. Because of its better precision, Norsk Hydro’s reference method is wet sieving using acetone (AS 2879.2) rather than water (ISO 8220), whereas the industry control methods, particularly in the smelter sector favour laser diffraction. Comparing different models of laser diffraction equipment resulted in somewhat different numerical values, see Table 4. Hamberg is continuing to examine a variety of approaches, targeting possible correlations between dusting behaviour during ship unloading and potroom dust measurement. An update of his work will be presented at the 5th Alumina Quality Workshop.

Table 4

Determination of the <20m SGA fraction by different laser diffraction techniques

Instrumentation and protocol

Sample #1

Sample #2

Laser/Microtrac , on <45 m fraction

0.8

1.2

Laser/Microtrac , on total sample

2.4

1.9

Malvern 2600C Particle Sizer on <45 m fraction, surfactant.

0.9

1.2

Malvern MasterSizer MS20 on total sample, no surfactant

1.8

1.3

Malvern 3600E Particle Sizer on <45 m fraction, surfactant.

1.1

1.5

Malvern Mastersizer MAS 5000, on <45 m fraction, surfactant.

0.9

1.9

5.3 Attrition Index (AI)

Loading and unloading of alumina and transport and dry scrubbing within smelters tends to break up alumina particles. As a consequence, the increase in the < 45m fraction under "standardised" attrition conditions has become an important empirical parameter. F. M. Kimmerle (Alcan) undertook to evaluate the attrition index methods used. The key parameters of the equipment and protocols varied considerably among the SGA-ISG members, see Table 5.

Table 5

Selected parameters used to determine attrition index

Analytical step

Range of conditions reported

Sub-sampling

manual blend, linear rifflers, rotary rifflers

Sample preparation

as received, humidity control, 105oC drying

Sizing

dry or wet sieving with varying Ro-Tap procedures, laser diffraction

Apparatus

Column

manual, automatic

straight, multi-diameter

Orifice diameter

381, 397, 400 m

Air pressure

56, 60, 64, 70, >75 psi

Flowrate

varying 5.6.to 7.6 L/min controlled to within 0.05 to 0.5L/min

Repeatability claimed

0.07 to 0.9%

The critical component of the attrition apparatus was found to be the flow through the attrition plate; see Figure 2 for a preferred design. Interlaboratory agreement of attrition index could be obtained if the flow-rate was adjusted to give a fixed value for a reference alumina (in the case illustrated in Figure 2, AI = 15%). Using semi-automated equipment with strict control of the weight of sample, the time of attrition and constant flow rate, intralaboratory repeatability of 0.2% was reported with interlaboratory reproducibility of 0.6% in round-robins involving seven different locations. In this study the sizing techniques appeared to have a limited effect on the final AI values as long as the same measuring technique was used before and after attrition of the sample. An inter-industry round-robin using a single source of attrition plates is planned for 1999. A recent MN/9 round-robin confirmed Alcan's flow-calibration approach and an Australian standard is now in the draft stage based on Alcan's procedure. When finalized, this standard will be considered by the SGA-ISG for presentation to SC7 as a new ISO standard.

Figure 2

Attrition Orifice Plate

5.4 Specific Surface Area

G. Yaxley (Comalco) undertook to examine the use of the Vulcan 3-G (2700) carbon black CRM (as used in AS 2879.4) to calibrate BET instrumentation. A round-robin involving eight laboratories and two alumina samples was organised using this CRM (and absolute measurements) to determine their surface area. The results are summarised in Table 6, where column 2 refers to absolute measurements of the CRM, ALU-10 and ALU-03 samples respectively and columns 3 and 4 represent the repeatability and reproducibility of the measurements at the 95% confidence level. While the intralaboratory repeatability for the CRM (r = 2.04) is acceptable, large differences were shown between laboratories, (R = 9.3) and consequently, the normalised SSA values showed far poorer interlaboratory agreement. Since several laboratories questioned the use of a different sample pre-treatment for the alumina versus the carbon black CRM and since the alumina samples gave more reproducible results in the first place, it was deemed retrograde to normalise the SSA value as specified in AS 2879.4. The round-robin also indicated that a minimum sample weight of 0.2 g and more precise sub-sampling instructions should be included. It is now planned to submit AS 2879.4 as an international standard, with these minor modifications; eliminating normalisation with respect to the carbon black CRM and using a new alumina reference sample for quality control.

Table 6

Round-Robin Specific Surface Area (m2/g)

Material

SSA (abs.)

r

R

Vulcan 3G-2700 (certified)

71.3

 

2.7

Vulcan 3G-2700 (found)

72.9

2.04

9.3

ALU-03

72.9

2.7

6.5

ALU-10

54.9

1.7

3.1

5.5 Alpha Alumina

Discussion of the MN/9 standard AS 2879.3: 1991 centred on the nature of the 100% a standard, sample preparation and the choice of diffraction lines used. Members noted that the intensity of the various a diffraction peaks, including the (012) and (116) lattice plane reflections, vary with the origin of the alumina. For fluid flashed alumina with low a content (3 to 5%) and consequently low peak intensity, variation in background subtraction has become more important and weak interference from sub-a forms of alumina and of a -sodium aluminate have been reported (Martin, 1985). Calcined alumina from various sources has been collected and SGA-ISG action awaits completion of an in-house evaluation of the combination of XRD peaks least sensitive to variations in calcining technology and origin of the SGA alumina.

5.6 XRF Analysis; Fusion Technique

Existing ISO standards determine the impurity levels in SGA using wet chemical determinations, whereas XRF analysis is the preferred technique in alumina refineries. Members examined the recently published standard AS 2879.7 and while supporting the use of borate glass disks (eliminating briquetting) questioned the use of a specific lithium tertraborate/metaborate flux mixture. Development work has been carried out in the meantime to include among the analytes the relatively volatile elements, sulphur and fluorine. A subcommittee meeting is planned for late 1998 to finalise these proposed extensions of AS 2879.7 and to organise a new round-robin to generate necessary statistical data before proposing it as an ISO method to replace the wet chemical determinations.

5.7 Production and Certification of Reference Materials

While it would be desirable to have NIST or other national laboratories produce SGA reference materials certified for key physical properties, it was recognised that this avenue would produce unacceptable delays in revision of several standards. To circumvent this problem, two candidate reference materials are currently being prepared by Alcan prior to certification by WG3 and MN/9 members. The first of these is for SSA (targeted value 80m2/g, with AI < 20), and the second is for Attrition Index (targeted value approx. AI = 15). Approximately 2000 bottles (100 g each) will be produced for SSA and 400 bottles (500g each) for AI. The first of these are planned to be distributed to WG3 and MN/9 members for testing in late 1998. If found acceptable as a reference material, the material will be divided among the participants at cost and then become available to others. This trial certification is expected to lead to certification for other properties and may even be extended to the common production of some simple equipment like attrition plates.

6.0 CONCLUSIONS

The SGA-ISG developed from an initial informal gathering in 1995 to become a structured ISO working group meeting at least once a year with the objectives of reaching agreement on the analytical standards to be used to characterise the physical and chemical properties of smelter grade alumina. This we intend to accomplish by updating or declaring obsolete the existing ISO methods, proposing new standards reflecting modern technology and promoting the production of certified reference materials through co-operative work by the industry.

Close collaboration with the Standards Australia MN/9 Committee work has minimised duplication of effort and expedited what has traditionally been a slow international process. International acceptance of these standards for the future is assured.

REFERENCES

Davies, T. J. and Mercer, A. E. (1993), Bulk Density, Proceedings, 3rd Alumina Quality Workshop, pp 186-199.

Ledru, B and Roach, G.I.D. (1993), Measurement of Smelter-Grade Alumina Properties, Proceedings, 3rd Alumina Quality Workshop, pp 200-212.

Martin, J. P. (1985), Determination of a Alumina in Alumina by XRD, Alcan Method 1109-85.

Schmidt-Hatting, W., (1997), Minutes of the Seattle Meeting of TC47/SC7, Sept.1997.

Mercer, A. and Warren, H., (1997), Sieve Analysis of Alumina by AS2879.6, Private communication to Standards Australia MN/9 Committee., and Minutes of 4th SGA-ISG Meeting Orlando 1997, App. 1.