Dear David
Delphine and I agree : the well known database of
Bob Downs is one of the one we use to look at the most often. A very
good one indeed

Philippe Dillmann
LAPA, LMC UMR5060 CNRS et SIS2M UMR3299 CEA/CNRS

Le 3 sept. 2010 à 18:13, "[log in to unmask]" <[log in to unmask]
U> a écrit :

> Thanks to all for very thoughtful responses to my initial provocation.
>
> Thilo makes a number of good points. Paying attention to the atomic
> ratios of
> metal ions to silica, as calculated by SEM, will indeed distinguish
> olivines
> from pyroxenes, and its commendable that he teaches students to pay
> attention
> to this. But this is not a general solution - for example, the
> ratios of metal
> ions to sulphur in many of the ore minerals are too close to each
> other to
> apply this approach. As Thilo well knows, you must use optical
> microscopy or
> Raman microscopy (together with chemistry) to identify many ore
> minerals.
>
> And I'm not convinced that SEM/EDAX is sufficient for identifying
> the iron
> oxides. Wüstite and magnetite are not the only iron oxides in iron w
> orking
> residues - there is of course haematite, maghemite is often seen,
> and students
> need to be able to recognize the several hydrated iron oxides that are
> ubiquitous in ores and corrosion products. Polarized optical
> microscopy is
> excellent for this - if you have a microscope equipped for both
> transmitted and
> reflected light and a sufficiently intense reflected light source
> (at least
> 100W). I'm interested to learn that Philippe Dillmann and Delphine
> Neff have
> compiled Raman spectra for the iron oxides in corrosion, and would
> like
> to hear
> more. (For those who use Raman, I should also draw your attention to
> the work of
> my colleague Bob Downs (Geosciences, University of Arizona) who is
> generating
> XRD, chemistry and Raman spectra of samples of all known natural
> minerals. The
> spectra for each mineral can be downloaded for free from http://rruff.info/
> ).
>
> This brings me to the least developed area of archaeometallurgy,
> which is the
> study of the ores used. (Thanks to Noel Gale for correcting my
> neglect of Rob
> Ixer - I have been using his excellent Virtual Atlas of Opaque and Ore
> Minerals
> for years). I think that the whole debate over whether arsenical
> copper
> results
> from accidental use of copper/arsenic ores or from deliberate
> additions of an
> arsenic-rich phase might have been settled long ago if
> archaeometallurgists had
> been systematic about investigating the composition and mineral phases
> of green
> minerals on ancient metalworking sites. Because this has not been
> done we have
> had instead a lot of unproductive speculation on this issue.
>
> Thilo is of course right that you can't train students in optical
> microscopy
> within the contraints of an MSc or PhD archaeometallurgy program in
> Britain. My
> solution to this is to recruit students into our PhD program who
> already have an
> undergraduate degree (at least) in Geosciences, and thus have some
> background in
> mineralogy and petrology. (It's much easier for science students to
> correct for
> ignorance of archaeology than for archaeology students to correct for
> deficiencies in science). Of course we don't have the same time
> constraints as
> in Britain - our PhD students take more formal coursework, have more
> time to
> develop skills, and have no sharp deadline for completion of the
> degree. In
> this respect our system is more like that in Germany than that in
> Britain.
>
> I suppose that the conclusion that I would draw from this is that
> you can't
> train students in everything, so archaeometallurgical research teams
> need to be
> made up of teams that cover all the appropriate skills. My plea is
> that these
> teams include someone skilled in optical petrography and ore
> microscopy
> - which
> is not usually the case in current Anglo-American archaeometallurgy.
>
>
> Quoting Thilo Rehren <[log in to unmask]>:
>
>> This is an interesting observation, and one that I share to some
>> considerable extent. In my opinion, one typically does what one has
>> learned
>> in an early / formative stage of one's career, and few people are
>> intellectually mobile enough to embrace different approaches later
>> in life,
>> unless they offer significant advantages. For historical reasons,
>> lots of
>> the German school of '-men' (mainly Bachmann, Hauptmann, Keesmann;
>> more
>> recently with the French addition of Dillmann) have this strong
>> mineralogy
>> background (Begemann is an exception, as a chemist); and I still
>> admire the
>> aesthetically most beautiful thin section micrographs that Ingo
>> Keesmann
>> used to present at conferences. It is no surprise then that they
>> set a path
>> that others in Germany followed. In the UK, scholars such as Gowland,
>> Tylecote, Northover and Salter have been the formative ones: all
>> with hard
>> core metallurgy backgrounds and much focussing on the metal and
>> artefacts,
>> while Morton and Wingrove with their slag studies never made much
>> of an
>> impact beyond their few but important papers they published on iron
>> slags.
>> (Ian Freestone, a mineralogist by training, is an exception, but he
>> quickly
>> moved on to ceramics first and then glass, so is more or less
>> 'lost' for
>> archaeometallurgy; Paul Craddock is a ?physicist? by training? and
>> appears
>> to have pretty much stayed clear of slags).
>>
>> My own experience - and one which I am guilty of passing on to my
>> students -
>> is a certain frustration with XRD, since too often I have seen people
>> happily interpreting the spectra 'as read', and finding quartz next
>> to
>> fayalite or magnetite next to metallic iron: phases which should
>> not coexist
>> in equilibrium and which are either mis-identified, or present due to
>> corrosion and dirt inclusions. A microscope-based method is
>> therefore always
>> my preference, since I see what I analyse, and whether it is a
>> primary or
>> secondary (=weathering) phase.
>> Since thin sections rarely give good information for the opaque
>> phases, and
>> in metal slags it is the opaques which carry at least half of the
>> important
>> information, I decided early on to work with polished blocks rather
>> than
>> thin sections (of course there are polished thin sections, and my
>> former
>> colleague Andreas Ludwig at the Bergbau-Museum in Bochum is a true
>> artist in
>> preparing those). Clearly, this is a direct reflection of my
>> background in
>> ore microscopy... But microscopy is an art more than a science, and
>> it takes
>> some years to get the level of expertise necessary to cover at
>> least the
>> main ore minerals AND metals AND metallurgical byproducts, and the
>> modern UK
>> university system simply doesn't allow this luxury of a sound in-
>> depth
>> training within an MSc or PhD programme. (This is an observation,
>> not an
>> endorsement of the practice.)
>> So what I try to teach our students is the very rudimentary basics of
>> optical reflected light microscopy of the main phases relevant for
>> their
>> particular research, and the judicious use of SEM-EDS analysis as a
>> suitable
>> (!) and relatively commonly available analytical tool. After all, an
>> ordinary SEM-EDS spectrum gives more than just peaks for iron and
>> silica; it
>> does give quantifications both as weight percent (commonly reported
>> in
>> archaeometallurgy, for historical reasons), and as atom percent.
>> Running
>> known stoichiometric compounds as test samples routinely gives
>> result such
>> as 48/52 at% S and Pb for galena (despite the peak overlap problem
>> of S and
>> Pb); surely this is close enough to recognize it as PbS (14 wt% S /
>> 86 wt%
>> Pb). For fayalite and other olivines, we teach that the sum of Me2+
>> ions
>> should be twice the sum of Me4+ (basically just Si4+) ions, and for
>> a member
>> of the pyroxene family, the ratio Me2+ to Si4+ (in atom percent,
>> not weight
>> percent!) should be (and typically is) close to 1. I am sure that I
>> have
>> seen various pyroxenes in European slags, but mostly in copper
>> smelting slag
>> (often more lime-rich than iron smelting slag). Of the top of my
>> head I
>> think that some of our Chinese students have pyroxens in their iron
>> smelting
>> slags, but again those are (very) lime-rich.
>> We proceed similarly with the free iron oxide (mostly wuestite
>> 'FeO' and
>> magnetite / spinel family) as a key indicator of redox conditions,
>> even in
>> heavily weathered samples full of hydroxides etc. in the cracks and
>> pores.
>> Of course, the EDS system does not allow to determine oxidation
>> states
>> directly; but the characteristic minor oxides such as Al3+, Cr3+
>> and Ti4+
>> are restricted to magnetite / spinels, while wuestite is typically
>> relatively clean (but not stoichiometric!), and has typically only
>> has low
>> levels of Me2+ minor compounds, such as Ca2+, Mg2+ or Mn2+. Crystal
>> or
>> aggregate shape is a further important discriminator, as is colour:
>> back to
>> the optical microscope, which as a matter of principle should be
>> done in
>> detail before any SEM work is undertaken. With the odd solid
>> solutions
>> prevalent in many slag systems, XRD data bases are often not able
>> to cope,
>> and throw up weird and wonderful phases, which are often chemically
>> impossible for the material in question.
>>
>> In summary: I agree with Dave's concerns about inappropriate
>> analytical
>> work, but this is not specific for SEM-EDS analyses, but true for any
>> method. Whatever you learn and apply as a method, it is crucial to
>> have
>> sufficient proficiency in it, and to know one's limits and whom to
>> ask if in
>> doubt: or better how to solve a questions through independent logical
>> thinking and literature research. For publications, it is crucial
>> that peer
>> review (either formally as for journals, or informally among
>> colleagues
>> prior to submission to a book chapter etc.) picks up any serious
>> issues. A
>> certain trans-cultural mobility is also quite healthy, to see and
>> learn the
>> approaches of those who work elsewhere. I certainly benefitted
>> enormously
>> from my half year with Peter Northover and Chris Salter, but would
>> never say
>> that I am as fluent in any of their skills as they are.
>>
>> Thilo
>>
>>
>> -----Original Message-----
>> From: Arch-Metals Group [mailto:[log in to unmask]] On
>> Behalf Of
>> [log in to unmask]
>> Sent: 01 September 2010 20:02
>> To: [log in to unmask]
>> Subject: iron silicates in slags
>>
>> I was recently reminded, when reviewing manuscripts for journals,
>> of the
>> striking differences between Anglo/American archaeometallurgists on
>> the one
>> hand, and German/Swiss/French archaeometallurgists on the other, in
>> the
>> matter of identification of minerals in slags and ores from
>> archaeological
>> sites. (As far as archaeometallurgists of other nationalities are
>> concerned,
>> I don't have enough data to offer an opinion). Although I
>> obviously belong
>> to the Anglo/American group, it is clear to me that the German/
>> Swiss/French
>> archaeometallurgists are the more reliable in their mineral
>> identifications.
>>
>> The reasons for this, I think, is that many of the G/S/F group come
>> to
>> archaeology from mineralogy or geology, whereas in the A/A world most
>> archaeometallurgists come to the subject from metallurgy/ materials
>> science,
>> or (increasingly commonly) have no background in science before
>> enrolling
>> for training in archaeometallurgy.
>>
>> One consequence of this difference is that in Anglo-American
>> archeometallurgy the main technique for investigation of slags and
>> ores is
>> the scanning electron microscope with energy-dispersive x-ray
>> analysis
>> (EDAX). Obviously this gives chemical composition, not mineralogical
>> identification. Sometimes this is combined with XRD for mineral
>> identification, but increasingly rarely. Using EDAX alone, mineral
>> identification is, in essence, just an informed guess.
>>
>> Let me give a real example. With iron smelting slags, there is a
>> widespread
>> tendency in A/A archaeometallurgy to assume that if you get X-ray
>> peaks for
>> iron and silica, then the phase is fayalite. But there are other
>> possibilities, namely the iron-rich clinopyroxenes and
>> orthopyroxenes. I'm
>> particularly puzzled by the fact that in African iron slags I quite
>> often
>> see orthoferrosilite (identified in thin section by petrographic
>> methods)
>> but that this has not, to my knowledge, ever been identified in
>> British iron
>> smelting slags. Is this a real difference between slags in the two
>> areas? Or
>> is orthopyroxene just not being identified in British slags? (I'm
>> not
>> picking on the Brits - for obvious reasons I don't have much
>> opportunity to
>> examine iron slags from the Americas!)
>>
>> Does it matter? Even though I don't have a German gene in my body,
>> I do
>> side with the continental archaeometallurgists in thinking that
>> full and
>> accurate description does matter, even if it has no obvious
>> consequences in
>> term of technological reconstruction. I think that students working
>> on
>> extractive metallurgy should be trained in optical crystallography
>> as well
>> as in SEM/EDAX.
>> Yet very few in A/A archaeometallurgy can use the petrographic
>> microscope
>> effectively, and even fewer can identify ore minerals in reflected
>> light.
>> These skills are much more often used in G/S/F archaeometallurgy,
>> and it
>> shows in the quality of their publications on extractive metallurgy.
>>
>> Petrographic and ore microscopy may decline in importance as Raman
>> microscopy becomes more developed - at the moment the limiting
>> factor is the
>> availability of Raman reference spectra for minerals - and Raman
>> microscopy
>> is certainly easier to learn to use. But I don't think that it will
>> entirely
>> replace the optical methods, and I think that we should be pushing
>> students
>> in A/A archaeometallurgy programs to learn some optical
>> crystallography and
>> become at least competent, if not expert, in these methods.