Towards the differentiation of non-treated and treated corundum minerals by ion-beam-induced luminescence and other complementary techniques

ORIGINAL PAPER

Towards the differentiation of non-treated and treatedcorundum minerals by ion-beam-induced luminescenceand other complementary techniques

H. Calvo del Castillo & N. Deprez & T. Dupuis &

F. Mathis & A. Deneckere & P. Vandenabeele &

T. Calderón & D. Strivay

Received: 7 November 2008 /Revised: 23 January 2009 /Accepted: 5 February 2009 /Published online: 26 February 2009# Springer-Verlag 2009

Abstract Differentiation of treated and non-treated gem-stones is a chief concern for major jewellery importcompanies. Low-quality corundum specimens coming fromAsia appear to be often treated with heat, BeO or flux inorder to enhance their properties as precious minerals. A setof corundum samples, rubies and sapphires from differentorigins, both treated and non-treated has been analysed atthe Centre Européen d’Archéométrie, with ion-beam-induced luminescence (IBIL) and other complementarytechniques such as Raman, proton-induced X-ray emission(PIXE), and proton-induced gamma-ray emission (PIGE).IBIL, also known as ionoluminescence, has been used

before to detect impurities or defects inside syntheticmaterials and natural minerals; its use for the discriminationof gemstone simulants or synthetic analogues has beenelsewhere discussed (Cavenago-Bignami Moneta, Gemología,Tomo I Piedras preciosas, perlas, corales, marfil. EdicionesOmega, Barcelona, 1991). PIXE has been frequentlyapplied in the archaeometric field for material character-isation and provenance studies of minerals (Hughes, Ruby &sapphire. RWH Publishing, Fallbrook, 1997; Calvo delCastillo et al., Anal Bioanal Chem 387:869–878, 2007;Calligaro et al., NIM-B 189:320–327, 2002) and PIGEcomplements the elemental analysis by detecting lightelements in these materials such as—and lighter than—sodium that cannot be identified with the PIXE technique(Sanchez et al., NIM-B 130:682–686, 1997; Emmett et al.,Gems Gemology 39:84–135, 2003). The micro-Ramantechnique has also been used complementarily to ion beamanalysis techniques for mineral characterisation (Novak et al.,Appl Surf Sci 231–232:917–920, 2004). The aim of thisstudy is to provide new means for systematic analysis ofcorundum gemstone-quality mineral, alternative to thetraditional gemmologic methods; for this purpose, a Spanishjewellery import company supplied us with a number ofnatural corundum samples coming from different places(part of them treated as explained above). The PIXEelemental concentrations of the samples showed largequantities of calcium and lead in some cases that can belinked to treatment with fluxes or lead oxide. The plot ofthe chromium and iron concentration grouped the samplesin various aggregates that corresponded to the differenttypes of corundum analysed. Micro-Raman complementedthe PIXE analysis corroborating the presence of lead oxidesbut the use of the PIGE technique was not successful forthe detection of beryllium due to the low cross section of

Anal Bioanal Chem (2009) 394:1043–1058DOI 10.1007/s00216-009-2679-y

Work presented at the International Symposium on LuminescenceSpectrometry 2008, September 7–11, Bologna, Italy

H. Calvo del Castillo (*) :N. Deprez : T. Dupuis : F. Mathis :D. StrivayCentre Européen d’Archéométrie—I.P.N.A.S.,University of Liege,Allée du 6 Août, 17, BAT. 15, Sart Tilman,4000 Liege, Belgiume-mail: [email protected]

A. DeneckereVakgroep Analytische Chemie, University of Ghent,Proeftuinstraat 86,9000 Ghent, Belgium

P. VandenabeeleDepartment of Archaeology and Ancient History of Europe,University of Ghent,Blandijnberg 2,9000 Ghent, Belgium

T. CalderónDepartment of Geology and Geochemistry,Universidad Autónoma de Madrid,Ctra. Colmenar km 15, Cantoblanco,28049 Madrid, Spain

the nuclear reaction chosen for its identification. IBIL wascapable of distinguishing between treated and non-treatedsamples of the same type based on the luminescent featuresof the materials.

Keywords Corundum . Ruby . Sapphire . IBIL . PIXE .

PIGE .Micro-Raman

AbbreviationsIBIL ion-beam-induced luminescenceIL ionoluminescencePIXE proton-induced X-ray emissionPIGE proton-induced gamma-ray emissionCEA-IPNAS

Centre Européen d’Archéométrie—Institut dePhysique Nucléaire Atomique et de Spectroscopie

Ulg Université de LiègeUAM Universidad Autónoma de Madrid

Introduction

Corundum (α-Al2O3) is considered to be the mostimportant mineral species after diamond in the frame ofgemmology. It has been known since antiquity and iscommonly named by its most remarkable gemstonevarieties: ruby and sapphire.

The α-Al2O3 mineral is not frequently found pure innature. It is typically located in granite, gneiss, micaceousor chlorinated schists, calcareous or dolomite rocks regu-larly accompanied by spinel, zircon (ZrSiO4) or tourmalinefor instance, and it carries little amounts of impurities suchas silica (SiO2), iron oxide (Fe2O3), manganese oxide (MnO2),titanium oxide (TiO2) and chromium oxide (Cr2O3) [1].Other impurities present may be magnesium, vanadium,copper or gallium.

Pure crystalline α-Al2O3 can be described as a hexag-onal close-packed structure of O2− ions where Al3+

occupies two thirds of its octahedral sites. This materialdoes not present colour by itself and it has no absorptionfrom 160 to 5,000 nm [2]; it is the existence of impuritiesplaced in the octahedral sites, in simple or coupledsubstitution, that rend different hues and tonalities to thegemstones, thus having the well-known crimson rubies(α-Al2O3: Cr

3+) or blue sapphires (α-Al2O3: Fe2+, Ti4+) for

instance. There are other tints however, such as the“padparadscha” (pink–orange sapphire) and more of themwithout any specific name like yellow (with Fe3+) or purple(Cr3+, Fe2+, Ti4+) corundum. Other sapphires are yellowdue to the existence of Mg2+ trapped-hole pairs. Structuraldefects can also give colour to corundum. The most typicaldefects in corundum are F-centres; these are defectsinvolving the absence of O2− that may or may be notsubstituted by the presence of one or two electrons.

Luminescent properties of corundum are also related to itsdefects; rubies and sapphires are luminescent in the visiblerange (sharp band at 694 nm) due to the presence of chromium,and blue sapphiremay show as well some luminescent featuresat higher wavelengths that have been attributed to iron andtitanium, though they have only been spotted in syntheticsapphire, where the quantities of these ions are superior to theamounts that are usually present in natural gemstones [3].Intrinsic defects manifest their luminescent properties withbroader bands around 350 nm. Rubies coming from specificplaces may also show phosphorescence.

Provenance of rubies

Historically, gemstone-quality ruby was mainly importedfrom Asia, Myanmar (also known as Burma or oldBirmania) being the most important producer followed byThailand. Other deposits are found in Sri Lanka, Cambodia,Vietnam, Pakistan, Afghanistan, Tanzania, Kenya, Zim-babwe, Russia, Greenland or the USA [2, 4].

Myanmar rubies come either from secondary clay andsand deposits that have been covered with sterile alluvialdeposits or from calcareous dolomite marbles [1]. ThoughMyanmar remains the number one producer, gemstonequality has decreased since Möng Hsu mines took overMogok exploitation [2].

Mogok has been the primary source of corundum in theworld for over 800 years; its rubies show the muchappreciated pigeon’s blood red shade. At first, Vietnamrubies appeared as a fine alternative to Burmese’s; Thairubies, though showing purer red colour than Mogok’s andno silks (TiO2 inclusions in form of needle or asterism),lack phosphorescent properties that make gemstones moreappealing once faceted. Sri Lanka rubies are too pale andKenya, Pakistan and Afghanistan, for instance, rarely yieldmaterial that is clean enough to use for faceting.

In 1991, Möng Hsu mines were first reported and duringthat decade the 90% of less than 3 ct rubies in the marketwere coming from Möng Hsu exploitations, in spite of thefact that these minerals show heavy fractures, silk cloudsand dark purplish cores that can make them look asrhodolite garnets more than rubies [2].

Treatment of Möng Hsu rubies

The inner bluish or purplish core inclusions present inMöng Hsu minerals have been attributed to high concen-trations of iron and titanium [5, 6]. These dark cores thatMöng Hsu rubies present are usually corrected throughdifferent treatments in order to sell them for jewellerypurposes, together with the sealing of their fractures.

The purplish core is treated subjecting the minerals tohigh temperatures. Heating creates or annihilates different

1044 H. Calvo del Castillo et al.

aspects of the appearance of corundum, depending on thetemperature and use of oxidising or reducing atmosphere: at1,600 °C, it is possible to remove silk and asterism inreducing atmosphere by diffusion of hydrogen [6]; MöngHsu rubies are treated with temperatures that range from1,500 to 1,850 °C during some hours. This makes the highconcentration of impurities in the dark cores diffuse morehomogeneously, thus improving their overall appearance.(Note that the melting point for ruby is 2,050 °C). Heatingis nevertheless an accepted treatment amongst jewellers; infact, no information is given nowadays in jewellery storesabout heating since it is assumed that, unless specified, allcorundum gemstones have been heated to some extent.Practically, the total of the rubies that come from Möng Hsuundergo this treatment.

Fractures are altogether something different; treatmentfor fractures involves nowadays partial melting of theruby’s surface with some flux such as borax (B2O6),calcium borate (Ca2B2O6 5 H2O), calcium or sodiumphosphate (Ca3(PO4)2, Na3PO4) or sodium feldspar forinstance to lower the melting point—other chemicals maybe used in its stead. Since it is a permanent irreversibleaction, it has to be documented. The treatment is known as“flux healing”: The heating of the rubies together withsome flux allows the surface to melt; the flux mixture fillsthe fracture dissolving its walls until the liquid in the cracksaturates with corundum solution coming from the originalmineral. Then, the “synthetic ruby” re-crystallizes in thecracks by controlled cooling, and the flux is subsequentlyremoved with acid (acid cleaning) since acids do not attackcorundum. As a result of this, the ruby improves itsdurability, but, given that the treatment involves a profoundchange of the gemstone, if acknowledged or spotted, theprice of the gemstone drops down.

There is no easy telling of a flux healing treatment;the only give away is the presence of glass pockets con-taining or not some air in the areas where the fracturesexisted, but then inclusions also occur in natural non-treated rubies.

The beryllium colour-diffusion treatment for corundumto obtain pink–orange corundum or cancel purplish coresin ruby [6]

Obtention of the “padparadscha” colour

The naturally occurring pink–orange corundum receives thename of “padparadscha”. This colour of corundum is morerare than ruby or blue sapphire and thus highly appreciated.In the year 2001, beryllium colour-diffusion treatment forsapphires was first performed in Chanthaburi (Thailand) toobtain this shade of corundum and by the same year thetreated pink–orange corundum appeared in the market

together with the real padparadscha; it was not until year2003 that the Chanthaburi Gem and Jewelry Associationreported the introduction of beryllium in corundum.

This treatment consisted of heating the sapphire togetherwith chrysoberyl (BeAl2O4). Beryllium (Be2+) acts in thesame way as magnesium (Mg2+) ions would do in thecorundum lattice except for the fact that they are smalleratoms and thus they are more soluble in the same givenmaterial.

In a normal blue sapphire, the concentration of titanium[Ti4+] and silicon [Si4+] is equal or higher than theconcentration of [Mg2+]:[Ti4+ + Si4+]≥[Mg2+]; there areunbalanced positive charges in the structure that have to becompensated with structural defects. (The naturally occur-ring concentration of beryllium is neglected for it is farsmaller than in the final treated corundum). The addition ofchrysoberyl as a source of beryllium will cause the situationto shift and make the structure have too many electrons:[Mg2+ + Be2+]≥[Ti4+ + Si4+]. This would not change thecolour of the sapphire in reduction conditions for therewould be oxygen vacancies to compensate the charge in themineral, and oxygen vacancies will produce no colour.However, a change occurs in oxidising atmosphere whereMg2+ and Be2+ will trap holes producing strong yellow toorange colouration since the high concentration of oxygeninvolves that there are not many oxygen vacancies present.Orange colour appears in fact when [Mg2+ + Be2+]>>[Ti4+ + Si4+] in oxidising conditions.

It has been described that the beryllium colouringdiffusion goes 2 mm deep in corundum in 40 h at1,800 °C. The more the beryllium penetrates, the moreorange is the colour of the material. If the corundum’soriginal colour is pink, the yellow–orange colouration dueto the treatment permeates most of the original pinksapphire, and the padparadscha colour is obtained.

The purple-core treatment in rubies

Bluish or purplish cores or inclusions in ruby are typicallyattributed to high concentrations of iron and titanium in thesame way these ions colour blue sapphire [2, 6]. Berylliumcolour diffusion can cancel visually the darkening spots inthe gemstone or eliminate them depending on how thetreatment is performed.

Using the before mentioned procedure for the manufac-ture of orange–pink sapphires only to create a yellowsurface is enough to cancel visually a dark core in a ruby.However, this sort of “surface disguise” can be easilyspotted. More difficult is the identification of a ruby thathas undergone the beryllium treatment in deep.

The beryllium treatment can be applied in a way as toachieve a full beryllium penetration that will make thewhole piece turn yellow under oxidising atmosphere since

Differentiation of non-treated and treated corundum minerals 1045

diffusion is only a matter of time and the adequateconditions. The yellow saturated colour due to trapped-hole Be2+ can be reduced by re-heating the sample in soft-reducing atmosphere; in such circumstances, the formationof oxygen vacancies takes over the trapped-hole Be2+

defects in the charge compensation process, so that redcolour from chromium predominates again and the purplishcore disappears [6].

Differentiating treated from non-treated corundum

Differentiating treated and non-treated corundum samplesthat have undergone the above- mentioned processes(colour diffusion, flux healing) is an important issue onthe day-to-day business of gemstone import companies andjewellers.

Standard gemmological testing includes studies ofrefractive index and specific gravity determination, as wellas pleochroism, absorption spectra and the observation ofthe response of the gemstones to ultraviolet (UV) radiation.None of these tests are conclusive. Refractive index,specific gravity, birefringence or pleochroism results donot vary significantly from treated to non-treated samples.

The identification of uneven colour distribution or colourzoning may indicate the presence of beryllium treatmentbut it is not enough to guarantee it. The study of zirconinclusions can only tell whether the material has beenheated more than usual (when only heat treatment isperformed), for then this mineral is found in opaqueirregular masses instead of transparent regular crystals.

UV–visible (Vis)–near-infrared spectroscopy is not use-ful and techniques such as Raman or infrared (IR) spec-troscopy are only able to detect heating processes but not tolink them to beryllium diffusion. Energy-dispersive X-rayfluorescence may detect abnormal quantities of calcium orzirconium that are not usually present in natural rubies butattributed to flux healing. Secondary ion mass spectrometryis one technique capable of determining the presence ofberyllium and has shown that the surface of beryllium-treated corundum contains higher concentrations of thiselement than non-treated one [7]. X-ray photoelectronspectroscopy has lately been used to study the cores ofMöng Hsu rubies [8].

The aim of this work is to try alternative techniques todifferentiate treated from non-treated corundum samples.For this purpose, a Spanish major jewellery and gemstoneimport company has provided our team with 78 mineralcorundum samples coming from different places, some ofwhich have been treated.

We have performed ion-beam-induced luminescence(IBIL) and other complementary techniques such as Raman,proton-induced X-ray emission (PIXE) and proton-inducedgamma-ray emission (PIGE) on these samples.

IBIL, also known as ionoluminescence, has been usedbefore to detect impurities or defects inside syntheticmaterials and natural minerals; its use for the discriminationof gemstone simulants or synthetic analogues has beenelsewhere discussed [3]. PIXE has been frequently appliedin the archaeometric field for material characterisation andprovenance studies of minerals [9–11] and PIGE comple-ments the elemental analysis by detecting light elements inthese materials such as—and lighter than—sodium thatcannot be identified with the PIXE technique [5, 12]. Themicro-Raman technique has also been used complementa-rily to ion beam analysis (IBA) techniques for mineralcharacterisation [4].

The final aim of this study is to provide new means forsystematic analysis of corundum gemstone-quality mineral,alternative to the traditional gemmologic methods. Resultsbased on chromium detection as a luminescent impuritywith IBIL, structural information supplied by micro-Ramanor possible compositional differences found with PIXE andPIGE techniques will be discussed in this work.

Experimental

Description of the samples

The 78 corundum samples supplied by Antonio NegueruelaS.A. are red, pink and yellow corundum minerals ofdifferent origins and sizes, some of which are faceted orpolished while others appear rough. In some cases, thecorundum minerals have undergone treatment whereasothers are not treated. The list of the samples and itscharacteristics along with the information the companyfurnished can be found in Table 1.

Experimental conditions

Less than year ago, we supplied the archaeometry in-airbeam line of our cyclotron at the Centre Européend’Archéométrie—Institut de Physique Nucléaire Atomiqueet de Spectroscopie (CEA-IPNAS; Université de Liège)with a new IBIL setup in collaboration with the Depto.Geología y Geoquímia (Universidad Autónoma deMadrid).

This setup consists of an Ocean Optics HR2000 opticfibre spectrometer connected on one side to a 1-mm-diameter optic fibre adequate to measure in the UV–Visrange that is placed 45° to the normal of the sample andapproximately 1 cm distant from it. See Fig. 1 for a sketchof the geometry of our setup. On the other side, thespectrometer is connected to a laptop where the spectraacquisition software OOIBase32 also from Ocean Optics isinstalled.


We developed an external trigger to control the irradia-tion of the samples with the OOIBase32 acquisitionsoftware since the IBIL signal changes while irradiationgoes on; typically, a decrease of the luminescence yieldtakes place with constant increasing proton dose, followinga second-order exponential curve [13]. Some mechanismsfor the decay of the IBIL signal have been elsewhereproposed [14].

A 3.12-MeV proton beam was used to excite theluminescence of the ruby samples. The intensity of the cur-rent measured in the extraction nose just before theextraction of the beam to the atmosphere varied between0.5 and 10 nA depending on the sample’s sensitivity towardsthe luminescent phenomenon.

We took ten spectra of each sample during the 10 s thatlasted its irradiation. In every case, we chose the secondmeasurement to compare the spectra to ensure that nospurious light was registered. Of course, we only comparedthe intensity of the luminescence for those spectra taken inthe same conditions of current intensity.

For the PIXE technique at the CEA-IPNAS cyclotron,we used the 3.12-MeV proton beam with a current intensityof 2 nA measured before extraction. X-rays were collectedin two Ultra-LEGe GUL0035P Canberra detectors placed40° each to the beam direction. One of them was used forthe detection of low-energy X-rays. Its resolution at5.9 KeV was of 140 keV and it was supplied with heliumflux and a lead diaphragm of 2-mm diameter in order todecrease the solid angle and to have a counting ratecomparable to that of the other detector. The other detector,configured for high-energy X-rays, had a resolution of

156 eV at 5.9 keV and was supplied with a 50-μm-thick Alabsorber in order to attenuate the X-rays coming from lightelements. Both detectors have a beryllium window of7.62 μm.

The quantification of the elements detected with thePIXE technique was carried out with the GUPIX softwarein its MS-DOS version and TRAUPIX developed by L.Pichon at the C2RMF for batch mode. The GUPIXsoftware is used to fit the PIXE spectra and extract thepeak intensities of the X-rays that correspond to thedifferent elements present in the sample. The programmeallows different approaches to calculate the elementalconcentrations from the X-ray peak intensities; we per-formed the analysis using the thick sample approach andmatrix and trace calculation, based on the previouscalculation of the efficiency of detection of the X-rays by

sampleBeam direction

Optic fibre

4

0.8cm Distance

from extraction

nose to sample

Distance

from sample

to optic fibre

~1cm

Fig. 1 Setup for IBIL technique at the in-air beam line of the CEA-IPNAS cyclotron

Table 1 Name and description of the corundum samples studied in this work

Sample number Origin or reference number Treatment Faceting Color

1, 1.1, 1.2, 1.3, 1.4 Möng Hsu Non-treated Rough Red, dark cores

2, 2.1, 2.2, 2.3, 2.4 6859 Ruby Non-treated Faceted Red

3, 3.1, 3.2, 3.3, 3.4 5510 Ruby Non-treated Faceted Red

4, 4.1, 4.2, 4.3, 4.4 101 Sapphire Non-treated Cabochon polished Yellow–orange

5, 5.1, 5.2, 5.3, 5.4 955 Sapphire Treated BeO Faceted Yellow–orange

6, 6.1, 6.2, 6.3, 6.4 Vietnam Treated heat Rough Pink

7, 7.1, 7.2, 7.3, 7.4 5165 Sapphire Do not know if treated Faceted Red

8, 8.1, 8.2, 8.3, 8.4 211 Sapphire Treated BeO Faceted Red–orange

9, 9.1, 9.2, 9.3, 9.4 Vietnam Twice treated in USA Rough Pink

10, 10.1, 10.2, 10.3, 10.4 Möng Hsu Treated heat Rough Red

11, 11.1, 11.2, 11.3, 11.4 Möng Hsu Twice treated BeO/flux Rough Red, dark cores

12, 12.1, 12.2 Morogoro, Treated heat in USA Rough Red

13, 13.1, 13.2, 13.3, 13.4 Morogoro Do not know if treated Rough Red

14, 14.1, 14.2, 14.3, 14.4 Ruby Treated possibly Borax/BeO Rough Red

15, 15.1, 15.2, 15.3, 15.4 7979 Ruby Treated lead Faceted (tear) Red

16, 16.1, 16.2, 16.3, 16.4 5383 Ruby Do not know if treated Faceted Red


the measurement of standards. Particularly, a diorite DR-Nstandard (ref. 95GOV-01) from the Association Nationalede la Recherche Technique of France was used for thispurpose. The PIXE concentrations of this standard havebeen checked with the given concentrations (AssociationNationale de la Recherche Technique) to ensure that theresults are correct, as it can be seen in Table 2. TheTRAUPIX software allows running the GUPIX programmein a batch mode and displays the concentration resultsdirectly in EXCEL files.

We also performed some μ-PIXE measurements at theAGLAE IBA facility in the C2RMF CNRS UMR171 inParis, in one of the Möng Hsu non-treated samples thatpresented a dark core to study its composition, and runPIXE and PIGE measurements on some of the samples withan ordinary beam size to contrast our results at CEA-IPNAS. Details about the AGLAE setup for in-air analysiswith these techniques can be found in [15].

The PIGE technique was performed both at the CEA-IPNAS cyclotron using a 3.12-MeV proton beam and at theAGLAE with a 3- and 4-MeV proton beam. The PIGE detec-tor at the CEA-IPNAS is an HPGe Canberra, placed at 60°.

For the μ-Raman measurements, we used a ReinshawSystem-1000 coupled with an Olympus BH-2 microscopeand objective lens ×5. A 785-nm laser transition with50 mW was used for the excitation of the sample.

Results and discussion

PIXE results

Quantification of PIXE spectra yielded approximately 42%average of Al, 17% of Si and certain amounts of otherimpurities such as (but not only): K, Ca, Ti, V, Cr, Fe andGa. The average concentrations of some elements can befound in Table 3 for the different types of sample.

The high concentrations of calcium in the samples ofMorogoro and Möng Hsu caught our attention, especiallyfor the Möng Hsu samples treated thermally and with BeO/flux. In this case, the concentrations reach 20% in weight.This sample is the only one that shows the presence ofstrontium in 0.1% weight. These high concentrations maybe explained through the flux healing process that takesplace in the surface usually with calcium compounds.Möng Hsu treated and non-treated samples also contain thehighest concentration of impurities among the samplesanalysed; they show large amounts of titanium and iron andare the sole minerals presenting manganese.

Samples labelled 7979—treated with lead oxide—yielded extremely high quantities of impurities too, suchas Zr, Ga or Pb, lead being the most prominent feature inthe PIXE spectra for high Z elements as it can be seen inFig. 2, thus making the technique an appropriate tool for the

DR-N diorite

Standard given values Uncertainty PIXE measured values

Al2O3 17.52% 0.14% 18.28%

CaO 7.05% 0.06% 7.15%

Fe2O3 9.70% 0.07% 9.84%

K2O 1.70% 0.03% 1.71%

MgO 4.40% 0.07% 4.04%

MnO 0.22% 0.01% 0.23%

Na2O 2.99% 0.06% 2.67%

P2O5 0.25% 0.02% Under limit of detection

SiO2 52.85% 0.19% 54.28%

TiO2 1.09% 0.02% 1.09%

Cr2O3 0.0058462% 4.38465E−04% Under limit of detection

SrO 0.04733044% 1.537393E−03% 0.0599%

ZnO 1.804902E−02% 6.2238E−04% 2.55E−02ZrO2 0.0016875% 0.001215% Under limit of detection

Al 9.27% 0.07% 9.68%

Cl 400 ppm 55 ppm 464 ppm

Mn 1,700 ppm 46 ppm 1,774 ppm

Ni 15 ppm 3 ppm Under limit of detection

Pb 55 ppm 4 ppm 78 ppm

S 350 ppm 1,334 ppm

Table 2 Comparison of thestandard concentrations forDR-N with the concentrationsobtained by PIXE for some ofthe elements present


Tab

le3

PIX

Eelem

entalcon

centratio

nsin

partspermillionforthedifferenttyp

esof

samples

analysed,g

iven

astheaverageforeach

type

ofcorund

umanalysed

with

theircorrespo

ndentstandard

deviations

(inthecase

where

itproceeds)andlim

itsof

detection

Sam

ples

Num

ber

of samples

analysed

Na(%

weigh

t)Mg(%

weigh

t)Al(%

weigh

t)Si(%

weigh

t)K

Ca

Ti

VCr

Mn

Fe

Zn

Ga

Sr

Zr

Pb

Mön

gHsu

(non

-treated)

5–

6.5±

1.2

41.1±

6.2

3.7±

3.7

1,08

6±

1,04

759

8±

689

1,08

5±

1,25

216

3±

763,02

8±

892

97± 104

10,605

±20

,636

23± 33

42±13

––

86± 48

LOD

00

0.1

4645

2845

2299

177

712

9

6859 (non

-treated)

5–

5.7±

0.04

46.7±

0.1

0.65

±0.05

118±

3710

7±

161

103±40

18± 21

1,73

7±

681

–2,82

7±

570

–25

±7

––

–

LOD

1.4

00

5044

3336

2010

6

5510 (non

-treated)

50.25

±0.20

5.8±

0.02

45.1±

1.0

11.3±

0.6

496±

321

307±

181

5,61

3±

3,46

298

± 151

3,96

0±

1,36

9–

–12

± 1747

±10

––

–

LOD

00.2

0.1

0.1

6557

3917

428

1210

101 (non

-treated)

50.19

±0.06

6.0±

0.09

46.4±

0.1

0.68

±0.06

263±

106

59± 56

65±25

–22

2±85

–5,56

2±

1,71

8–

31±8

––

–

LOD

50.16

000.1

0.1

044

3928

1813

7

955(Be)

5–

5.91

±0.25

46.6±

0.4

0.65

±0.15

131±

4122

± 1451

±40

19± 21

73±31

–5,23

6±

1,311

–47

±16

––

–

LOD

0.13

0.1

0.04

4235

2725

2013

7

Vietnam

(heat)

50.86

±0.21

8.2±

0.1

43.8±

0.2

0.84

±0.13

166±

7918

2±

6298

±85

88± 163

380±

286

–119±80

33±3

58±18

––

–

LOD

0.21

0.1

0.1

065

5040

3930

97

5

5165

4–

5.9±

1.0

45.9±

1.2

1.1±

0.3

176±

133

–63

4±

718

511±21

02,64

0±

1,13

5–

178±

194

–28

0±

108

––

–

LOD

0.4

0.1

0.1

225

131

144

121

50.00

30.00

221(Be)

5–

5.3±

0.3

46.5±

0.6

0.72

±0.14

299±

275

862±

1,47

22,89

5±

5,27

3–

1,47

0±

474

–5,33

3±

1,01

5–

65±69

––

–

LOD

0.2

0.1

012

495

7067

6568

68

Vietnam

(twice

treated

USA)

3–

4.5±

1.6

46.3±

0.1

1.5±

0.6

359±7

196±

194

465±72

–2,31

9±

1,111

–73

7±

647

30± 52

157±

49–

––

LOD

0.5

0.2

025

826

017

210

644

4332

Mön

gHsu

(treated

heat)

25.6±

6.0

3.2±

1.2

36.1±

11.6

3.0±

2.6

550±

330

1,64

1±

714

3,06

8±

7565

4±

406,64

1±

2,61

335

0±

8110

,827

±13

,749

30± 42

138±

4748

±68

104±

65–

LOD

10.5

0.2

0.1

258

188

168

238

180

334

9544

3012

816

6

Mön

gHsu

(treated

heat/flux)

15.1

–20

.37.1

1,44

420

2,63

73,82

5–

4,97

360

227

,893

190

206

1,08

385

4–


Tab

le3

(con

tinued)

Sam

ples

Num

ber

of samples

analysed

Na(%

weigh

t)Mg(%

weigh

t)Al(%

weigh

t)Si(%

weigh

t)K

Ca

Ti

VCr

Mn

Fe

Zn

Ga

Sr

Zr

Pb

LOD

0.96

0.4

0.2

715

502

654

387

513

109

6432

258

378

Morog

oro

(treated

heat

USA)

2–

6.4±

1.5

43.9±

1.2

1.2±

0.3

820±

1,16

02,33

1±

861

2,91

7±

2,03

2–

5,07

4±

1,24

4–

1,55

9±

276

54± 76

101±

40–

––

LOD

0.5

0.2

0.1

466

271

185

244

6249

31

Morog

oro

(dono

tkn

owiftreated)

2–

7.5±

0.8

44.2±

0.3

–46

5±

541,87

1±

2,29

3–

–6,01

0±

1,24

1–

534±

262

–13

7±

183

130±

183

369±

346

–

LOD

0.1

045

538

427

891

7831

935

5

Treated

possibly

Borax/BeO

1–

7.4

41.6

1.3

–58

63,37

3–

12,862

–1,18

6–

168

––

–

LOD

0.8

0.3

0.2

522

489

547

149

104

7979 (treated

lead

oxide)

23.5±

3.9

8.1±

1.6

38.2±

6.7

0.7±

0.5

––

3,02

6±

4,02

1–

5,81

2±

1,42

2–

5,75

7±

2,05

0–

298±

111

–3,65

3±

4,85

321

,814

±6,96

5

LOD

3.2

20.4

0.8

2,55

61,92

72,15

623

618

085

220

,733


identification of lead treatment. This sample showed alsohigh concentrations of Fe, Cr and Ti.

The large amounts of magnesium present in all thesamples are due to the bad resolution of the magnesium andaluminium bands in the spectra. We have compared theseresults obtained at the CEA-IPNAS with the ones acquiredfor some of the minerals at the AGLAE to check this factthat will be corrected in future works.

Plots of the concentration of Cr against the concentrationof Fe have been used before to determinate the origin ofrubies [10, 11]. According to the chromium–iron plot of oursamples in Fig. 3, the gemstones analysed group togetheraccording to the different types of corundum that we havestudied, though the unknown origins do not really matchthe samples whose provenance is known.

We compared our PIXE results with the data publishedby T. Calligaro et al. [11]. T. Calligaro et al. analysed alarge set of ruby samples finding different groups of them

that corresponded to several origins and may be ‘probablylinked to three geological forming contexts’. These groupshad different concentrations of Fe and Cr and were thefollowing:

& Group I: Burma, Vietnam& Group II: Sri Lanka, Afghanistan, Kenya, Vietnam& Group III: India, Cambodia, Thailand, Madagascar,

Vietnam

When displaying the means of Cr and Fe concentrationfor each type of corundum sample that we analysed withPIXE, we found that sample 5165 of unknown originbelonged to group I and that in group II there were samplescoming from Morogoro (Tanzania, neighbouring country toKenya) and Vietnam. Just one type of sample coming fromMorogoro belonged in this group (the samples for whichwe ignore if any treatment has been done), and just onetype of sample coming from Vietnam was included in this

Fig. 2 PIXE spectra and fit of the sample number 15.3 of type 7979 treated with lead oxide. Bands of some relevant elements have been marked

Fig. 3 Plot of PIXE concentra-tions in parts per million ofchromium against iron in thesamples measured


group, and it was the samples twice treated in the USA; thesamples treated with heat also coming from Vietnam do notbelong to any of the groups described by Calligaro et al. Ingroup III, our samples came from unknown origins andMorogoro (samples that were treated with heat). Möng Hsusamples did not belong in any group, though we could haveexpected them to fit the group I since Möng Hsu is inBurma. All of this can be clearly seen in Fig. 4; our data arenot in good agreement with the works by T. Calligaro et al.This can be due to our comparatively small number ofsamples or to the difference existing between the mineseven in the same regions.

μ-PIXE mapping results

We performed μ-PIXE mapping in one of the samples(sample 1.2 non-treated, coming from Möng Hsu; Fig. 5).This ruby possesses a dark round purplish core right in thecentre that is placed apparently next to the surface. We

carried out the measurements bearing in mind the previousreferences in the literature to the composition of thesecores, in which high concentrations of iron and titaniumwere stated [5, 6]. To our surprise, as Fig. 6 shows, we didnot find such significant amounts of these ions but largequantities of chromium.

The image corresponding to the total number of countsshows that the dark core contains higher quantities ofmetals—the number of counts is higher (yellow colour onthe lower part of the image). This means that part of thecore is placed on the surface. The other part goes more indeep so the proton beam does not reach it. This is thereason why the yellow features do not resemble the shapeof the inclusion.

Regarding the composition of this dark core, the imageon the right upper corner of Fig. 6 shows the distribution ofiron (green), chromium (red) and titanium (blue). Iron isevenly allocated throughout the sample whereas chromiumand titanium appear anti-correlated; the core shows large

Fig. 4 Mean values of Cr andFe for the samples analysed inthis work, marked in groups thatwere described by T. Calligaroet al. [11]

Fig. 5 PIXE analysis of theMöng Hsu non-treated sample1.2 at AGLAE


concentrations of chromium and scarce titanium while theopposite occurs in it the rest of the sample. This can be seenon the PIXE spectra taken of the inclusion (on the left ofFig. 6) and of another area that does not correspond to theinclusion (on the right of Fig. 6).

PIGE results

As a first approach to the PIGE detection of beryllium, wetried the gamma emission of this element at 718 keV fromthe 9Be(α,γ)10B reaction at 3 MeV. However, no emissions

of the samples were registered; in fact, not even in a pureberyllium foil could beryllium be detected at that energydue to its small cross section (∼103 Nγ/μC sr at 3.1 MeV).

We thus run some tests in the AGLAE at 4 MeV, relyingon the 9Be(p, α1)

6Li reaction with gamma emission at3,562 keV, whose cross section is higher (∼106 Nγ/μC sr at4.2 MeV). Beryllium emission at 3,562 keV could be thennoticed in the thin beryllium foil in the form of a broadband (due to Doppler effect as the beryllium nuclei recoil)but not on a beryllium-treated sample (sample number 5-995 sapphire), its spectrum being the same as the one of a

Fig. 6 The image on the leftupper corner corresponds to thetotal number of counts (bluestands for a little number ofcounts whereas red is a largenumber of counts). On the rightupper corner, the same imagewith the map of Fe (green), Cr(red) and Ti (blue). Underneath,on the left bottom corner, thespectrum of the dark core of thesample. On the right bottomcorner, the spectrum of a non-dark area

Fig. 7 PIGE spectra for a pureberyllium foil at 3 and 4 MeV,together with the spectra of aberyllium-treated sapphire(sample 5-995 sapphire) and anon-treated one (sample 8-221sapphire)


Fig. 8 Raman spectra of MöngHsu, Morogoro and Vietnamsamples. Bands marked inyellow correspond to α-Al2O3


non-treated sample (sample number 8-221 sapphire), as itcan be seen in Fig. 7.

This indicates that the beryllium concentration of theberyllium-treated sample is under our detection limits forthat reaction. Further work should be conducted for thedetermination of the limits of detection of light elements byPIGE at higher energies.

μ-Raman results

One random mineral was chosen within every group ofsamples for μ-Raman analysis and several measurementswere performed in different areas. We will compare here thespectra obtained for the samples from known origins suchas the ones coming from Möng Hsu, Vietnam or Morogorodisplayed in Fig. 8.

The Raman spectra of the Möng Hsu sample number 1(non-treated) and sample number 11 (treated with BeO/Flux) show bands related to α-Al2O3 at 350, 375, 414 and641 cm−1 for sample number 1 and at 376, 413, 640 and745 cm−1 for sample 11. Sample number 10, correspondingto a Möng Hsu mineral treated with heat, does not show α-Al2O3 bands but bands that correspond to another com-pound: minium (Pb2O3). High concentrations of lead havebeen detected with PIXE for this sample as well as for otherMöng Hsu treated ones but its concentration is 30 timessmaller than the one found for sample 5.3 (7979, treatedwith lead oxide) for instance.

We must take into consideration the fact that the area ofanalysis is different in μ-Raman and PIXE; since the PIXEshows the same amounts of lead for either treated and non-treated Möng Hsu samples, in sample number 10, we mayhave come across some minium inclusion in the μ-Ramanmeasurements.

The Raman spectra of the Vietnam samples (number 6subjected to heat and number 9 twice treated in the USA)also feature bands attributed to α-Al2O3. In the case of theMorogoro samples, though number 13, which we do notknow if it has been treated, shows bands related to α-Al2O3, sample number 12 treated with heat in the USA isnot Raman active, meaning it does not correspond to α-Al2O3 or other phases of Al2O3. Nevertheless, this sampledoes not show any important variations in the concen-trations obtained by PIXE.

IBIL results

The IBIL spectra registered in every case the usual signalsof chromium in octahedral coordination, corresponding totransitions 2E2–

4A2 (known as the R-lines) and 2T2g–4A2

(N-lines) The shape of these signals was expected for all thesamples except for sample number 6.4 (heated, comingfrom Vietnam), as it can be seen in Fig. 9.

The bands that appear in the spectrum of sample 6.4have been elsewhere attributed to clusters of this elementrather than single substitutions of aluminium, N-lines thatbecome more intense when there are close chromium atomsin the structure [16]. The PIXE concentrations obtained forchromium in this sample are within the regular chromiumconcentrations that we have found in other corundumminerals; the only abnormal concentrations found in thiscase are calcium (3,500 ppm) and iron (5,100 ppm), thehighest found in all the samples measured. Iron (II) and(III) in octahedral coordination show luminescence typical-ly over 700 nm too, and both ions have energy levels closeto that of chromium in corundum. Though we do not havemeans to assure this, our results suggest that the lumines-cence of chromium may be related to that of iron in thiscase.

In order to find out the appropriate conditions to capturethe IBIL spectra of the corundum samples, the lumines-cence of one mineral was measured within each type, whileirradiated with different proton intensity currents rangingfrom 0.5 to 10 nA. There is a strong dependence betweenthe intensity current of the proton beam and the yield of theluminescence; as a higher amount of particles reach thematerial, a larger number of chromium ions are excitedinside the structure. An example of this can be seen inFig. 10 for sample number 1 (Möng Hsu non-treatedsample).

The adequate intensity currents were chosen for thesamples of the same origin in order to compare theirspectra. Möng Hsu rubies proved to be the most lumines-cent ones; in fact, the lowest intensity current we couldachieve at the CEA-IPNAS cyclotron (0.5 nA) was enoughto saturate in some cases our spectrometer.

Fig. 9 Spectra corresponding to two of the corundum samplesmeasured. In red, the spectra of one of the rubies were measured.All the samples analysed showed the same band structure except forone of the samples coming from Vietnam and subjected to heattreatment (sample number 6.4)


We controlled that no decay of the luminescence tookplace during the 10-s irradiation in any case since the decayof the luminescent intensity is related to the damage(reversible or permanent) of the material. Luminescenceof the R-lines of chromium has been proven to decreasewith the fluence of protons and helium ions in themegaelectron volt range, through the annihilation of thesecentres [17, 18]. In none of the samples analysed was suchdecay observed—not even pushing the beam current up to13 nA as it can be seen in Fig. 11, where the decrease of theluminescent signals is due to the shutting of the beamthrough the external trigger around the ninth second.

We studied different aspects of the 694-nm signal ofchromium such as the possible shifts in wavelength or thevariations in its full width at half maximum (FWHM) dueto treatment.

We would expect large shifts in the wavelength if therewas a change in the coordination number of chromium,which is not very likely to occur since, even in the casethat there was a phase transition from α-Al2O3 to any ofits polymorphic forms, octahedral sites would still bepresent. Small shifts of the emission wavelength mayoccur, though, due to the distortion of those octahedralsites with the treatment. This would cause too an en-largement of the band width (FWHM) due to the loss ofsymmetry induced by the diffusion of foreign elementsinside the structure (where the samples have undergone thistreatment) or to the heating–cooling process. However, noconclusions could be extracted from these data in neitherways; no displacement or enlargement of the signals hasbeen observed considering the 0.9-nm resolution of ourspectrometer.

As Fig. 12 shows, the slight difference in the wavelengthof emission found in the treated samples in relation to thenon-treated ones cannot be accounted for since the sen-sitivity of our spectrometer will not allow it; as for theFWHM, no variations were found.

The reason for this may rely on a non-sufficientwavelength resolution of our spectrometer or simply onthe fact that neither the diffusion nor the high temperaturesor flux healing treatments affect much the luminescentsignal that way.

As for the intensity of the 694-nm chromium emission,we were only able to compare the set of samples comingfrom Möng Hsu since they are the only ones for which wehave non-treated and treated specimens. Plotting theluminescence measurements performed in the 15 MöngHsu samples (non-treated, heat treated and twice treatedwith BeO/Flux) during 3-nA proton irradiation, we foundthat the intensity of the 2E2–

4A2 signal is in all the caseshigher for the non-treated samples, as it can be seen inFig. 13.

The high temperatures to which in any case (heat orBeO/flux treatment) the samples have been subjected couldhave led to some loss of crystallinity that would explain thelower luminescent response of the material. An enlargementof the FWHM of the signal would be nevertheless expectedto accompany such a loss due to chromium ions partiallyplaced in octahedral distorted sites; but since no enlarge-ment of the FWHM has been perceived with our IBILsetup, we cannot conclude that the treatments have led tothe loss of crystallinity.

Another plausible explanation for the smaller chromiumemissions in beryllium-treated corundum is the quenchingof the luminescence by mechanisms related to the beryllium

Fig. 11 Intensity of the luminescent bands of sample number 1(Möng Hsu non-treated ruby) measured during the 10-s irradiationwith a 13-nA proton beam. As before, the 694-nm signal has beenexcluded for it saturated the spectrometer

Fig. 10 Intensities of the different bands that can be seen in thespectra of a Möng Hsu non-treated ruby (sample number 1) capturedduring subsequent irradiations with different current intensities for theproton beam. Note that the 694-nm band corresponding to thetransitions 2E2–

4A2 is not plotted since it saturated the spectrometer


ions through the variation of the electronic levels in thecorundum structure. This is supported by the fact that thereis a change of colour to orange in the ruby layers reachedby the diffused beryllium. In the band model of corundum,the levels of chromium are displayed higher in energy thanthe beryllium ones (assuming that beryllium would act asMg2+). Be2+ traps holes next to the valence band in order tocompensate its excess of electrons and be isoelectronic withAl3+; however, in the presence of chromium, the hole getstransferred to this other element that is higher in energy,and this generates the orange coloration. The transfer of thehole to chromium occurs as a charge mechanism compen-sation for Cr4+ that appears from the oxidation of Cr3+ [6].

The luminescence of Cr3+should be affected by thepresence of beryllium; the lesser Cr3+ there is in the treatedareas of the material, the lesser is the luminescence yield.

The energy supplied by the proton beam would allow therecombination of electron–hole pairs and leave the Cr4+ inan excited state, being then less Cr3+ in the structurestreated with BeO; while the luminescence yield of Cr3+ issmaller, a new signal due to Cr4+ luminescence shouldappear. Luminescent emission of Cr4+ appears neverthelessat higher wavelengths—around 1,200 nm. Our next task inthis regard is to compare the relative intensities of Cr3+ andCr4+by performing luminescence measurements in the IRregion and confirm that the emission of Cr4+appears.

In this way, the presence of beryllium may induce thequenching of the luminescence in the external layers of thesample through the creation of Cr4+centres. Note that 40 hof treatment lead to a 2-mm layer of treated material; thismeans that the internal part of the mineral would be stillemitting but luminescence would be always less intense.

The loss of intensity in the samples that have undergoneonly thermal treatment could be explained by the annihi-lation of the chromium centres that cannot be stabilisedthrough formation of Cr4+. This would also explain thereason why we do not see any changes in the FWHM of thechromium transition.

Conclusion

The different analyses run on the corundum samplesyielded varied information about their composition. PIXEproved to be a useful technique to detect lead oxidetreatment for instance, and based on the concentrations ofchromium and iron it was possible to distinguish thedifferent groups of samples—for those of which we haveinformation on the origin, we are able to say that differentorigins make different groups. The μ-PIXE maps showedhigh concentrations of chromium in the dark core of a MöngHsu ruby instead of large amounts of iron or titanium.

Fig. 12 Emission wavelengthand FWHM of the transitions2E2–

4A2 for the different MöngHsu rubies. The intensity of theproton beam varies from 1- to0.5-nA proton beam in the non-treated samples to 3 nA in thetreated ones; this does not haveany repercussion in the mea-surement of the position of thesignal and does not affect theFWHM (there is no correlationbetween the intensity or the areaof this signal and the FWHM).The resolution of the spectrom-eter appears as an error bar andis 0.9 nm

Fig. 13 Comparison of the intensity for the 2E2–4A2 chromium

transition in Möng Hsu rubies, when irradiated with a 3-nA protonbeam. Note that some of the points in the figure are overlapping andthat, when 4,000 counts are reached, it means the signal saturated


PIGE measurements did not rend good results for thedetection of beryllium, the cross section of the reaction 9Be(p,α1)

6Li being the highest of all the possible ones under4 MeV. We should study the cross sections for this reactionat higher energies in order to explore its possibilities forberyllium detection.

The μ-Raman technique pointed out the presence ofminium (Pb2O3) in one of the samples that was not treatedwith lead oxide (a Möng Hsu sample, heated) but containedhigh concentration of lead in comparison to the rest of thecorundum minerals analysed—this is due perhaps to thefact that the areas analysed with μ-Raman are smaller thanthose analysed with PIXE, and we may have come acrossan area where there were lead oxide inclusions.

When we are dealing with techniques as different as μ-Raman and PIXE, we must always bear in mind that thesize of the spot analysed is not the same; where PIXE givesa general view of an area about 1 mm2 with which it is notalways possible to exclude the inclusions that may be pres-ent in the material, μ-Raman uses smaller areas and thus itmay be possible to avoid them (if they can be spotted out).

No difference between treated and non-treated MöngHsu rubies could be found by analysing the emissionwavelength or FWHM of the 2E2–

4A2 transition with theIBIL technique; we believe this to be due to the resolutionof the spectrometer. However, it has been proven that theluminescence yield is higher for non-treated samples than itis for those samples treated at high temperatures with orwithout beryllium oxide diffusion.

The reason for this is related to the defects introduced byberyllium oxide in the corundum structure that may causethe oxidation of Cr3+ to Cr4+; our next work will be focusedon measuring the luminescent signals of non-treated andtreated corundum in the IR range, which is the region inwhere Cr4+ is luminescent.

All in all, the combination of techniques provided abetter characterisation of the corundum minerals analysed;PIXE and μ-Raman are complementary and work very welltogether and it has been shown in this work in the samplescontaining lead oxide; however, no differentiation ofheating or beryllium treatment could have been madewithout the IBIL technique.

Acknowledgements The authors would like to acknowledge Anto-nio Negueruela S.A. for providing the corundum samples for thiswork. Also to technicians A. Marchal, M. Philippe and A. Holsbeekfor their support at the CEA-IPNAS cyclotron during the measure-ments and to technicians M. Clar and Said Rakkaa for theircontribution to the development of the external trigger. The CEA-IPNAS research group would like to thank J. Salomon and L. Pichonfor the measurements performed in the AGLAE. This project issupported by the Inter-university Attraction Pole Programme P6/16—Belgian State—Belgian Science Policy and the Spanish AECIprogramme no. A9332/07.

References

1. Cavenago-Bignami Moneta S (1991) Gemología, Tomo I Piedraspreciosas, perlas, corales, marfil. Ediciones Omega, Barcelona (InSpanish)

2. Hughes RW (1997) Ruby & sapphire. RWH Publishing, Fallbrookhttp://www.ruby-sapphire.com (2008)

3. Calvo del Castillo H, Ruvalcaba JL, Calderón T (2007) AnalBioanal Chem 387:869–878

4. Calligaro T, Colinart S, Poirot JP, Sudres C (2002) NIM-B189:320–327

5. Sanchez JL, Osipowicz T, Tang SM, Tay TS, Win TT (1997)NIM-B 130:682–686

6. Emmett JL, Scarratt K, McClure SF, Moses T, Douthit TR,Hughes R, Novak S, Shigley JE, Wang W, Bordelon O, Kane RE(2003) Gems Gemology 39:84–135

7. Novak SW, Magee CW, Moses T, Wang W (2004) Appl Surf Sci231–232:917–920

8. Archiwawanich S, Brack N, James BD, Liesegang J (2006) ApplSurf Sci 252:8646–8650

9. Osipowicz T, Tay TS, Orlic I, Tang SM, Watt F (1995) NIM-B104:590–594

10. Calligaro T, Mossmann A, Poirot JP, Querré G (1998) NIM-B136–138:846–850

11. Calligaro T, Poirot JP, Querré G (1999) NIM-B 150:628–63412. Calligaro T, Dran JC, Poirot JP, Querré G, Salomon J, Zwaan JC

(2000) NIM-B 161–163:769–77413. Calvo del Castillo H, Millán A, Beneitez P, Ruvalcaba-Sil JL,

Calderón T (2008) Rev Mex de Fisica 54(2):93–9914. Agulló-López F, García G, Olivares J (2006) NIM-B 249:118–12115. Dran JC, Salomon J, Calligaro T, Walter P (2004) NIM-B 219–

220:7–1516. Basun SA, Meltzer RS, Imbusch GF (2007) J Lum 125:31–3917. Inouye A, Nagata S, Toh K, Tsuchiya B, Yamamoto S, Shikama T

(2007) J Nucl Mats 367–370:1112–111618. Skurqtov SA, Jong Gun K, Stano J, Zagorski DL (2006) NIM-B

245:194–200


http://www.ruby-sapphire.com

Towards the differentiation of non-treated and treated corundum minerals by ion-beam-induced luminescence and other complementary techniques

Documents