MEROHEDRAL TWINNING IN THE ZEOLITE PHlLLIPSITE

AND THE FORMATION OF MERLINOITE DOMAINS

MARK HOWARD BADHAM

A thesis submitted to the Department of Geological Sciences

in conformity with the requirernents for

the degree Master of Science

Queen's University

Kingston, Ontario, Canada

September 1997

Copyright O Mark Howard Badham, September 1997 National Library Bibliothèque nationale of Canada du Canada Acquisitions and Acquisitions et Bibtiographic Services services bibliographiques 395 Wellington Street 395. nie Wellington OttawaON K1AON4 Ottawa ON K1A ON4 Canada Canada

The author has granted a non- L'auteur a accordé une licence non exclusive licence allowing the exclusive permettant à la National Library of Canada to Bibliothèque nationale du Canada de reproduce, loan, distribute or sell reproduire, prêter, distribuer ou copies of this thesis in microfonn, vendre des copies de cette thèse sous paper or electronic formats. la forme de rnicrofiche/nlm, de reproduction sur papier ou sur format électronique.

The author retains ownership of the L'auteur conserve la propriété du copyright in this thesis. Neither the droit d'auteur qui protège cette thèse. thesis nor substantial extracts fkom it Ni la thèse ni des extraits subsîantiels may be printed or otherwise de celle-ci ne doivent être imprimés reproduced without the author's ou autrement reproduits sans son permission. autorisation, i

A bstract

A zeolite from Cava Nuova, Monte Somma, Vesuvio, Italy, origindy identified as the rare mineral merlinoite &Ca&3&Si,0a)-24H,0, has been studied by powder and single- crystal X-ray diffraction, optical microscopy, and electron micro probe.

The powder pattern shows d-space iines unique to meriinoite at - 10A and -4.45A, however, it also shows d-spacings that are found only in the structurally similar mineral phillipsite (YNa,Ca),,(Si,Al)gO,~ 6H20.Precession work reveals that the reciprocal lattice of this sample matches the stmcture of phillipsite that is merohedraily twinned. The twin law has been identifieci as a 90' rotation about the a axis as illustrated by the morphology of the crystals; the precession-film lattice patterns; and indirectly by the fact that the crystal structure could not be refined from CAD4 difEiactometer data collected during the study.

It is postulated that twlluiing in this phillipsite is responsible for the formation of merlinoite domains within the ciystal. In phillipsite, the siliwn/aiuminurn tetrahedra are linked to form incomplete doublarings in a sinusoidd or "wavy" ribbon structure. The merlinoite fiarnework is similar, except that it is based on complete double-rings. It is shown that at twin boundaries in the sarnple being studied, complete rings could form, and that repetitive twinnllig results in the exact merlinoite structure. ii

The formation of rnerlinoite in phillipsite by twinnuig has implications for the vaiidity of thermodynamic data calculateci in 1990 for synthetic merlinoite. The produas of the synthesis were reported as merlinoite and zeolite ZK-19 (a syithetic phiiiipsite-like phase) that were identsed by powder pattern alone. Since ZK-19 does not have the -6.4A line of normal phillipsite, it would be very difncult by powder diaction alone to distinguish pure merlinoite fiom an intergrowth of twinned ZX- 19 and merlinoite. The thermodynarnic data calculated may therefore be based on a compound which is not pure merlinoite. iii

Acknowledgements

The author would like to thank several people who have helped him during the long, drawn-out process of completing a part-time Master's Degree. First and foremost, I would Wce to thank my wife, Sheryl, who never gave up on me and constantly offered me encouragement to wry on with this study. This work is dedicated to her, and to my daughter Torie whose life began almost at the same tirne as this project did all those years ago. My supervisor, Dr. R. C. Peterson, always had helpful suggestions when problems arose, and he was understanding when my farnily and my niIl-the job often occupied more of my time than my research did. Alan Grant kept the X-ray equipment operating, and on more than one occasion went out of his way to make sure 1had the equipment cor@prations that 1 needed to complete this study. Thanks are also due to Forrest

Cureton of Tucson, Arizona, who supplied the sample.

This research was supported by an National Science and Engineering Research Councii gant to Dr R.C. Peterson. iv

Table of Contents v

References ___CI -- -

Appendices

Appendk 1: Merlinoite Stmcture Refinement Files: LATCON

Appendii II: Theorectical D-spacings for Phillipsite -

Appendix III: Microprobe Analyses with Balance

Error Calculations -- -_HI-_- Vita------_---_

List of Tables

Table 1: Powder Pattern of Analcime Cornpareci to

JCPDS Card #4 1- 1478 -____- ----_

Table 2: Powder Pattern of Dioside Compared to

JCPDS Card #Il-654------O--

Table 3 : Powder Pattern of Ankerite and Calcite wmpared

To JCPDS Cards #3 3-282 and #5-586------

Table 4: Merlinoite: Gandolfi and Difractometer X-ray Cornparison----

List of Figures

Figure 2- 1: The Double Crankshaft Structure------5

Figure 2-2: The Crystal Structure of Merboite------8

Figure 2-3 : The Crystal Structure of Phillipsite- -- O--- 10

Figure 2-4: Schematic Cornparison of Merlinoite and

PhiMipsite Structures------pu---i------12 vi

Figure 2-5: Unit Cell Choices in Philiipsite-

Figure 3-1: Whole-rock X-ray Powder Pattern

Figure 4- 1: Gandolfi X-ray Powder Patterns of Sample

Figure 4-2: X-ray Powder Diffiactometer Trace of Sample

Figure 4-3 : Zero-level Precession Film Indexed with the

Merlinoite CeU.

Figure 4-4: Cone Axis Film ----

Figure 4-5: Upper-Ievel Precession Film Indexed with the

Merlinoite CeU- -

Figure 46: Zero-level Precession Film Indexed with the

Phillipsite Ceii

Figure 4-7: Upper-level Film Indexed with the

Phillipsite Ceii -_----

Figure 4-8: Cornparison of Precession Films Taken at Right Angles------

Figure 4-9: Reciprocal Lattice Points Due to Twinning-----

Figure 4- 10: Phillipsite Crystal Morphology Diagrams- --

Figure 4- 1 1: Photomicrograp h of Crystal ------

Figure 5- 1: Electron Back-scatter Images of Microprobe Samples

Showing Analysis Areas -

Figure 6- 1: Twinning Causing Merlinoite Domains in Philii psite------

Figure 6-2: Photomicrograph Down a Showing Complex Twinnùig---- -1-

Chapter 1. Scope and Purpose of Study

This study focuses on a sarnple wntaining smd (Clmrn) glassy translucent crystals reportedly identifiai by X-ray difidion as the rare zeoiite merlinoite (Forrest Cureton, personal communication) fiom a new locaiity at Cava Nuova, Monte Somma, Vesuvio,

Napoli, Itaiy. These samples are currently being marketed by several mineral dealers as

"particularly fine crystais of merlinoite that are much better than the type locality material."

Preliminary Gandoln X-ray dfiaction studies on this sarnple by Vaiyashko and McGlade

(unpublished course work, 1995) did show key lines at 10A and 4.47A that are essential in identifjmg merlinoite, but several other lines were also present in the pattern which can not be indexed using the merlinoite space group and ceii. Precession photos also showed anornaIous difEaction spots, and extinction conditions that are not consistent with the

Immm space group assigneci to merlinoite.

Due to the weii-developed nature of the crystals in this sample, a çhidy was initiated to investigate the crystallography and fùrther refine the crystal structure of merlinoite. The ultirnate goal at the outset of this project was to explain the extra Iines in the powder difEaction pattern and the precession photo anomalies by either assigning a new space group and redefining the crystal structure, or by identifjmg a new mineral species.

Methods of investigation employed in this work include X-ray powder difiaction, single- -2- crystal d-action ( CAD4 automated difEactometer, and Buerger precession carnera), electron micro probe, and optical microsco py and goniometry. -3-

Chapter 2. Literature Review

2.1 History and Significance of Zeolite-group Minerals

Natural zeoiite-group rninerals are characterized by a three-diensional dumino-silicate fiamework (tektosilicates) with open channels in the structure containing loosely-bound water molecules and exchangeable cations (Gottardi and Galli, 1985). Today, they are known fiom many geological environments, notably of hydrothermal origin in volcanic rocks and as microcrystalline masses formed by rock-water chernical reactions in rocks of volcanic-sedimentary origin.

The first use of the narne "zeolite", which loosely translates from its Greek root words as

"boiling stone", dates back to Cronstedt, 1756, who used the tem to describe minerals which expel water when heated (Gottardi and Gdi, 1985). One of the rernarkable and economically useful features of zeolites is that this expulsion of water is reversible, as the mineral readily re-hydrates at room temperature with no structural changes. In fact, this property is essential for a minera1 to be included in the zeolite group. More irnportantly fiom a commercial viewpoint, dehydrated (or "activated") zeolites will adsorb not only water but a variety of other molecules by a process coined "molecular sieving" by McBain in 1932 (Mumpton, 1977).

In the 1950's the study of zeoiite-like phases began in emest when their industrial applications as ion exchangers and molecular sieves were recognized. Much of the research today centres on synthetic phases that are tailored as specific molecular sieves. -4-

Although there are roughly 30-40 natural zeolites known, nearly one hundred unique zeolite phases have been synthesized which have no hown natural counterparts

(Mumpton, 1977). Today, natural zeolite phases are used as mers in paper, rubber, and plastic; lightweight aggregates in mortars; ion exchangers for water treatment; dietary supplements for fami animals; and in fertilizers. Synthetic phases have been taiiored as molecular sieves to remove amrnonia and other water impurities; as odor control products for household and agriculturd use; and as a soi1 substitute for hydroponic applications.

2.2 Stmctural Classification of Zeolites

Several topologicd ciassification schemes have been proposed in the past to describe the crystal structures of rninerals of the zeolite group (Gottardi and Galli, 1985). EssentiaiIy, zeolite structures cm be broken down into groups bas& on the geometric shapes formed by the Mages of the siiicon/aluminum tetrahedra. The tetrahedra are designated as the

Primary Building Units (PBUYs),and the various chahs, rings, double-rings, and higher- order polyhedrai cages formed are designated as Secondary Building Units (SBUYs).

In this study, the classification proposed by Gottardi and Galli (1985) will be followed.

They divide zeolites into 6 structural groups: 1. Fibrous Zeolites, 2. Zeolites with singly connected 4- ring chains, 3. Zeolites with doubly comected 4-ring chains, 4. Zeolites with

6-rings, 5. Zeolites of the mordenite group, 6. Zeolites of the heulandite group. 2-1 (b) SilicoB/aluminum tetrahedra 2-1 (a) Two silicon/aluminum tetrahedra pointing towards each other. pointing up linked to two pointing down

Silicon or aluminum

h

2-1 (c) Perspective and plan view of the double n crankshafl fomied by fi C) alternathg arrangement (a) and (b)

c, Note: Atomic sizes and bond lengths are not to scale.

Figure 2-11 Schematic Representation of The Double Crankshaft The tetrahedral Primary Building Units in zeolites like merlinoite are linked to form double crankshafk in the structure. In some parts of the crankshaft, two corner-sharing tetrahedra that point upwards are joined to two other corner-sharing tetrahedra which point downwards (fig 2-la). At other places in the shaft, the tetrahedral vertices face each other (fig 2-lb). Alternating arrangement (a) and arrangement (b) results in the double-crankshaft, shown by joining the midpoints of the cations in the tetrahedra These crankshafts are side-linked within the structure to form a framework with long open channels. -6-

2.3 Previous Work on Merlinoitc and Similar Minerais

Merlinoite K,Ca,(AigSi,O,)-24Ha and the structurdy similar minerais harmot orne

@&K)1.2(si,Ai)8û16-6H20,weiisite (Ba,Ca,KJ(Ai,Si,JO,, 6H20,and phillipsite

(~Na,Ca),dSi,Al)8016*6H,O fdl into Group 3: Zeolites with doubly connected 4-ring chains. In these minerals, tetrahedra share corners to form long chains (Figure2-1).

DEerences in symmetry in these minerals anse from the variety of ways that the chains can be "cross-linked" to form a three-dimensional hework, These four minerals are often described as being based on a "double crankshaft" composed of hked chains.

Smith (1 978) has outlined 17 different theoreticai ways in which these chains can be linked to form fiameworks, one of which exists in merlinoite and a second in the isostmctural minerals harmotome, wellsite, and phillipsite.

Although synthetic analogues of merlinoite-like structures had been prepared as early as

1956 with "phase K-M" (Barrer and Baynham, 1956) and "Linde-W" (Miton 1961), a natural occurrence was not reported until 1977 fiom Cupaello, Rieti, Italy ( Passaglia et al., 1977). This was foilowed in 1979 with the discovery of an occurrence in Hôwenegg,

Hegau, Germany (Alberti et al., 1979) and in 198 1 £tom Khibira MassK Kola Peninsula, former USSR (Khomyakov et al., 1981). It has also been reported in manganese nodules fiom the Indian Ocean (Mohapatra and Sahoo, 1987), and corn Searles Lake, California

(Donahoe et. al., 1990), a saline, aikaline lake environment.

In spite of its apparent rady in nature, the experirnental work of Donahoe et. al. (1985) -7- have show that merlinoite is quite easy to synthesize. The synthesis takes place at 80°C in high pH clear-solutions containing sodium hydroxide, potassium hydroxide, alurninum chloride hexahydrate, and silicic acid. Donahoe et. ai. show that phillipsite and merlinoite readily precipitate in 5 to 7 months, as determineci by powder X-ray dfiaction of the precipitate. The synthetic phillipsite mineral produced is actuaîly more like the zeolite ZK-

19 synthesized by Kühl(1969) than a naturai phillipsite since it lacks the -6.4A line found in most phillipsite patterns. The pH of the solution was found to have the greatest eEect on the phases produced, with merlinoite being favoured in high pH (13.5 5-1 3.7 1) and phillipsite in pH's dom to approximately 10. Furthemore, merlinoite formation appears to be more likely in solutions with Iower ratios of Ca/Na.

Originally, the crystal stmcture of merlinoite was solved and refined using the Rieti type- locality material to a residual R-value of 9.3% (Galli et. al., 1979). Due to the small size of the crystals, the crystal morphology and other important characteristics of the mineral are known oniy in a generd way. Based on single-crystal X-ray dEaction, the rnineral was assignecl the orthorhombic space group Irnmm (#71) with ceU dimensions a =

14.1 l6(7)A 6 = 14.229(6)& c = 9.946(6)&

The unit ceil and structural arrangement of merlinoite is shown explicitly and schematically in Figure 2-2 to show the arrangement of the crankshaft chahs. Four chahs are arranged parauel to the c crystallographic axis and side-linked to form open, octagond double-rings in the centre of the ceil. Eight tetrahedra making up one half'of the ring point upwards a mtermolecule C Potassium (K) atom b .aicium (Ca) atom 1 SWtetra hedron

Double-ring structure

=.. ,= 'crankshaft chahsn

Figure 2-2 (a) The Crystal Structure of Merlinoite &Ca, [ASSi,OJ .24 H,O

Merlinoite crystallizes in the orthohombic system (Immm) with a=14.116. btI4.229. ~9.946À. For simplicity, oniy one of the eight-msmbered 'double rings' is show in this drawing to illustrate the orientation of the channels along c fomed by the "crankshaft chains' (see figure 2-1) of SVAI tetrahed ra.

Figure 2-2(b) Merlinoiteviewed along c (the channel mis)

A full unit-cell view down c (with adjacent atoms drawn) shows the linked-ring channels and the pseudo-tetragonal symmetry of merlinoite. The crankshaft chain axis is perpendicular to the page. Schematically, the structure is shown to the rïght with the chains represented by squares and the rings by octagons. The central ring is slightly above Me plane fomied by the rings at the corners of the cell. and linked to them by crankshaft Vamps" which dip downward into the page. -9- dong c, and are joined to eight downward-pointing tetrahedra from above. The rings deviate slightly from a perfect octagon, thus lowe~gthe symrnetry fiom tetragonai to orthorhornbic.

The X-ray powder pattern of merlinoite is very similar to that of phiilipsite. In fact, the dominant distinguishing features of merlinoite are the presence of extra lines at 10.02A and 4.475q and a lack of the 6.4A line shown in the phiiiipsite pattern. The structure of phillipsite was used as a starting point for solving the stmcture of merlinoite by Galii et.al

(2 979).

The topology of the structure of phillipsite was first solved by Steinfïnk (1962) using

"direct methods" electron density Fourier projections. The framework reported is the same as that solved almost siiultaneously for harrnotome by Sadanaga et al. (196 1) using heavy-atom Patterson methods to locate the barium atoms, and then direct methods to complete the structure. The symmetry of philiipsite was reported by Steinfink as orthorhombic B2mb while harmotome was attributed rnonoclinic P2, symmetry by

Sadanaga. The fiamework proposed by both authors is virtually identicai. Subsequent work by Rinaidi et. al (1974) revealed that both minerais are actudy monoclinic, cvstallizing in P2, for phillipsite and P24m for harmotome.

The stmchire of phiiiipsite (Fig 2-3) is also based on chain-like crankshafts of SilA tetrahedra. The differences from the merlinoite structure arise because of the way in Water molecule Potassium (K) atom Calcium (Ca) atom SUAI tetrahedron

\ pseudo double-ring channel direction

Figure 23 (a) The Crystal Structure of Phillipsite, KCa(Si&UJO,,GH,O

Phillipsite crystallizes in the monoclinic system (P2,) with a=9.865, b=14.30, ~8.668A.This serni-perspective view along b shows the usual beta angle of 124-12S0, and the unit ceIl outline. Two sets of channels are fomed by incomplete 'double ringsn in this structure. an irregularly-shaped channel along b and a more regular channel along a (see fig 2-3(b)).

Figure 2-3 (b) Rotation ofthe PhillipsiteStructure to showthe Relationship to Meriinoite

When the SUAI incomplete double-ring along a in phillipsite is rotated into the plane of the page, the relationship to the merlinoite framework becomes clearer. Rather than facing downwards like the rest of the upper rnembers in each ring. two of the tetrahedra face up to join the ring above. The arrangement is shown schematically to the right by drawing lines joining the central atoms in the tetrahedra which are in the same plane. Essentially. one of the four "chains' of each ring (see Figure2-1) is offset by a distance of al2 perpendicular to the plane of the page creating a "wavy ribbonnpattern rather than a complete ring. -11- which the chains are side-linked. In merlinoite, four chains are iinked in rings fonning charnels with ahost (but not quite) four-fold symmetry. That is, the whole ring can be generated fbrn one chah operateci on by a simple rotational four-fold axis in the centre of the channel. In phillipsite however, one of the chains is moved dong the channel axis by a distance of a/2. The "rings" are not completely closed because of this and the pattern formed is often described as a "wavy ribbon" shape (Figure2-3@)) This has the net effect of destroying any syrnmetry operators dong the a axis, and reducing the symmetry to monoclinic compared to the orthorhombic symmetry of merlinoite.

A side-by-side cornparison of the two schematic structures as viewed down the channei axes in Figure 2-4 illustrates the different linkages between the chains in the structures, and also shows the basic similarities of the two structures.

The syrnmetry of phillipsite, hamotome, and weUsite had long been a subject of debate, especialiy during the nineteenth century (Cemy, 1964). Some scientists advocated monoclinic symmetry, while others were convincecl that they were orthorhombic. Most of the confusion was probably due to the complex twinning often seen in philiipsite and other zeolites. Based on optical and morphological studies, scientists of the time eventualiy agreed on monoclinic symmetry for these minerais.

Even after the use of X-rays for structure determination became possible however, it can be seen by Steinfink's 1962 paper that other interpretations of the syrnrnetry were still Figure 24 (a) An idealized schematic of the merlinoite structure to show crankshaft linkges

The structure is shown on the right looking down c using perfect octagons to represent the rings foned by crankshaft chains (squares). Rings at the sarne height in the cell are shown in green. The left-hand diagram shows a view in the plane of the complete ring formed by four crankshafts. The arrows show the 'direction of dipnof that part of the chain. The 'dip arrowsncan be seen to al1 point away frorn, or toward, each ring.

d Wavy Rbbon' -structure --+

Figure 2-4 (b) An idealized schematic of the phillipsite structure to show crankshaft tinkages

Looking down the "pseudbnng axis" a of phillipsite, a similar pattern is seen when campareci to merlinoite, however, the direction of dip for the chains is different. The diagrarn on the left attempts to show a typical incornplete ring in phillipsite made of crankshafts. The left-hand chain has been shifted "up" by haif its repeat distance. lnstead of its turning into the page to complete the ring, this chain is turned out of the page. The result is a "wavy ribbonn(Gottardi and Galli, 1985) structure shown in the diagram in green. -13- being made. Cern9 (1964) outlined the probable causes of the confusion, and upheld the monoclinic symmetry. As shown in Figure 2-5, the motifformed by the identipoints in the phiIlipsite structure can easiiy be described by a monoclinic cell, or as a pseudo centred- orthonet with almost orthorhomie symmetry. In the two cells, the a and b axes are cornmon, but the pseudo-orthorhombic c axis is equal to the vector sum of 2c + a. Given the approximate dimensions of the rnonoclinic phillipsite ceIl as a= 9.8& ~4.3A,and

~8.674and P=124* it can be seen through simple trigonometry that c,, becomes

14.30A and the angle between a and c is aimost exactly 90'. This construction explains nicely how Steidïnk was able to refine the stnicture of phiIlipsite in an orthorhombic space group BZmb. Because of the geometry, and probably due to twinning in the crystal

(Hofian et. al., 1977), the X-ray data measured in a B-centred orthorombic ce11 could be used to deduce the approximate framework structure. After Cerny, 1964

Figure 2-5 Direct-space unit cell choices in a repeating net with dimensions like phillipsite

As in most direct-space nets. several choices of unit cell are always possible. Because of the ce11 dimensions of phillipsite-type minerals, either a monoclinic ce11 or a pseudo centred-orthonet can be drawn. The pseudo centred-orthonet has a beta angle almost exactly equal to 90amakingit very close to tnie orthorhombic symmetry. According to Cemy (1964), this makes rneasurements made in the pseudo-arthorhombic cell almost exactly the same as those measured in the tnie monoclinic cell. -15-

Chapter 3. Description of Sample

3.1 Host Rock Minerdogy

A detailed petrographic study of the rock in which these minerals occur was not the airn of this study, however, the mineralogy of the rock was studied by X-ray dBaction to gain a basic understanding of the host. A thin-section was not cut, but a representative portion of the rock (wedge-shaped, roughiy 5 mm by 4 mm across by 2 mm at the thickest part of the wedge) was crushed and macroscopic rnineral grains were isolated by hand-picking under a binocular microscope. These grains were investigated with single crystal Gandolfi and powder-diffraction Guinier X-ray techniques. The very fine-grained groundmass was also powdered, and a whole- rock X-ray was done on a Siemens powder diffractometer to detennine its gross mineralogy.

The host rock is of volcanic origin, with the glassy crystals investigated in this study occurring within 0.5-3 millimeter vugs throughout the rock. The majority of the rock is composed of crearn-to-white coloured blebby crystais forming masses up to 2 millimeter across within a fine-grainai overall purplish-tinged light-grey groundmass.

A single grain of the cream-coloured mineral was mounted on a glas fibre with epoxy, and investigated using the Gandolfi technique. In this technique the rnineral does not have to be ground in order to determine the powder diffraction pattern, unlike other diffraction methods which rely on the random orientations of many particles in a powder to produce full cones of diffraction on the film. A single crystal is mounted on a geared sarnple-holder -26- which spins the crystal continuously about two axes, effectively bringing alf lattice planes in the crystai into the diffraction position at some point. This simulates the random orientation of a powder, but needs a much smaller volume ofsample. Measurement of the film, exposed for 12 hours using zirconium-filtered cobalt Ka radiation, showed that the minerai is another zeolite, analcime (NaAlSi,OgH,O). The observed d-spacings are tabulated and compared to the JCPDS card 19-1 180 in Table 1. The ongin of anaicime in volcanic rocks is a subject of debate. It is thought not to be a primary rnineral, but may form by alteration (sodafication) of Ieucite (KAlSi,OJ which crystallized from the original magma (Gottardi and Galii, 1985).

The only other macroscopicalIy distinct mineral in the rock is a light green mineral with a waxy appearance. A small sample was extracted, ground to a powder, and investigated with a Guinier focusing X-ray camera using rnonochromated copper Ka radiation. The lines on the film were measured with a dedicated Guinier film measunng device, and correcteci d-spacings (Table 2) were calculated with a spreadsheet. The calculated pattern matches the mineral diopside (CaMgSi,O,).

Using the sobare package XPLOT (Raven, 1995), the major lines in the whole-rock X- ray were matched with three minerais (Figure 3-1). Not surpnsingly, given the results of the est Gandolti film, analcime is a major component of the rock. Hematite (FqO,) is also present, accounting for the "purplish" colour of the rock. The third rnineral is anorthite (CaAi2Si20,), a plagioclase feldspar. Table 1: Measured Powder Pattern of Analcime cornpared to JCPDS Card 41-1478

(Intensity values in d column are approximate estimates only) -18-

Table 2: Measured Powder Pattern of Diopside compared to JCPDS Card #Il-654

(Intensity values in d ,wcolumn are approximate estimates ody)

.------

d mea,um, (4 d, mca,,md (A) (4 intensiîy (htensity) diopside (Intensity) - 1 Figure 3-4: Whole-rock X-ray Powder Patbm

The peaks present in a wholerock groundrnass X-ray are shown to match a mixture of analcime. hematite, and anorhite. Undoubtedly there are other minerals in the rock, but these are the most abundant phases. -20-

3.2 Mineralogy of the Vugs

The vugs which host the glassy crystals are very simple, mineralogicdly. The vug walls are lined with blocky7 white crystals, and covered intermittently by a orange-brown coating. The merlinoite crystals and the orange-brown coating appear to be the last phases to have crystailized. Since the thin orange-brown coating was extremely difftcult to separate f?om the white crystals, a quantity of sample containing both phases was ground to a powder and X-rayed in the sarne Guinier camera described previously. The measured pattern (tabulateci in Table 3) matches a mixture of ankente

Ca(Fe2',~g,Mn)(C0,), and calcite (CaCO,). A subsequent test of the vugs with 10% dilute hydrochioric acid produced an effervescent reaction, coniïrming the presence of carbonate minerals.

The mineralogy of the wgs at Cava Nuova is similar to merlinoite occurences at the type locality Rieti, Italy, and at the Howenegg, Gemany, locality. AU three localities have calcite as a major component of the vugs. A paragenetic sequence could not be detedned for the Rieti locality, but at Howenegg the sequence is calcite-merlinoite- amicite (K2Na,Al,Si,0,, =5H20)(Alberti et. al., 1979). This squence, and the carbonate- merlinoite sequence at Cava Nuova, agree with the experimentai findings of Donahoe et. ai. (1985). Their experimental results show that synthetic rnerlinoite fonns in solutions with high pH and low Ca/Na ratios whereas philiipsite is favoured in high CaMa solutions.

The precipitation of carbonate minerals first would remove Ca £iom the solution, and favour the formation of merlinoite. -21-

Table 3: Measured Powder Pattern of Ankerite and Calcite compared to JCPDS

Cards 33-282 and 5-586

(Intensity values in d column are approximate estimates only)

(Intensity) (4 Intensity calcite ankerite (intensity) -22-

Chapter 4. Crystallography

4.1 X-ray Powder Diffraction Study

This study was initiateci by prelVNnary work (Valyashko and McGlade, unpublished course work, 1995) which showed extra lines in the powder pattern of a mineral reportedly identitied as merlinoite (Forrest Cureton, personal communication). Valyasko and McGladeYsGandolfi powder pattern is show in Figure Cl(Pattem 2). Due to the long exposure tirne required for such a small crystal, the central portion of the film is fogged by air-scattered radiation, however, the 10A line of merlinoite is visible on the original film. Aiso present is the line at 4.47q another key line in identwg rnerlinoite.

The key line that could not be indexed in the pattern is present at approxirnately 6.4A.

Early in this study it was recognized that this is a line present in the phillipsite pattern, a fact not realized by Vaiyashko and McGlade. Even after this was deterrnined, the anomalous difiaction points present in the precession camera work (Section 4.2.2) provideci enough impetus to carry on with the investigation.

A Gandolfi X-ray camera was modified so that it can be flushed continuously with helium during an exposure. The helium effectively elirninates air scatter within the camera and reduces the "fogging" of the film in the low two-theta angle region during a long scan.

The signal-to-noise ratio is increased dramatically, and the low-angle region is much clearer. The results of a 48 hour exposure on the crystal used for subsequent precession (1) (2) merlinoite phillipsite d, d, (A) d&) #29-989 #39-1375

10.17 merl 10.0 - 8.14 merl 8.15 phi1 8.11 7.08 merl 7.08 phil 7.16 6.35 - phi1 6.42 5.34 merl 5.36 phil 5.38 4.98 merl 4.98 phi1 4.94 phil 4.67 4.45 merl 4.48 - 4.24 merl 4.29 phi1 4.29 4.09 merf 4.07 phil 4.06 phil 3.92 3.646 merl 3.66 phil 3.688 merl 3.34 phil 3.43 3.238 merl 3.24 phil 3.21 3.1 59 merl 3.1 8 phii 3.1 36 2.930 merl 2.935 phi1 2.929 2.722 mer1 2.72 phil 2-712 2.676 mer1 2.67 phil 2.68 2.535 merl 2.507 phi1 2.53 2.400 mer1 2.391 phi1 2.388 2.321 merl 2.354 phi1 2.34 2.244 phil 2.239 2.161 merl 2.18 phil 2.158 2.054 merl 2.065 phi1 2.057 1.963 med 1-973 phi1 1.984 1.912 mer1 1.855 card end 1.768 mer1 1-764 1-71 5 merl 1-717 1.676 merl 1.671 1.591 merl 1.592 1.558 meri 1.558 1.538 mer1 1.515 1-481 merl 1A88 1.363 merl 1.371 1.319 merl 1.317

Pattern 1 Pattern 2 Figure 4-1: Gandolfi powder patterns showing both merlinoite and phillipsite lines

The film on the rigtrt (pattern 2) was prepared by Valyashko and McGlade (1995). They used a larger crystal which gave sharper lines, but the film is quite dark due to long exposure time. Although they do not show up well on this electronic image, the lines indicated by the red arcs were obsenred on the original film.

The film on the left (pattern 1) was done during this study in a helium atmosphere to reduce fogging of the low two-theta area. The crystal used is the same one from which precession films were prepared. The unique meriinoite lines at approximately 10A and 4.47A (yellow) are visible on both f lms. as are lines that are exclusively phillipsite (green) such as 6.39A. Instrument: Stoe transmission step-scan difftactometer. Settings: 45kV, 32ma, rnonochrornated CuKa, radiation. 120 secondslstep count tirne, 0.02"two-theta step size.

Figure4-2: X-ray Diffraction Powder Pattern

A small quantity of crystals hand-picked under a binocular microscope was ground in alcohol, and placed in a 0.5 millimeter diarneter capillary tube and X-rayed. The red and blue lines show the positions of the diffraction peaks for meriinoite and phillipsite (respectively) from JCPDS cards 29- 989 and 39-1 375. As can be seen, the pattern shows a diffuse peak near 10A. and a more distinct peak at 4.47A which are diagnostic lines for merlinoite. The pattern also shows a welldeveloped peak at 6.365A which is present in phillipsite but should be absent in merlinoite. The rernainder of the peaks match lines present in both rninerals, with the exception of an anomalous large peak at 2.212 which cannot be explained. work is show in Figure 4- t (Pattern 1).

As there was not enough sample for a regular powder dsactometer investigation, a small

quantity of crystals was hand-picked under a binocular microscope, ground to a powder in

aicohol, and X-rayed. The instrument used was a Stoe transmission-geometry automated

diffiactometer which required that the sample be mounted in a OSmm diameter glas

capillary. Graphite-monochrornated copper radiation (45kv, 32ma tube operation) was

used. X-rays were wunted for 120 seconds at each 0.02' step in an effort to reduce the

signal-to-noise ratio in the scan which covered the range &om 5-80' two-theta. The results of the X-ray are shown in Figure 4-2. The d-spacings in the pattem, automatically calculated by the program XPLOT (Raven, M., 1995), compare weU with the values measured on the Gandolfi films (Table 4). Table 4: Merlinoite Gandolfi and Diffractometer X-ray Pattern Cornparison

dolfi d-spacing (A) Rractometer d-spacing (A) -27-

Both the Gandolfi nIms and the powder diflkactometer trace show the key lines for

merlinoite. However, both also show characteristic lines for the stmcturaliy sirniiar

minerd philiipsite.

4.2 Single Crystal Work

4.2.1 CAD4 Automated Diffractometer

Early in the study, a single crystal of dimensions O. lOmm X O. lOrnrn X O. I4mm was

mounted on a glas fibre with epoxy and examined using a Enraf-Nonius CAD4 automated dfiactometer. Graphite-monochromated molybdenum Ka radiation was used, and data were collected at 45kV and 26ma.

Morphologically, the crystai does not resemble the rnerlinoite crystals shown by Galli et al.

(1 977) and so the orientation of the unit cell in this crystal was unknown. Using the

CAD4 software, a merlinoite-iike cell was chosen for data collection with rough initiai dimensions aci4.163(9)q b= 14.20(1)k c=9.920(5)4 and alpha= 90.12(6)' , beta=

90.04(5)O, and gamma= 90.00(6)0 calculated £tom 17 centred and indexed reflections.

Data were coiiected hmO to 30' theta, collecting a full sphere fiom O to 13' and the unique data (118 sphere for orthorhombic syrnrnetry) from 13-30 degrees. Including the intensity standard reflections which are measured multiple tirnes during data collection, a total of 5524 reflections were measured. Psi-scan data (for absorption corrections) were coiiected for six reflections in anticipation of a structure refinement. -28-

The XTAL 3.2 structure refinement software package (Hail, S.q Flack, H.D., Stewart,

J.M., editors, 1992) was used to sort the reflection lis& initially specifjing a primitive orthorhombic Iattice to check extinction conditions in the data, The results of the sort showed no conditions on hkl(113 present with 607.5 counts and sigma of 10.6); no restrictions on hkO, Okl, hl?/,;restrictions on ho0 @=Zn); restrictions on OkO (k2n); and restrictions on 001 (1=2n).

The 110 reflection (responsible for the unique 10.02A line in merlinoite) is present in the sorted list ( 156.8 counts with sigma of 2.6), but the overail extinction conditions shown are not consistent with the reported space group of Immm (Galli et. al, 1979) for merlinoite. The only possible space group with these conditions is P2,2,2, (#19).

Refinement of the structure in this space group using the atomic positions reported for merlinoite was unsuccessful. In the hopes that P2,2,2,could somehow be an approximate subset of 1- an attempt was made to translate the atomic positions reported for merlinoite in Immm to a P2,2,2, symmetry. A spreadsheet was used to caiculate aii symmetrically equivalent positions for atoms in both groups, but none of the atoms from

Immrn could be matched with positions in P2,2,2,.

At this point, many atternpts were made to refine the data dected as merlinoite with hmm syrnmetry. Using the program ADDREF @avenport, G., Hall, S., 1992 in XTAL

3.2 Reference Manual, 1992) in the XTAL 3 -2 package (1 992) the space group was specified as Immm to remove those reflections in the CAD4 data file that are not -29-

consistent with the extinction conditions of merhoite. 25 17 reflections were rejected

fiom the data file during this step for violating the systematic absence conditions of

Immm. Next, the program SORTRF (Hall, S., Spadaccini, N., Stewart, J., 1992 in XTAL

3.2 Reference Manual) was employed, and the reflections were sorted with Friedel pairs

averaged. The sorted list created for subsequent refinement contained 1642 unique

reflections with the following statistics:

Reflection packets output to archive 1642

Reflection statistics h k 1 theta

min O O O 2.03

max 19 19 13 29.89

Reflection counts for rcodel rcode2 rcode3 rcode4 rcode5

1170 472 O O O

SUM(del(Y))/SUM(Y) equiv.s .O351 1365 replicates

SW(sig(Y) /SUM(y) .O433

The low value of the SUM (del (Y))/Sum (Y) hdicates a fairly close agreement between the accepted reflections and the symmetry of merlinoite.

LATCON (Schwanenbach, D. and King, G., 1992 in XTAL 3.2 Reference Manuai) was used to further refine the lattice parameters by restricting the alpha, beta, and gamma angles in the ceil to exactly 90' as required by orthorhombic symmetry. The CAD4 data collection software does not constrain these angles when it calculates the ceii dimensions and angles. The centred reflections used for the original data coUection on the CAD4 were entered, and the resuits of the ceil parameter refhement are as follows:

DIRECT ELEMENTS

VECTORS IN AC' 1 -- A, B, C, ALPHA, BETA, GAMMA, COSA, COSB, COSC AND THEIR E.S. D-S 14.1882 14.2214 9.9501 90.0000 90.0000 90.0000 .O1 53 .O1 00 ,0212 .O000 .O000 .O000

CELL VOLUME 2007.70

The input reflection List and full results of LATCON are included in Appendu 1.

Refinements based on reported rnerlinoite atomic positions were never successful. The best R value achieved was 3 1.9% after a very controlled refinement using CRYLSQ

(Olthof-Hazekamp, R., 1992 in XTAL 3 -2Reference Manual) in XTAW .2(1992) to slowly refine sale and overail temperature factor one at a tirne. Mer this, an attempt was made to refine isotropie temperature factors for each atom and site positions and occupations. Wethe R-factor dropped to 27.3% after 4 cycles of refinement, the site populations and temperature factors were unreasonable for many of the atoms. An acceptable residual value for a structure refinement is on the order of 10% or less, and refinement based on the rneriinoite structure was clearly not going to be possible.

At this point, the only alternative seemed to be that this was a totally new crystal structure. The search for a crystal structure which could satisfy the rneasured reflections was attempted by "direct methods" and by Patterson techniques for al1 possible space groups. Mer many months of trying, no reasonable structure couid be found using the XTAW.2(1992) package, or SheLr86 (Sheldrick, G.M. 1986).

When aii attempts to rehe the structure as merlinoite or as a new structure failed, it was decided to investigate the possibiiity that this mineral is actually phillipsite. The crystal was mounted once again on the CAD4, and data were collected on a ce1 rneasu~g a=9.940(6)q b= M. 15(2)A, c= 8.656(5)A, and beta= 125.01(4)O. A total of 288 1 reflections were measured in the range hm0-20' theta. Using the data reduction prograrns ADDREF and SORTRF from XTAL3 -2 (1 W2), the raw data was reduced to a sorted list with the following statistics:

Reflection packets output to archive 936

Reflection statistics h k 1 theta

min O O -8 2.48

max 9 13 6 19.97

Reflection counts for rcodel rcode2 rcode3 rcode4 rcode5

874 62 O O O

Subsequent attempts at refinernent of this data set aiso met with hstration and failure however, as the lowest R-value achieved was even higher than that of the rnerlinoite refinements. -32-

Based on the crystal morphology, it was thought eariy in the study that twinning might cause problems. When two data collections seemed to have no problems with hding the cells specified however, it was decided that twinning was probably not an issue. Oniy after months of unsuccessful work with the data were possible problems due to twinning addressed again. Precession camera work was initiated to answer the questions.

4.2.2 Precession Camera Work

Buerger precession photos were taken using Polaroid fiIm and zirconium fiitered molybdenum Ka radiation A new crystal of dimensions 0.07mrn X 0.07mm X 0.25mrn was selected and mounted on a glas fibre using epoxy. The new crystal was selected, instead of using the one fkom the CAD4, study because it has better developrnent of the crystal faces which aided in orienting it on the precession camera. Optical goniometry was employed to aiign the pnsm zone of the crystal parallel to the fibre axis, and the first face of the pnsm was positioned perpendicular to the X-ray beam. Mer several orientation fiims, the crystal was aligned properly and the zero-level film (Figure 4-3) was prepared and indexed with the merlinoite cell. As the extinction conditions tabulated at the bottom of Figure 4-3 show, the symrnetry is consistent with the Immm space group of merlinoite at this point. Ce11 dimensions measured on the photo show b = 14.2~4and c = 9.94A. Instrumental Settinas 40kV, 15ma, Mo& Zr fitter p=25O, r,=15, screen disbnce=32rnm

Crystal Orientation b-

X-rays directed along -a,, direction

Measured Cell Dirnensio --merlinoite 1 a= nia

Figure 4-3 Zero Level Precessionfilm indexed with merlinoite cell

The zero level film can be described and indexed with an orthorhombic merlinoite cell with dimensions (direct space) b = 14.21A, c = 9.94A. The cell geometry and extinction conditions shown (seetable below) are consistent with the lmrnrn space group reported for rnerlinoite.

Extinction Conditions Observable on this Film ! 1 Meriinoite Cell l 1 j hkl: l 1 Okl: k+l=Sn 1

/ OkO: k=2n

Consistent with lmmm space group Figure 4-4 Cone Axis Film.

The circular diffraction cones shown on the original film have been traced in this image to show the positions more clearly. The first ring (radius Ft,) is due to the direct bearn intersectingthe film while it precesses at the uangle of 10'. Since the uangle can be set more accurately than the film distance s on the precession camera, the relationship & = s tan u was used to accurately determine s by assuming u is 10' and R,, measured as 7.18rnm. The s distance calculated in this way is 41.4mm. Using this distance and the relationship d-' = (cos u - cos(tan-'(RJs))) 10.71 07, the d' values were calculated as follows:

WNi the inaccuracies usually present in a cone-axis photo, this is sufficient to show that the spacing along the axis is approximately 14& as expected from the merlinoite cell dimensions. Instrumental Settinas 40kV, 15ma, Mo& Zr filter p=2O0. r,=15, screen distance=29mm film advance= 3mm

Crvstal Orientation

b-

X-rays directed aloi -a,, direction

Figure 46 First Upper Level Precessionfilm Iklindexed with merlinoite cell

The deaction spots shown on the first upper level are indexed with the merlinoite cell The indexing in the merlinoite cell shows no extinction conditions for hkl reflections, which is in violation of the lmmm space group assigned to rnerlinoite (should be h+k+f=2n). Note that weak diffraction spots which showed up on the original film but not this electronic image have been indexed as obsenred. Extinction Conditions Observed on this Film

Meriinoite Cell l r -- - - -

/ hkl: No restrictions l ! Okl: k+l=Zn lfiom zero level)

j OkO: k=2n (from zero level) 1 001: I=2n (from zero level) Violates lmmrn symmetry -36-

A cone-axis photo (Figure 4-4) was taken next, to determine the lattice spacing perpendicular to the b*c* plane in preparation for photographing the upper level kl. A cone-axis photo is generdy not accurate enough to rneasure precise ce11 dimensions, but it is sufncient to determine that the spacing is between 14.15A and 14.7A in this crystal.

Since the CAD4 data file indicated a spacing of about 14.19k this value was used to set the precession camera to image the upper level.

Indexing of the upper level -M with the merlinoite ceU shows violations of the Immm space group, as tabulated in Figure 4-5. The presence of such reflections as 3 1, and i5 1 violate the constraint of h+k+l = 2n fiom Imrnrn.

Carefitl inspection of the reciprocal lattice net in the zero-level fih showed that it could also be indexed with a phillipsite cell (Figure 4-6). The transformation matrix to convert fiom the merlinoite indexing to the phillipsite indexing scheme is:

AU reciprocal lattice points show on the zero-level film can be indexed using the phillipsite cd. The lattice shows no restrictions on reflections, as tabulated at the bottorn Instrumental Settincis 40kV. 15ma, Mo& Zr filter p=25', rs=l5, screen distance=32mm

Crystal Orientation c,, b,,

X-rays directed along -a,, or -b,, direction

Measured Cell Dimensions p hiliipsite

Figure 4-6 Zero-level precession film indexed with phillipsite ceII

The zero level film from Figure 4-3 can also be described and indexed with a phillipsite cell with dimensions (direct space) a = 9.94A, c= 8.66A. and a p angle of 124.8 degrees. Cornparison with JCPDS card 39-1 375 shows that the intensity distribution is consistent with this indexing (e.g. 207 is listed as one of the 'strong" reflectionson the JC PDS card) . Extinction Conditions Obsewable on This Film Phillipsite Cell hkl: - I Okl: - h01: No restrictions 1 hkO: - ! hOO: No restrictions 1 OkO: - I 1 001: NO restrictions ! Consistent with PZ, space group Instrumental Settings 40kV, 15ma, Mo& Zr fiiter l.r=20°, r,=l5, screen distance=29mm film advance= 3mm

Crystal Orientation

X-rays directed along -a,, or -b,, direction

Figure 4-7 First Upper-Level Precession film hi1indexed with philfipsite cell

This upper level film hil has benindexed with the phillipsite cell. It can be seen that only every second diffraction spot can be explained with the phillipsite cell, with anomolous spots "halfway betweenn along the c mis. Subsequent precession work shows that these spots are due to twinning in phillipsite (see Figure 4-8 and Figure 4-9). Note that weak diffraction spots which show up on the original film but not this electronic image have been indexed as observed.

Extinction Conditions Obsewed on This Film I PhiIlipsite Cell 1! / hkt: No restrictions 1 Okl: No restrictions 1 i h01: No restrictions lfrom zero level) 1

i hOO: No restrictions (from zero level) +I OkO: - 1 : 001: No restrictions (from zero level) 1 Consistent with P2, symrnetry -39-

of Figure 4-6, which is consistent with the space group P2, of phillipsite. Direct space ce11

dimensions measured fiom the film are a = 9.946i ,c = 8.66A , and P = 124.8'. These

compare favorably with ceil dimensions and angles reported for phiiii psite.

Indexing of aii of the Iattice points on the upper-level photo with the phillipsite ceIl was

not possible. As show in Figure 4-7, only every second spot on the film in the c direction

can be indexed. This problem was solved with the next set of precession films.

Originaiiy, the crystal was mounted on the glas fibre so that only the dia1 axis of the precession carnera needed to be rotated in order to put the plane containhg the b* axis into the difiaction position. Essentialiy, since the a%* plane was perpendicular to the X- ray beam, a rotation of the crystallographic gamma angle (90°in orthorhombic and monoclinic cells) on the did should have resulted in a photo showing the b* axis.

Surprisingly, as show in Figure 4-8, the resulting photo is identical to the first zero-level film taken (Figure 4-3). The upper-level photos are identical as well, as shown in Figure

4-8.

The fact that photos taken at 90" dial rotations are identical points out that this crystal is in fact twinned. Twinning was suspected early in the study simply by the morphology of the crystals, but when the CAD4 software had no trouble finding a reasonabie unit cell, the possibiiity of twlluiing seemed remote. Figure44 Camparison of precession photostaken at right anglesto each other

Zero and first upper-level precession photos taken at dial rotations of 90" to each other are compared side-by-side. The photos al1 show a'c* of phillipsite or bec*of meriinoite indicatirtg that there is a Win plane in the crystaf at a 45' angle. Relative to the phillipsite ceIl. this plane has indices (021). -41-

Twinning is a very cornplex, and to some degree poorly understood phenomenon in

minerals. In twinned minerals, two or more orientations of the crystal stmcture are

superimposeci that are related by a syrnmetry operator. The symmetrical relationship and

cornrnon ongin of the crystal lattice are the cnteria for distinguishing twins fiorn simple

"multiple crystal" growths. Mirror planes, centres of symmetry, and rotation axes can ali

act as twin operators. Ofien, twinning occurs when the unit ceiI has higher symmetry than the mineral itseK an example of which is shown in this study. It has been show that a pseudo-orthorhombic (and in this case nearly tetragonal) ce11 can be used to descnbe

(monoclinic) phillipsite.

Domay and Donnay (1974) have discussed the difnculty of recognizing twinned structures in some minerais. Before the advent of X-ray crystdography, twins could only be recognized and studied when there was an external expression of the twinning. These extemal expressions are usually obvious symmetry-related intergrowths, or syrnmetrically aîtached crystals. Precession methods help in the recognition of more subtle twins by actually showing the superimposeci reciprocai lanices. Even with the use of X-rays though, there are still wes where twinning is extremely difncult to detect.

Domay and Donnay (op. cit.) have divided &vins into two overall groups: twinning by twin-lattice symmetry (TLS), and twinning by twin-lattice quasi-symmetry (TLQS). The distinction made is based on the symrnetry elements present in the crystal lattice compared to the symmetry of the ciystd itself For example, £tom a pure geometry -42-

viewpoint, a crystai lattice can approach perfect orthorhombic or tetragonal symmetry

while the crystai structure itself (due to atomic arrangement) is monoclinic. In twinning by

TLS, the twh operator eda symmetry element of the lattice. In the preceding

example of an orthorhombic or tetragonal crystal lattice, the twin operator by TLS would

be the two-fold or four-fold rotation axes respectively. After twinning, the twin-lattice

points are perfectly superimposeci on the original-lattice points. Catti and Ferraris (1 976) define this type of twinning as "twinning by merohedry". When the lattice deviates slightly fiom the "perfect" higher symmetry, twinning cmstill take place using the pseudo- symmetry element. The resulting twin lanice however is offset from the original by an amount relative to the deviation f?om penect symrnetry show in the crystal lattice. This is classineci as %ng by hvin-lattice quasi-symrnetry by Domay and Do~ay(op. cit.)

On precession films, TLQS is easy to identify. Two distinct orientations of the reciprocal

Iattice are visible, evidenced by "doubling" of the lattice spots. On the other hand, if the twinned and untwinned lattice points are exactly superimposeci due to the geometry of the cd, then there is no "doubling" of the spots and the twinning is by TLS (merohedry).

Twinning by twin-lanice symmetry is often seen in monoclinic rninerals which have beta angles very close to 90". As an example, Domay and Donnay (op. cit.) and Hoffinan et. al. (1973) point to Steinfink's (1962) solution of the structure of phillipsite in an orthorhombic space group when it is actuaily monoclinic. Steinfink's crystal was pseudo- orthorhombic and hwuied by TLS. -43-

The phillipsite in this study is another example of twinning by TLS. As shown on the

precession photos and in the CAD4 data collection, a unit cell can be dehed (the

"merlinoite" ceil) which is pseudo-orthorhombic and very nearly tetragond ( la( a 161, a, P,

y, 90"). Examination of the crystal morphology (Figure 4- 1 1) shows that each crystai is

acnially an interpenetrating twin. The composition planes appear to be mirrors with Miller

indices (021) and (021) relative to the monoclinic (P=124.80 phillipsite ceU.

Interpenetrating twins are often better described by rotation twin operators rather than

mirrors, however. A very similar case of twinning is shown by staurolite, which is

comrnonly found as cross-shaped interpenetrant twuis. Hurst et. al. (1956) showed that

staurolite is monoclinic with a p angle of nearly 90°, and that it twins by a 90' rotation

about [100]. The resulting crystal shows two mirror planes at right angles parallel to the a

axis, exactly the situation that is seen in the phiIlipsite from Cava Nuova.

In summary, the hvinning by TLS seen in the sarnple being studied wi be described as a

four-fold rotation about the direct-space a axis of phillipsite. A compact notation for this twin law is [100],. Because of the pseudo-tetragonal lattice in this direction, the difiFiaction spots of the twin lattice are exactly superimposed on the original lattice in the zero layer within the resolution limits of the precession camera. There is no doubling of the spots on the zero-layer precession photos because of this fact.

Most importantly, this "twin model" cm be used to explain the extra reciprocal lattice -44- points in the upper-levei films. Figure 4-9 illustrates the effects of the [100], twin law on the reciprocal lattice. Since the rotation axis is wincident with direct-space a, it affects reciprocai lattice points within planes parailel to the b*c* plane. Ifwe examine such an upper-level plane (Figure 4-9@)), we see that the rotation tums the structure 90' about the a axis. When the two structures are superimposed in Figure 4-9(c), it can be seen that extra reciprocal lattice points are created in odd-numbered levels along b* (hll, Ml, etc.).

Due to the geometry of the philiipsite ce11 in this crystai, these "twin" spots are located alrnost exactly halfway in between the "normal" reciprocal lattice points. The rnirror planes (021) and (021) are also created as a result of the rotation.

The zero-level film showing a*c* (Figure 4-6) is largely unaffiected by twinning (at least at the resolution of the precession camera films) because it is an "even numbered laye? along u,and the twin lattice points wül simply be superimposed on existing points. As supporting proof, a theoretical calculation of the d spacings using the program REFLEX

(Hay, 1995) in a crystai with the ce11 parameters of this sample is included in Appendk II.

It shows that the vector [020]* with length 7.105A (the d-spacing for the (020) plane) would be exactly superirnposed on [001]* with length 7. t 1 lA for example.

The twin mode1 explains not only the distribution of the lattice points, but also the diniculties with renning the structure from the CAD4 data. Many of the reflections in the reciprod lattice that were assumed to be single diffraction points are actually the sum of two reflections. This changes the intensiîy at each point, which makes it impossible to cf 4-9 (a) A perspective view of the phillipsite reciprocal lattice with direct-space axes Beta* angl a=9.94A. b=14.21A. c= 8.66A and f3=124.8°. of 54.2" In order to understand the effect of a direct- space twin plane of (021), the tattice must be viewed down the crystallographic a direction (perpendicular to the bec* plane). 4-9(b) shows the first upper level along a* (the 1kl plane). First Upper luyer

4-9 (b) A scaîe view of the 1kl plane in phillipsite as seen down direct-space axis a for cornparison with the precession film hil from Figure 4-7. Twin planes with indices (021) and (027)due to the [IOO], rotation are shown in green. It is important to note that the origin (on the zero-level) falls exactly halfway between reciprocal lattice points along cl' due to the geometry of the phillipsite cell (as explained in Figure 2-5). This is why the rnirror planes do not cross on a lattice point in odd-nurnbered upper-level planes along direct-space a. Results of the Winning are shown below, and cornpareci to the upper- level precession photo.

Direct-mace a axis -+ b* axis -+

w lattice points due to twinning< A

Figure 4-9 Reciprocal Lattice Points Due to Twin Rotation

The blue spots in the diagrarn at the bottom nght are due to the 90°twin rotation. In sorne cases, lattice points are superimposed on top of existing ones, but in other cases a lattice point is created exactly haif way in between existing lattice points. The upper level diffraction pattern to the left (from Figure 4-7) can be completely explained by twinning in phillipsite by a 90' rotatior! about a. -46-

calculate meaningful structure factors for the least-squares rehement routine.

In sumrnary, the minera1 being studied is actually a twinned philiipsite. The action of the

twin rotation creates a composite reciprocal lattice that has the dimensions and shape of the merlinoite cd, but which does not show the extinction conditions of its space group.

As shown by the attempts at structure refinement, the "twinned stmchire" created is ironically more like merlinoite than phillipsite. Crystals of phillipsite and harmotome almost always occur as cornplex twins, often as fourlings and occasionally as eightlings and higher-rank twhs (Galli and Gottardi, 1985).

Forms wmrnonly present on the crystals are (001) and (OlO), with (1 10) terminations.

These simple forms are usuaiiy acted upon by various twin planes forming very corn plex- looking crystais. Figure 4-10(a) shows some of the cornmon forms of twimed phillipsite ciystals.

Crystals of phillipsite fiom Cava Nuova exhibit rare morphology for this rnineral, and this may have been a factor in their initial identification as merlinoite. The crystals show an elongated, pseudo-tetragonal prism paralle1 to the a crystallographic axis. Carefùl examination of the terminations (Figure 4-1 1) shows that each crystal is actually an interpenetrant twin. None of the crystals examined showed any striations or re-entrant lines on the prism faces however, and the smail size of the teminations made it difficult to recognize the twinnllig originaily. The faa that these crystals are twinned explains the pseudo-tetragonal shape of the prism zone, even though the rnineral itself is monoclinic.

Although no explicit study on the morphology of phillipsites nom different environrnents could be found, a visual survey of articles with photographs shows that the morphology is dxerent. Hydrothermally-formed phillipsite in wgs in basalts tend to have the (1 10) tenninations as outlined in Gottardi and Galli (1985). Phillipsite fTom sedimentary Figure 610(a) Common Morphology of tmnned phillipsite crystals

Phillipsite crystals are very rarelyuntwinned. Crystals comrnonly show the forms (001) and (010) teminated by (110) and (110) Twin planes, shown with dashed lines in the above diagram, create cornplex fourling (first crystal) and eightling (second and third crystals) twins from these simple foms.

(loi) Figure 4-(0(b) Theoretical untwiannedand observed twin Phillipsite crystals.

The crystal on the left shows the theoretical untwinned foms of the phillipsite from Cava Nuova. The-Cava Nuova crystals are unusual for phillipsite because they are teminated by (100) and (101). Although these foms have been found elsewhere, the morphology shown in 4-10 (a) is more common. The crystal on the right is a Win produced by specwing 90" rotation about [IOO]. It matches the morphology of the crystals in this study exactly, as shown in the photomicrograph in Figure 4-11. Figure 4-11 Photornicrograph of crystal used for precession, powder diffraction, and rnorphology studies.

This crystal of phillipsite shows the interpenetrant twin morphology outlined in Figure 4-IO(b). All crystals in the sample exhibit the same form. The photograph is slightly dark since the crystals are actually colourless and transparent. The photograph was taken in oil (R.I. 1.55) using planepolarized light with a Nikon Microphot- FXA. -50- deposits in the oceans (Sheppard, 1970) and synthetic philiipsite grown in the lab

(Donahoe, 1985) seem to show the (100) and (loi) terminations most often. The rnorphology of the crystals at Cava Nuova is definitely unusual for the hydrothennal environment in which it is found.

Aithough extremely rare, other examples of this twin morphology have been found.

PassagIia et. al. (1985) reported one example fkom Saint-Jean-le-Centenaire, Ardèche,

France. The crystals ocair in wgs in basalt, and match the morphology of the Cava

Nuova samples exactly. Although they offer no explanation of the twin law, they refer to it as a "Marsburg hwi" which is the name given to sirnilar crystals originaily found at

Asbach, Germany in the late 1800's (Passaglia et. al, 1985).

Using the crystal drawing software package Shape (Dowty, 1995), the crystal in Figure 4-

10 @) was drawn using the axes determined frorn precession and CAD4 work. The drawing of the untwinned crystal (which is rarely seen in phillipsite) is made up of the simple foms (0 10) and (00 l), uncharacteristically terminated by (101) and (100). The drawing of the twimed fom of the crystal (which exactly matches the samples fiom Cava

Nuova, Figure 4-2) was produced by speciSing the twin operator [100],. The resulting crystal has four individuals, each showing the foms (OlO), with terminations (100) and

(101).

Before the crystais were drawn, an optical goniorneter was used to measure the angles -51- between the faces. Due to the small size of the crystafs, accurate measurements could be made only on the pnsm faces. The crystal was adjusted so that the prism formed by (001) and (010) faces was pardel to the horizontal circle of the goniorneter. In this orientation, the angles between the pnsm faces could be rneasured directly using only the dial axis with the horizontal circle held in its "zero" position of 3 13' 30'. The four faces gave did readings of 18' IS', 108' 20', 198' 30', and 288' 15'. By subtraction, this gives angles between the faces of 90° 5', 90' IO', and 89' 45', and 90' 0'. In the monoclinic system, these angles are exactly 90'.

An attempt was made to measure the angles between the terminal faces, but the srnaii siie of the faces made it impossible. One set of measurements gave horizontal circle readings of 77" 30' and 9' 30' yielding 68' for (100)~(10ï).The "Shape" software package reports the calculateci angle as 69-88'. Measurements muld not be made for the other two terminal faces on the natural crystal. -52-

Chapter 5: Chemistry

Zeolites are characterized by a fkmework of silicon and aluminum tetrahedra with loosely bound "extra-firamework" exchangeable cations and water rnolecuIes. A general formula for ail zeolites has been defined (Gottardi and Ga,1985) as:

in which the atoms within the square brackets produce the fiamewort and all others are extra-framework atoms, cations and water molecules. Since the cations easily substitute for one another, many zeolite species cannot be represented by a single rigid chemical formula as the oniy real constraint in the general formula is that SitAi.

Gottardi and GalIi (1985) have proposai a balance-error formula based on the general chemid formula to check the validity of chemical analyses, as follows:

If the balance error exceeds 10 in any anaiysis, it shodd be discardeci as unreliable.

A study of the crystai chernistry of phillipsites (Galli and Ghittoni, 1972) showed a large variability in the composition of phillipsite. Silicon occupancy in the tetrahedral sites can -53-

vaiy nom 57% to 74% (with the balance occupied by aluminum) and while potassium is

always present, sodium and calcium range f?om abundant to almost absent. The paper

also notes that some analyses in the literature are unreliable, with poor balance-error

sumrnations.

For this study, two sarnples were mounted in epoxy-based microprobe "mini-mounts".

The first is a single crystal mounted in an orientation to analyze for chernical variation fiom the nrn to the core of the crystal. The second mounted sarnple is a tiny matrix chip with crystals. Both samples were very difficult to polish accurately because of the small size of the crystals (a third sample was ground completely away during the polishing process).

The mples were analyzed in the Queen's University Electron Microprobe Laboratory using an ARLSEMQ electron microprobe in energy dispersive system (EDS)mode. An accelerating voltage of 15kV with a barn current of 40 nano-amps was used for al1 analyses. A Tracor-Northem system ninning the general analysis program "TASK" was used for data acquisition and Bence-Albee corrections, with the spectra collected between

O and 10.23 kev over a count tirne of 200 seconds per a&dysis.

The silicate-glas microprobe standard S-204 was used as a reference standard since no zeolite, or zeolite-like, standards were available. A preliminary analysis on the mounted samples showed a complete absence of barium which, although not essential, was -54-

expected in this sample due to its initiai identification as merlinoite. Based on this fact, and

an acceptable andysis for potassium on the probe-standard OR-1 (= S-75, adularia) when

analyseci as an unknown, it was decided to continue to use S-204 as a standard.

It was reaiizeci early in the analysis that a point-beam anaiysis would not be possible.

Under an intense beam with a count the of 200 seconds, Na and K atoms migrate out of the excitation area resulting in inaccurate anaiysis totals for these atoms. Once the beam was set to a raster-scan size of roughly 20 pm, the analyses were thought to be acceptable. As subsequent balance-error calculations showed (Appendix m),the raster size should probably have been set higher, or the count time shortened, for more accurate analysis of sodium and potassium levels.

A total of 23 analyses was done on the crystals under investigation (Figure 5-1).

Subsequent balance-error calculations on the analyses showed that of that number, only 5 were acceptable. AU analyses, with atomic proportions recaiculated to philfipsite and merlinoite formulae, are presented in Appendix III with the calculated balance-error.

Low chi-squared values of approximately 2.60 - 4.37 for these 5 analyses indicates a good

"fit" of the calculateci-to-observed spectra using oniy the elernents in the standard 10- element silicate analysis. Badon this facf the water molecule proportions in the formula were calculated by difference in the totais fiom 100%. The acceptable analyses gave chernical forrnulae as follows: In general, these are potassium and sodium enriched, but calcium-poor phillipsite samples.

Silicon atoms occupy approximately 70% of the tetrahedral sites, which is at the high end

of the range reported by Gaiii and Ghittoni (1972). In terms of chemistry, this sample is

unusual but not out of line with other phillipsite analyses as reported by Gaili and Ghittoni

(op. cit.). Their analyses show that in high-silica members of the group there are reiatively few Na, Ca, K., and Ba cations in the structure, an observation which is dso shown in t hese analyses.

In terms of overall chemistry, these analyses are closer to the compositions of deep-sea phillipsites (Sheppard et. al., 1970) fiom the Indian and Pacific Oceans. The Cava Nuova samples have an Si/A ratio ranging fiom 2.27 to 2.86 with an average of 2-47. This fits into the range reported by Sheppard (op. cit.) of 2.44-2.79. Deep-sea phillipsites also tend to have high K and Na values, and low Ca compared to hydrothermal phillipsite.

Yet again, the similarity to merlinoite K,C~(~Si,O,)-24H20 is show in the chemical Figure 5-1 (a): Back-scattered Electron lmage of MRL-1.

Analysis areas are shown &y approximately 20pm squares, representing the raster size of the beam. Analyse

Figure 5-1(b): Back-scattered Electron Image of MRL-3

Analysis areas are shown by approximately 20pm squares, representing the raster size of the beam. Analyses with acceptable balanceerrors are shown in green. Analyses 24 and 25 are analcime. Since the formulae of zeolites are extremely variable however, the underlying framework structure is the definitive key. As show by precession work, this sample is dominantly a twinned phillipsite stmcture. -58-

Chapter 6. Discussion of Results and Conclusions

This mineral shows characteristics of two very similar minerals: merlinoite and p hillipsite.

Results of the precession work done on a single crystal point ovenvhelmingiy to the fact that the mineral is a twinned philiipsite. Morphologidy, the crystals can be explained again by the twinned phillipsite rnodel.

The major problem with this sample being interpreted solely as phillipsite is show in the powder daaction pattem prepared using the same crystal as was used for the precession work. The powder pattem shows definitive lines characteristic of both mineral species within a single crystal. While single-crystal experiments are greatly affectecf by twinning in a crystal, the powder pattern does not generally show effects of twinning. And so the problem becomes, how can the merlinoite lines be explained in the powder pattern?

The simiIarities between the phillipsite and merlinoite structures were noted by Galli et. al.

(1979) in their original paper on merlinoite. They state that the ribbon structure in phiIlipsite can easily be transfonned into the "octagond ~g"structure of merlinoite by a simple shifk or slippage of the stmcture dong the (phillipsite) b axis. They go on to Say that if this slippage occurs to a large extent, a crystal could form which has characteristics of both minerals.

Figure 6-1 shows another way in which merlinoite-like domains might form in the minerai being investigated in this study. It has already been shown that these crystals are twinned -59-

by [100],. At a hvin transition, the "wavy ribbon" structure of phiiiipsite takes a right-

angled tum. As a wnsequence of this tum in the structure, cornpiete ring structures

(similar to those found in merlinoite) form dong the (021) plane. The same twin rotation

acting on the "tumed" ribbon causes more rings to form. Lfcarried to the extreme in a

single area, either at the core of the crystal or dong the (021) composition planes, this

repetitive twinnllig can produce the topology shown in Figure 6-l(c) which is exactly the

merlinoite structure. In effect, domains of merlinoite are created within the phillipsite, resulting in a crystai which has characteristics of both minerais.

In order for this mechanism to account for merlinoite lines to show up in the powder pattern however, the twinning has to take place over an appreciably large volume in the crystal. The width of a dmaction peak incrûases, as the thickness of the crystal decreases

(Cullity, 1978). The Sherrer formula, t = 0.9A1 (B cosû,) can be used to estimate the particle size of very small crystals. The thickness t is related to the wavelength A, the width of the difiaction peak B, and the theta angle at which the peak occurs. Assuming copper radiation and a theta angle of 4-41' (the 10A 1 10 plane in merlinoite), a peak with a measurable width of 1' requires a crystal thickness of 79.74A perpendicular to the 110 reflecting plane. In this direction, a single merlinoite ceil is approxirnateiy 20A across, and so a minimum of 4 unit celis could theoretically mate a recognizeable dfiaction peak.

The exarnple serves to show that a relatively smaii volume of merlinoite in phillipsite would show up in a powder pattern. Figure 6-2 shows a cross-sectionai view of a twinned crystal down the a axis in both plane-polarized light, and crossed polars. The major (021) \3- *d Figure 6-1 (a) The schematic phillipsite structure viewed down a

Figure 6-l(b) Composition Plane (021) formed by [100], twin rotation

Figure 6-1 (c) Repetitive twinning results in the merlinoite structure.

Figure 6-1 Twinning causing merlinoite-like domains in phillipsite

The diagram at the top shows the Wavy ribbon* framework structure of phillipsite. If the structure twins along (021) as shown in (b), the ribbon makes a 90' turn at that point. Along the composition plane, domains are created which have the full ring structure similar to those seen in the meriinoite framework.

In order to create the merlinoite structure exactiy, twinning must occur in the "tumed ribbon" as well, as shown in (c). Figure 6-2 shows that twinning is cornpiex and extensive in the crystals being studied from Cava Nuova, and repetitive twinning as shown in (c) is likely in these samples, especially near the core. -61- composition planes due to the [IOO], twin axis can clearly be seen as diagonal iines running across the entire crystal. More importantly, the individuals between the major twin boundaries show a very cornplex extinction pattern under crossed polars which is reminiscent of the twinning occasionaiiy seen in thin sections containing twinned feldspars.

This shows that the twinning is not confineci exclusively to the major (021) twin boundaries in the crystais being studied. Mertinoite domains are most likely spread throughout the entire volume of the crystal and concentrated in the core in large enough volumes to result in difEaction lines on the powder pattern.

The chernical analyses can not be used to differentiate between merlinoite and phillipsite due to extreme variability in the chemistry of phillipsite . The fact that the formula of the

Cava Nuova material is very close to merlinoite, however, may actually have contributed to the abundant twinning in the crystal leading to the formation of merlinoite domains. A sûnilar phenomenon is seen in feldspars within the sarne fiarnework topology where different twin laws occur at dBerent compositions (albite twinning and pericline hvinning for example). Plane-Polarized Light

Crossed Polars

Figure 6-2: Photomicrograph Down a Showing Cornplex Twinning in PhiIlipsite

This pair of photos are of the same crystal in plane-polarized light (PPL) and under crossed polars (XPL). The ?PL view shows obvious zoning within the crystal. Under XPL, the crystal shows the overall Win plane (021) and an extremely complex pattern of twinning throughout each quadrant. Domains of merlinoite are most likely spread throughout the crystal due to this twinning.

The photographs were taken in oil (R.I. 1.55) using a Nikon Microphot-FXA. -63-

Chapter 7: Recommendations for Further Study

As has already been discussed, the morphology of the crystals, the difnculties encountered during the precession work and the CAD4 diftiactometer study, and the powder pattern of this sample can be explained by the twin model. In order to actually prove that this mechanism is responsible, the volume occupied by the merlinoite domains should be investigated. Aithough possibly not sensitive enough, a Rietveld anaiysis of the powder pattern may give a rough idea of the percentages of merlinoite and phillipsite present if enough sarnple could be obtained. The Rietveld technique is a method of structure rehement based on powder patterns, rather than single crystal studies. The use of a powder would circumvent the difficulties encountered in the CAD4 data collection due to the overlapping twin reflections. Rietveld refinement ailows the simultaneous refinement of multiple phases within one powder, and cm be used to detennine percentages of ail phases present in the powder. Perhaps the powder pattern could be refmed using this technique spec@ng that both phillipsite and merlinoite are present. The overlap of most of the peaks in merlinoite and phillipsite would probably be a major problem in using this method, however.

Now that the twin law has been determined for this sarnple, it may be possible to refine the structure using the twinned crystal data. Computer programs such as SHELX93 and newer versions of XTAL are available which can calculate structure factors taking the twinning into acwunt. In order for them to be successful however, the relative amounts of each twin must be known. -64-

The orientation of the merlinoite domains within these phillipsite crystals is the most important consideration in proving the validity of the twin hypothesis presented in this study. The merlinoite is present either as "twin domains" spread throughout the ciystal, or it could be that the crystai is zoneci f?om core to rim with "shells" of merlinoite and philiipsite. Since there is no way to distinguish the two phases chernically, a microprobe cannot be used to show the zoning. A transmission electron microscopy (TEM) study should be carried out to image the ample at the unit ceil level. h this way, it could be ascertained whether the merlinoite domains are in fact grouped dong twin boundaries, and not simply present as physically separable overgrowth or core zones within the phillipsite crystal.

Lastly, the crystals produced in the merlinoite synthesis experiments of Donahoe et. ai.

(1985 and 1990) need to be studied more thoroughly. It is possible that the "merlinoite" formed is actually present in a twi~edphiIlipsite-like phase similar to the mineral in this study. The Cava Nuova material was previously identified as merlinoite by powder X-ray dEaction in soite the presence of the obvious philli psite line at -6.4A. Dohahoe et. al.

(1 985) note that the "p hiIlipsite" produced in their expetiments match synthetic ZK- 19

(Kühl, 1969) which lacks this d-spacing. Because of this, there is no way to distinguish a twimed phillipsite-like structure with merlinoite domains by X-ray powder difiaction aione. Single-crystal precession and difiactometer work, and optical microscopy as in this study are needed to determine the ctystal structure explicitly. There is no mention in the papers of any identification techniques other than powder diffraction being applied to -65- the synthetic material. Consequently, it is possible that the precipitates are not pure merlinoite, and so the thermodynarnic data presented in Donahoe et. al. (1 990) although accurate may not be valid for true merhoite. References

Alberti, A, Hentschei, G., Vezzaiini, G., (1979) Amicite, a new naiural zeolite. Neues. Jahra. Miner. MonaMtrh. 1979 pp 135- 144.

Barrer, R.M., Baynham, J.W.,(1956) The hydrothenal chemistry of the silicates. Part W. Synthetic potassium aluminosilicates. J. Ch.Soc. (Zond) 1956 pp 2882-2891

Catti, M., Ferraris, G., (1976) Twinning by Merohedry and X-ray Crystal Structure Determination. Acta. Cryst. A32 , pp. 163- 165.

Cerny, P., (1964) The phillipsite-wellsite-harmotomesymmetry: orthorhombic or monoclinic? Neues. Jahrb. Miner. Momtsh. 1964, pp 129- 134

Cullity, B.D., (1978) Elements of X-ray Dfiaction (2.6 Edition), Addison-Wesley Publishing Company Inc.

Donahoe, R., Liou, J.G., (1985) An Expenmental Study On The Process of Zeolite Formation. Geochim. et Cornochim. Acta 49, pp. 2349-2360.

Donahoe, R., Liou, J.G.,Hemingway, B.S.,(1990) Thermochemical data for merlinoite: 2. Free energies of formation at 298.15 K of su synthetic sarnples having various SVAI and Na/(Na + K) ratios and application to saline, aikaline lakes. Am. Min. 75, pp. 201- 208.

Donnay, G., Donnay, J.D.H, (1974) Classification of Triperiodic Twins. Cm. Min. 12, pp 422425.

Dowty, E., (1995) SHAPE for Wmdows, Kingsport, Tennessee, USA.

Fleischer M, Mandarino J, (editors) (199 1) Glossary of Mineral Species 1991. published by The Mineralogical Record Inc., Tucson, Arizona.

Galli, E., Ghittoni, A.G., (1972) The Crystal Chernistry of PhiIlipsites. Am. Min. 57, pp 1125-1 145.

Gottardi, G., Galli, E., (1 985) Natural Zeolites. 1985 (book) Springer-Verlag.

Hall, S.R, Flack, H.D., Stewart, J.M., editors(l992) XTAL 3.2 Reference Manual, University of Western Australia.

Hay, D., (1 995) Crystallographic Utility Prgrams. CSIRO Division of Materials Science and Technology, Australia Hurst, V.J., Do~ay,J.D.H., Do~ay,G., (1956) Staurolite Twinning. Min Mag. XXlQ pp. 145-163.

Hofian, E.L., Do~ay,J.D.H., Donnay, G., (1973) Symmetry and Twinning of phillipsite and Hannotome. Am. Min. 58, p. 1105 (ahtract)

Khomyakov, AP., Kurova, T.A.,Muravishkaya, G.I., (1981) Merhoite, First Occurrence in the USSR (in russian). DokL AMNd SSSR 263, pp. 978-980.

Kühl, G.H.,(1 969) Synthetic Phillipsite. Am. Min. 54, pp. 1607-16 12.

Milton RM, [to Union Carbide Corp](196 1) Zeolite W, Br Patent 864, 707; US patent 3, 012, 853

Mohapatra, B.K., Sahoo, R.K., (1987) Merlinoite in Manganese Nodules From the Indian Ocean. Min Mag. 51, pp. 749-750.

Mumpton F.A., (1977) Naturai zeolites. In ''Mineralogy and Geology of Natural Zeolites" MS A Short Course Notes, vol 4, pp 1- 1 5.

Passaglia, E., Gaüi, E., Gottardi, G.,Vezzaiini, G., (1985) An anornaious phillipsite from Saint-Jean-le-Centenaire, Ardèche. Bull. & MN>.108, pp. 7 19-724.

Raven, M., (1995) XPLOT powder difiaction software package. CSIRO, Australia

Rinaldi, R., Pluth, J.J., Smith, J.V., (1974) Zeolites of the Phillipsite Fady. Refinement and Crystal Stmctures of Phillipsite and Harmotome. Acta. Cryst. B30 pp 2426-2433

Sadanaga, R., Mammo, F., Takeuchi, Y., (1961) The Crystal Stmcture of Harmotome, B~AI,Si,,O,,.12H2O. Acta Cryst. 14, pp 1153-1 163.

Sheldrick, G.M., (1 986) SHELX86 structure refinement program.

Sheppard, RA, Gude, A. J., Grifnn, J.J., (1970) Chernical Composition and Physical Properties of Phillipsite From the Pacinc and Indian Oceans. Am. Min. 55, pp. 2053- 2062

Smith, J.V., (1978) Enurneration of 4-comected 3-dimensional nets and classifjcation of hework silicates. II Perpendicular and near-perpendicular linkages fiom 4.8*, 3.1 22, and 4.6.12 nets. Am. Min 63, pp 960-969.

Steinfink, H., (1 962) The crystai structure of the Zeolite, Phillipsite. Acta. Cry~l.15, pp 644-65 1 Appuidix 1 Structure Refinement files : Merlinoite

LATCON input reflections and results

HKL 2T (OBS) 4 3 1 14.960 -4 2 O 12.840 -2 4 O 12.840 O 8 O 23.040 O02 8.180 -4 -2 O 12.880 -4 1 -1 12.580 -4 3 1 14.960 -5 O 1 14.900 -5 O -1 14.980 051 14.980 -2 -4 O 12-860 -1 O 3 12.640 O -5 1 14.880 251 16.040 2 -5 1 16.000 O -8 O 23.060 ERROR OF FIT .3137 FOR 14 DEGREES OF FREEDOM DIRECT ELEMENTS VECTORS IN A** 1 -- A, B, C, ALPHA, BETA, GAMMA, COSA, COSB, COSC AND THEIR E.S.D-S 14.1882 14 .2214 9.9501 90.0000 90,0000 90.0000 .O153 .O100 .O212 .O000 .O000 .O000 CELL VOLUME 2007.70 Appendu II

Theoretical d-spacings in phiiiipsite with cddimension d.94b=14.2 1, c=8.66A P= i24.8O. No extinction conditions were applied (i.e. OkO = 2n are extinct in aetual pattern)

Monoclinic Effective Camera Diameter = 1 14.59 156 Wavelength = 1.54056 Maximum Bragg Angle = 30 No calibration parameters applied

Lattice Parameters

Appendix III

Microprobe Analyses with caiculated Balance Error Note: Analyses with acceptable balance-errors are listed firg

Analysis 4: MRL 1 LOWER CORNER OF XL (RASTER) 200 SECS CHISQD= 2.73

Oxie wt.% Formula(l6 oxygens) Formula(64 Oxygens) Si02 53.70 Si: 5.96 Si: 23.88 A1203 15.86 Al: 2.08 Al: 8.30 Ca0 2.32 Ca: 0.28 Ca: 1.11 Mg0 0.00 Mg: 0.01 Mg: 0.05 Na20 0.4 1 Na: 0.09 Na: 0.35 K2C) 8.93 K: 1.26 K: 5.06 total 8 1.22

Balance Error EYo (> 10 means analysis is unreliable): 7.54

Analysis 12: MRL 1 GRID ACROSS XL AT 2000X RASTER 200 SECS CHISQD= 2.62 Oxide wt.% Formula(l6 oxygens) FormuIa(64 Oxygens) Si02 55.42 Si: 5.69 Si: 22.76 A1203 19.3 Al: 2.34 Al: 9.34 Ca0 4.08 Ca: 0.45 Ca: 1.80 Mg0 0.0 Mg: 0.0 Mg: 0.0 Na20 O. 84 Na: 0.17 Na: 0.66 la2 !kQQ K: 1.18 K: 4.72 totai 88.65

Balance Error E% (NO means anaiysis is unreliable): 4.14

Anaiysis 13 : MRL l#9 AT 2000X RASTER 200 SECS CHISQD= 4.37

QXkk wt.% Formula(l6 oxygens) Fomula(64 Oxygens) Si02 55.21 Si: 5.59 Si: 22.35 A1203 20.63 Al: 2.46 Al: 9.84 Fe0 O. 15 Fe: 0.01 Fe: 0.05 Ca0 4.25 Ca: 0.46 Ca: 1.84 Mg0 0.0 Mg: 0.0 Mg: 0.0 Na20 0.96 Na: 0.19 Na: 0.75 _K20 8.84 K: 1 -44 K: 4.58 totd 90.06

Balance Error E% (X 0 means analysis is unreliable): 9.16

Analysis 22: MRL3 #1 RASTER BOTTOM LEFT CORNER ski& EL% Formula(l6 oxygens) Formula(64 Oxygens) Si02 55.71 Si: 5.70 Si: 22.82 A1203 19.45 Al: 2.35 Al: 9.39 Ca0 3.57 Ca: 0.39 Ca: 1.56 Mg0 0.00 Mg: 0.0 Mg: 0.0 Na20 O. 7 Na: 0.14 Na: 0.55 K20 9.46 K: 1.24 K: 4.94 total 88.88

Balance Error E% Q10 means analysis is unreliable): 8.89

Analysis 23: MRL3 #2 RASTER 200 SECS CHISQD= 3.53

(3xiBe Formula( 16 oxygens) Fomula(64 Oxygens) Si02 Si: 5.61 Si: 22.43 Ti02 Ti: 0.009 Ti: 0.038 A1203 Al: 2.36 Al: 9.42 Ca0 Ca: 0.36 Ca: 1.44 Mn0 Mn: 0.008 Mn: 0.03 Mg0 Mg: 0.01 Mg: 0.0 Na20 Na: 0.20 Na: 0.78 K20 K: 1.52 K: 6-08 total

Balance Emor E% (>IO means analysis is unreliable): -4.15 -75-

Unacceptable balance-error analyses

Andysis 1: MRL 1 #1 Centre of xi Raster Scan. 200 seconds ChiSqd=2.14

OxiSe wt.% Fornula.( 16 oxygens) Fornula (64 Oxygens) Si02 55.61 Si: 5.80 Si: 23.18 Ti02 o. 10 A1203 19.26 Al: 2.36 Al: 9.46 Ca0 3.82 Ca: 0.43 Ca: 1.71 Mg0 0.05 Mg: 0.01 Mg: 0.03 Na20 0.68 Na: 0.14 Na: 0.55 K20 5.29 K: 0.70 K: 2.81 total 84-8 1

Balance Error E% (>IO means analysis is unreliable): 38.33

Analysis 2: MRL 1 #2 TOP RIGHT CORNER OF XL 200 SECS CHISQD= 2.1 1 (raster)

Qld.!z wt,% Formula(l6 oxygens) Formula(64 Oxygens) Si02 59.87 Si: 5.69 Si: 22.77 Ti02 0.08 A1203 21.29 Al: 2.38 Al: 9.54 Ca0 3 -82 Ca: 0.39 Ca: 1.56 Mg0 0.09 Mg: 0.01 Mg: 0.05 Na20 1.12 Na: 0.21 Na: 0.83 K20 8.73 K: 1.O6 K: 4.23 total 94.92

Balance Error E% (> 10 means anafysis is unreliable): 15.35

Analysis 3: MRLl TOP RIGHT CORNER OF XL 200 SECS CHISQD= 3.05 (point beam)

Oxide wt.% Fonnula(l6 oxygens) Formula(64 Oxygens) Si02 61.63 Si: 5.74 Si: 22.96 AI203 22.48 Al: 2.47 Al: 9.87 Cr203 0.09 Ca0 4.20 Ca: 0.41 Ca: 1.68 Mg0 0.06 Mg: 0.01 Mg: 0.03 Na20 O. 53 Na: 0.10 Na: 0.38 &a 5.83 K: 0.69 K: 2.77 total 94.82

Balance Error E% (> 10 means analysis is unreliable): 50.1

Analysis 5: GRID ACROSS MRLl XL (RASTER)200 SECS CHISQD= 2.85 {Beam was blanked for part of analysis}

Oxide Formula( 1 6 oxygens) Fomula(64 Oxygens) Si02 Si: 5.63 Si: 22.53 A1203 AI: 2.75 Al: 11.00 Fe0 Fe: 0.04 Fe: 0.17 Mn0 Mn: 0.02 Mn: 0.06 Ca0 Ca: 0.21 Ca: 0.83 Mg0 Mg: 0.04 Mg: 0.05 Na20 Na: 0.09 Na: 0.36 K20 K: 0.53 K: 2.11 total

Balance Error E% 010 means analysis is unreliable): 147.16

Analysis 6: GRID ACROSS MRLl XL #2 RASTER 200 SECS CHISQD= 2.00

Oxide wt.% Formula(l6 oxygens) Fonnula(64 Oxygens) Si02 54.22 Si: 5.73 Si: 22.92 Ti02 O. 13 Ti: 0.01 Ti: 0.04 A1203 19.37 Al: 2.41 Al: 9.65

Cr203 0.11 Cr: 0.01 Cr: 0.04 Ca0 2.94 Ca: 0.33 Ca: 1.33 Mg0 0.15 Mg: 0.02 Mg: 0.10 Na20 0.52 Na: 0.11 Na: 0.42 Ica LE! K: 0.97 K: 3.88 total 84.61

Balance Error E% (>IO means analysis is u~eliable): 34.94 Anaiysis 7: GRID ACROSS MRLl Xi, #3 RASTER 200 SECS CHISQD= 3.60

QZkk Formula(l6 oxygens) Formula(64 Oxygens) Si02 Si: 5.69 Si: 22.76 Ti02 Ti: 0.005 Ti: 0.02 A1203 Al: 2.40 AI: 9.62 Ca0 Ca: 0.45 Ca: 1.80 Mg0 Mg: 0.0 Mg: 0.0 Na20 Na: 0.13 Na: 0.52 K20 K: 1.O0 K: 3.97 totai

Balance Error E% (>IO means andysis is unreliable): 18.96

Analysis 8: MRLl GRID ACROSS XL #4 (RASTER) 200 SECS CHISQD= 1.9 1

QXkk Formula(l6 oxygens) Fonnula(64 Oxygens) Si02 Si: 5.80 Si: 23.22 Ti02 Ti: 0.007 Ti: 0.03 A1203 Ai: 2.38 Al: 9.54 Ca0 Ca: 0.44 Ca: 1.75 Mg0 Mg: 0.0 Mg: 0.0 Na20 Na: 0.10 Na: 0.41 Ka2 K: 0.63 : 2.52 total

Balance Error E% 010 means anaiysis is unreliable): 47.94

Analysis 9: MRLl GRID ACROSS XL #5 (RASTER) 200 SECS CHISQD= 2.34

Shi& Formula(l6 oxygens) Fomula(64 Oxygens) Si02 Si: 5.73 Si: 22.92 A1203 Al: 2.46 AI: 9.83 Fe0 Fe: 0.01 Fe: 0.04 Ca0 Ca: 0.54 Ca: 2.18 Mg0 Mg: 0.0 Mg: 0.0 Na20 Na: 0.08 Na: 0.33 K: 0.52 K: 2.08 total

Balance Error E% (> 10 means analysis is unreliable): 45.13 Analysis 10: MRLl GRID ACROSS XL #6 (RASTER) 200 SECS CHISQD= 4.60

Oxide Formula(l6 oxygens) Formula(64 Oxygens) Si02 Si: 5.77 Si: 23.08 Ti02 Ti: 0.01 Ti: 0.05 A1203 Al: 2.42 Al: 9.67 Ca0 Ca: 0.37 Ca: 1.48 Mg0 Mg: 0.01 Mg: 0.04 Na20 Na: 0.12 Na: 0.50 K20 K: 0.74 K: 2.97 total

Balance Error E% (> 10 means analysis is unreliable): 48.20

Analysis 11: MRLl GRID ACROSS XL #7 (RASTER) 200 SECS CHISQD= 9.20

Oxide wt.% Forrnula(16 oxygens) FormuIa(64 Oxygens) Si02 59.13 Si: 5.68 Si: 22.73 A1203 2 1.33 Al: 2.42 Al: 9.66 Ca0 3 -56 Ca: 0.37 Ca: 1.47 Mg0 0.12 Mg: 0.02 Mg: 0.07 Na20 0.87 Na: 0.16 Na: 0.65 K2C) 9.00 K: 1-10 K: 4.42 total 94.02

Balance Error E% (>IO means anaiysis is unreliable): 18.81

Analysis 14: MRLl #10 AT 2000X RASTER 200 SECS CHISQD= 2.23

Oxide Formula(l6 oxygens) Fomula(64 Oxygens) Si02 Si: 5.81 Si: 23.22 Al203 Ai: 2.35 AI: 9.41 Cr203 Cr: 0.02 Cr: 0.06 Ca0 Ca: 0.39 Ca: 1.57 Mg0 Mg: 0.0 Mg: 0.0 Na20 Na: 0.11 Na: 0.42 K20 K: O. 79 K: 3.17 total

Balance Error E% (>IO means analysis is unreliable): 39.57 Analysis 15: MRL 1 #11 AT 2000X RASTER 200 SECS CHISQD= 3 -25

Dxide Y!&% Formula(l6 oxygens) Formula(64 Oxygens) Si02 59.11 Si: 5.72 Si: 23.86 Ai203 21.68 Al: 2.47 Ai: 9.88 Ca0 4.60 Ca: 0.48 Ca: 1.91 Mg0 0.07 Mg: 0.01 Mg: 0.04 Na20 0.54 Na: 0.10 Na: 0.41 Ka2 5.37 K: 0.66 K: 2.65 total 91.37

Balance Error E% (> 10 means analysis is unreliable): 42.25

Analysis 16: MRL 1 # 12 AT 2OOOX RASTER 200 SECS CKtSQD= 3 -78

Oxide Formula(l6 oxygens) Formula(64 Oxygens) Si02 Si: 5.75 Si: 22.99 Ai203 Ai: 2.42 Al: 9.67 Fe0 Fe: 0.02 Fe: 0.08 Ca0 Ca: 0.45 Ca: 1.79 Mg0 Mg: 0.0 Mg: 0.0 Na20 Na: 0.13 Na: 0.54 K2C) K: 0.69 K: 2.78 total

Balance Error E% (>IO means analysis is unreliable): 40.04

Analysis 17: MRLl #13 AT 2000X RASTER 200 SECS CHISQD= 3.72 ski& Formula(l6 oxygens) FomuIa(64 Oxygens) Si02 Si: 5.65 Si: 22.61 Ti02 Ti: 0.005 Ti: 0.02 Ai203 Al: 2.44 Al: 9.78 Ca0 Ca: 0.46 Ca: 1.85 Mg0 Mg: 0.01 Mg: 0.0 Na20 Na: 0.15 Na: 0.59 KZ(r K: 0.95 K: 3.78 total

Baiance Error E% (>IO means analysis is unreliable): 19.39 Analysis 18: MRLl#14 AT 2000X RASTER 200 SECS CHISQD= 3.60

Oxide Formula(l6 oxygens) Formula(64 Oxygens) Si02 Si: 5.70 Si: 22.82 A1203 Al: 2.42 Al: 9.70 Mn0 Mn: 0.006 Mn: 0.02 Ca0 Ca: 0.40 Ca: 1.62 Mg0 Mg: 0.0 Mg: 0.0 Na20 Na: 0.18 Na: 0.70 Km K: 0.93 K: 3.72 total

Balance Error E% (> 10 means anaiysis is unreliable): 26.76

Analysis 19: MRL 1 # 15 AT 2000X RASTER 200 SECS CHISQD= 7.94 ski& Formula(l6 oxygens) Formula(64 Oxygens) Si02 57.83 Si: 5.63 Si: 22.52 A1203 2 1.46 Al: 2.46 Al: 9.85 Ca0 3.75 Ca: 0.39 Ca: 1.56 Mg0 0.08 Mg: 0.01 Mg: 0.05 Na20 O. 76 Na: 0.14 Na: 0.57 K20 %xi K: 1.16 K: 4.64 total 93.23

Balance Error E% (> 10 means analysis is unreliable): 16.62

Analysis 20: MRL1#16 AT 2000X RASTER 200 SECS CHISQD= 3.30

QXae EL'% Formula(l6 oxygens) Formula(64 Oxygens) Si02 55.8 1 Si: 5.63 Si: 22.52 A1203 20.87 Al: 2.48 AI: 9.92 Ca0 4.06 Ca: 0.44 Ca: 1.76 Mg0 0.07 Mg: 0.01 Mg: 0.04 Na20 O. 54 Na: 0.11 Na: 0.42 K20 8.LS K: 1.O5 K: 4.21 total 89.53

Balance Error E% (>IO means analysis is unreliable): 20.67 Analysis 2 1 : MRL 1 # 17 AT 2000X RASTER 200 SECS CHISQD= 3.47 ch.& FomuIa(l6 oxygens) Formula(64 Oxygens) Si02 Si: 5.77 Si: 23.09 A1203 Al: 2.35 Al: 9.42 Mn0 Mn: 0.02 Mn: 0.08 Ca0 Ca: 0.45 Ca: 1.81 Mg0 Mg: 0.01 Mg: 0.04 Na20 Na: 0.12 Na: 0.49 K20 K: 0.77 K: 3.07 total

Balance Error E% (X0means analysis is unreliable): 29.44

End of philiipsite analyses. Analysed a few other grains on matrix chip.

Andysis 24: MRL3 #4 RASTER (unknown minerai identified as analcime) 200 SECS CHISQD= 2.02

_Oxide wt.% Formula (6 oxygens) Si02 50.1 Si: 1.99 A1203 21.7 AI: 1.02 Ca0 0.71 Ca: 0.03 Mg0 0.0 Mg: 0.0 Na20 10.32 Na: 0.80 K20_ 2.24 K: 0.1 1 total 85.06

Balance Error E% (> 10 means analysis is unreliable): 4.90

Analysis 25: MRL3 #5 RASTER (unknown minera1 identified as analcime) 200 SECS CHISQD= 1.68

QZkk wt.% Formula (6 oxygens) Si02 53.44 Si: 2.03 A1203 22.54 Al: 1.01 Fe0 0.27 Fe: 0.008 Ca0 0.34 Ca: 0.01 Mg0 0.0 Mg: 0.0 Na20 10.3 1 Na: 0.76 _K2_0 1.09 K: 0.05 total 87.99

Balance Emor E% (> 10 means analysis is unreliable): 20.01

End of Anaiysis-