RHEOLOGY OF POROUS RHYOLITE by GENEVIEVE ROBERT B.Sc. (Honours), McGill University, 2005 A THESIS SUBMITThD IN PARTIAL FULLFILMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE in THE FACULTY OF GRADUATE STUDIES (Geological Sciences) THE UNIVERSITY OF BRITISH COLUMBIA (Vancouver) March 2008 © Genevieve Robert, 2008 ABSTRACT I describe an experimental apparatus used to perform deformation experiments relevant to volcanology. The apparatus supports low-load, high-temperature deformation experiments under dry and wet conditions on natural and synthetic samples. The experiments recover the transient rheology of complex (melt ± porosity ± solids) volcanic materials during uniaxial deformation. The key component to this apparatus is a steel cell designed for high-temperature deformation experiments under controlled water pressure. Experiments are run under constant displacement rates or constant loads; the range of accessible experimental conditions include: 25 - 1100 °C, load stresses 0 to 150 MPa, strain rates 106 to 102 i, and fluid pressures 0-150 MPa. I present a suite of high-temperature, uniaxial deformation experiments performed on 25 by 50 mm unjacketed cores of porous (-0.8) sintered rhyolitic ash. The experiments were performed at, both, atmospheric (dry) and elevated water pressure conditions (wet). Dry experiments were conducted mainly at 900 °C, but also included a suite of lower temperature experiments at 850, 800 and 750 °C. Wet experiments were performed at —650 °C under water pressures of 1, 2.5, 3, and 5 IVJPa, and at a fixed PH2O of 2.5 MPa for temperatures of -385, 450, and 550 °C. During deformation, strain is manifest by shortening of the cores, reduction of porosity, flattening of ash particles, and radial bulging of the cores. The continuous reduction of porosity leads to a dynamic transient strain-dependent rheology and requires strain to be partitioned between a volume (porosity loss) and a shear (radial bulging) component. The effect of increasing porosity is to expand the window for viscous deformation for dry melts by delaying the onset of brittle deformation by -50 °C (875 °C to 825 °C). The effect is more 11 pronounced in hydrous melts (--0.67 — 0.78 wt. % H20) where the viscous to brittle transition is depressed by --140 to 150 °C. Increasing water pressure also delays the onset of strain hardening due to compaction-driven porosity reduction. These rheological data are pertinent to volcanic processes where high-temperature porous magmas I liquids are encountered (e.g., magma flow in conduits, welding of pyroclastic materials). 111 TABLE OF CONTENTS ABSTRACT ii TABLE OF CONTENTS iv LIST OF TABLES vi LIST OF FIGURES vii PREFACE viii ACKNOWLEDGEMENTS x CO-AUTHORSHIP STATEMENT xii CHAPTER I: Introduction 1 1.1 Context 1 1.2 Previous studies 1 1.3 Goals and approach 5 1.4 References 7 CHAPTER II: The fluid cell 9 2.1 Introduction 9 2.2 Experimental apparatus 10 2.2.1 Fluid cell 11 2.2.2 Temperature calibration 13 2.3 Calibration for viscosity 14 2.4 Volcanological experiments 15 2.4.1 Materials 15 2.4.2 Experiments 19 2.4.3 Textural analysis of experiments 23 2.5 Discussion 24 2.6 Acknowledgements 27 2.7 Appendix 2.A: Melt viscosity of the Rattlesnake Tuff ash 28 2.8 References 30 CHAPTER III: Deformation experiments 34 3.1 Introduction 34 3.2 Experimental methods 37 3.2.1 Experimental apparatus 37 3.2.2 Fabrication of experimental cores 38 3.2.3 Pre-experimental sample characterization 41 3.3 Experimental results 44 3.3.1 Overview 44 3.3.2 Dry high-T experiments 44 3.3.3 Wet high-T experiments 46 3.3.4 Textural analysis of experimental cores 48 3.4 Post-experimental physical properties 50 3.4.1 Porosity 50 3.4.2 Water content 54 3.5 Analysis of experimental results 56 iv 3.5.1 Effect of temperature and PH2O 56 3.5.2 Analysis of strain 60 3.5.3 Effective viscosity 63 3.6 Discussion 67 3.7 Acknowledgements 75 3.8 Appendix 3.A: Correction for dwell-time effects 75 3.9 Appendix 3.B: Water contents of samples 78 3.10 References 82 CHAPTER IV: Discussion 86 4.1 Water 86 4.2 Experimental design modifications 90 4.3 Temperature gradient 91 4.4 Pore size distribution and pore shape 92 4.5 References 94 CHAPTER V: Summary 96 APPENDIX A: Cell design 98 APPENDIX B: Data acquisition 100 APPENDIX C: Experimental data 101 APPENDIX D: Data processing 103 v LIST OF TABLES Table 2.1 Summary of calibration and deformation experiments, including conditions, properties, and composition of samples 16 Table 2.2 Measured values of viscosity for glass from melted Rattlesnake Tuff ash and VFT coefficients (A, B, C) 25 Table 3.1 Chemical composition of the Rattlesnake Tuff ash 40 Table 3.2 Experimental conditions used in deformation experiments and geometry of sample cores pre- and post-experiment 42 Table 3.3 Measured values of density and porosity for pre- and post-experiment sample cores 43 Table 3.4 Analysis of strain 45 Table 3.B Values of H20 and LOT (wt.%) for post-experiment cores 80 vi LIST OF FIGURES Figure 2.1 Experimental apparatus 12 Figure 2.2 Temperature and viscosity calibration 18 Figure 2.3 Pre- and post-experimental products 20 Figure 2.4 Experimental results 22 Figure 3.1 Overview of previous experimental studies 35 Figure 3.2 Starting experimental materials 39 Figure 3.3 Summary of experimental data 47 Figure 3.4 Textural evolution of samples due to deformation 49 Figure 3.5 Nature and distribution of porosity in sample cores 52 Figure 3.6 Volume strain 55 Figure 3.7 Effect of temperature and PH2O 59 Figure 3.8 Analysis of strain 62 Figure 3.9 Summary of apparent viscosity 64 Figure 3.10 Textural comparison of samples run under dry and wet conditions 69 Figure 3.11 Glass transition and relaxation timescale 73 Figure 3.A Systematic corrections to experimental data 77 Figure 3.B Bulk water contents of experimental samples 79 Figure 4.1 Proportion of isolated porosity with deformation 87 Figure A.1 Water cell design 99 Figure C.1 Experiment RSO3 102 vii PREFACE This research comprises two complementary manuscripts prepared for publication in peer-reviewed international scientific journals. Chapter II is published in the American Mineralogist, under the title “High-temperature deformation of volcanic materials in the presence of water”. I am senior author, and my co-authors are 3. K. Russell, Daniele Giordano, and Claudia Romano. Cliff Shaw (University of New Brunswick) and Luigi Burlini (ETH Zurich) were journal reviewers. This chapter presents the design and calibration of a new apparatus to run uniaxial deformation experiments on volcanic materials under temperature and water pressure conditions relevant to volcanologic processes. The original design of the apparatus is by Daniele Giordano with technical advice from Oliver Spieler. Ray Rodway is responsible for machining the apparatus and helping with subsequent design changes. My experimental work was the basis for making design modifications to improve the performance of the apparatus. Chapter III has been submitted for publication under the title “Rheology of porous volcanic materials: High-temperature experimentation under controlled water pressure” in a special volume of Chemical Geology (8th Silicate Melt Workshop, Eds. D.B. Dingwell, R. Moretti, P. Richet), and is currently under review. I co-authored the manuscript with 3. K. Russell and Daniele Giordano. Chapter III presents a series of high-temperature deformation experiments run on porous aggregates of sintered volcanic ash under both wet and dry conditions. The experiments are organized to show the effects of (i) water pressure, and (ii) temperature on the rheological behaviour porous volcanic materials. viii Pre- and post-experimental physical properties of samples, including length, radius, mass, density, total, connected, and isolated porosity are reported in chapter III, as well as characteristic textures of the run-products, bulk water content and whole rock chemistry. Whole rock analyses of starting materials and run-products, including bulk water, were performed by ALS Chemex. Karl-Fischer Titration analyses of water content on the samples were performed by Daniele Giordano at ETH Zurich. Chapters IV and V provide a discussion of the entire research program, including a summary of the main results and the potential avenues for future work, respectively. The discussion also addresses issues that were not necessarily considered prior to or during experimentation. Four appendices are used to include detailed cell design, complete data sets, and data processing methods. ix ACKNOWLEDGEMENTS Financial support for my M.Sc. was provided by an NSERC PGS-M Scholarship. Costs to build the experimental apparatus were met by an NSERC RTI Grant “High temperature experiments on porosity and permeability evolution in volcanic systems” held by J.K. Russell, G.M. Dipple, and L.A. Kennedy. Operational costs for the research were covered by an NSERC Discovery Grant held by J.K. Russell. Nils, I wouldn’t have made it through any of this without you, and I definitely wouldn’t have had this much fun. I owe you a lifetime supply of Sortilège, and I shall deliver it myself, wherever in the world you may be. Stephen, thank you for your great listening skills, and thanks for all the fish. Krista, R-E, Curtis, Jackie, thanks for letting me shuffle your workspace whenever i so desired. You have all contributed to my success by keeping me in equilibrium. I am grateful I have such great people around me making this academic experience a great life experience too. I have made friends here I wish to keep for life. Kelly, I want to thank you for giving me so many great opportunities and for your guidance and unconditional support throughout this Masters, but especially for countless scientific discussions where I felt like a colleague rather than a student. Lori, it is always a pleasure to discuss ideas, problems and results from my experiments with someone who understands how many hours of work one little piece of experimental data actually represents. Mark, you brought thoughtful and unexpected arguments to our scientific discussions, making me think and investigate further. Thank you to Daniele, Ben and Steve for being such enthusiastic experimental lab colleagues, and to Ben especially for bringing a different perspective to my research, x always being available and interested, and for being such a great friend on top of it all. I am forever grateful to Ray Rodway for providing the technical support that made it all possible. En terminant, je tiens a remercier tout particulièrement ma famille. Papa, maman et Polo, tout au long de mes etudes, votre soutien et vos encouragements m’ auront permis de réaliser mes ambitions. Je vous aime. xi CO-AUTHORSHIP STATEMENT This thesis comprises two complementary manuscripts prepared for publication in peer-reviewed international scientific journals. Chapter II is published in the American Mineralogist, under the title “High-temperature deformation of volcanic materials in the presence of water”. I am senior author, and my co-authors are J. K. Russell, Daniele Giordano, and Claudia Romano. Chapter II presents the design and calibration of a new apparatus to run uniaxial deformation experiments on volcanic materials under temperature and water pressure conditions relevant to volcanologic processes. The experimental cell was conceptualized by my supervisor (J.K. Russell) and was originally designed by Daniele Giordano with technical advice from Oliver Spieler. Ray Rodway is responsible for machining the apparatus and helping with subsequent design changes. My experimental work was the basis for making design modifications to improve the performance of the apparatus. I performed all calibration experiments, all experiments on natural materials, and all experimental data reduction and analysis. Chapter III has been submitted for publication under the title “Rheology of porous volcanic materials: High-temperature experimentation under controlled water pressure” in a special volume of Chemical Geology (8th Silicate Melt Workshop, Eds. D.B. Dingwell, R. Moretti, P. Richet), and is currently under review. I am senior author, and my co-authors are J. K. Russell and Daniele Giordano. Chapter III presents a series of high-temperature deformation experiments that I performed on porous aggregates of sintered volcanic ash under both wet and dry conditions. In chapter III, I report measurements I made of pre- and post-experimental physical properties of samples, including length, radius, mass, density, total, connected, xii and isolated porosity, as well as characteristic textures of the run-products, bulk water content and whole rock chemistry. The whole rock analyses of starting materials and run-products, including bulk water, were performed by ALS Chemex. Karl-Fischer Titration analyses of water content on the samples were performed by Daniele Giordano at ETH Zurich. I am responsible for data reduction and analysis of the physical and chemical properties of the experimental samples, as well as for the reduction and analysis of the rheological data obtained from the deformation experiments. xiii CHAPTER I: Introduction 1.1 Context Experimental volcanology is an expanding field of research driven by new methods for exploring volcanic processes through high-temperature experimentation (Dingwell, 1998; Gardner, 1999; Tinker et al., 2004; Quane et al., 2004; Grunder et al., 2005). Dynamic deformation of complex volcanic materials (melt ± crystals ± pores) in the laboratory is of great interest because of the direct applications to the flow of volcanic materials, notably in volcanic conduits, during lava transport, and during welding of pyroclastic volcanics. Porous and hydrous volcanic materials are of special interest because of the ubiquity of water, and consequently bubbles, in volcanic systems. There are several sets of high-temperature experiments on natural volcanic materials that have been performed under dry conditions (e.g., Yagi, 1966; Bierwirth, 1982; Bagdassarov and Dingwell, 1992; Quane, 2004). However, performing similar experiments under controlled water pressures is inhibited by the technical difficulties involved (Friedman et al., 1963; Grunder et al., 2005). Thus, establishing the rheology of hydrous volcanic materials remains one of the principal challenges in volcanology (Grunder and Russell, 2005). 1.2 Previous studies Bierwirth (1982) studied the compaction and welding of the rhyolitic Bandelier Tuff ash, New Mexico and dacitic air fall deposit from Mount St. Helens, Washington, under dry conditions and temperatures between 650 °C and 800 °C. His experiments were conducted on jacketed samples of loose ash, at constant load pressures between 0.72 1 and 3.62 MPa. Higher loads and higher temperatures resulted in more compaction (greater porosity loss). Bierwirth developed an equation to describe the compaction of Bandelier Tuff ash. The equation expresses strain, decomposed into time and strain rate, as a function of density and material properties, which depend on a temperature- dependent activation energy. Bagdassarov and Dingwell (1992) performed uniaxial deformation experiments on core samples of vesiculated natural obsidian from Little Glass Butte, Oregon. They used constant stress (510 to i0 Pa) uniaxial deformation experiments (E = 0.01-0.015) to determine the viscosity of the samples with low (-0-0.5), moderate (—P0.25-0.35), and high (-4165) pore fractions at temperatures near the glass transition (—850 °C) of the melt. They observed a decrease of apparent viscosity with increasing pore fraction and fit their experimental data to a viscosity () vs. porosity (1) relationship of the form: (1.1) using a dimensionless constant C of 22.4. Lejeune et al. (1999) performed uniaxial deformation experiments on calcium aluminium synthetic silicate melt samples. They vesiculated the synthesized melts in air to obtain low (t = 0-0.13) to moderately (1 = 0.32-0.47) porous samples. Deformation experiments were conducted at temperatures ranging from 830 to 960 °C, and at a constant stress varying from 1.1 to 67.7 MPa. The experiments of Lejeune et a!. clearly show that the apparent viscosity of porous melt decreases with increasing porosity. The measured decrease in viscosity due to the addition of 47% porosity in their experiments corresponds to a viscosity change caused by an increase in temperature of 10 °C. 2 Quane (2004) used both soda lime silica glass beads as an analogue for silicate melt and natural ash from the Rattlesnake Tuff, Oregon, to investigate the rheology of porous volcanic materials via a series of dry, high-temperature experiments conducted at constant displacement rate or constant load. Results from the experiments are also reported in Quane and Russell (2005) and Quane and Russell (2006). Cores of sintered beads or Rattlesnake Tuff ash were fabricated to produce large cores. The physical properties of each core were fully characterized before and after each experiment. The displacement rates used in the glass bead core deformation experiments ranged between 2.510 and ii03 cmls, and the loads between 5 and 50 kg for temperatures of 535, 550, 600, and 650 °C and starting porosities between —27 and -37%. The displacement rates used in the Rattlesnake Tuff ash experiments ranged from 1 .25 i04 to 5.0 i04 cmls, and the loads ranged from 22.5 to 90 kg for temperatures ranging from 800 to 900 °C and starting porosities ranging from -.-70 to 80%. The rheology of both materials was found to be strain dependent, and the changes in temperature to have a much greater effect on rheology than changes in load or displacement rate. In the glass bead experiments, strain accumulates dominantly by porosity loss at low amounts of total strain. At higher strain, radial strain becomes more important. In contrast, deformation of natural ash cores shows radial bulging to be dominant at lower amounts of strain with porosity loss becoming more important at higher amounts of total strain. Strain accommodated by porosity loss is described by the following relationship: (1.2) 1-f where is initial total porosity and I is final total porosity. The results of Quane and Russell (2005) suggest that, in analogue and especially in high-porosity natural materials, 3 porosity distributions control the mechanisms and extent of welding. They developed a constitutive relationship relating porosity to melt viscosity at constant temperature by an empirical factor (a), reflecting the starting porosity of the material, the geometry and character of individual glass clasts and the ability of individual clasts to rearrange or rotate during deformation to describe the rheological behaviour of the Rattlesnake Tuff ash: flehloe (1.3) where 1e is the sample viscosity, rio the melt viscosity and the sample porosity. Experiments investigating the rheology of porous hydrous volcanic materials in which water pressure is controlled independently of the load applied to the sample or the rate of deformation are few. Friedman et al. (1963) published the only set of deformation experiments performed on natural volcanic ash in which water pressure was independently controlled. They investigated the viscosity of crushed porous rhyolite glass at temperatures between 400°C and 850 °C and at water pressures between 0 and 6.89 MPa. Most experiments were performed at temperatures above 485 °C and water pressures below 2.07 MPa, conditions at which welded ignimbrites can form. The deformation experiments of Friedman et al. were performed on jacketed samples of loose ash from the Bandelier rhyolite tuff, New Mexico. They estimated the initial porosity of the samples to be —50% on the basis of geometry, and they controlled temperature, load, and water pressure for the duration of each experiment. The sample was brought to temperature by a resistance furnace that surrounded the lower part of the experimental assembly. Fluid pressure was controlled with a hand pump, and load was applied to the sample by placing weights on a lever that was connected to the piston used to deform the 4 sample. They recorded the compaction rates of the ash and compared the results to compaction rate curves for Pyrex glass under dry conditions to obtain viscosity values for the Bandelier Tuff ash. For a given experimental temperature, Friedman et al. (1963) report an increase in viscosity with increasing strain or with reduction of porosity, and compactions rates were observed to be faster at higher water pressures. It should be noted that the figures in Friedman et al. (1963) are mislabelled to indicate incorrectly that compaction rate decreases with increasing water pressure (cf. Sparks et al., 1999). The experiments reported in this thesis are the only other wet experiments that investigate the rheology of porous volcanic materials. 1.3 Goals and approach The objectives for this project were: (i) to build a deformation apparatus capable of holding water pressures relevant to volcanic processes at high-temperature, (ii) to calibrate the apparatus for viscosity and temperature, (iii) to run wet and dry experiments on porous volcanic materials, and (iv) to use the resultant data to understand porosity in collapsing volcanic materials. Specifically, constant displacement rate, parallel-plate deformation experiments (Gent, 1960) were performed on the porous cores of ash from the Rattlesnake Tuff, at high temperatures and at controlled water pressures, in a new apparatus designed for high-temperature, uniaxial deformation experiments in the presence of water. Strain in the experiments is expressed by a shortening and radial increase of the sample, and a reduction in porosity from the pre-experimental values. These goals are organized as two manuscripts. The first two objectives were met and are published as an article in American Mineralogist (Robert et al., 2008). The latter 5 two parts of the project are in a manuscript submitted to a special volume of Chemical Geology (Robert et al., In Review). Because of the chosen thesis format, the appendices to this thesis are critical and more of the technical background, methods used and raw data from the experiments are presented there. The final design of the cell is presented in the first appendix. The second appendix explains the data acquisition process, and contains all the experimental data files. The raw experimental data is compiled in electronic format in the third appendix. The MATLAB code used to process the experimental data is in the final appendix. Some repetition in the introductory sections of Chapters II and III is unavoidable as each chapter is a separate manuscript for different publications. 6 1.4 References Bagdassarov, N.Sh., Dingwell, D.B., 1992. A rheological investigation of vesicular rhyolite. Journal of Volcanology and Geothermal Research 50, 307-322. Bierwirth, P.N., 1982. Experimental welding of volcanic ash. B.Sc. Honours Thesis, Monash University, 74.p. Dingwell, D.B., 1998. Recent experimental progress in the physical description of silicic magma relevant to explosive volcanism. In: Gilbert, J.S. and Sparks, R.S.J. (eds.) The Physics of Explosive Volcanic Eruptions, Geological Society, London, Special Publications 145, 9-26. Friedman, I., Long, W., Smith, R.L., 1963. Viscosity and water content of rhyolite glass. Journal of Geophysical Research 68, 6523-6535. Gardner, J.E., Hilton, M., Carroll, M.R., 1999. Experimental constraints on degassing magma; isothermal bubble growth during continuous decompression from high pressure. Earth and Planetary Science Letters 168, 201-218. Grunder, A., Russell, J.K., 2005. Welding processes in volcanology: insights from field, experimental, and modeling studies. Journal of Volcanology and Geothermal Research 142, 1-9. Grunder, A.L., Laporte, D., Druitt, T. H., 2005. Experimental and textural investigation of welding: effects of compaction, sintering, and vapor-phase crystallization in the rhyolitic Rattlesnake Tuff. Journal of Volcanology and Geothermal Research 142, 89-104. Lejeune, A.M., Bottinga, Y., Trull, T.W., Richet, P., 1999. Rheology of bubble-bearing magmas. Earth and Planetary Science Letters 166, 7 1-84. 7 Quane, S.L., 2004. Welding in pyroclastic materials, PhD Thesis, University of British Columbia, 2O8p. Quane, S.L., Russell, J.K., Kennedy, L.A., 2004. A low-load, high-temperature deformation apparatus for volcanological studies. American Mineralogist 89, 873- 877. Quane, S.L., Russell, J.K., 2005. Welding: insights from high-temperature analogue experiments. Journal of Volcanology and Geothermal Research 142, 67-87. Quane, S.L., Russell, J.K., 2006. Bulk and particle strain analysis in high-temperature deformation experiments. Journal of Volcanology and Geothermal Research 154, 63-73. Robert, G., Russell, J.K., Giordano, D., Romano, C., 2008. High-temperature deformation of volcanic materials in the presence of water. American Mineralogist 93, 74-80. Robert, G., Russell, J.K., Giordano, D., In Review. Rheology of porous volcanic materials: High-temperature experimentation under controlled water pressure. Chemical Geology Special Issue, 8t Silicate Melt Workshop. Sparks, R.S.J., Tait, S.R., Yanev, Y., 1999. Dense welding caused by volatile resorption. Journal of the Geological Society, London 156, 2 17-225. Tinker, D., Lesher, C.E., Baxter, G. M., Uchida, T., Wang, Y., 2004. High-pressure viscometry of polymerized silicate melts and limitations of the Eyring equation. American Mineralogist 89, 1701-1708. Yagi, K., 1966. Experimental study on pumice and obsidian. Bulletin of Volcanology 29, 559-572. 8 CHAPTER II: The fluid cell’ 2.1 Introduction Experimental volcanology is an expanding field driven by new methods for exploring volcanic processes through high-temperature experimentation (Dingwell, 1998; Gardner, 1999; Tinker et al., 2004; Quane et al., 2004; Grunder et al., 2005). High- temperature experiments are used to retrieve data on the rheological behaviour of natural melts (e.g., Dingwell et al., 1993; Dingwell, 1998; Richet and Bottinga, 1995; Giordano et al., 2004), the properties of pyroclastic materials (e.g., Friedman et al., 1963; Bierwirth, 1982; Quane et al., 2004; 2005; Giordano et al., 2005), the conditions attending explosive collapse of lava and domes (e.g., Spieler et al., 2004), and the mechanisms of fragmentation processes in volcanic conduits (Tuffen et al., 2003; Kennedy et al., 2005). The explosive or effusive behaviour of volcanic systems is governed by magma rheology, which largely reflects the abundance and nature (e.g., dissolved vs. exsolved) of volatile components. However, the rheological properties of volcanic materials in the presence of a fluid phase as ubiquitous as water in volcanic systems remain poorly known (Bagdassarov and Dingwell, 1992; Lejeune et al., 1999; Stein and Spera, 1992). This gap in knowledge results from the technical difficulties in designing and running the appropriate experiments. Establishing the rheology of hydrous volcanic materials, therefore, remains one of the principal challenges in volcanology (Grunder and Russell, 2005). ‘A version of this chapter has been published. Robert, G., Russell, J.K., Giordano, D., Romano, C., 2008. High-temperature deformation of volcanic materials in the presence of water. American Mineralogist 93, 74-80. 9 The purpose of this paper is two-fold. First, we describe a new experimental cell for high-temperature deformation experiments of samples under controlled fluid pressures. The “fluid cell” can be used with the Volcanology-Deformation-Rig (VDR; Quane et al., 2004) for rheological studies of volcanic materials (e.g., pumice, ash, lava) over T-PH2O conditions pertinent to volcanological processes. Second, we report on a series of experiments used to: a) calibrate the apparatus; and to b) explore the properties (e.g., viscosity) of natural pyroclastic materials at volcanic T-PH2O conditions. These data are critical for the understanding of a variety of volcanic processes such as: welding and compaction of ignimbrites; fragmentation and annealing of magma in volcanic conduits; flow of volcanic domes; and amalgamation and flow of clastogenic lavas. 2.2 Experimental apparatus The VDR (Fig. 2. la) was designed to explore the rheology of volcanic materials by performing high-temperature, low-load (<1136 kg) deformation experiments at constant load, or displacement rate, or at controlled load rates (Quane et al., 2004). The apparatus comprises a GeoComp LoadTracll reinforced “T”-frame equipped with a step- motor that moves a lower platen upwards at specified rates, or applies a prescribed load. An S-beam type load transducer measures load; displacement is measured by a linear variable displacement transducer (LVD transducer). A commercially purchased Zircar® fiber-insulated heater furnace with helically-wound Fe-Cr-Al alloy resistance wire elements allows for temperatures up to 1100 °C. The main attributes of the VDR are that it accommodates large (D < 7.5 cm; L < 10 cm) sample cores and covers temperatures, 10 load stresses and strain rates consistent with natural volcanic processes (see Fig. 3 in Quane et al., 2004). The original VDR was restricted to high-temperature experiments at ambient atmospheric conditions. We have now built a steel, sealable cell (Fig. 2. lb) that allows high-temperature experimentation at elevated fluid pressures. The fluid cell can be used in the VDR after only minor modifications of the original assembly. The current sample assembly comprises, from bottom to top, a lower cooling plate, a stainless steel holder for a ceramic spacer, the sample cell, and a stainless steel spacer (Fig. 2.1 a). The upper steel spacer has a machined lip that aligns the water cell to the central axis of the rig. 2.2.1 Fluid cell All parts of the fluid cell are machined out of a corrosion-resistant, high- temperature stainless steel (grade 310) suited for experimentation involving fluids. The cell can operate at temperatures of 25-1100°C and fluid pressures of 0-150 MPa. The sample chamber is a 25 cm tall cylinder with a wall thickness of 1 cm and outside diameter of 5 cm. It can accommodate sample cores up to 3 cm in diameter and 10 cm in length. An internal piston is connected to the VDR by a 31 cm long piston shaft that slides out of a sealed opening at the top of the fluid cell (Fig. 2.1). The VDR controls displacement of the piston and, thus, deformation of sample. The cell is sealed metal-on-metal at either end. The lower and upper metal seals are fastened to the sample chamber by socket head cap screws. The top seal has a long, narrow neck used to align the piston shaft. A high-temperature Viton® 0-ring, cooled by a water-cooling jacket, provides a tight seal on the piston shaft. Dow Corning® high 11 piston 4—thermocouple o-ring—— :1 cIamp— sample Figure 2.1 (A) Volcanology-Deformation-Rig (VDR) modified from Quane et al. (2004) for experiments using the fluid cell. (B) Cross-section of fluid cell and sample arrange ment. Detailed line diagrams and parts list for the water cell can be found at http://www.eos .ubc.calresearchlinfrastructure/cesl.html 0 thermocouple valve and / transducer I complex —to H20 system load transducer fiber insulation — ceramic __1 spacer cooling plate to temperature controller displacement to computer valve and transducer complex in valve sample chamber _— 12 temperature lubricant is applied to the 0-ring to minimize friction on the piston shaft during experiments. Normally, the piston-shaft slides into the cell under its own weight and, for dry experiments, all applied load is used to deform the sample. The upper part of the piston shaft is threaded so that the top of the piston can be removed without taking apart the top of the cell. The outer diameter of the lower piston head is slightly less (<1 mm) than the inner diameter of the cell which ensures that: (i) there are no frictional effects between piston head and cell wall during an experiment, and that (ii) the piston head does not create an impermeable barrier to the fluid phase; fluid pressure is hydrostatic and equal on either side of the piston. The piston shaft is hollow and allows a thermocouple to be placed at the lower piston head (e.g., 2 mm above top of sample). The valve and transducer complex comprises a safety valve, a water pressure transducer, an air valve, and a line for introducing distilled water via a 2-way manual fluid pressure intensifier. The intensifier serves as a fluid delivery system that allows the operator to control fluid pressure during the experiment by adding or removing fluid (Fig. 2.1). The system can be used to compensate for slow leaks or to allow for experiments having a variable or cyclical fluid pressure (e.g., degassing or fluid pressurization events). 2.2.2 Temperature calibration A factory-built fiber insulated tube furnace is used to heat the sample assembly and fluid cell. The lower cooling plate and ceramic spacer have a hole drilled in their centers to accept a type K thermocouple (Fig. 2.1) which controls temperature at the base of the cell. The thermocouple inside the piston reads temperature at the top of the sample 13 and helps monitor vertical temperature gradients within the sample. The top and bottom of the furnace are stuffed and wrapped in fiber insulation to minimize temperature gradients. Vertical temperature profiles in the sample were measured experimentally to find a sample position that minimizes thermal gradients. The experiments used standard cores (2.54 cm x 5 cm) of dacite lava that had vertical (0.5 cm in diameter) holes drilled down their centre. A special piston that allows the thermocouple to slide down the shaft, out the piston head, and into the sample core was used for measuring the temperature profiles. Steady-state temperatures were achieved (1 hour dwell time) and temperatures were measured at 12.5 mm increments from the base of the sample to the piston/sample interface. On the basis of these experiments, the minimum temperature gradient is achieved by having the bottom of the sample positioned 65 mm above the base height of the tube furnace (Fig. 2.2a). The maximum gradients are 4 °C over 4 cm and 8.5 °C over 5 cm with no signs of strain localization related to temperature gradients in our experiments to date. 2.3 Calibration for viscosity Calibration experiments were performed on solid glass cores (10 mm x 25 mm) of NIST (NBS) standard reference material (SRM) 717a (borosilicate glass) under constant load and dry conditions at temperatures (550-600 °C; see Table 2.1). The temperature gradient along the length of these cores is °C. The shear viscosity of the cores is computed for a given applied load (F; N), sample volume (V; m3), sample length at time t 14 (L; m), and rate of shortening (ãLIãt; m s’) using the no-slip (Eq. 2.1) and perfect-slip (Eq. 2.2) models of Gent (1960) (cf. Dingwell et al., 1993): 2irL5F c9L (2.1)3V—(2rL + V) and ?‘,(Pa s) = (2.2) respectively. Based on the geometry of the run-product cores (i.e. little bulging) we chose the perfect-slip end-member model to compare viscosity values from the deformation experiments to the temperature-dependent viscosity curve for NIST 7 17a glass (Fig. 2.2b). The shaded field on the curve indicates the is uncertainty on the standard glass. The uncertainty on each experimental determination of viscosity (boxes) includes variations in temperature during the experiment (Table 2.1). Our calibration experiments reproduce the viscosity of the standard well and suggest an experimental accuracy of 0.2 log units. 2.4 Volcanological experiments 2.4.1 Materials High-temperature deformation experiments were performed on fabricated cores of ash from the Rattlesnake Tuff: a high silica rhyolite (SiO2 >75%; Table 2.1; cf. Streck and Grunder, 1995). The ash is sieved to a 0.6-2 mm size fraction (coarse ash) and cores are sintered by heating loose ash in a mold (2.54 cm x 8 cm) at 900 °C for 20 minutes. Samples are trimmed to —5 cm lengths creating cores with a 2:1 aspect ratio (Fig. 2.3a). There is little change in composition after sintering (Table 2.1). 15 * Co m po sit io n o fR at tle sn ak e as h co re s as S i0 2, T i0 2, A l 203, Fe O (T ), M nO ,M gO ,C aO , (i) Po st -s in te rin g: 77 .6 4; 0. 17 ; 12 .4 8; 1. 17 ; 0 .0 7; 0. 00 ;0 .3 1; 3. 38 ;4 .6 2; 0. 01 ; 0 .1 5. (ii )P os t-e xp er im en ta l: 77 .1 7; 0. 16 ; 12 .8 1; 1. 16 ;0 .0 7; 0. 00 ; 0 .3 0; 3. 44 ; 4 .6 5; 0. 00 ; 0 .2 4. aV a l ue s o fs tr ai n ca lc ul at ed fro m m ac hi ne di sp la ce m en t( Am ), sh or te ni ng o f co re (A l). bV al ue s o fo rig in al an d fin al to ta lp or os ity . eV al ue s o fs he ar v isc os ity as st re ss o v er st ra in ra te (1 )a n d fro m pe rfe ct sli p m o de l( ri 1 3 ) . T ab le 2. 1 Su m m ar y o fc al ib ra tio n an d de fo rm at io n ex pe rim en ts, in cl ud in g co n di tio ns ,p ro pe rti es , a n d co m po sit io n* o f sa m pl es . N o. Se t-U p T P (H 20 ) Lo ad Al /A t St ra in a Po ro sit yb V isc os ity (P a) C (°C ) (M Pa ) (N ) (rn s’ ) m 1p s rtf O2 Ce ll 87 8± 1 D ry — 1. 25 .1 06 0. 30 0. 30 0. 73 0. 64 10 6. 61 0b 03 — rtf o4 Ce ll 65 6± 10 3. 3- 1. 7 — 1. 25 10 6 0. 30 0. 34 0. 72 0. 63 10 8. 21 0b 01 — rtf O5 Ce ll 64 5± 5 3 — 2. 50 .1 06 0. 30 0. 35 0. 73 0. 69 i 0 7° - i 0 9 4 — n ist O l V D R 56 2± 13 D ry 48 .4 — 0. 16 0. 15 — — — 10 10 1 10 b0 2 n ist 03 V D R 57 1± 11 D ry 48 .4 — 0. 25 0. 24 — — — 10 98 10 99 n ist 04 Ce ll 57 5± 2 D ry 48 .4 — 0. 30 0. 30 — — — 10 96 10 97 n ist 05 Ce ll 56 8± 1 D ry 48 .4 — 0. 24 0. 23 — — — 10 9. 81 0b 00 N a 20 , K20 , P205, LO T: Figure 2.2 Calibration results for VDR and fluid cell. (A) Calibration of thermal gradient across 4 cm (grey) and 5 cm (hatch) sample cores. (B) Results of experiments plotted against known viscosity (upper inset) of NIST glass cores (lower inset). Main figure shows expected values of viscosity for NIST glass over experimental range of temperatures (shaded grey). Viscosity values derived from: (i) dry experiments performed in the VDR (open rectangles), and (ii) experiments performed in the fluid cell under dry conditions (closed rectangles). See Table 2.1 for experimental conditions and results. 17 54 C-) 0 . 52 0 0 1 0 560 600 T(°C) 620 600 580 560 540 Cl) cr3 Q10 0 0) 0 12.5 1 0000/T(K) Figure 2.2 See previous page for figure caption. 570 580 590 T(°C) 11 9 11.5 12 18 The sintering process causes point annealing of shards and forms a highly-porous, floating, shard-supported framework (Fig. 2.3b). Cores of ash comprise curvilinear and Y-shaped bubble wall shards, complete vesicles (e.g., bubble shards), smaller proportions of pumiceous shards, and up to 1% crystals. There are two types of bubble shards (Fig. 2.3b): (i) a population of thick-walled vesicles characteristic of the original ash, and (ii) a subordinate population of thinner-walled vesicles produced during the sintering process. The latter population may have resulted from nucleation and growth of new bubbles or, more likely, represent original closed bubbles (isolated porosity) that expanded during the heating and sintering of the cores. Total porosity of sintered cores is slightly in excess of 70% (Table 2.1). The cores have an essentially isotropic texture (e.g., little to no fabric). The shards do not appear deformed except around bubbles that grew during sintering, where the shards appear to conform to the shape of the thin-walled bubbles. 2.4.2 Experiments Three unjacketed experiments were run in the fluid cell system under constant displacement rate (_.106 m s’) and to strains of -3O% (Table 2.1). The dry experiment (rtf2) was at 878 °C; two experiments under -3 MPa PH2O (rtf4 and rtf5) were performed at 656 °C and 645 °C, respectively. The corresponding experimental run-products are shown in Figure 2.3c, d. Figure 2.4a shows the relationship between applied load stress and total strain for experiments rtf2 and rtf5. The data have been filtered to compensate for the fact that the high sampling rate captures the oscillations of the step motor that drives piston displacement. Smoothing the data before processing gives a more accurate record of the changing properties (e.g., rheology) of the system with increasing strain. 19 Figure 2.3 Pre- and post-experimental products (Table 2.1). (A) Sample core of sintered ash used in deformation experiments. (B) Scanning electron micrograph of thin-section of sintered core of Rattlesnake Tuff ash (e.g., starting material). (C) SEM backscattered electron micrographs of thin section of (C) run-product rtf2 and (D) run product rtf4 (load stress had a vertical orientation in these images). Ash particles are light grey and pore space is dark grey to black. White boxes highlight two populations of bubbles: thick-walled and thin-walled (see text). 20 For the dry experiment, increased load is required to maintain a constant rate of displacement. In order to achieve 30 % shortening of the core, load stress increases from 0.1 to 0.6 MPa. This increase in load stress, at constant displacement rate, is a direct indication of the transient properties of the core during progressive deformation. Increasing strain causes shortening of the core by porosity reduction (volume strain), which has the concomitant effect of increasing the effective strength (e.g., viscosity) of the material. This trend is as described by Quane et al. (2004) and Quane and Russell (2005) in their deformation experiments on cores of glass beads. We have also calculated the apparent viscosity of the core sample as a function of strain (Fig. 2.4b). Once steady deformation is achieved, viscosity rises from Pa s during the first 1-2% of strain, to i’° Pa s at 30% strain. The increase in viscosity during deformation is broadly consistent with constitutive relationships established for the viscosity of hot porous aggregates (e.g., Sura and Panda, 1990; Bagdassarov and Dingwell, 1992; Quane and Russell, 2005). The deformation path of the dry core shows a steady increase in viscosity of -1 order of magnitude for a porosity reduction of 10%. Experiments run at fixed PH2O (3-3.2 MPa) use similar displacement rates, achieve similar values of strain, but are performed at -200 °C lower temperatures. At the same temperatures and strain rates, under dry conditions, experiments on Rattlesnake ash cores produce brittle (rather than viscous) deformation. The load in the wet experiments is corrected for the effects of FH2O by subtracting a constant value of load from the data set; that value is recorded at time zero before any deformation has occurred. At PH2O < 3 IVIPa piston friction is negligible. At higher water pressures a correction is needed; future 21 0. 11 0.1 0.2 0.3 Figure 2.4 Results of two experiments on cores of Rattlesnake Tuff ash: (i) rtf2 (dry at 878 °C), and (ii) rtf5 (wet at 645 °C). (A) Stress evolution vs. strain for two experiments and corresponding water pressure for rtf5 (see text). (B) Calculated effective viscosity vs. total strain for data in (A). (C) Results compared to independent values of melt viscosity (solid line) and Tg for melt (e.g., i 1012 Pa s). Effect of porosity is to reduce viscosity of the melt. Open symbols show results of two experiments (see B); arrows indicate direction of increased strain. (D) Melt viscosity (solid line) as shown in (C). Dashed lines are calculated effects of dissolved water (0.1, 0.25, 0.5, and 1.0 wt.% H20; Giordano et al. 2008) and grey squares show depression of Tg for hydrous melts. Experi mental results from two deformation experiments are plotted as in (C) (see text). The star represents the viscosity of the Rattlesnake melt at 645 °C having 0.73 wt.% water (see text). 0.5 0.4 0.3 Cl) ci) 0.2 0.1 /1/ P(H2O)forrtf5/ — (dry) -D F’)0 2D 10 Cl) C9 0 0)8 C 7 0 rtf2 0.1 Strain 0.2 Strain V) CD C 0 6 8 10 12 1 0000/T(K) 6 8 1 0000/T(K) 10 12 22 work includes calibrating this effect to allow for experimentation at higher >50 MPa PH2O. Experiment rtf5 shows substantially different behaviour than seen in the dry experiment (Fig. 2.4a). The load stress curve for rtf5 is essentially flat with increasing strain, implying that a single critical load is required to maintain a constant displacement rate throughout the entire experiment (Fig. 2.4a). The load stress shows a maximum of 120 kPa at small amounts of strain (<10%), decreases to 80 kPa after 15-20% strain and then increases slightly to 120 kPa at 30% strain. The calculated effective viscosity is also nearly constant (109.2 to i0 Pa s) despite the core undergoing 30% strain via porosity loss (Table 2.1). The presence of a fluid pressure not only lowers the material strength (< 100 kPa vs. 100 - 600 kPa), but also compensates for the expected increase in viscosity due to lower temperature (645 °C vs. 878 °C) and delays the onset of “strain hardening” of the sample as porosity is reduced. 2.4.3 Textural analysis of experiments Samples rtf2 (dry; Fig. 2.3c) and rtf4 (wet; Fig. 2.3d) have undergone similar strain (-30%) and have lost identical amounts of porosity (Table 2.1). Porosity in deformed samples occurs as intraclast voids between annealed shards, as bubble voids (thick- and thin-walled) and as smaller (<0.01 mm) isolated pores in vitric clasts (Fig. 2.3c, 2.3d). After 30% strain, the samples develop a pronounced planar fabric caused by rotation and flattening of shards to create a foliation. The intensity of the foliation is virtually identical in the two experiments (cf. Fig. 2.3b vs. 2.3c and 2.3d). Thick- and thin-walled bubble shards Fig. 2.3b) exhibit quite different behaviours during 23 deformation. The thinner-walled bubbles show much higher degrees of flattening than do the thick-walled bubbles; this disparate behaviour is independent of bubble size although, in general, smaller bubbles are less deformed. Another form of strain localization occurs in curvilinear and Y-shaped shards that are near flattened bubbles. These shards show a stronger alignment and higher degree of deformation than do shards away from large deformed bubbles. In summary, the SEM images of run-products from the dry and wet experiments provide no obvious means to differentiate between them. It is somewhat enigmatic that very similar run-products were produced even though: (i) the dry experiment was performed at -.2OO °C higher temperature; under dry conditions the 645 °C experiment would not support viscous deformation, and (ii) the dry experiment showed a continuous increase in load stress and viscosity as a function of progressive strain, whereas the wet experiment underwent the same amount of strain and porosity reduction but showed little to no strain hardening or increase in effective viscosity. 2.5 Discussion Our experiments address the viscosity of highly vesicular (70%) melts (Fig. 2.4c, 2.4d). The highest viscosity achieved in the dry (rtf2) experiment after 30% strain is i00’4 Pa s, which is close to the viscosity of the Rattlesnake melt at this temperature (1O08 Pa s; Fig. 2.4c; Table 2.2 and see Appendix A). The results of the deformation experiment performed at -3 MPa PH2O and 645 °C (rtf5; Fig. 2.4c) indicate apparent viscosities of 109.2 - i0 Pa s. The viscosity of the Rattlesnake Tuff melt (anhydrous) 24 Table 2.2 Measured values of viscosity for glass* from melted ash from Rattlesnake Tuff ash and VFT coefficients (A, B, C). T(°C) Log Ti Expta 917.80 9.92 MP 975.25 9.14 MP 1421.67 4.60 CC 1446.28 4.43 CC 1470.89 4.26 CC 1495.50 4.09 CC 1520.11 3.93 CC 1544.72 3.78 CC 1569.33 3.62 CC 1593.94 3.48 CC 1618.55 3.34 CC A B C VFT -7.43 19,766 52.9 * Composition of glass by EMP as Si02 (77.42), Ti02 (0.14), Al203 (12.22), FeO(T) (1.39), MnO (0.08), MgO (0.04), CaO (0.31), Na20 (3.44), K20 (4.96), P205 (0.01). a See Appendix 2.A. 25 extrapolated to this temperature would be iO’54 Pa s (Table 2.2; Appendix 2.A), which is substantially higher than observed. Although the wet experiment experiences the same strain as the dry experiment, its effective viscosity remains very much lower than the viscosity of the corresponding dry melt. Both porosity and dissolved water serve to reduce effective viscosity. Dissolved water causes a strong decrease in viscosity and is most pronounced in melts having the highest values of viscosity (e.g., low temperature). There are several ways in which fluid pressure might operate to reduce the effective viscosity of these samples during deformation. Firstly, elevated fluid pressure will cause hydration of the shards comprising the cores. The cores are very porous, feature high surface area to volume ratios, and were held above Tg for several hours (Figs. 2.4c, 2.4d). The calculated effects of H20 on melt viscosity are shown by the dashed curves in Fig. 2.4d (Giordano et al., in Press). These elements suggest that H20 may be dissolved into the glass shards causing a reduction in the viscosity of the framework material and, thus, a reduction in the apparent viscosity of the deforming core. At 3 MPa the maximum (equilibrium) dissolved water content for the shards would be 0.73 wt% (VolatileCaic 1.1; Newman and Lowenstern, 2002). This would reduce the viscosity of the melt at 645 °C from i0’ to iO”3 Pa s (Fig. 2.4d; star symbol), which remains substantially higher than the observed apparent viscosity of rtf5 (1092 - Pa s; Fig. 2.4d). This suggests that the low apparent viscosity recorded by experiment rtf5 reflects the combined effects of an elevated H20 pressure and a residual porosity. Secondly, the presence of the fluid phase itself (rather than dissolved H20) may also cause a reduction in effective viscosity. The presence of a fluid will create a pore 26 fluid pressure (Pfljd) that can lower the (dry) strength (ai.) of the sample such that the effective strength is: 0eff = Gdry — Pflujd (Terzaghi, 1943). Furthermore, the high porosity ensures that virtually all the interfaces between shards are wetted by H20 vapour which may allow for development of hydroxylated monolayers (Schlegel et al., 2002). The hydroxylated monolayers may serve as a lubricant to the glass shards allowing shards to glide past each other without having to deform internally. This is analogous to rock systems in which partial wetting of crystals by a melt phase facilitates grain boundary sliding (de Kloe et a!., 2000). Our experiments demonstrate the importance of porosity and the fluid phase during high-temperature deformation processes. They show that the combined effects of porosity and a fluid (H20) phase greatly expand the window for viscous deformation of volcanic materials. The viscosity recorded by experiments under 3 MPa PH2O (e.g., rtf5) is too low to be ascribed solely to the effects of residual porosity or to elevated dissolved water contents. 2.6 Acknowledgements This research is funded by the Natural Sciences and Engineering Research Council (NSERC) via the Research Tools and Instruments program (JKR), the Discovery Grants program (JKR), and the PGS fellowship program (GR) and by the Italian Dipartimiento della Protezione Civile (2004-06 Agreement, Instituto Nazionale di Geofisica e Vulcanologia — INGV). Chemical analyses of cores of Rattlesnake Tuff ash were generously provided by Steve Quane. We thank Don Dingwell for lab privileges to measure viscosity at the LMU, Munich, Germany. We also thank P. Ardia at ETH 27 Zurich for microprobe analysis of silicate glasses (e.g., fused samples of Rattlesnake Tuff). The manuscript benefited from critical reviews by Luigi Burlini and Cliff Shaw. Finally, we would like to especially thank UBC’s Earth & Ocean Sciences machinists Ray Rodway and JOrn Unger. 2.7 Appendix 2.A: Melt viscosity of the Rattlesnake Tuff ash The viscosity of melted Rattlesnake Tuff ash (Streck and Grunder, 1995) was measured independently by concentric cylinder and micropenetration techniques at the LM(J Munich. The Vogel-Tamman-Fulcher equation (cf. Richet and Bottinga, 1995) has been fit to the data to model the temperature dependence of viscosity for the dry and non- vesicular melt (Table 2.2). Concentric cylinder and micropenetration techniques measure viscosity in the ranges (10’-10 Pa s) and (1081012 Pa s), respectively, and are calibrated against NIST SRM 717a glass. Homogeneous melts were prepared by fusing samples in a thin-walled Pt-crucible in a MoSi2 element furnace (1 atm and 1500 — 1650 °C). The original glass shards contained minor water, which caused vesiculation during fusion. The sample was kept in the melting furnace for more than 1 week until all bubbles had escaped. The sample was then transferred to the concentric cylinder viscometer furnace and a stirring spindle was used to stir the melt. The spindle was periodically lifted out of the melt to determine when the melt was free of crystals and bubbles. Concentric cylinder measurements were performed once the melt was devoid of crystals and bubbles. The crucible was removed from the furnace and allowed to cool in air to quench the sample to a glass. The composition of the glass was determined by electron microprobe analysis using the JEOL JXA 8200 device at ETH-Zentrum, Zurich (Table 2.2). The 28 sample was then cored to produce 3 mm thick, doubly polished disks for low-temperature measurements of viscosity using micropenetration techniques (Giordano et al., 2005). Measurements were performed under Argon atmosphere using a modified Bähr 802 V vertical push-rod dilatometer (Dingwell et al., 1993; Giordano et al., 2004; 2005), and the samples were held at temperature for 1 hour to achieve structural relaxation before each measurement. Shear viscosity (ii) was calculated as described by Pocklington (1940) and Toboisky and Taylor (1963). 29 2.8 References Bagdassarov, N.Sh., Dingwell, D.B., 1992. A rheological investigation of vesicular rhyolite. Journal of Volcanology and Geothermal Research 50, 307-322. Bierwirth, P.N., 1982. Experimental welding of volcanic ash. B.Sc. Honours Thesis, Monash University, 74p. De Kloe, R., Drury, M.R., van Roermund, H.L.M., 2000. Evidence for stable grain boundary melt films in experimentally deformed olivine-orthopyroxene rocks. Physics and Chemistry of Minerals 27, 480-494. Dingwell, D.B., 1998. Recent experimental progress in the physical description of silicic magma relevant to explosive volcanism. In: Gilbert, J.S. and Sparks, R.S.J. (eds.) The Physics of Explosive Volcanic Eruptions, Geological Society, London, Special Publications 145, 9-26 Dingwell, D.B., Bagdassarov, N.S., Bussod, G.Y., Webb, S.L., 1993. Magma rheology. In: Luth, R.W. (ed) Experiments at high pressure and applications to the Earth’s mantle, Mineralogical Association of Canada, Short Course Handbook 21, 131- 196 Friedman, I., Long, W., Smith, R.L., 1963. Viscosity and water content of rhyolite glass. Journal of Geophysical Research 68, 6523-6535. Gardner, J.E., Hilton, M., Carroll, M.R., 1999. Experimental constraints on degassing magma; isothermal bubble growth during continuous decompression from high pressure. Earth and Planetary Science Letters 168, 201-218. Gent, A.N., 1960. Theory of the parallel-plate viscometer. British Journal of Applied Physics 11, 85-87. 30 Giordano, D., Romano, C., Papale, P., Dingwell, D.B., 2004. The viscosity of trachytes, and comparison with basalts, phonolites, and rhyolites. Chemical Geology 213, 49-61. Giordano, D., Nichols, A.R.L., Dingwell, D. B., 2005. Glass transition temperatures of natural hydrous melts: a relationship with shear viscosity and implications for the welding process. Journal of Volcanology and Geothermal Research 142, 105-118. Giordano, D., Russell, J.K., Dingwell, D.B., In Press. Viscosity of magmatic liquids: A model. Earth and Planetary Science Letters. Grunder, A., Russell, J.K., 2005. Welding processes in volcanology: insights from field, experimental, and modeling studies. Journal of Volcanology and Geothermal Research 142, 1-9. Grunder, A.L., Laporte, D., Druitt, T. H., 2005. Experimental and textural investigation of welding: effects of compaction, sintering, and vapor-phase crystallization in the rhyolitic Rattlesnake Tuff. Journal of Volcanology and Geothermal Research 142, 89-104. Kennedy, B., Spieler, 0., Scheu, B., Kueppers, U., Taddeucci, J., Dingwell, D.B., 2005. Conduit implosion during Vulcanian eruptions. Geology 33, 58 1-584. Lejeune, A.M., Bottinga, Y., Trull, T., Richet, P., 1999. Rheology of bubble-bearing magmas. Earth and Planetary Science Letters 166, 71-84. Newman, S. and Lowenstern, B., 2002. VolatileCaic: a silicate melt-H20-C0 solution model written in Visual Basic for Excel. Computers and Geosciences 28, 597- 604. 31 Pocklington, H.C., 1940. Rough measurement of high viscosities. Proceedings of the Cambridge Philosophical Society 36, 507—508. Quane, S.L., Russell, J.K., 2005. Welding: insights from high-temperature analogue experiments. Journal of Volcanology and Geothermal Research 142, 67-87. Quane, S.L., Russell, J.K., Kennedy, L.A., 2004. A low-load, high-temperature deformation apparatus for volcanological studies. American Mineralogist 89, 873- 877. Richet, P., Bottinga, Y., 1995. Rheology and configurational entropy of silicate melts. In Mineralogical Society of America, Reviews in Mineralogy 32, 67-93. Schlegel, M.L., Nagy, K.L., Fenter, P., Sturchio, N.C., 2002. Structures of quartz (1010)- and (1011)-water interfaces determined by X-ray reflectivity and atomic force microscopy of natural growth surfaces. Geochimica et Cosmochimica Acta 66, 3037-3054. Sparks, R.S.J., Tait, S.R., Yanev, Y., 1999. Dense welding caused by volatile resorption. Journal of the Geological Society, London 156, 2 17-225. Spieler, 0., Kennedy, B., Kueppers, U., Dingwell, D.B., Scheu, B., Taddeucci, J., 2004. The fragmentation threshold of pyroclastic rocks. Earth and Planetary Science Letters 226, 139-148. Stein, D.J., Spera, F.J., 1992. Rheology and microstructure of magmatic emulsions; theory and experiments. Journal of Volcanology and Geothermal Research 49, 157-174. 32 Streck, M.J., Grunder, A.L., 1995. Crystallization and welding variations in a widespread ignimbrite sheet; the Rattlesnake Tuff, eastern Oregon, USA. Bulletin of Volcanology 57, 151-169. Sura, V.M., Panda, P.C., 1990. Viscosity of porous glasses. Journal of the American Ceramic Society 73, 2697-2701. Terzaghi, K., 1943. Theoretical Soil Mechanics. John Wiley and Sons, New York, NY., 510 pages. Tinker, D., Lesher, C.E., Baxter, G. M., Uchida, T., Wang, Y., 2004. High-pressure viscometry of polymerized silicate melts and limitations of the Eyring equation. American Mineralogist 89, 1701-1708. Tobolsky, A.V., Taylor, R.B., 1963. Viscoelastic properties of a simple organic glass. Journal of Physical Chemistry 67, 2439—2442 Tuffen, H., Dingwell, D.B., Pinkerton, H., 2003. Repeated fracture and healing of silicic magma generate flow banding and earthquakes? Geology 31, 1089-1092. 33 CHAPTER III: Deformation experiments’ 3.1 Introduction Many volcanic processes involve the production and growth of gas-filled bubbles, the connection of bubbles to produce permeability, and the subsequent collapse of the bubbles. These cycles of bubble growth and collapse are important elements in processes as diverse as magma ascent, transition from explosive to effusive volcanic eruption, fragmentation processes in volcanic conduits, dome growth and collapse, and the inflation, collapse, and welding of pyroclastic density currents. Despite its obvious importance for understanding and modelling volcanic processes, our knowledge of the rheological properties of porous magmas is incomplete. Compaction and sintering of particulate materials in the ceramics industry has provided insights on the effects of porosity on the viscosity of composite materials (Fig. 3.1; Rahaman et al., 1987; Ducamp and Raj, 1989; Sura and Panda, 1990) and, more importantly, has inspired experimentation on materials pertinent to volcanology (Fig. 3.1). For example, there are a now a number of high-temperature experimental studies on synthetic melt systems that elucidate the rheological behaviour of porous melts (i.e., Stein and Spera, 1992; Lejeune et al., 1999; Quane et al., 2004). There also are a smaller number of parallel experimental studies on natural volcanic materials (i.e. Friedman et al., 1963; Bierwirth, 1982; Bagdassarov and Dingwell, 1992; Quane, 2004; Quane and Russell, 2005) and these studies report a wide range of rheological behaviours (see Quane and Russell, 2005; Grunder and Russell 2005 for reviews). ‘A version of this chapter has been submitted for publication. Robert, G., Russell, J.K., Giordano, D., In Review. Rheology of porous volcanic materials: High-temperature experimentation under controlled water pressure. Chemical Geology Special Issue, 8th Silicate Melt Workshop. 34 1Figure 3.1 Compilation of previous experimental studies on deformation of porous melts or glasses as the relative viscosity (ii,.), taken as the ratio of apparent viscosity of the porous system (lapp) and the viscosity of the melt (1ieit)’ vs. total porosity of the system (I). Studies are grouped as deformation of (i) porous ceramic glasses or glass powders (dashed curves; (R): Rahaman et al., 1987; (D&R): Ducamp and Raj, 1989; (S&P): Sura and Panda, 1990); (ii) bubbly or porous synthetic melts (solid curves; (S&S): Stein and Spera, 1992 and (Q&R): Quane et al., 2004; and solid circles: Lejeune et al., 1999); and (iii) porous natural melts (bold, solid curves; (B&D): Bagdassarov and Dingwell, 1992; (Q): Quane et a!., 2005). 0. 35 Figure 3.1 summarizes results from some of these experimental studies, including experiments on natural and synthetic melts. These experiments comprise two end- member approaches: (i) deformation experiments on porous samples in which bubbles are suspended in a coherent melt/glass phase (Bagdassarov and Dingwell, 1992; Stein and Spera, 1992; Lejeune et al., 1999), or (ii) experiments deforming porous samples in which the pores exist between the particles that constitute the solid framework, including sintered ceramic particles, glass beads, or volcanic ash (Friedman et al., 1963; Bierwirth, 1982; Rahaman et al., 1987; Ducamp and Raj, 1989; Sura and Panda, 1990; Quane, 2004). In the latter case the particles, themselves, may or may not be porous. Ultimately all hydrous melts vesiculate at or near the Earth’s surface to produce bubble-rich melts, which commonly continue to expand to the point of fragmentation. How the increase in porosity affects the viscosity of the magma remains unclear. Here, we use high-temperature (T) uniaxial compression experiments on cores of volcanic material to elucidate the rheological behaviour of high porosity magmas. Our program uses an apparatus that allows for deformation (i.e. compaction) experiments on porous cores of sintered volcanic ash at high-T and under controlled water pressure (PH2O). These experiments cause a reduction in the porosity of the ash-core samples and a concomitant change in their rheological properties. Most strain can be ascribed to volume loss by pore destruction (volume strain); however, we also show that radial expansion of the sample cores (shear strain) becomes increasingly important at high values of strain. Our results also demonstrate that, at the timescales of these experiments, the window of viscous deformation is expanded substantially by increasing porosity. Under dry conditions, the temperature limits of viscous deformation for highly porous 36 cores of ash are reduced by --50 °C. Experiments under PH2O of 2.5 MPa also show that increasing porosity expands the window of viscous deformation by --140-150 °C, depending on the water content of the melt. These results have implications for the processes governing the welding of ignimbrites (Sparks et a!., 1999), fragmentation cycles in volcanic conduits (Tuffen et al., 2003; Kennedy et al., 2005) and the formation and flow of clastogenic lavas (Manley, 1996; Wolff and Sumner, 2000). 3.2 Experimental methods 3.2.1 Experimental apparatus All experiments presented are performed using the Volcanology-Deformation-Rig (VDR) at the University of British Columbia in conjunction with a water cell specifically designed for high-temperature experimentation at volcanic conditions in the presence of water (Quane et al., 2004; Robert et al., 2008). The cell can operate at temperatures up to 1100 °C and fluid pressures of 0-150 IVIPa; sample sizes can be up to 30 mm in diameter and 100 mm in length. Deformation experiments on sample cores can be performed under a constant load (<1135 kg) or at constant displacement rate (5 10 to 2.5 102 cmls). The VDR’ s computer system records time, load, displacement; the water cell is equipped with a transducer that records water pressure continuously. A detailed description of the experimental apparatus and its calibration for recovering melt viscosity can be found in Robert et a!. (2008). The original design for the bottom of the cell has been modified slightly to prevent leaks. Specifically, the bottom seal remains metal-on-metal, but we have developed a more efficient tightening mechanism that provides an even pressure distribution around the entire lower seal. 37 Detailed line diagrams and a list of parts for the water cell can be found at http://www.eos.ubc .calresearchlinfrastructure/cesl.html. 3.2.2 Fabrication of experimental cores The deformation experiments are performed on cores created by sintering volcanic ash collected from the Rattlesnake Tuff (Streck and Grunder, 1995). The ash is sieved to a 0.6-2 mm size fraction and cores are sintered by heating the loose ash in a mold (2.54 cm by 8 cm) at 900 °C for 20 minutes. Samples are trimmed to -5 cm lengths creating cores with a 2:1 aspect ratio (Fig. 3.2a). Table 3.1 reports chemical composition data of the Rattlesnake Tuff ash for: (i) natural (non-sieved) ash; (ii) sintered ash, pre and post-experiment; and (iii) fused ash (glass). These measurements show that there is little change in composition after sintering (Table 3.1). Cores of ash comprise curvilinear and Y-shaped bubble wall shards, complete vesicles (e.g., bubble shards), smaller proportions of pumiceous shards, and up to 1% crystals. The sintering process causes point annealing of shards and forms a highly porous, floating, shard-supported framework (Fig. 3.2b). The cores produced by the sintering technique have an essentially isotropic texture and show no foliation or preferred orientation of shards (Fig. 3 .2a, b). The sintered cores feature two types of bubble shards (Fig. 3.2): (i) a population of thick-walled vesicle shards characteristic of the original ash, and (ii) a subordinate population of thinner-walled vesicle shards. The latter population appears to be produced during the sintering process and represent originally closed vesicles (isolated porosity) that expanded during heating or new bubbles that nucleated and grew during fabrication. The ash particles are not deformed during 38 Figure 3.2 Starting experimental materials. (A) Photograph of fabricated core (2.5 cm by 5 cm) of Rattlesnake Tuff ash used in high-T deformation experiments. (B) SEM photomicrograph of pre-experiment core of sintered ash showing proportions of ash (light grey) to pore space (black) and the diversity of ash particles, including: bubble walls, glass shards, and pumiceous fragments. Large, round, thin-walled bubble shards are likely a product of vesiculation of hydrous shards during the sintering process. Smaller, thick-walled bubble shards are a direct product of the original fragmentation. 39 Ta bl e 3. 1. Ch em ic al co m po sit io n o ft he R at tle sn ak e Tu ff as h. O xi de Li te ra tu re a SQ 00 00 b SQ 00 01 c G RR S3 O d SQ 08 21 be R S_ m el t Si ev ed Si nt er ed Si nt er ed Po st- ex p’ t X R F EM P S i0 2 77 .1 1 73 .9 9 76 .7 9 76 .1 2 76 .1 1 76 .2 7 76 .2 9 T i0 2 0. 12 0. 15 0. 17 0. 14 0. 16 0. 20 0. 14 A 1 203 11 .7 7 12 .1 2 12 .3 4 13 .1 6 12 .6 3 12 .8 6 12 .0 4 Fe O (T ) 1. 45 1.1 1 1. 16 1. 00 1. 14 1. 36 1. 37 M nO 0. 09 0. 07 0. 07 0. 08 0. 07 0. 08 0. 08 M gO 0. 00 0. 04 0. 00 0. 07 0. 00 0. 06 0. 04 Ca O 0. 35 0. 29 0. 31 0. 29 0. 30 0. 31 0. 30 N a 20 3. 70 3. 26 3. 34 3. 83 3. 39 3. 60 3. 39 K2O 5. 23 4. 42 4. 57 4. 59 4. 59 4. 74 4. 89 P205 0. 01 0. 02 0. 01 0. 01 0. 00 0. 01 0. 01 H 20 + 0. 00 3. 30 0. 15 0. 14 0. 24 0. 07 0. 00 To ta l 99 .8 3 98 .7 7 98 .9 1 98 .4 3 98 .6 3 99 .5 6 98 .5 4 a A na ly sis fro m St re ck an d G ru nd er (19 97 ). b N at ur al R at tle sn ak e Tu ff as h sie ve d to co ar se as h fro m Qu an e( 20 04 ). C R at tle sn ak e Tu ff as h af te rs in te rin g fro m Qu an e ( 20 04 ). d R at tle sn ak e Tu ff as h af te rs in te nn g (th is st ud y). e R at tle sn ak e Tu ff as h fro m po st- ex pe rim en ta lc o re (D RY ) f ro m Qu an e( 20 04 ). X R F an al ys is o fR at tle sn ak e Tu ff m el tu se d in co n ce n tr ic -c yl in de r m ea su re m en ts (cf .R ob er t e ta l., 20 08 ). g EM P an al ys is o fR at tle sn ak e Tu ff m el t u se d in co n ce n tr ic -c yl in de rm ea su re m en ts (cf .R ob er t e ta l., 20 08 ). sintering except around bubbles that formed or expanded during heating; there, shards are bent around thin-walled bubbles (Fig. 3.2b). One unexpected result of the sintering process is that the shards become extensively fractured and pitted (Fig. 3.2b); the microfracturing may result from rapid cooling when the samples are removed from the sintering oven. 3.2.3 Pre-experimental sample characterization The physical properties of each sample are measured prior to running the deformation experiments, including: geometry, density and porosity. The volume of these highly porous cylindrical sample cores is calculated from averages of replicate (n=10) measurements of diameter and length (Table 3.2). This volume and the sample mass are used to compute the bulk density (Pbulk) of the core (Table 3.3: 0.37-0.43 g/cm). Skeletal (or framework) density (Psiceietai) is obtained by measuring sample volume via helium pycnometry and ranges from 0.97-1.40 g/ cm3 (Table 3.3). Connected porosity QIconnected) is calculated from skeletal and bulk density from the relationship: connected = 1 — (3.1) Pskeletal We obtained values of dense rock equivalent (DRE) density for sintered materials by crushing three sintered cores and performing pycnometry on the resulting powders. These cores for DRE measurements were also fabricated in the way described above; we assume all experimental cores to have the same average DRE density (2.36 g/cm3). Using this average value for powder density we compute total and isolated porosity as: otal1 Pbulk (3.2) Ppowder 41 Ta bl e 3. 2. Ex pe rim en ta lc o n di tio ns ” u se d in de fo rm at io n ex pe rim en ts an d ge om et ry b o fs am pl e co re s pr e- an d po st- ex pe rim en t. 2. 5. 10 6 5. 26 8 2. 43 0 1. 26 0 1. 36 5 2. 5. 10 6 5. 21 0 1. 10 0 1. 26 7 1. 41 7 2. 5. 10 6 5. 16 5 2. 03 5 1. 27 8 1. 37 7 2. 51 O 5. 06 9 3. 60 9 1. 27 2 1. 33 0 - 5. 13 7 4. 95 0 1. 27 8 1. 27 7 2. 5. 10 6 5. 15 2 3. 70 3 1. 27 1 1. 28 4 2 .5 10 5. 19 6 2. 38 1 1. 27 2 1. 31 4 2. 5. 10 6 5. 20 0 3. 72 7 1. 27 9 1. 30 1 2. 5. 10 6 5. 04 0 3. 57 8 1. 28 5 1. 32 4 2. 5. 10 6 4. 42 0 1. 94 2 1. 27 7 1. 36 8 2. 5. 10 6 5. 27 7 1. 12 5 1. 27 8 1. 44 6 2 .5 1 0 4. 89 1 - 1. 29 4 - 2. 5.1 11 6 5. 20 3 1. 15 5 1. 28 0 1. 39 1 2. 5. 10 6 5. 26 9 - 1. 28 2 - 2. 5. 10 6 5. 23 2 - 1. 28 1 - 2. 51 11 6 5. 20 8 3. 69 1 1. 28 5 1. 27 6 2. 51 11 6 5. 29 5 2. 53 4 1. 24 1 1. 29 8 2. 51 11 6 5. 31 7 2. 35 4 1. 30 6 1. 39 1 2. 54 11 6 5. 32 1 2. 61 2 1. 32 1 1. 34 1 2. 51 11 6 5. 41 7 - 1. 28 2 - 2. 51 0- s 5. 40 7 2. 65 0 1. 29 2 - 2. 51 11 5. 24 3 0. 81 7 2. 59 2 3. 14 2 _ _ _ V 0 V 11 .38 11 .3 5 26 .2 7 14 .2 2 9. 69 9. 65 26 .2 8 6. 94 11 .3 1 11 .2 4 26 .4 9 12 .1 3 10 .8 8 10 .8 4 25 .7 6 20 .0 5 10 .9 3 10 .71 26 .3 6 25 .3 6 10 .4 5 10 .4 5 26 .1 5 19 .1 8 10 .5 6 10 .5 8 26 .4 3 12 .91 10 .8 3 10 .8 6 26 .7 0 19 .8 2 10 .3 1 10 .0 9 26 .1 3 19 .7 1 8. 84 8. 68 22 .6 5 11 .41 10 .4 6 10 .2 7 27 .0 6 7. 39 10 .1 7 8. 50 25 .7 4 - 10 .5 2 10 .51 26 .7 7 7. 02 11 .0 9 8. 25 27 .2 2 - 11 .1 3 - 26 .9 6 - 11 .1 4 11 .11 27 .0 2 18 .8 8 11 .7 8 11 .7 6 25 .6 3 13 .41 11 .4 7 11 .03 28 .4 8 14 .3 2 11 .61 11 .59 29 .1 6 14 .7 5 12 .0 7 11 .2 6 27 .9 5 - 12 .0 2 10 .0 8 28 .3 7 - 11 .2 8 11 .2 2 11 0. 67 25 .3 4 N o Se t-U p t T P (H 20 ) zM /A t l 1 r0 rf m0 m f LJ RS O3 Ce ll 10 72 8 64 0± 4 1 RS O4 Ce ll 15 76 8 68 1± 20 5 RS O5 Ce ll 12 28 0 65 9± 9 1 RS O7 Ce ll 52 26 65 4± 4 1 RS O9 Ce ll - . 54 00 [25 -65 0] — 2.5 RS 1O Ce ll 52 23 64 7± 7 5 RS 11 Ce ll 10 21 2 65 9± 13 5 R S1 2 Ce ll 50 67 66 2± 8 3 R S1 3 Ce ll 51 21 65 4± 6 2. 5 R S1 4 Ce ll 88 60 65 0± 9 2. 5 R S1 5 Ce ll 15 72 3 66 6± 18 2. 5 R S1 6 V D R 96 50 80 0± 15 0 R S1 7 V D R 15 64 2 90 0± 15 0 R 51 8 Ce ll 10 59 0 45 2± 12 2. 5 R S1 9 Ce ll 10 51 2 38 6± 16 2. 5 RS 2O V D R 52 08 90 0± 15 0 R S2 1 V D R 10 60 8 90 0± 15 0 R S2 2 Ce ll 10 71 0 55 0± 14 2. 5 R S2 3 V D R 10 66 8 85 0± 15 0 R S2 4 V D R 10 88 4 75 0± 15 0 R S2 5 V D R 10 83 85 0± 15 0 R S2 9 V D R 17 45 7 90 0± 15 0 a Ti m e (t) in s; te m pe ra tu re (T )i n ° C; w at er pr es su re (PH ,o) in M Pa ;d isp la ce m en tr at e (A IJA t) in m is. b D im en sio ns o fc o re s: (1: le ng th (cm ); r: ra di us (cm ); V: v o lu m e (c m 3) ) be fo re (i.e ., l )a n d af te r( i.e . l )e x pe rim en ta lr u n s. Ta bl e 3. 3. M ea su re d v al ue s o fd en sit y an d po ro sit y fo r p re - a n d po st -e xp er im en ts am pl e co re s. D en sit ya Po ro sit yb 1T bu lk bu lk sk ele tal sk ele tal , to ta l to ta l co n n ec te d , co n n ec te d T iso la te d , j iso la te d 1 0 Po Pf Po Pf ‘ ‘ f ‘ ‘ 0 ‘ t’ f ‘ ‘ o ‘ ‘ f RS O3 0. 43 3 0. 79 4 1. 29 2 2. 11 3 0. 81 6 0. 66 4 0. 66 5 0. 64 1 0. 15 2 0. 02 3 RS O4 0. 36 9 1. 39 1 1. 16 4 2. 17 7 0. 84 4 0. 41 1 0. 68 3 0. 36 1 0. 16 1 0. 04 9 RS O5 0. 42 8 0. 92 8 1. 10 9 2. 18 3 0. 81 9 0. 60 7 0. 61 4 0. 57 5 0. 20 5 0. 03 2 RS O7 0. 42 2 0. 54 2 1. 14 3 1. 75 3 0. 82 1 0. 77 0 0. 63 0 0. 69 1 0. 19 1 0. 07 9 RS O9 0. 41 5 0. 42 2 1. 11 9 1. 97 6 0. 82 4 0. 82 0 0. 62 9 0. 78 5 0. 19 5 0. 03 5 RS 1O 0. 40 0 0. 54 5 1. 10 0 1. 79 5 0. 83 0 0. 76 9 0. 63 6 0. 69 6 0. 19 4 0. 07 3 RS 11 0. 40 0 0. 82 0 1. 01 6 1. 87 8 0. 83 1 0. 65 3 0. 60 7 0. 56 3 0. 22 4 0. 08 9 R S1 2 0. 40 6 0. 54 8 1. 12 2 1. 89 0 0. 82 8 0. 76 8 0. 63 9 0. 71 0 0. 19 0 0. 05 8 R S1 3 0. 39 5 0. 51 2 1. 11 6 1. 77 5 0. 83 3 0. 78 3 0. 64 6 0. 71 2 0. 18 7 0. 07 1 R S1 4 0. 39 0 0. 75 9 1. 15 8 1. 99 1 0. 83 5 0. 67 8 0. 66 3 0. 61 9 0. 17 2 0. 06 0 R S1 5 0. 38 7 1. 39 1 1. 07 4 2. 20 5 0. 83 6 0. 41 1 0. 64 0 0. 36 9 0. 19 6 0. 04 1 R S1 6 0. 39 6 - 1. 12 8 - 0. 83 2 - 0. 64 9 - 0. 18 3 - R S1 7 0. 39 3 1. 49 7 1. 04 2 2. 24 5 0. 83 3 0. 36 6 0. 62 3 0. 33 4 0. 21 1 0. 03 2 R S1 8 0. 40 8 - 1. 07 4 - 0. 82 7 - 0. 62 0 - 0. 20 7 - R S1 9 0. 41 4 - 1. 10 5 - 0. 82 5 - 0. 62 6 - 0. 19 9 - RS 2O 0. 41 2 0. 58 8 1. 12 0 1. 76 9 0. 82 5 0. 75 1 0. 63 2 0. 66 7 0. 19 3 0. 08 3 RS 21 0. 46 1 0. 87 7 1. 39 6 2. 04 5 0. 80 5 0. 62 8 0. 67 0 0. 57 1 0. 13 5 0. 05 7 R S2 2 0. 40 3 0. 77 0 1. 10 3 2. 26 5 0. 82 9 0. 67 4 0. 63 5 0. 66 0 0. 19 5 0. 01 4 R S2 3 0. 39 7 0. 78 6 1. 05 2 1. 81 2 0. 83 2 0. 66 7 0. 62 2 0. 56 6 0. 20 9 0. 10 1 R S2 4 0. 43 2 - 1. 10 5 - 0. 81 7 - 0. 60 9 - 0. 20 8 - R S2 5 0. 42 4 - 1. 01 2 1. 85 0 0. 82 0 - 0. 58 1 - 0. 23 9 - R S2 9 0. 41 0 1. 77 1 0. 97 4 2. 26 6 0. 82 6 0. 24 9 0. 57 9 0. 21 8 0. 24 7 0. 03 1 a B ul k an d sk el et al de ns ity o f p re - ( Po )a n d po st- (o f) ex pe rim en ta l c o re s in g/ cm 3 To ta l, co n n ec te d an d iso la te d po ro sit y o f p re - ( ) an d po st- () ex pe rim en ta l c o re s. Pbulk — Pbulk isolated — Pslcele:at Ppowder (cf Michol et al., 2008). The total porosity of sintered cores varies from -0.80-0.84 and comprises both connected (-0.58-0.67) and isolated (-0.14-0.25) fractions (Table 3.3). 3.3 Experimental results 3.3.1 Overview A total of 21 deformation experiments were performed on sintered cores of Rattlesnake Tuff ash, including eight at atmospheric pressure (dry) and 13 at controlled water pressure (wet). The experimental conditions for the 21 experiments are summarized in Table 3.2. The same constant displacement rate (2.5 i03 mmls) was used in all deformation experiments except for sample RS25, which was deformed 1 order of magnitude faster than the others (Table 3.2). The experiments run under dry conditions are used to illustrate the effects of porosity on melt rheology and establish a baseline response against which we compare results from wet experiments. The PH2O experiments provide a closer approximation to nature in that they involve deformation of porous cores at temperatures and fluid pressures commonly found in volcanic environments. 3.3.2 Dry high-T experiments Four experiments were performed at atmospheric pressure conditions (dry) and at a temperature of 900°C (Table 3.2). The four dry experiments RS2O, RS21, RS 17 and RS29 were deformed to strains () of 0.25, 0.5, 0.75, and 0.82 respectively (Fig. 3.3a; 44 Table 3.4. Analysis of strain. Strain a b Et St c d No RSO3 0.500 0.539 0.454 0.148 RSO4 0.750 0.789 0.735 0.201 RSO5 0.585 0.606 0.539 0.139 RSO7 0.250 0.288 0.221 0.085 RSO9 0.000 0.036 0.024 -0.002 RS1O 0.250 0.28 1 0.266 0.020 RS11 0.500 0.542 0.513 0.062 RS12 0.250 0.283 0.260 0.034 RS13 0.250 0.290 0.229 0.059 RS14 0.500 0.561 0.486 0.128 RS15 0.750 0.787 0.722 0.219 RS16 0.500 - - - RS17 0.750 0.778 0.737 0.153 RS18 0.500 - - - RS19 0.500 - - - RS2O 0.250 0.291 0.300 -0.014 RS2I 0.500 0.521 0.474 0.085 RS22 0.500 0.557 0.477 0.119 RS23 0.500 0.509 0.494 0.029 RS24 0.500 - - - RS25 0.500 0.5 10 - - RS29 0.822 0.844 0.769 0.319 a Total strain from machine displacement. b Total strain from shortening of the core. C Total strain from porosity reduction. d Total strain from radial increase. 45 Tables 3.2-3.4). All four experiments show a smooth increase in load stress with increasing strain to 0.5 followed by a much steeper increase in load stress with additional strain. The resulting stress-strain relationships for each experiment are nearly identical as shown by the overlapping curves in Figure 3.3a because the starting materials were virtually identical (i.e., diameter, porosity, etc.; Table 3.3). These results indicate the high degree of reproducibility of our experimental methods. 3.3.3 Wet high-T experiments A total of 13 experiments were completed at elevated water pressures of: 1, 2.5 and 5 MPa (Table 3.2). All three different PH2O series were all performed at —650°C. Experiments performed at elevated water pressure but temperatures < 650 °C are reported in Tables 3.2-3.4, and are discussed in a later section. The three experiments RS7, RS3, and RS5 were run at —650 °C and PH2O 1 MPa (Table 3.2; Fig. 3.3b) and samples were deformed to Et of 0.25, 0.5 and 0.6, respectively (Table 3.4). Overall the stress-strain relationships are similar to those observed in the dry, high-temperature experiments. Load stress shows a smooth and continuous rise with increasing total strain. The stresses achieved during these experiments are about an order of magnitude lower than observed in the dry high-T experiments. However, significant rises in stress (—20%) are achieved at lower amounts of strain (25-40%) than was observed in the dry experiments (>60%). At this strain rate, the 1 MPa PH2O experiments show an exponential rise in stress at —0.55. Three experiments were conducted at —650 °C and 2.5 IVIPa (Table 3.2; Fig. 3.3c). Samples RS13, RS14 and RS15 were deformed to = 0.25, 0.5 and 0.75, respectively 46 Figure 3.3 Summary of experimental data (cf. Table 3.2; Table 3.4) plotted as stress (a) vs. strain (8). (A) Data recorded from high-temperature (—900°C) dry series of experi ments. Lower temperature (—650 °C) experiments performed under water pressures of: (B) H2O = 1 MPa, (C) H2O = 2.5 MPa, and (D) P1120 = 5 MPa. Controlled P1120 experi ments (i.e. B, C, D) were terminated after —25%, 50% and 75% total strain except for sample RSO5 (B) which recorded —60% strain (Table 3.2). Atm (Dry) at 900°C R A 0.5 S29 . 0.25 RS211 j7 RS17 RS2O 0 0.25 0.5 Et RS2O RS21 a 30 C’, a.. 20 10 Co C’, a 0.25 0.5 8 0.75 C’, 0 0.5 8 0.5 8 47 (Table 3.4). The increase in water pressure from 1 MPa to 2.5 MPa allows for continuous strain at substantially lower load stresses (<2 MPa). Moreover, there is little to no significant rise in stress over the interval 0 to 0.3. At Et > 0.5 the load stress required to sustain deformation increases but remains low (— 1.5 IVIPa at r = 0.75) relative to the dry and PH2O 1 MPa experiments. Three deformation experiments were run at —645-680 °C and a water pressure of 5 MPa (Table 3.2; Fig. 3.3d). Samples RS1O, RS11, and RS4 were deformed to Et = 0.25, 0.5 and 0.75 respectively. Experiment RS 10 ( = 0.25) shows no increase in stress over the total deformation path, and the stress required for deformation is near the resolution level of the apparatus. The intermediate strain experiment (RS1 1; E = 0.5) shows a saw tooth pattern recorded during experiment that is due to sharp fluctuations in water pressure around an average value of 5 MPa. Overall, the deformation path shows a slight, relatively linear increase in stress from -—0.4 to 0.8 IVIPa. To a first approximation, these experiments show that, at PH2O = 5 MPa, continuous deformation can proceed with no increase in load stress despite the fact that porosity is being reduced from —80 to 40%. Only at very high values of total strain (E,> 0.7) where porosity is less than 50% is there a hint of increasing stress with progressive strain. 3.3.4 Textural analysis of experimental cores Figure 3.4 illustrates the textural evolution of samples with progressive strain. At values of 0.25, deformed samples are still highly porous (—0.75-0.78) and cracks that were evident in fabricated cores are still present. Much of the porosity remains intact and there is little annealing of shards. Ash shards show mainly point sintering. However, 48 Figure 3.4 Textural evolution of samples during high-T deformation (H2O = 2.5 MPa) of cores represented by photographs of shortened cores and SEM images showing the microstructures associated with flattening, folding, and annealing of ash-sized particles (e.g., bubble walls, glass shards, and pumice) and parallel loss of pore space. Images are for experiments of: —25% (A), 50% (B), and 75 % (C) strain. E=O.25 49 compared to the post-sintering samples, shards in the experimental run products are clearly deformed and show more warping and re-orientation perpendicular to loading. Deformed shards show extension cracks on the outer curvature radius. At higher values of strain ( 0.5), samples have a porosity between -O.63-O.67. Shards are intricately folded and alignment of shards and flattening of pumice is apparent (Fig. 3,4b). The shards also show signs of annealing; cracks in the shards resulting from the fabrication process are not as common and show signs of healing. Figure 3.4c illustrates the textural evolution of the ash cores after 75% strain (Et = 0.75). Porosity is reduced to —0.37-O.41. Most shards are highly deformed, folded and flattened; pumices are also collapsed and flattened. Shards are collapsed and annealed into coherent masses such that the boundaries between welded shards are hard to distinguish. Particles have a strong preferred alignment developed perpendicular to the compression direction. Extension cracks are absent from deformed shards in high strain samples. 3.4 Post-experimental physical properties 3.4.1 Porosity Total, connected, and isolated porosity was measured before and after each experiment, using the methods explained in the experimental methods section (3.2.3; Table 3.3). As discussed above, post-sintering (pre-experimental) sample cores have a total porosity varying between 0.81 and 0.84; connected porosity dominates but there can be up to —0.2 isolated porosity (see Table 3.3). Values of total, connected, and isolated porosity for the pre- and post-experimental cores are plotted in Figure 3.5. 50 Figure 3.5 Nature and distribution of porosity in pre- and post-experiment sample cores. The presence of water during the deformation experiments does not affect the distribution of the porosity; there is no difference in the trends for dry and wet experiments in terms of porosity. We therefore make no distinction between the two. (A) Measured values of total porosity (Fe) are plotted against values of connected (Is: circles) and isolated (Li: squares) porosity for pre-experiment (filled symbols) and post-experiment (open symbols) cores. The overall reduction in I features an initial decrease in and parallel increase in Ic followed by a steady loss of alone. A single “dwell-time” experiment (—120 minutes) was performed to track the porosity changes (open vs. filled triangles) prior to starting the deformation experiment (see text). (B) Isolated porosity vs. connected porosity. Pre-experiment values of porosity (grey circles) are near constant plotting between 15O-It lines 0.8 and 0.85, but comprise different proportions of isolated and connected porosity. Porosity values of post-experiment cores are grouped by total strain and highlighted by labelled () grey ellipses. Triangles denote “dwell-time” experiment (as in A). 51 1 A 7 0.8 ao -e-0.6 0 () -e- 0.4 0.2 D C QD 0.8 Z 0.6 ø.::.” V N Z z N\ 7 N /‘ fl A NZNU.Lt 7N N 7 N N7 N N 7 NN N N 0.2 VV •05 •.ø.25.0:82 O.7•5 • NIN ‘0 0.2 0.4 0.6 0.8 Figure 3.5 See previous page for caption. 52 Every experiment begins with a —2 hour equilibration period (“dwell-time”), designed to allow the cell and sample to reach the experimental temperature and water pressure. We ran a “dwell-time experiment” to assess the extent and nature of physical changes occurring during the dwell time and prior to the onset of deformation (see Appendix 3.A). In this experiment, the sample was taken to experimental conditions (—650°C and — 2.5 MPa PH2O), given enough time to equilibrate at these conditions, and was cooled back down to room temperature. The physical properties of the resulting core were then measured to quantify the extent of change (Tables 3.2 and 3.3). In addition to a systematic reduction in core length (Appendix 3.A), the major change in the sample concerns the distribution of porosity. During the dwell time total porosity is conserved (from 0.824 to 0.820), however, there is a shift in the proportions of isolated and connected porosity. The data show that, during dwell time, isolated pores are destroyed while the total abundance of connected porosity increases (triangles in Fig. 3.5). One explanation for this pattern is that, during dwell time, isolated pores become connected either by coalescence or microfracturing. Figure 3.5 shows the evolution in porosity during the deformation experiments. After the dwell time, the sample retains a total porosity of 0.8 comprising both connected (> 0.65) and a residual isolated (<0.1) porosity. As deformation proceeds, total porosity is reduced continuously; however, the porosity reduction is mainly at the expense of connected porosity. After the initial decrease in isolated porosity that occurs during dwell time, deformation to very high (70-80%) values of strain produces no further change in isolated porosity. 53 The materials we are experimenting on are extremely porous (1 0.7), and it is reasonable to assume that most of the strain is accommodated by volume loss. Knowing the initial and final porosity of a sample, the amount of strain due to volume loss (pore destruction) can be calculated using the following relationship: (3.4) l-f where t is initial total porosity and I is final total porosity (Quane and Russell, 2005). Figure 3.6 shows that most of the deformation in our experiments can be ascribed strictly to volume loss (e,), but that at values of strain -0.7, the run-products are more porous than predicted by Eq. (3.4). This suggests that deformation mechanisms other than volume strain (strain from porosity loss) are active during the compaction. Further evidence for more than one mechanism of deformation being active is provided by the “onset” of significant radial increase of the samples at high amounts of total strain. 3.4.2 Water content We measure the bulk water content of our experimental run products to verify the amount of water incorporated into the sample during deformation. Any amount of water dissolved into the glass will reduce the viscosity of the melt and contribute to the overall strength reduction of the sample. We found that the timescale of our experiments (- 1.5 to 4.5 hours after initial dwell-time) is sufficient to allow water dissolution into the glass. The bulk water content results are presented and discussed in Appendix 3.B (Table 3.B). 54 -e t Figure 3.6 Measured values of total porosity (J) are plotted against total machine strain ()• The black bar (top left) denotes the range of initial porosities of cores for the entire suite of experiments (0.804 to 0.844). The grey shaded band represents the field of model curves of decreasing porosity calculated for a known initial porosity as a function of increasing strain and assuming pure volume strain. The model curves converge to zero porosity as E approaches a value equal to the initial porosity. Volume strain is sufficient to explain these data until E -60% where the run-products may have higher porosity than predicted. This suggests that deformation mechanisms other than volume strain (strain from porosity loss) are active during compaction (see text and Fig. 3.8). 0 0.5 1 55 3.5 Analysis of experimental results 3.5.1 Effect of temperature and PH2O. We ran six additional experiments to document the effects of temperature on the rheological behaviour of these porous volcanic materials. Three dry experiments, involving samples RS23, RS16 and RS24, were carried out at 850, 800 and 750 °C, respectively (Fig. 3.7a). The three experiments deformed samples to = 0.5 and provide data that map the temperature boundary between ductile and brittle deformation for dry porous samples at the characteristic deformation timescale of these experiments (texp). Sample RS23 shows a smooth increase in stress with increasing strain, but compared to the 900 °C dry experiment (RS21), the stress increase with strain at 850 °C is more pronounced. The lower temperature experiments (750 and 800 °C) show strikingly different behaviours than observed in the higher temperature experiments. Instead of a smooth increase of stress with increasing strain, they display a saw tooth pattern with sharp (20%) stress build-ups followed by quasi-instantaneous stress drops (Fig. 3.7a). Run-products for samples deformed at those low temperatures are extensively fractured, often broken-up in smaller pieces, or exhibiting fractures (see picture insets in Fig. 3.7). This suggests that, at the timescale of these experiments, the transition from ductile to brittle deformation of the dry porous cores occurs at between 850 and 800 °C. The effect of water pressure is summarized in Figure 3.7b by comparing results of a 900 °C dry experiment to results from parallel experiments involving more than 50% strain at 3 different water pressures. The dry experiment (900 °C) was performed at a temperature -250 °C higher than any of the experiments performed at water pressures of 1, 2.5, and 5 MPa. Increasing water pressure during deformation significantly reduces 56 sample strength, and allows for large amounts of deformation to be obtained with minimal stress imposed. The high temperature experiment plots between the 1 and 2.5 MPa PH2O experiments and illustrates the trade-off between temperature and water pressure. At low water pressures and --650 °C, the porous cores carry more stress than the dry core at 900 °C, while at —650 °C and water pressures greater that 2 MPa the material is substantially weaker (Fig. 3.7b). We also ran three high-PH2Oexperiments at reduced temperatures to document the temperature-controlled transition from ductile to brittle deformation in the wet systems. Samples RS22, RS18 and RS19 were deformed at 550, —450 and —385 °C, respectively, at the same displacement rate of 2.5 i03 mrnls, to a total strain of 0.5. A PH2O of 2.5 MPa was chosen as an intermediate response between the two end-member water pressures used in our experiments. Sample RS22 (550°C) showed a smooth increase in stress with increasing strain, but stress was --3 times higher than recorded in the —650 °C, 2.5 MPa PH2O experiment (RS14) at the same values of e. Samples RS18 and RS19 showed a pattern of brittle deformation characterized by sharp rises and drops in stress with increasing strain, but without stress ever building up over 1 IVIPa. The pattern is very similar to that of the dry experiments (Fig. 3.7a) but the stress drops are not as pronounced. On the basis of these response curves, we suggest that for these experiments, the transition from ductile to brittle behaviour occurs at between 550 and 450 °C. We expect that the brittle-ductile boundary will shift to lower temperatures with increasing PH2O or with lower rates of displacement (e.g., longer experimental timescales). 57 Figure 3.7 Experimental data used to illustrate the effects of temperature and PH2O on sample rheology. (A) Experimental data plotted as stress (0) vs. strain () for atmospheric dry experiments run to —50% strain and performed over a range of temperatures (750-900 °C). Experiments at 850-900 °C show o vs. E patterns that are consistent with ductile deformation. Dry experiments run at 750 and 800 °C show much more complicated patterns and are characterized by a 3-fold increase in stress and cyclical rises and drops (30-40%) in stress suggesting deformation by brittle fracturing. Run-products for samples deformed at those low temperatures are extensively fractured, often broken-up in smaller pieces, or exhibiting fractures (see picture insets). (B) Results of wet deformation experiments (PH2O: 1 to 5 MPa; T: —650 °C) compared to data from dry, high-T (900 °C) experiment (RS17; Fig. 3.3). The main effects of PH2O are to reduce the strength of cores and to permit ductile deformation at temperatures well below the effective Tg of the dry cores (see Fig. 3.7a and text). 58 A0.1 0.2 0.3 0.4 -T 3 c 0 1 00 0 Figure 3.7 See previous page for caption. 0.5 0.4 59q“ 3.5.2 Analysis of strain During these deformation experiments, total strain (es) is given by: = L0—L, (3.5) where L0 is the initial sample length and Lf = L0 — (total machine displacement). During deformation samples get shorter and increase in radius (Table 3.2). We calculate volume strain () using Eq. (3.4), and radial strain (Er) from the initial and final sample radius: 2 8r=1 (3.6) rf Most samples show uniform radial increase along their entire length, but bulging (greatest radial increase at the mid-point of sample length) is observed in run-products taken to high total strain (Et > 0.6). Figure 3.8 illustrates how the strain is progressively partitioned between volume (E) and radial strain (Er). At low values of total strain, the two metrics are sub-equal (Eq, - Et) and plot near the 1:1 line indicating that most of the observed strain is being accommodated by volume loss (Fig. 3.8a). However, as total strain increases the departure from the 1:1 line increases, as does the calculated amount of radial strain (Fig. 3. 8b). These patterns clearly show that the total strain, as manifest by shortening of the core, cannot be fully accommodated by porosity reduction (Em: volume strain) but requires radial bulging (Er: shear strain). The proportion of shear strain to volume strain increases with increasing strain (Fig. 3.8a-c). The combination of volume and shear strain (E + Er) can exceed the total strain as represented by shortening of the core (Fig. 3.8c-d). At low values of strain the combination of volume and shear strain are more or less equal to total strain represented 60 by shortening of the core. However at values of total strain above --0.5, the combination of + Er is greater than the total strain (data plot above 1:1 line; Fig. 3.8d) and the deviation increases with increasing e. These patterns indicate the nature of coupling between the two strain mechanisms. At low values of strain, where pore fraction>> solid fraction, total strain is mainly accommodated by volume loss and radial strain is minimal. There is also little to no evidence for coupling between these two strain mechanisms (i.e., volume vs. shear strain) and they may operate independently. However, at higher amounts of strain (Et> 0.6), where porosity < 0.6, strain is accommodated by volume loss and a significant component of radial bulging. This behaviour is also expressed in Figure 3.6 where the measured residual porosity departs from the model vs. E curves at Et > 0.6. Figure 3.8c shows the changes in proportions of E to Er with increasing total strain. At low values of strain E, is substantially greater than Er, however, the proportion of Er increases steadily with increasing strain. In fact, at values of Et> 0.5, the summation of and Er exceeds the total strain computed from shortening of the core (Fig. 3. 8d), suggesting that there is strong coupling between the volume strain and shear strain. Moreover, at Et > 0.8 the combined values of E,+ 6r fall above the iso-strain contour for 1.0 (Fig. 3.8c) and incremental increases in strain are dominated by radial expansion rather than by volume loss. The implication is that once there has been sufficient strain (Et 0.6) to reduce porosity to a critical value ( <0.6) subsequent compaction (shortening of core) is accommodated by porosity reduction (volume strain) and concomitant radial bulging 61 0.6 C) 0.4 0.2 CO 0 1 Figure 3.8 Analysis of strain in experimentally deformed cores. (A) Total strain () as recorded by piston displacement is plotted against the strain computed from porosity lost(s). Values of (volume strain) increase linearly with but are always less than total strain. (B) Total strain () plotted against strain ascribed to increase in cross-sectional area (i.e., radius) of the deformed core (Er) Values of Er are always smaller than total strain but increase markedly with increasing total strain. (C) Values of plotted against Er• Dashed lines are iso-EL contours (e.g., = E + (D) Values of plotted against the sum [ - Erj• Data plotting above the solid line suggest coupling of and Er (see Fig. 3.8c); the extent of coupling is proportional to the distance each point is above the 1:1 line and increases with total strain. :c 08 • • %. •%%% %. %. ‘% ‘% %. %. %.. ••% r . \ p.. %. •% % %. ‘% ‘% ‘7 •%. *%__“ •% • %.. ‘Ss/% S’S s ‘S ‘S ‘S ‘S ‘S ‘S ‘S ‘S ‘S ‘S ‘S ‘S 0.5 E 62 (shear strain). At this point, progressive strain comprises volume strain (porosity reduction) that is dependent on a component of shear strain; the degree of coupling between volume and shear strain is indicated by the upwards departure from the 1:1 line in Figure 3.8d. In summary, the high-T deformation experiments elucidate three potential strain regimes: (i) at low values of strain ( = < 0.5) where porosity > 0.6, most strain is accommodated volume strain and a subordinate amount of independent (or weakly coupled) shear strain; (ii) at intermediate values of strain ( --0.5-0.6), where porosity 50-60%, shear strain becomes increasingly important and volume and shear strain are at least weakly coupled; and (iii) at high values of strain (> 0.7) where porosity drops to below 40%, volume strain and shear strain are strongly coupled as evidenced by [Ec+Er]/Et> 1.0. 3.5.3 Effective viscosity The digital data recorded in each experiment provide load stress, total and incremental displacement (strain) at each time step and, thus, incremental and total strain rate. These data allow us to compute the apparent viscosity of the sample during deformation as a function of total strain (Fig. 3.9). The apparent viscosity (Tlapp) of the sample is the viscosity of the porous aggregate of volcanic ash at the experimental conditions, and is calculated as: r1app (3.7) where a is stress and is the total strain rate. In general, these experiments show the 63 12 Atm(Dry)at900°C ‘ o RS2O l 3) o I 0 — ‘0 0.2 0.4 0.6 0.8 1 P0=5 MPa (04 - 0 (1.0) C3 RS111 D. 659°C RSO4 °10 (1.5)) 9 (2.0) 647°C — D 0.8 1 0.2 0:4 0.6 0:8 Figure 3.9 Summary of apparent viscosity calculated as load stress over total strain rate, plotted as a function of total strain. (A) Calculated apparent viscosity from high- temperature (—9OO°C) dry series of experiments. Calculated apparent viscosity for lower temperature (—65O °C) experiments performed under water pressures of: (B) H9O = 1 MPa, (C) H2O = 2.5 MPa, and (D) H2o = 5 MPa. Open circles at = 1 represent model melt viscosity for water content in parentheses (wt.% H20). Grey gradient shading and dashed vertical line represent the onset of significant radial strain observed in run prod ucts. 0 0.2 0.4 0.6 0.8 1 12 0 0.2 0.4 0.6 64 porous cores of ash to have a strain-dependent behaviour where, under the constant displacement rate constraint, stress increases with increasing strain (i.e., strain hardening). The strain dependent rheology of these samples is a reflection of the porosity reduction due to compaction. The strain hardening is most pronounced at high values of strain (e.g., > 0.5; Fig. 3.3) where increases in strain cause high rates of porosity reduction (see Fig. 3.6). In our ductile experiments, the increase in stress with increasing strain can track the increase in apparent viscosity of the porous melt samples during deformation due to porosity reduction. We have demonstrated that deformation of our samples is expressed in, at least, two different ways: (i) volume strain due to porosity loss; and (ii) shear strain manifest by an increase in sample radius. We, therefore, recognize that the apparent viscosity values we have plotted in Figure 3.9 are a product both of volume strain and a shear strain, and that the contributions of these components varies as a function of total strain. Moreover, the total experimental strain rate will also comprise varying proportions of volume strain rate and shear strain rate. Volume strain rate dominates up to values of total strain of 0.6, whilst at total strain > 0.6, shear strain rate is expected dominate, and Eq. (3.7) becomes a cruder approximation of viscosity. The switchover between the two strain regimes is illustrated schematically in Figure 3.9 by a diffuse boundary (and shading) at -0. 6 total strain. Values of apparent viscosity calculated for samples deformed under dry conditions at high temperature are self-consistent and, again, demonstrate the reproducibility of the technique used in this study (Fig. 3.9a). Overall, porosity reduces the viscosity of the sample. The dry deformation experiments clearly show a strain 65 dependent behaviour that translates into a rise in apparent viscosity from 109.1 to 1011.9 Pa s over the full range of porosity reduction from 0.8 to 0.25. Over the interval 0 to 0.5 viscosity rises from 109.1 to 10100 compared to melted Rattlesnake Tuff which has a viscosity of 10102 at 900 °C. At values of strain > 0.5, Eq. (3.7) cannot be used to model viscosity accurately because of the substantial component of shear strain (radial increase). Post-experiment analysis of samples shows that the cores have water contents that generally exceed values predicted by standard 1120-melt solubility models (see Appendix 3.B). The measured values may represent a combination of chemically dissolved and mechanically trapped (e.g., nanopores) water. For the purposes of analysis we have assumed that during deformation the melt fraction of the samples contains, at a minimum, the H2O content predicted by the Newman and Lowenstem (2002) model. For a temperature of 650 °C, the model predicts values of 0.42, 0.67 and 0.95 wt.% H2O for 1, 2.5 and 5 JVJPa PH2O, respectively, The apparent viscosity of low PH2O (1 MPa) system is also clearly strain- dependent and shows an increase in apparent viscosity from iO” to 10106 Pa s over the Et interval 0 — 0.5. At higher values of strain the calculated apparent viscosity rises markedly due to dominance of shear strain and the breakdown of Eq. (3.7). At 1 MPa, the melt is expected to have 0.42 wt.% dissolved water and a melt viscosity at 650 °C of 1012 Pa s. It is apparent that the porosity reduces the effective viscosity by up to 2 orders of magnitude. These experiments probably could not be run effectively on the bubble free melt because these temperatures are below the glass transition temperature (Tg) of the melt (e.g., r 1012 Pa s). 66 The same effects are not observed at higher water pressures (2.5-5 MPa). Instead, these experiments suggest near constant values of apparent viscosity until Et exceeds 0.6. For example, the corresponding high-strain (e 0.75) experiments for 2.5 and 5 IVIPa PH2O show only slight rises in viscosity of 109598 and 10b0.0b02 Pa s, respectively over 0-0.5. Given the water contents predicted by Newman and Lowenstern (2002) for the experiments RS15 (0.67 wt. %) and RSO4 (0.95 wt. %) we expect melt viscosities at 650 °C of 10112 and iO’°5 Pa s. Using these limiting values for H20 solubility, it appears that for all water pressures the porous melts have substantially lower viscosities than their hydrated melt (non-porous) equivalents (see white circles in Fig. 3.9). At higher values of PH2O, the effects of strain hardening are greatly reduced (Fig. 3.3) and the apparent viscosity of the porous melts remains approximately constant over most of the deformation (i.e. compaction) history. 3.6 Discussion Our experiments explore the transient rheology of particulate porous natural melt from high (-4J.8) to moderate (—0.25) pore fractions under both atmospheric pressure conditions and at elevated water pressure. During sample deformation strain is accommodated by: (i) shortening of the sample core; (ii) reduction in sample porosity; and (iii) increase in the radius of the sample core. The strain is achieved via a combination of volume strain (reduction of pore space) and shear strain (radial expansion). The relative contributions of these two mechanisms to vary as a function of strain; at high strain (Et> 0.6) and relatively low pore fractions (<0.6) shear strain begins to dominate and both mechanisms are strongly coupled. Our unjacketed 67 deformation experiments are in that way analogous to deformation occurring in an unconfined ignimbrite sheet able to flow freely horizontally, perpendicular to the loading direction due to gravity. The deformation experiments clearly document the strongly strain-dependent rheology of these cores of volcanic ash. The rate of strain hardening increases rapidly as strain increases and probably mirrors the increasing role of shear strain as compaction proceeds. At high values of total strain, samples develop a strong foliation from alignment of glass and pumice shards, consistent with the major increase in shear strain. The presence of a fluid phase (i.e. PH20) appears to reduce the extent of strain hardening. We observe no apparent textural differences between samples deformed at dry, higher temperature (900 °C) conditions and lower temperature (650 °C) experiments performed at elevated PH2O. Figure 3.10 comprises thin section (A) and SEM photomicrographs (B) of sample RS17 resulting from dry compaction at 900 °C and to 75% strain. Corresponding images are shown (Fig. 3. lOc, d) for sample RSO4 which derives from an experiment run at -650 °C, under 5 MPa PH2O, and to 75% strain. The run products are indistinguishable from one another. This demonstrates that different experimental (i.e. environmental) conditions (Table 3.2) can produce distinct compaction paths (Fig. 3.3), yet yield virtually identical products. The main difference in the run products is their measured water contents: 0.15 wt.% for RS17 vs. 1.61 wt.% for RSO4 (see Table 3.B); the run-products have porosities of 37% and 41% respectively. This has implications for natural systems, wherein the features of welded volcanic deposits are used to deduce the nature of compaction and welding processes. Welding intensity in pyroclastic deposits is a reflection of emplacement conditions of the deposit 68 Figure 3.10 Textural comparison of samples run under dry and wet conditions. Scan of polished thin section (A), and SEM photomicrograph (B) for sample RS 17. Scan of polished thin section (C), and SEM photomicrograph (D) for sample RSO4. Both samples were deformed to 75% strain. Sample RS17 was deformed under dry conditions, at 900 °C, and sample RSO4 was deformed at = 5 MPa, and -65O °C. 69 (e.g., emplacement temperature and accumulation rate), physical and chemical properties of the materials (e.g., porosity, composition and water content of the melt), and dynamic feedbacks during welding (e.g., destruction of porosity, water resorption) (Smith, 1960a,b; Guest and Rogers, 1967; Riehie et a!., 1995; Sparks et al., 1999). These environmental parameters can combine in a multitude of ways to generate the same overall intensity of welding. Ideally, we hope that there are features that can be observed in the field that can be used to gauge the relative roles of these parameters (e.g., T, PH2O, load) (Grunder and Russell, 2005; Russell and Quane, 2005). The results above cast some doubt on this anticipation, in that material with virtually identical physical and textural properties has resulted from two distinct end-member processes: (i) hot dry compaction, and (ii) cool, wet compaction. Therefore, there is likely to be no unique solution for the conditions required to develop a specific welding intensity or facies. This insight serves to highlight the over-simplification of many early and existing models of the welding process in pyroclastic deposits and welding facies distribution, where temperature and load are the only conditions considered (e.g., Ross & Smith, 1961; Ragan & Sheridan, 1972). We suggest that models of welding zonation development, and critically, welding profiles (Reihle et al., 1995) from which porosity and permeability information is inferred, be re-examined to account for the effects of porosity and water pressure before, during and after welding has occurred. Part of this study demonstrates the pronounced effect of temperature on the rheological behaviour of these cores of volcanic ash. Under dry conditions, and a constant displacement rate of 2.5 i0 mm/s (strain rate - i0 s1), experiments conducted at 850 °C or higher produced stress-strain relationships consistent with viscous 70 deformation (Figs. 3.3, 3.7a). The same experiments performed at 800 °C or lower produced stress-strain patterns indicative of brittle relaxation (Fig. 3.7a). These data imply that, at the timescales of our experiments, the rheological glass transition temperature (Tg) (marking the temperature boundary between ductile and brittle behaviour) resides at between 800-850 °C (Fig. 3.1 la). The two viscous experiments (RS 16 and RS23) were used to extract values of effective viscosity at identical values of = 0.25 where the porosity is still very high (—0.75-0.78; Table 3.3). These values are plotted at their experimental temperatures (filled circles; Fig. 3.11) and used to define an Arrhenian curve having the same slope as the melt viscosity (open circles) and representing the temperature dependence of viscosity for dry porous cores of Rattlesnake Tuff ash. We have adopted and plotted a mid-range (e.g., 800 to 850 °C) value for the Tg of 825 °C (Fig. 3.1 la; dashed vertical line). The same viscosity data are plotted in terms of their characteristic relaxation timescales by scaling the melt viscosity to the bulk shear modulus (Dingwell, 1995). The intersection of the apparent Tg and the viscosity curve for the dry porous melt (Fig. 3.1 la) implies an average experimental timescale (texp) of —4 s (Fig. 3.1 lb). This is illustrated by a grey horizontal dashed line on Figure 3.1 lb. Where the porous sample has a characteristic relaxation timescale (‘tr) shorter than texp, the experimental response will be viscous. Conversely, at lower temperatures (i.e., <825 °C) samples will have values of tr that are larger (i.e. longer) than texp; under these conditions the rate of building stresses in the core (texp) is faster than the capacity of the sample to relax viscously (tr). This results in brittle failure of the sample (Fig. 3.7a, 3.1 lb). The horizontal arrow marks the intersections of the experimental timescale (texp) with viscosity curves for the Rattlesnake Tuff melt and the same melt with —75% porosity 71 Figure 3.11 Summary of variations in viscosity and relaxation time scale for the Rattlesnake Tuff melt resulting from temperature, dissolved water content and porosity. (A) Viscosity of the dry, porous Rattlesnake Tuff melt (taken at Et=O.25) as a function of temperature, and compared to the viscosity of melt alone. Solid line is based on experimental measurement of anhydrous melt (Robert et al. 2008); dashed line is an Arrhenian fit to the experimental viscosity data (this study). Low temperature, brittle experiments are represented by filled squares, and viscous, higher temperature experiments by filled circles. The vertical, dashed grey line is the effective glass transition temperature for the dry, porous system (825 °C). (B) Relaxation timescale (see text) of the dry, porous melt as a function of temperature. Experimental data and conditions as in (A). The characteristic experimental timescale (4 s; see text) is shown as a dashed grey horizontal line. The expansion of the viscous deformation field due to viscosity is illustrated by a grey arrow (see text). (C) Viscosity of the wet (PH2O=2.5 MPa), porous Rattlesnake Tuff melt (taken at Et=O.2S) as a function of temperature, and compared to the model viscosity (Giordano et al., In Press) of the hydrous melt (wt.% H20 in parentheses), for solubility of 0.67 wt.% at 650 °C and 0.78 wt.% at 550 °C (Newman and Lowenstern, 2002). The vertical, dashed grey line is the effective glass transition temperature for the wet, porous system (528 °C; see text). (D) Relaxation timescale of the wet, porous melt as a function of temperature. Experimental data and conditions as in (B). The characteristic experimental timescale (same as in C) is shown as a dashed grey horizontal line. The expansion of the viscous deformation field due to porosity is illustrated by a grey arrow (see text). 72 Figure 3.11 See previous page for caption. T(°C) 1395 975 725 12 8 4 16 U) 0 0 r 0) C U) ct 0 0 1 0) 0 102 100 U) -2 — 10 1 106 108 1 U) 1 108 1 0000/T(K) 6 8 10 1 0000JT(K) T(°C) 975 725 560 440 350 P0120) = 2.5 MPa — ;;;;;;- C.- :/ co — c’J , U, ._ — ,- — 8 4 8 10 12 14 16 8 10 12 14 16 1 0000/T(K) 1 0000/T(K) 73 (Fig. 3.1 ib); namely their respective glass transition temperatures. Under dry conditions, the addition of porosity expands the window for viscous deformation by -5O °C. In a similar manner, we have explored the effect of porosity on the viscosity of hydrated melts. The viscosity of hydrated melts are calculated (Giordano et al., In Press) for melts having fixed water contents consistent with their experimental conditions (solid lines, Fig. 1 ic): 2.5 IVIPa PH20 and 650 (0.67 wt.%) and 550 °C (0.78 wt.%). Parallel curves have been drawn through the experimental data points taken from the hydrous deformation experiments that showed a viscous response. These curves represent the temperature dependence of the effective viscosity of these hydrated porous cores. The two lower temperature hydrous experiments (--450 and —385 °C) that gave brittle responses are plotted as squares (Fig. 3.1 ic), and suggest that the rheological glass transition temperature for these hydrous cores resides between 550°C and 450°C. We have used the apparent average timescale of the experiment (Texp 4s) to constrain the Tg of the most hydrous sample to 528 °C (Fig. 3.1 lc). Under a water pressure of 2.5 IVIPa, the addition of porosity to these hydrated melts increases the field for viscous deformation by 140-150 °C (grey arrow in Fig 3.1 id). Increasing the displacement rate by 1 order of magnitude at 900 °C also pushed the material into the field of brittle behaviour (RS25). Increasing displacement rate causes an order of magnitude decrease in the experimental timescale (_.101 or 102 s) and the characteristic relaxation timescale of the sample staying constant. The result is that experiments that originally featured samples with tr % assumption is there is no change in area % displacement to strain (m to dimensionless) strain=dispmllim; %total strain %%% Fit for displacement rate % AX=B % X=A\B === the solution to this is the slope with a 0 intercept % ts=A % slope=X == I will call XT1disprate_fit” % dispm=B disprate_fit = ts\dispm % disprate_fit units are rn/s dispm_fit = ts.*disprate_fit; %this is the calculated displacement using the fit strain_fit = dispm_fit/lim; strain_rate_fit = disprate_fit/lim; 105 plot(dispm,dispm_fit, ‘-r’) pause plot(ts,dispm_fit, ‘-.b’) pause plot(ts, strain_rate_fit, ‘-g’) pause %%% Smooth stress (Pa) data plot(ts, stress_Pa, ‘Ok’) hold on windowSize=[ 1,2,3,5,7,15,20] for i=1:7 stress_Pa_filter=filter(ones( 1 ,windowSize(i))/windowSize(i), 1 ,stress_Pa); plot(ts,stress_Pa_filter,color(i)); disp(’size of filtering window’) windowSize(i) xxx=input(’accept filtering window by typing desired window size (reject by hitting return)’,’s’) if xxx == [1 continue elseif xxx > 0 break end end hold off stress_Pa_final=stress_Pa_filter; plot(strain, stress_Pa_final/1000000, ‘-k’); xlabel(’\epsilon_{ total }‘) ylabel(’\sigma (MPa)’) xlim([0 1]); disp(’Save figure?’); ANSWER=input(’hit RETURN for YES; any NTJMBER for NO ‘); if isempty(ANSWER) == 0; ‘do not save’ else saveas(gcf, input(’.fig file name? ‘,‘s’), ‘fig’) end 106 format bank visc_nolog=stress_Pa_final./strain_rate_fit; visc_filter=log 1 O(stress_Pa_final./strain_rate_fit); plot(strain, vise_filter, ‘-k’); xlabel(’\epsilon_{ total }‘) ylabel(’log_{ 1O}\eta_{eff} (Pa s)’) pause disp(Save figure?’); ANSWER=input(’hit RETURN for YES; any NUMBER for NO ‘); if isempty(ANSWER) == 0; ‘do not save’ else saveas(gcf, input(’.fig file name? ‘,‘s’), ‘fig’) end % compare viscosities obtained from “raw” data % and smoothed data visc_filter_max=log 1 0(max(stress_Pa_final)/strain_rate_fit); visc_max=log 1 0(max(stress_Pa)/strain_rate_fit); viscosity=[visc_max; visc_filter_max] pause 107