Complex conductivity of volcanic rocks and the geophysical mapping of alteration in volcanoes

Highlights

• Induced polarization is sensitive to the alteration of volcanic rocks.

• Induced polarization of volcanic rocks can be understood in terms of a dynamic Stern layer model.

• The effect of temperature and pyrite (or magnetite) content can be understood in terms of a simple model.

Abstract

Induced polarization measurements can be used to image alteration at the scale of volcanic edifices to a depth of few kilometers. Such a goal cannot be achieved with electrical conductivity alone, because too many textural and environmental parameters influence the electrical conductivity of volcanic rocks. We investigate the spectral induced polarization measurements (complex conductivity) in the frequency band 10 mHz–45 kHz of 85 core samples from five volcanoes: Merapi and Papandayan in Indonesia (32 samples), Furnas in Portugal (5 samples), Yellowstone in the USA (26 samples), and Whakaari (White Island) in New Zealand (22 samples). This collection of samples covers not only different rock compositions (basaltic andesite, andesite, trachyte and rhyolite), but also various degrees of alteration. The specific surface area is found to be correlated to the cation exchange capacity (CEC) of the samples measured by the cobalthexamine method, both serving as rough proxies of the hydrothermal alteration experienced by these materials. The in-phase (real) conductivity of the samples is the sum of a bulk contribution associated with conduction in the pore network and a surface conductivity that increases with alteration. The quadrature conductivity and the normalized chargeability are two parameters related to the polarization of the electrical double layer coating the minerals of the volcanic rocks. Both parameters increase with the degree of alteration. The surface conductivity, the quadrature conductivity, and the normalized chargeability (defined as the difference between the in-phase conductivity at high and low frequencies) are linearly correlated to the CEC normalized by the bulk tortuosity of the pore space. The effects of temperature and pyrite-content are also investigated and can be understood in terms of a physics-based model. Finally, we performed a numerical study of the use of induced polarization to image the normalized chargeability of a volcanic edifice. Induced polarization tomography can be used to map alteration of volcanic edifices with applications to geohazard mapping.

Introduction

Geophysical methods and the methods developed in the framework of hydrogeophysics are increasingly used on volcanoes to determine the distribution of material properties (e.g., seismic velocities, bulk density, electrical properties) with application to volcanic eruption forecasting, geohazard mapping, and hydrogeology. Electrical conductivity tomography can be used to image the 3D distribution of the electrical conductivity of volcanoes (e.g., Johnson et al., 2010; Revil et al., 2010; Rosas-Carbajal et al., 2016; Gresse et al., 2017). Electrical conductivity of volcanic rocks depends on two contributions. The first is a bulk contribution controlled by the water content and pore water salinity. It corresponds to electrical conduction through the connected pore network of the material (e.g., Archie, 1942). The second contribution is an interfacial contribution called surface conductivity. This contribution is associated with conduction in the electrical double layer coating the mineral grains (e.g., Revil et al., 2002 for volcanic rocks and Waxman and Smits, 1968 for sedimentary materials). This electrical double layer consists of the Stern layer (with weakly or strongly adsorbed counterions depending on their affinity for the mineral surface) and the diffuse layer. Since electrical conductivity of volcanic rocks depends on these two contributions, electrical conductivity tomography is rather difficult to interpret and therefore cannot be used as a stand-alone geophysical method (see discussion in Bernard et al., 2007; Komori et al., 2010; Kemna et al., 2012; Usui et al., 2016; Soueid Ahmed et al., 2018a, Soueid Ahmed et al., 2018b). For instance, a volcano can be a highly conductive body because of the high salinity of the pore water or a high degree of alteration (or both). Induced polarization can be used to discriminate these effects as long as the content in metallic particles (e.g., pyrite and magnetite) is not too high (i.e., 1% in vol.).

In the context of the present paper, it may be useful to recall what we mean by alteration. Classically, the alteration of volcanic rocks is produced by the circulation of hydrothermal fluids and involves the replacement of primary igneous glass and minerals (e.g., plagioclase, pyroxene, amphibole) by secondary minerals that are stable at the thermodynamic conditions of alteration (e.g., Bonnet and Corriveau, 2007). We are especially interested by the case where these secondary minerals are clay minerals (kaolinite, chlorite, illite, and smectite, see Honnorez et al., 1998). Note that some volcanic rocks can also be altered through surface weathering.

Complex conductivity characterizes the reversible storage of electrical charges in rocks (Schlumberger, 1920; Bleil, 1953; Seigel, 1959), a process known as (induced) polarization. This “polarization” is a low frequency (<10 kHz) characteristic of rocks that is unrelated to the dielectric polarization phenomena observed at higher frequencies (>10 kHz) (e.g., Revil, 2013). The origin of the low-frequency polarization of rocks is generally associated with the polarization of the electrical double layer around the mineral grains and can be described with two interrelated parameters: the quadrature conductivity and the normalized chargeability (see Revil et al., 2017b). The quadrature conductivity corresponds to the imaginary component of the complex conductivity. The normalized chargeability measures the dispersion of the in-phase conductivity with the frequency. Since the polarization and surface conductivity are both controlled by the properties of the electrical double layer, it is therefore unsurprising that these parameters are also interrelated (Revil et al., 2017b). This relationship is very important to interpret electrical resistivity and induced polarization tomographies that can be carried out at the scale of volcanic structures. As a side note, measurements of induced polarization can be performed in the field with the same equipment used for electrical resistivity tomography (e.g., Kemna et al., 2012). Induced polarization may therefore appear as an attractive method to study volcanic and geothermal systems.

The key questions we want to address are (1) Is the relationship developed in a previous paper for basaltic rocks (Revil et al., 2017a) valid for all types of volcanic rocks? (2) Is the cation exchange capacity (CEC) a proxy of the alteration of the volcanic rocks? (3) How do the surface and quadrature conductivities of volcanic rocks depend on their cation exchange capacity (CEC) and specific surface area? (4) How can the quadrature conductivity and the normalized chargeability be related to each other? (5) How is the polarization affected by temperature and the volume content of metallic particles? (6) Can we image electrical conductivity and normalized chargeability of a volcanic edifice and offer a combined approach to interpret these data? Answers to these questions are important to use induced polarization tomography to image the alteration of volcanic rocks. Since alteration is responsible for the weakening of the mechanical properties of volcanic rocks (see Pola et al., 2012; Frolova et al., 2014; Wyering et al., 2014; Heap et al., 2015), our study has strong implications regarding the mapping of geohazards in volcanic environments (Day, 1996; Reid et al., 2001; Finn et al., 2001; Reid, 2004).

We investigate here the complex conductivity of a set of 85 new core samples from five active volcanoes in the world: Merapi and Papandayan in Indonesia, Furnas in Portugal, Yellowstone in the USA, and Whakaari (White Island) in New Zealand. The laboratory measurements were collected in the frequency range 10 mHz–45 kHz with a very sensitive impedance meter. The interpretation of the laboratory data will be based on the dynamic Stern layer polarization model developed in Revil (2013) and recently updated in Revil et al., 2017a, Revil et al., 2017b, Revil et al., 2018.