Comment on “Rapid fluid disruption: A source for self-potential anomalies on volcanoes” by M. J. S. Johnston, J. D. Byerlee, and D. Lockner


  • A. Revil

    1. Department of Hydrogeophysics and Porous Media, European Center for Research in Environmental Geosciences, CNRS-CEREGE, Aix-en-Provence, France
    Search for more papers by this author

[1] Johnston et al. [2001] were concerned with a new generation mechanism (the “rapid fluid disruption” effect, RFD) to describe the self-potential anomalies measured above active volcanoes and geothermal systems. Self-potential signals represent the electrical potential distribution obtained at the ground surface using at least two nonpolarizable electrodes. One is kept at a reference station, and the other is used the scan the ground surface electrical potential. These anomalies result from some polarization mechanisms occurring at depth. Johnston et al. [2001] advanced an explanation of such polarization mechanism, which is based on a new set of experiments they developed in their laboratory. These experiments, which involve the injection of cooler liquid into very hot and dry rock samples, are new. The behavior of the system, where there is a large temperature gradient within the sample, and where the fluid undergoes a phase change, is different from earlier electrokinetics experiments [e.g., Ishido and Mizutani, 1981; Jouniaux and Pozzi, 1997; Lorne et al., 1999a, 1999b], which have been performed on water-saturated samples at uniform temperature below boiling. So it is reasonable to ask whether new mechanisms for generating electrical potentials may be at work. However, the authors claim that these experiments cannot be interpreted with the help of the electrokinetic explanation, in which the electrical field is generated by the fluids (gas and water) flowing in the connected pore space of the porous rock sample. They emphasize several times in their paper that the electrokinetic effect does not provide the correct polarity and amplitude to explain both field and laboratory observations.

[2] The origin of self-potential anomalies is an important subject because of the applications to monitor the internal activity of volcanoes an possibly forecast volcanic activity. While the subject is very exciting, I think that there are a number of major flaws in the interpretation provided by Johnston et al. [2001]. These flaws arise mainly from a basic misunderstanding of the electrokinetic effect. I show below that the electrokinetic explanation provides both the correct polarity and amplitude to explain self-potential anomalies recorded at the ground surface of active volcanoes and that the same phenomenon can be used to interpret the experiments reported by Johnston et al. [2001]. Consequently, this yields very different conclusions from Johnston et al. [2001].

[3] In their abstract, Johnston et al. [2001] claim that the electrokinetic effect is not able to provide the “correct polarity” of the self-potential field recorded at the ground surface of active volcanoes. They suggest that “the relative amplitudes of RFD potentials and electrokinetic potentials to be dramatically different and the signals are opposite in sign”. Furthermore, they claim in their abstract that “the RFD effect of positive sign in the direction of gas flow dominates.” It follows that according to Johnston et al. [2001] the electrokinetic effect usually generates a negative electrical potential in the flow direction. I can prove easily that the previous statements are wrong and they probably result from some confusion of the authors about the electrokinetic effect. Indeed, in most natural conditions the excess of charge located in the vicinity of the pore water/ mineral interface is positive when the minerals are silicates or alumino-silicates (see Figures 1 and 2). This positive electrical charge in the electrical diffuse layer is extremely well documented [e.g., Avena and De Pauli, 1996; Lorne et al., 1999a, 1999b; Pengra et al., 1999; Revil and Leroy, 2001]. The drag of the positive charge of the electrical diffuse layer with the pore fluid flow generates a positive electrical potential in the flow direction. This fact has been checked by laboratory experiments (e.g., Ishido and Mizutani [1981] and Jouniaux et al. [2000] for silica and volcanic rocks), theoretical investigations [Pride, 1994; Revil et al., 1999a, 1999b; Lorne et al., 1999a, 1999b; Revil and Leroy, 2001], and field data [e.g., Trique et al., 1999b].

Figure 1.

(a and b) Zeta-potential in clays versus pH. Experimental data are from 1, Poirier and Cases [1985]; 2, Avena and De Pauli [1996], 10−3 M; 3, Avena et De Pauli [1996], 4 × 10−3 M; and 4, Lorenz [1969] (0.6–2) × 10−3 M. (c) At pH > pH(pzc), the fixed charge occurring at the surface of alumino-silicates is negative. The counter charge located in the electrical diffuse layer is opposite in sign to the surface charge.

Figure 2.

Sketch of the triple layer model (TLM) at the water/mineral interface. The locked charge of the clay minerals (surface charge plus the charge due to the isomorphic substitutions) is compensated by mobile charges present in the electrical double layer in the vicinity of the mineral/water interface. The double layer includes the Stern layer of adsorbed counterions and the diffuse layer. The potentials φ0, φβ, and φd of the triple layer model are the surface potentials in the 0 plane, the potential at the β plane, and the inner potential of the electrical diffuse layer in absence of external applied electric field, respectively; C1 and C2 are the electrical capacitances assumed to be constant in the regions between planes. The outer Helmholtz plane (OHP) characterizes the inner part of the electrical diffuse layer. A negative zeta-potential implies a negative fixed charge at the mineral surface. The net charge in the diffuse layer is opposite in sign to the surface charge due to electrostatic interactions between the fixed charge and the ions from the electrolyte.

[4] The fact that the electrokinetic mechanism produces a positive electrical potential in the flow direction can be understood in terms of electrical double (or triple) layer theory. The electrical double layer is a generic name used to describe the electrochemical interactions occurring at the interface between a mineral and the pore water. All the minerals react in water. Indeed, there are chemical reactions between the hydroxyl surface groups (e.g., silanol > SiOH and aluminol > AlOH) present at the surface of silicates and alumino-silicates (like clays and zeolites) and the pore water. These sites can accept or loose protons, and some chemical compounds can get adsorbed on these sites. As a result of these reactions, the surface develops a net charge. At a pH greater than a critical value denoted pH(pzc) (where pzc is for “point of zero charge,” see Figure 1), the surface charge of silicate and alumino-silicates is negatively charged and the zeta-potential is negative. This fixed surface charge is counterbalanced by a mobile charge in the so-called electrical diffuse layer (Figure 2). The diffuse layer has a net charge of opposite sign to the surface charge of the mineral (to maintain a global electroneutrality in the system). The flow of pore water drags most of the mobile positive charge of the electrical diffuse layer in the flow direction. Note that the situation is opposite for pH below the point of zero charge, which is characterized by a net surface charge equals to zero. Therefore, it follows that in most natural conditions corresponding to pH > pH (pzc) ∼ 0–5, the electrical potential generated at the macroscopic scale is positive in the flow direction. In hydrothermal systems associated with active volcanoes, pH can range from low (∼2) to high (>10). However, it is generally in the range 5–8, and therefore the zeta potential is usually negative.

[5] In two-phase flow conditions (flow of gas and water), there is an enhancement of the generated electrical field as shown by Revil et al. [1999b] except if the water saturation is below a critical level corresponding to the irreducible water saturation. This has been recently confirmed by laboratory experiments (L. Jouniaux, personal communication, 2001).

[6] At the macroscopic scale, the intensity of the electrokinetic coupling is described by the electrokinetic coupling coefficient defined by C = Δψ/Δp, where Δψ/L is the macroscopic (measurable) electrical field generated by the flow induced by the pore fluid pressure gradient Δp/L (L is the length of the sample, ψ is the electrical potential, and p is the pore fluid pressure). For the electrokinetic effect, sign(C) = sign (ζ) (where ζ is the zeta potential shown in Figures 1 and 2 at the interface between the mineral and the pore water). Therefore ζ < 0, C < 0, and this yields Δψ ∼ (−1) Δp, so the electrical potential is positive in the flow direction because in the flow direction, Δp < 0.

[7] In contrast, Johnston et al. [2001] claim the opposite is true, namely, that the electrical potential generated by the electrokinetic phenomenon is negative in the flow direction. I suggest, therefore, that their comments on the application of the electrokinetic effect to explain the self-potential pattern in active volcanoes are wrong. As I mentioned previously, there is no doubt that the electrical potential produced by the electrokinetic effect is positive in the flow direction. There are at least two types of flow on an active volcano. The first one is due to the thermal convection of the groundwater due to the thermal buoyancy generated by the heat associated with the presence of magmatic bodies inside the system. This creates vorticities in the flow pattern and an upflow of hydrothermal fluids above the magmatic body in the central part of the system. A positive self-potential anomaly is generated by this upward flow.

[8] A second type of flow is gravitational in nature and due to the topography of the water table. This gravitational flow pattern generates negative potential anomalies on the flanks of the volcano and positive self-potential anomalies at the base of the volcano as again an excess of positive charge is dragged along with the pore fluid flow. This explain very well the pattern of the self-potential-anomalies observed on the Miyake-Jima Volcano (see their Figure 6) as discussed by Revil et al. [1999b] and Revil et al. [2002]. Johnston et al. [1999b, p. 4327] wrote “many of the recent models…postulate complex groundwater circulation systems…for which, there are no supporting field observations.” I cannot agree with this sentence. The flow pattern used by all the authors cited by Johnston et al. [1999b] represent our current understanding of the flow pattern in active volcanoes such as La Fournaise volcano (Réunion Island [see Michel and Zlotnicki, 1998, and references therein]). This is also consistent with the flow pattern simulated numerically around a dike injection [see, e.g., Revil et al., 1999b, Plate 1]. So I return to my first statement: The polarity produced by the electrokinetic effect is consistent with field observation.

[9] To obtain an order of magnitude estimate of the efficiency of electrokinetic conversion, we can compute the effect of a rapid temperature increase in a water-saturated rock volume upon the intensity of electrical potential generated in drained conditions. For this purpose, I introduce a sensitivity coefficient CT between the electrical potential generated through electrokinetic coupling and temperature:

equation image

where J represents the total electrical current density (in A m−2), p is the pore fluid pressure (Pa), T is the temperature (in K), and ψ is the electrical potential (in V). The coefficient CT represents the electrical potential increase through electrokinetic coupling per unit degree Celsius of temperature increase in a water-saturated porous material when the pore water is permitted to drain freely. According to thermoporoelastic theory [e.g., Palciauskas and Domenico, 1989], the variation of pore fluid pressure with temperature is given by

equation image

where the Biot's coefficient are defined by

equation image
equation image
equation image
equation image

where mf represent the mass of pore water per unit rock volume in a reference state, σ = (σ11 + σ22 + σ33)/3 is the mean confining stress, ϵ is the bulk deformation, β is the bulk compressibility of the porous material, βS is an isothermal grain compressibility, ξ is the Biot coefficient, and αf is the isobaric thermal expansion coefficient for fluid (in °C−1) [see Palciauskas and Domenico, 1989]. The coefficient 1/R represents a measure of the change in water content for a given change in pore fluid pressure when the porous material is permitted to drain freely, 1/Q represents a measure of the amount of water which can be forced into a porous material under pressure while the volume of the material is kept constant, and αm (in °C−1) describes how much fluid mass of water is expelled out of the porous material when the temperature is increased by (TT0). According to equations (1) and (2), CT can be evaluated with

equation image

where C (<0) is the electrokinetic coupling coefficient discussed above. Basaltic and mudstone-type of materials would form the two end-members materials (hard and soft), which are generally forming the volcanic edifice. Palciauskas and Domenico [1989] give R αm = 0.88 and 0.18 MPa °C−1 for a basalt and a mudstone, respectively, and Qb ξ + αm) = 1.03 and 1.14 MPa °C−1 for a basalt and a mudstone, respectively. For a basalt, we obtain CT ∼ −10 mV °C−1 with a pore fluid conductivity σf = 1 S m−1 (corresponding to a highly mineralized water) and CT ∼ −100 mV °C−1 with a pore fluid conductivity σf = 0.1 S m−1. Therefore a temperature change of 300°C by comparison with the background temperature would lead to an electrical source term comprised between −3 to −30 V. Therefore the thermohydromechanical coupling represents an extremely efficient process to generate transient self-potential signals in active volcanoes. From the previous considerations, it is reasonable to assume that the electrokinetic effect can generate huge positive self-potential anomaly in the flow direction.

[10] I want to point out that I have no objections to the use of the RFD mechanism to explain part of the self-potential anomalies observed at the surface of active volcanoes. However, I think that the best candidate to explain the pattern, polarity, and intensity of self-potential anomalies in active volcanoes is the electrokinetic mechanism. In addition, I also point out that whereas the electrokinetic mechanism can be quantified from microprocesses and rock texture [Pride, 1994; Lorne, 1999a, 1999b; Revil et al., 1999a, 1999b], Johnston et al. [2001] failed in providing a way to quantify the so-called RFD effect. In addition, I will show now that most of the experimental results obtained by Johnston et al. [2001] can be also interpreted in terms of electrokinetic coupling rather than in terms of RFD mechanism.

[11] On the basis of my understanding of the experiments reported by Johnston et al. [2001], the water initially in the pore space of the sample is pushed out by the gas pressure set up in the upper reservoir. There is therefore a flow of the pore water out of the sample as it is “pushed out by the gas pressure” according to the authors themselves. A simple electrokinetic-based mechanism is consistent with the experiment results reported in their Figure 5. Indeed, a positive self-potential anomaly is expected in the direction of fluid flow as long as there remains water in the pore space of the sample. When all the water has been expelled from the sample, the flow of gas alone is unable to generate any self-potential of electrokinetic nature because the water saturation decreases below the irreducible water saturation [see Revil et al., 1999b, section 3.5]. About this experiment, Johnston et al. [2001, p. 4331] write “This forced removal of water molecules from the rock generates a positive potential… The voltage pulse generated by the gas flow was not at all surprising, since it is similar to the process of charge generation during spray electrification.” Actually, there is no need of any RFD mechanism to explain the data reported in their Figure 5. The coupling coefficient (Δψ/Δp) resulting from the measurements reported in their Figure 5 is roughly −100 mV MPa−1. Note that the minus sign is coming here from the fact that the pore fluid pressure and the electrical potential are measured at the two opposite sides of the sample (see the experimental setup of Johnston et al. [2001, Figure 2]). The intensity of the coupling is consistent with the range reported by Jouniaux et al. [2000] (−35 to −4900 mV MPa−1) for the electrokinetic coupling coefficients of volcanic materials and the fact that C is nearly temperature independent according to the measurements made by Revil et al. [2002].

[12] There is actually a way to test the nature of the signal recorded by Johnston et al. [2001]. It is known that the presence of alumina ions has a dramatic impact upon the voltage produced by electrokinetic mechanisms [e.g., Ishido and Mizutani, 1981; Lorne et al., 1998]. Johnston et al. [2001] should have repeated the experiments with alumina-doped brines with a small concentration of Al3+ ions (say 10−4 or 10−3 mol L−1). If Johnston et al. [2001] are correct, no change could occur in the measured potential. If I am correct, the amplitude of the generated potential will be seriously altered or even the polarity of the signal could be reversed due to strong adsorption of alumina at the surface of minerals (indeed, this adsorption generates a strong variation of the zeta potential, which controls the electrokinetic coupling effect).

[13] As electrokinetic phenomena produce a positive electrical potential in the flow direction, there is no problem to explain the experimental results of Johnston et al. [2001, Figures 3 and 4b] in light of this mechanism. Not only is the polarity consistent, but so is the magnitude of the phenomenon. The magnitude of the coupling mechanism is defined by the ratio between the voltage and the pressure differential. The coupling coefficient determined from the data reported in Figure 3 of Johnston et al. [2001] is −800 mV MPa−1 when the temperature has decreased to 100°C and the water is liquid This is consistent with the range reported by Jouniaux et al. [2000] (−35 to −4900 mV MPa−1 for water-saturated samples). In Figure 3, I have reported electrokinetic coupling coefficient of various rocks versus the electrical conductivity of the pore fluid used for the experiment. It can be be inferred from Figure 3 that the electrokinetic coupling coefficient is mainly controlled by the electrical conductivity of the pore water and not so much by the texture and lithology. The value −800 mV MPa−1 is inside the range of values obtained in rock samples with a relatively low pore water conductivity as used in the experiments reported by Johnston et al. [2001]. Therefore the voltages (polarity and intensity) shown in Figures 3 and 4b of Johnston et al. [2001] are consistent with the electrokinetic effect. There could be a RFD mechanism present, but the experiments provided by the authors do not allow the reader to discriminate between both effects. The experimental setup used by Johnston et al. [2001] is thus poorly designed to discriminate between competing effects. Note that there is also a small problem with their Figure 3. It seems that the temperature trend shown in the magnification of the lower part of the figure is not consistent with that showed in the upper part.

Figure 3.

Electrokinetic coupling coefficient versus pore water electrical conductivity for various rocks. (1) Crushed basalts from ODP Hole 395A (A. Revil and D. Hermitte, unpublished work, 2001), (2) crushed Fontainebleau sandstones [from Lorne et al., 1999a], (3) Berea and Bandera sandstones [from Pengra et al., 1999], (4) carbonates [from Pengra et al., 1999], (5) glass beads [from Pengra et al., 1999], and (6) consolidated zeolitized volcaniclastic rock samples [Revil et al., 2002]. A total of 83 measurements are reported here.

[14] The voltage reversal shown in Figure 4a of Johnston et al. [2001] is interesting, and I think that it can be again understood in terms of electrokinetic effect. When the pore water (liquid) enters into the heated sample, part of the pore fluid is vaporized. As the volume of water injected into the sample is higher here than in the experiment reported in Figure 3 of Johnston et al. [2001], it appears that only a fraction of the pore water is vaporized in this experiment. Such a phenomenon increases strongly the local fluid pressure due to expansion of the vaporized water as the vapor phase would take more space than the liquid phase. Such a local increase of the pore pressure can reverse temporarily the pressure gradient. This would explain the second fluid pressure increase at t = 17 s [Johnston et al., 2001, Figure 4a, bottom]. Such a fluid pressure reversal would also lead to an electrical potential reversal according to the electrokinetic effect, which is observed in Figure 4a of Johnston et al. [2001].

[15] There is a final point that I believe to be wrong in the paper by Johnston et al. [2001]. Johnston et al. [2001, p. 4329] note “in volcanic rocks where the temperatures at even moderate depths exceed 100°C, resistivity will approach 106 ohm m.” All electrical resistivity tomography performed to date in active volcanoes lead to the conclusion that below the water table, the volcanic materials are quite conductive with electrical resistivity much below 106 ohm m. Actually, experiments by Revil et al. [2001] showed that surface conductivity associated with the pore water/mineral interface is very high in altered volcaniclastic materials (≫ 10−3 S m−1). It follows that even a small amount of moisture is enough to make the porous rock relatively conductive. In active volcanoes, the value proposed by Johnston et al. [2001] (106 ohm m) is totally unrealistic in most cases accounting for the conditions in which water is vaporized in pressured environments.


[16] This work is supported by the French National Research Council (CNRS) and the Ministère de la Recherche et de l'Education Nationale (MENRT, ACI-Jeune 0693 to young scientists, 1999, and ACI “Eau et Environnement 2001”). I thank the two referees for their very helpful comments.