Experimental studies of seismoelectric conversions in fluid-saturated porous media

Transcription

1 JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 105, NO. B12, PAGES 28,055-28,064, DECEMBER 10, 2000 Experimental studies of seismoelectric conversions in fluid-saturated porous media Zhenya Zhu, Matthijs W. Haartsen, and M. N. Toks6z Earth Resources Laboratory, Department of Earth, Atmospheric, and Planetary Sciences Massachusetts Institute of Technology, Cambridge, Massachusetts Abstract. When seismic waves generate a relative fluid-solid motion in a fluidsaturated porous medium, the moving charges (streaming current) in the electric double layer induce an electroma. gnetic (EM) field. This paper first experimentally confirms that the coupling between the seismic wave and the electromagnetic field in the kilohertz range is electrokinetic in nature. Seismoelectric signals are measured in homogeneous cylindrical porous rock samples and multilayered models. The seismoelectric signals in homogeneous rock are electric fields that move along with the acoustic wave. The mechanism of the seismoelectric conversion is completely different from the piezoelectric effect of quartz grains. The seismoelectric sensitivity with respect to salinity of the saturant has been experimentally determined. The amplitude of seismoelectric signals increases as the saturant conductivity decreases. The seismoelectric effects are generated by two different mechanisms. Both the EM radiation and the electric potential generated at an interface and within a porous medium, respectively, were measured as the P wave, at ultrasonic frequencies, passes through the layered models. Our experimental results demonstrate that seismoelectric effects exist and are measurable in the kilohertz range. The paper concludes with a comparison of experimental data and modeled data in a threelayer porous model. Seismoelectric measurements could be an effective means of obtaining transport coefficients such as hydraulic permeability and other porous rock properties. 1. Introduction When a porous material is saturated with a fluid electrolyte, an electric double layer is formed at the boundary between solid and fluid. One layer is a bound layer, the other is a diffuse-charge layer with some mobile charges. When a seismic wave propagates in a fluidsaturated two-phase material, a relative motion of the charges is generated in the double layer. Moving charges induce an electromagnetic field. On the other hand, if an electric field is applied at both ends of a fluidsaturated porous rock sample, the collision between the solid and fluid with mobile charges generates a seismic wave, These phenomena are referred to seismoelectric and electroseismic conversions. In a homogenous porous material the conduction current exactly balances the streaming current induced by the seismic wave. The induced field is simply a sta- tionary electromagnetic field that moves along with the seismic wave [Pride and Haartsen, 1996]. If seismic waves traverse a contract in electrical and/or mechanical properties, the imbalance in the currents induces an electromagnetic wave that can be measured remotely [Haartsen and Pride, 1997]o Field data have been recorded demonstrating that such seismic to EM conversions may be detected in the near-surface sedimentary layers of the Earth [Martner and Sparks, 1959; Thompson and Gist, 1991, 1993; Butler et al., 1994, 1996; Mikhailov et al., 1997]. Laboratory experiments have been performed to measure streaming potentials caused by a pressure gradient through a porous rock and to determine the zeta potential or enhanced conductivity through the electrical double layer along the grain surfaces [Morgan et al., 1989; Jouniaux and Pozzi, 1995]. Zhu et al. [1999] investigated the electrokinetic conversions in fluid-sa- turated borehole models. Measurements of transient streaming potentials were performed in the quasi static 1Now at Shell UK Exploration and Production, Aberdeen, limit by Chandler [1981]. Dynamic measurements of Scotland, UK streaming potentials and streaming currents are discussed by Packard [1952], Groves and Sears [1975], Copyright 2000 by the American Geophysical Union. Sears and Groves [1978], Cerda and Kiry [1989], and Paper number 2000JB Zhu et al. [1994]. Their experiments are based on si /00 / 2000 JB $ nusoidally alternating fluid flow through capillaries and 28,055

2 28,056 ZHU ET AL.: SEISMOELECTRIC CONVERSIONS porous media to measure electrokinetic phenomena at low frequencies. In order to investigate the seismoelectric effects and to compare them with theoretical predictions, we set up an experimental system with an ultrasonic source and electrode receiver. The seismoelectric fields are measured with cylindrical and layered models in the kilohertz frequency range. Experiments were made to study the relationship between the amplitude of the seismoelectric field and the exciting acoustic field or the resistivity of the fluid-saturated porous medium. Our studies focus on the two kinds of the seismoelectric effects, namely, the radiating electromagnetic waves and the stationary seismoelectric field accompanying the acoustic waves within a homogeneous porous medium. The experimental results in layered models are compared with theoretical modeling. Because the seismoelectric effects are closely related to the fluid properties and flow in a porous medium, the seismoelectric measurements could be very important for directly investigating fluid-saturated formations in petroleum geology. This could be done both at the surface by recording the seismic-to-em converted waves generated at some interfaces at depth, and in borehole by measuring the seismoelectric potentials generated around the borehole formation. Seismoelectric mea- surements in a borehole could become a new logging method for the petroleum industry. 2. Experiments in a Cylinder Model Figure I shows a diagram of the experimental setup. A cylindrical sample saturated with water in a vacuum system is placed between the acoustic source and receiver transducers. An acoustic wave is excited in a cylinder (Berea or Coconino sandstone), which is 1.25 cm in diameter and 10 cm in length. Table I lists some properties of water-saturated sandstones and other materials. Most of the parameters in Table I were measured in our laboratory, except the permeability of the sandstones. The electrodes used for the electric field measure- ments are made of conductin glue (silver epoxy). They are 0.2 cm in diameter and separated by I cm. A wire with a sharp pin-head touches the conducting glue on the sample surface along its axis and measures the electric potentials. The electric signals are sent through an amplifier and filter and are digitally recorded. A metal film shields the whole sample to prevent outside electric influences. Because this cylinder is very slender, a pure extensional or flexural wave can be generated by a P or S wave source. The acoustic transducers are driven by a square electrical pulse with an amplitude of 500 V. The width of the pulse is adjustable, and it is usually adjusted to the half period of the recorded acoustic wave. We measured the seismic motions of the cylindrical sample with a small compressional transducer and calculated the velocity from the slope of the first arrival of recorded acoustic waves. When the source is a P wave transducer, the velocity of the acoustic wave propagating along the cylinder is 2400 m/s at about 18 khz. When the source is a S wave transducer, it is about 1150 m/s. Because the rock sample is a slender cylinder, the piston vibration of a P source at the end of the cylinder mainly generates a low-frequency extensional wave. Computer Preamplifier Oscilloscope (DATA 6000) ;. lcml I I I High Voltage --- Generator I I I 1 I Wave Source Figure 1. Diagram of the experimental setup for measuring the electric field generated by an acoustic signal in a rock sample. The rock sample is a cylinder, 10 cm in length and 1.25 cm in diameter.

3 ZHU ET AL.' SEISMOELECTRIC CONVERSIONS 28,057 Table 1. Physical Parameters of the Water-Saturated Samples Sample Density, g/cm 3 Porosity, % Vp, m/s Vs, m/s Permeability, Darcy Berea Sandstone Coconino Sandstone Glued Sand I Glued Sand II Lucite The velocity of the extensional wave is lower than the P velocity. The shear vibration of a 5' transducer mainly generates a flexural wave whose velocity is lower than the Rayleigh wave phase velocity. According to these measured velocities the main modes excited by P wave and 5' wave transducers should be the extensional and flexural waves, respectively. The center frequencies of these acoustic waves are about 20 khz, even though the normal center frequencies of the P and $ transducers are 250 and 500 khz, respectively. Therefore there are still some high-frequency components with smaller energy and propagating at P or 5' velocity along the cylinder. Figure 2a shows the seismoelectric signals generated by a compressional wave in the water-saturated Berea sandstone cylinder. The top trace (dotted line) in the plot is the acoustic wave received by the compressional transducer at the other end of the cylinder. The center frequency of these signals is approximately 18 khz. The phase velocity determined by the slope of the signals is about 2400 m/s, which is close to the extensional travel velocity. From the travel time of the first arrivals we know that the signals are not the electric influence of the source pulse. Comparing these signals with the acoustic waveform received at the other end of the sample (the top trace in the plots), we can see that the shape and the arrival time of the acoustic wave approximately coincide with those of the electric signals. They are not exactly in coincidence because the receivers, transducer and electrode, are different, and the acoustic transducer may receive more acoustic modes propagating along the rock cylinder. The comparison between the acoustic wave and electric signal shows that the electric signals are generated by the extensional wave propagating along the cylinder sample. Figure 2b shows the electric signals, generated by the shear transducer, received by electrodes at the surface of the cylindrical Coconino sample. The center frequency and the phase velocity of the signals recorded in Figure 2b are 27 khz and about l150 m/s, respectively. This velocity is close to that of the flexural wave in the cylinder. Even though different orientations in particle motions are excited by extensional and flexural waves, both modes generate seismoelectric signals in a fluidsaturated porous sandstone. The received electric signals are related only to the acoustic wave arriving at the point of measurement. In other words, the electric signals are not received at other points where the medium is not disturbed by the acoustic wave. This phenomenon confirms that the received electric signal is a stationary, localized potential, which does not propagate independently within the 10 (a) Electric signal Acoustic wave../ \ Time (psec) Electric signal Acoustic wave I loo pv 250 o (b) Time (psec) Figure 2. (a) Seismoelectric signals generated by a P wave transducer in a water-saturated Berea sandstone sample. The top trace (dotted line) is the acoustic waveform (extensional wave) received by a P wave transducer at the other end of the sample. (b) Seisrnoelectric signals generated by a S wave transducer in a water-saturated Coconino sandstone sample. The top trace (dotted line) is the acoustic waveform (flexural wave) received by an S wave transducer at the other end of the sample.

4 28,058 ZHU ET AL' SEISMOELECTRIC CONVERSIONS sandstone cylinder. An acoustic wave in a homogeneous porous sample does not act as a radiating antenna since the induced streaming current and conduction currents within the pulse cancel each other [Pride and Haartsen, 1 ] Test of the Seisrnoelectric Origin of the Signals We performed a series of experiments to confirm the seismoelectric origin of the signals. The experiments were designed to rule out the possibility of experimental errors. The first measurement was taken on a dry sandstone sample using the setup shown in Figure 1. The other experiments used layered models to determine if seismoelectric signals can be generated at interfaces between Lucite and air, water, dry porous rock, or unconsolidated sand, respectively. Piezoelectric coupling needs to be distinguished from the seismoelectric conversion. The quartz grain is a piezoelectric crystal. When quartz is strained, an electric potential is induced along its electric axis. Because the orientations of the quartz grains are not exactly in the same direction, the piezoelectric effect is very weak in a sandstone block, and the phase and amplitude of the piezoelectric field depend on the distribution and the orientation of the quartz grains in each part of the sandstone z > loo ß 80 = 60 E ß t 40 2O I I I I I W, ater-saturated i " i i I! ' 0 1 O (a) Voltage (V) [ ß.. i... ß ß,!.! ;, i... Berea :.!...--' ].. o--coconin, 0.., i,... :... ;--; '-... ::'7... :......? :r ;./..:_...,,'". ' i ,. - ' : :. ß! : ; : Figure 3 shows the electric signals received at the "... ;...!... i '. i i : ' surface of a dry Berea sandstone sample excited by a P! I I i I ' i I i! ': i ß wave transducer (Figure 1). The Berea sandstone was ''400' 6 )0' 800' '1000 dried by heating it in a vacuum oven for 8 hours. From (b) Resistivity ( *m) the moveout of the first arrivals in the signals we know that these electric signals are generated by the acoustic wave. Not only the amplitudes but also the phases of Figure 4. (a) Normalized amplitudes of the seismoelectric signals generated by a P wave transducer in the these first arrivals vary at different measurement points water-saturated Berea sample as a function of the voltalong the cylindrical sample. The maximum amplitude age exciting the P wave transducer. (b) Amplitudes of is less than 20 ttv, which is much smaller than the the seismoelectric signals generated by a P wave trans- ducer in the water-saturated Berea and Coconino rock samples as a function of the resistivity of the watersaturated sample. I20 pv Time (pse½) Figure 3. Electric signals generated by a P wave transducer in a dry Berea sandstone sample. The phases and amplitudes of these first arrivals inconsistently vary with the measurement points. seismoelectric signals measured in the wet rock sample, namely, 100/ V in the Berea sandstone and 40/ V in the Coconino sandstone. The different phases of the first arrivals in Figure 3 confirm that this phenomenon is not a problem of coupling between the transducer and the sample. Bad coupling decreases the amplitudes of the acoustic or electric signals but does not change the phase of the received signals. The phase velocity calculated from the first arrivals in Figure 3 is about 3100 m/s, which is larger than the velocity of an extensional wave and is close to the P wave velocity in the rock sample. The waveforms in Figure 3 are piezoelectric signals induced by the P wave propagating along the sample. The comparison of Figure 3 with the seismoelectric signals in Figure 2a shows that the mechanism

5 ß ZHU ET AL. ø SEISMOELECTRIC CONVERSIONS 28,059 Lucite P-wave 5 ø Source (fo= 100 khz) E 2 cm U (a) 25 cm Acoustic Receiver Position 5 10 Electrode Position 5 10 I,, O.O5. _ (b) 020-J! Figure 5. (a) Diagram of a layered model with (top) a Lucite block and (bottom) a Berea sandstone sample. The seismoelectric responses of a P wave source are measured at the interface between the two layers. (b) Acoustic waveforms received by a small P wave transducer at the bottom surface of the Lucite block before the rock block is placed against the bottom surface. (c) Seismoelectric signals recorded with a time delay of 10/ s at the interface between the Lucite block and water-saturated Berea sandstone sample. The amplitude is normalized by 100/ V. of seismoelectric conversion in a fluid-saturated porous medium is different from the piezoelectric effect. In the layered models made of two layers of materials, a Lucite block is used to eliminate the interference of the source pulse with the electric signal. We checked the electric field generated by the P wave on the Lucite surface. No electric signal was received at the interface between the Lucite and air or water (free surface). We also measured the electric fields at the interfaces where the Lucite block contacts dry Berea sandstone or watersaturated, unconsolidated, loose sand. In all of these cases we did not receive any electric signal above the system noise level. The experiment with the loose sand indicates that the seisrnoelectric effect needs a porous frame to have a relative fluid flow generated in the pore space by the seismic waves. In the loose sand case, the grain displacement and the relative displacement between solid and fluid are small at high frequencies, so the induced electric signals are very weak Seismoelectric Signal Sensitivity to Rock Resistivity The dependence of the seismoelectric signal amplitude on the resistivity of the rock sample is experimentally investigated using the setup shown in Figure 1. Figure 4a shows the relationship between the voltage of the electric pulse exciting the P wave transducer and the amplitude of the seismoelectric signal normalized by its maximum. The plot shows that the amplitude of

6 28,060 ZHU ET AL.: SEISMOELECTRICONVERSIONS the seismoelectric signals is directly proportional to the made, with electrodes placed inside the porous layers, amplitude of the acoustic waves. to measure the seismoelectric fields generated at inter- The resistivity of a fluid-saturated porou sandstone faces and within the porous medium. The purpose of varies with the molarity of the saturating fluid, and the this section is to measure the conversions when seismic amplitude of seismoelectric signals varies with the resis- waves traverse contrasts in medium properties. When a tivity of the sandstone. The amplitude dependence with seismic pulse straddles an interface, a dynamic current respect to the resistivity of the fluid-saturated Berea or imbalance is created, resulting a localized charge sep- Coconino sandstone is shown in Figure 4b. When the aration across the interface. This charge separation acts resistivity of the sample increases, the amplitude of the as a source for independently radiating EM fields that seismoelectric signals generated by the same acoustic can be measured in the medium as precursory events of wave increases. In our experiments we changed the resistivity of the saturated rock sample by changing the salt concentra Sandstone-Layered Model tion of the saturating fluid from pure water to brine. This layered model depicted in Figure 5a is composed When the rock is saturated by pure water, its resistiv- of a Lucite block and a piece of water-saturated Berea ity is not infinite due to the conductivity of the natural sandstone rock. A compressional source transducer rarock. In the resistivity range of 50 to 1000 f m, the diates P waves at an angle of 45 ø with respecto the amplitude of the seismoelectric conversion is propor- interface between the Lucite and the sandstone block. tional to the resistivity. If there are very few ions in the The center frequency of the acoustic wave is about 100 saturating fluid, the seismoelectric signal must be very khz. The Lucite block delays the acoustic wave arriving weak. In one of our experiments, the sandstone sample on the interface and eliminates the electric interference was saturated by alcohol, and no seismoelectric signal of the exciting pulse (500 V) with the seismoelectric sigwas observed. Theoretical studies on dynamic streamnals. At the interface between the Lucite and the rock, ing currents predicts this behavior [Haartsen and Pride, 10 measurement points made of conductinglue were 1997]. set up with a 2-cm spacing. Special care was taken to ensure that no air bubbles were entrapped at the inter- 3. Experiments in Layered Models face and that all of the gaps between the Lucite and the sandstone were saturated with water. We performed additional measurements with layered models. The first layer (Figures 5a and 6) is always Measurements began with the acquisition of the acousa Lucite elastic block that insulates the source trans- tic waves by a small standard P wave transducer at each ducer and its electronics from the porous media. We measurement point on the Lucite block bottom, before used a natural Berea sandstone mounted on the Lu- the rock layer was added to the model. Figure 5b shows cite block, with electrodes glued onto the elastic-porous the received acoustical waveforms which are normalized interface. Two artificial porous medium models were by their maximum amplitude in this plot. (The amplitudes in all the following plots are normalized by the maximum amplitude in each plot.) Then the layered Acoustic Source model was constructed as shown in Figure 5a. Figure (fo=100khz) 5c shows the electric signals which are generated by the acoustic wave and received at the electrodes connected to the measurement patches at the interface between the Lucite block and the Berea sandstone layer. In or-! der to avoid the interference of the high-voltage pulse exciting the acoustic wave, a 10 / s time delay (post- ^=ou,.= Lucite trigger) was applied in recording the electric signals in Figure 5c. By comparing these signals with the acoustic Receiver E "10 Setl waveformshown in Figure 5b, we see that the variations of the arrival times, shape, and amplitude of the electric signals are in good agreement with those of the acoustic waves. We conclude that thes electric signals o %,% Setl,,- t Ill SetIV the acoustic waves. are due to electrokinetic effects. Glued sand! Figure 6. Artificial epoxy-glued sand layer model. Some properties are listed in Table 1, and the resistivity of the bottom layer is 280 f m. Four sets of electrodes are embedded within the homogeneous porous layer (glued sand) Artificial Porous Layered Model In order to measure the seismoelectric signals inside a homogeneous porous medium, we made a layered model composed of a Lucite block and an artificial sand. Sand was glued with epoxy and mounted againsthe bottom surface of the Lucite block. Some properties of

7 ZHU ET AL' SEISMOELECTRIC CONVERSIONS 28,061 the epoxyed sandstone were measured in our laboratory and are given in Table 1. Four sets of electrodes made of thin wires were connected at the interface and within the glued sand with a vertical spacing of I cm and a horizontal spacing of 2 cm (Figure 6). The resistivity of the water-saturated glued sand is 280 f m. A P wave source transducer is placed in the center of the model. Figure 7 shows the acoustic waveforms received by a small P wave transducer on the Lucite bottom sur- face before the sand was glued on the bottom surface. And the waveforms were recorded with a time delay of 20 tts. The first arrivals (around ms) are the direct compressional waves while the secondary arrivals (around ms) correspond to a multiple reflection within the Lucite block. Figures 8a and 8b show the waveforms of the electric signals received by the electrode sets I and IV, respectively. The amplitudes are normalized by 25 ttv. The first arrivals of the electric signals shown in Figure 8a, and their amplitude variations resemble those acoustic waves shown in Figure 7. This similarity tends show that the acoustic wave contains electric signals that travel with the acoustic wave speed and are measurable at electrodes in contact with the porous medium. Figure 8b shows the two kinds of seismoelectric signals. At the time (around ms) the acoustic wave reaches the elastic-porous interface, it is converted into an EM wave which travels with the EM wave speed in the porous medium to the electrode layer sets where they are recorded. The high velocity of the EM waves results in almost identical arrival times 0.1 Acoustic Receiver Position 5 10 Figure 7. Normalized compressional waveforms generated by a P wave source and received by a small P wave receiver at the bottom surface of the Lucite block (Figure 6). The waveforms are recorded with a time delay of 20 tts. at the electrodes of set IV (the signals have almost no move-out). After the acoustic wave passes through the interface, it arrives at the electrode set IV accompanied by a stationary seismoelectric signal which has a higher amplitude and lower frequency than the signals converted at the interface. In Figure 8b the signals around ms are the seismic-to-em converted waves, and the event arriving around ms in the middle of the sections the electric field moving along with the acoustic waves. Even though the EM field and seismic pulse have identical frequency content when the seismic pulse is at the generating interface, the EM disturbance loses almost no frequency content while traveling through the medium, while the seismic pulse and associated electric signals lose much of their higher frequency content while traveling through the porous media. 4. Seismoelectric Experimental Data and Modeling in Multilayered Models In order to compare the seismoelectric signals recorded in a porous medium with theoretical modeling, we have built the multilayered model displayed in Figure 9. Two sand layers glued with epoxy, glued sands I and II, are added at the base of the Lucite block. The physical properties of the two porous layers are shown in Table 1. The sand of layer I is fine white sand with uniform grain diameter. The sand of layer II is brown sand with various grain diameters. Electrode set I is made of conducting glue at the interface between the Lucite and the glued sand I. Set II is buried in glued sand I, and set III is positioned at the interface between glued sands I and II. There are 10 measurement points or wires with a 2-cm horizontal spacing in each set. The resistivities of the water-saturated glued sands I and II are 280 and 420 f m, respectively. A P wave source transducer with a dominant frequency of 100 khz is placed at the center of the top surface of the Lucite block. Figure 10a shows the electric potentials of the seismoelectric signals received by electrode set II. We performed theoretical calculations for the medium properties and the parameters shown in Figure 9 and Table 1. For the details of the calculation method, the reader is referred to Haartsen [1995] and Haartsen and Pride [1997]. The salinity is determined by matching the measured bulk conductivity to the calculated conductivity. We have assumed that the ratio of bulk and shear frame moduli is 0.9. The acoustic source is a ver- tical point force. We have calculated the seismoelectric fields generated by the P waves [Haartsen et al., 1995]. The horizontal component Es of the seismoelectric field at electrode set II is shown in Figure 10b. We see low-frequency and large-amplitude components in both the measured data and modeled results (Figure 10). These electric fields traveling with corn-

8 28,062 ZHU ET AL' SEISMOELECTRIC CONVERSIONS Electrode Position (Set I) ' 0 Electrode Position (Set IV) 5 10 g0.05 g , O.lO. (a) (b) Figure 8. (a) Seismoelectric signals generated by a P wave source and received by electrode set I between the Lucite and glued sand. The amplitude is normalized by 25/ V. (b) Seismoelectric signals generated by a P wave source and received by electrode set IV. The amplitude is normalized by 25/ V. The signals around ms are the EM wave converted at the interface. The signals around ms are the seismoelectric signals generated by the acoustic waves arriving the set IV. The waveforms are recorded with a time delay of 20/ s. pressional wave speed through homogeneous porous medium (glued sand I) are caused by charge separations generated by pressure gradients inside the seismic pulse. Before the P wave arrives at the electrode set II, highfrequency, small-amplitude signals are recorded at 41/ s that is the P wave traveltime through the Lucite block. These signals are the EM waves converted from the P wave at the interface between the Lucite and sand I. Although the P wave source transducer with 'a diam- Acoustic Source (fo=100khz) eter of 3.81 cm cannot be considered as a vertical point force, the comparison between the arrival times, polarities and frequencies of the measured and modeled data confirm the two kinds of seismoelectric effects. One is generated at the interface between Lucite and glued sand and can be received in medium where the seismic wave does not arrive or disturb. The other is generated in a homogeneous medium and can be received only within the area disturbed by the seismic wave. The amplitude variation of the signals in Figures 10a and 10b is different because the measurement records the electric potential associated with the seismoelectric signals, whereas the numerical modeling calculates the horizontal component of the seismoelectric field. Lucite Set I Set II Set III Figure 9. Multilayered model made of artificial epoxyed sands. Their density, porosity, P and S velocities, and permeability are listed in Table 1. The resistivities of the two porous layers, glued sand I and II, are 280 and m, respectively. The electrode set II is embedded within the homogeneous porous layer, glued sand I. 5. Conclusions This paper describes and discusses seismoelectric laboratory experiments in homogeneous, porous cylindrical models or layered models in the kilohertz frequency range. The measurements are verified to be of electrokinetic origin in a fluid-saturated porous medium with a mechanism which is different from the piezoelectric effect. The amplitudes of the seismoelectric signals are directly proportional to the acoustic wave amplitude and increase with the resistivity of the medium, i.e., decrease with the salinity of the pore fluid. No measurable seismoelectric signals are recorded in vacuum dried sandstones. The seismoelectric signals measured along a homogeneous porous rock sample (Berea and Coconino sandstones) are electric fields that move along with the ex-

9 ZHU ET AL.' SEISMOELECTRIC CONVERSIONS 28,063 antenna position O, I I I i i 0 1 antenna position O O0. loo ß._ ß 200- t 200-t,,.....! (a) (,b) Figure 10. (a) Seismoelectric signals measured by the electrode set II and (b) the corresponding horizontal electric components of the seismoelectric field calculated theoretically with the multilayered model shown in Figure 9. The high-frequency components at 41 / s are the EM waves converted at the interface between the Lucite and the glued sand I. The low-frequency and large-amplitude components are generated by P waves within glued sand I. tensional (excited by a P wave transducer in a rod) and flexural wave (excited by a S wave transducer in a rod) modes. Both modes contain compressional wave constituents that cause the diffuse layer counter ions to accumulate and/or deplete in its peaks and troughs, and thus electric fields internal to the wave field are created. Independently propagating EM waves are not generated. Seismoelectric measurements were performed in layered porous models. The corresponding seismic pulses were measured. In addition, seismoelectric conversions were measured in the layered models. These conversions occur at contrasts in medium properties where the seismic pulse straddles the interface, which results in a lo- calized dynamic charge separation across the interface. This dynamic charge separation acts as a source for the EM disturbances that are measurable in the medium not disturbed by the seismic wave. These conversions have the same frequency content as the driving seismic pulse at the interface. The measurements made in the artificial epoxy-glued sand show that the frequency content of the seismoelectric signals associated with the seismic pulse is lower than the frequency content of the converted signal. We numerically modeled a conversion at an elastic porous interface. The numerical model using a vertical point force excitation simulated the arrival times and polarity of the seismoelectric signals. Also, a higherfrequency content of the converted signal is observable as compared to the electric signal associated with the seismic pulse which is strongly attenuated with distance traveled in the glued sand. Acknowledgmcnt.. Wc would like to thank A. II. Thompson, Steven R. Pridc. and T R. Madden tbr their valuable suggestions and ttscfitl discussions This study is supported by the ])cpartmcnl o1' ]5 crgy grant 1)] -1:(;()2-931:I( [ 322 References Butler, K., Russell, R., A. Kepic, and M. Maxwell, Mapping of a stratigraphic boundary by its seismoelectric response, Paper presented at Symosium on the Applications of Geophysics to Engineering and Environmental Problems (SAGEEP), Environ. and Eng. Soc., Boston, Butler, K., R. Russell, A. Kepic, and M. Maxwell, Measurement of the seismoelectric response from a shallow boundary, Geophysics, 61, , Cerda, C. M., and N. C. Kiry, The use of sinusoidal streaming flow measurements to determine the electrokinetic properties of porous media, Colloids Surfaces, 35, 7-15, Chandler, R., Transient streaming potential measurements on fluid saturated porous structures: An experimental verification of Biot's slow wave in the quasi-static limit, J. Acous. Soc. Am., 70, , Haartsen, M. W., Coupled electromagnetic and acoustic wavefield modeling in poro-elastic media and its appli- cations in geophysical exploration, Ph.D. thesis, Mass. Inst. of Technol., Cambridge, Haartsen, M. W., and S. R. Pride, Electroseismic waves from point sources in layered media, J. Geophys. Res., 102, 24,745-24,769, Haartsen, M. W., Z. Zhu, and M. N. ToksSz, Seismoelectric

10 28,064 ZHU ET AL.: SEISMOELECTRIC CONVERSIONS experimental data and modeling in porous layer models at ultrasonic frequencies, paper pp5.10 presented at SEG 65th Annual International Meeting, Soc. of Explor. Geophys., Houston, Groves, J., and A. Sears, Alternating streaming current measurements, J. Colloid Interface Sci., 53, 83-89, Jouniaux, L., and J.P. Pozzi, Streaming potential and permeability of saturated sandstones under triaxial stress: Consequences for electrotelluric anomalies prior to earthquakes, J. Geophys. Res., 100, , Martner, S. T., and N. R. Sparks, The electroseismic effect, Geophysics, œ, , Mikhailov, O., M. W. Haartsen, and M. N. ToksSz, Electroseismic investigation of the shallow subsurface: field measurements and numerical modeling, Geophysics, 62, , Morgan, F. D., E. R. Williams, and T. R. Madden, Streaming potential properties of westerly granite with applications, J. Geophys. Res., 9, 12,449-12,461, Packard, R., Streaming potentials across glass capillaries for sinusoidal pressure, J. Chem. Phys., 21, , Pride, S. R. and M. W. Haartsen, Electroseismic wave properties, J. Acous. $oc. Am. 100, , Sears, A., and J. Groves, The use of oscillating laminar flow streaming potential measurements to determine the zeta potential of a capillary surface, J. Colloid Interface Sci., 65, , Thompson, A. H., and G. A. Gist, Electroseismic prospecting, paper EM2.1 presented at SEG 61st Annual International Meeting, Soc. of Explor. Geophys., Houston, Thompson, A. H., and G. A. Gist, Geophysical applications of electrokinetic conversion: Leading Edge, 12, , Zhu, Z., C. H. Cheng, and M. N. ToksSz, Electrokinetic conversion in a fluid-saturated porous rock sample, paper SLi.1 presented at SEG 64st Annual International Meeting, Soc. of Explor. Geophys., Los Angeles, Zhu, Z., M. W. Haartsen, and M. N. ToksSz, Experimental studies of electrokinetic conversions in fluid-saturated borehole models, Geophysics, 6, , M. W. Haartsen, Shell UK Exploration and Production, i Altens Farm Road, Nigg, Aberdeen, Scotland, UK. (M.Haartsen@openmail.vedc713.dukepabe.simis.com) M. N. ToksSz, and Z. Zhu, Earth Resources Laboratory, Department of Earth, Atmospheric, and Planetary Sci- ences, Massachusetts Institute of Technology, Cambridge, MA (toksoz@mit.edu; zhenya@erl.mit.edu) (Received November 8, 1999; revised June 23, 2000; accepted September 6, 2000.)