Stable Isotope Geochemistry; A Tribute to Samuel Epstein © The Geochemical Society, Special Publication No.3, 1991 Editors: H. P. Taylor, Jr., J. R. O'Neil and I. R. Kaplan Oxygen isotope studies of zeolites: Stilbite, analcime, heulandite, and clinoptilolite-I. Analytical technique* XIAHONGFENGt and SAMUELM. SAVIN Department of Geological Sciences, Case Western Reserve University, Cleveland, OH 44106, U.S.A. Abstract-An analytical technique for measuring 8180 values of the zeolites stilbite, analcime, heu- landite, and clinoptilolite has been developed and tested. The framework oxygen of these minerals can be isotopically analyzed with a good precision using a standard dehydration procedure developed in this study. The analytical accuracy, however, is impaired by isotopic exchangebetween the framework oxygen and channel water, which is unavoidable during the process of dehydration. The errors thus introduced into the isotopic analyses can be corrected empirically using a calibration curve generated by controlled experiments. INTRODUCTION of performing isotopic analyses of zeolites. The 0180 THEOXVGENISOTOPICcomposition of a mineral is value of the framework (i.e., aluminosilicate) oxy- a function of both the 180/60 ratio of the ambient gen of a zeolite can be measured only after the re- fluid and the temperature of the environment in moval of channel water. Second, the occurrence of which it formed. If the mineral is not in equilibrium labile channel water within the open framework with its environment, its isotopic composition is structure of zeolites raised suspicions that the also affected by the extent to which it exchanged framework oxygen of these minerals might be sus- with the environment prior to the time it was col- ceptible to post-formational isotopic exchange. If lected. Therefore, the isotopic composition of a so, their isotopic compositions would be oflimited mineral may carry some information both about usefulness in studies of diagenesis and metamor- the environment in which it formed and about its phism. This paper is a report of the development subsequent history. These concepts have been the and testing of a technique for precise, accurate 18 basis for numerous studies of paleoenvironment and measurement of 0 0 values of zeolites. The pres- of diagenesis and other low temperature processes. ervation of isotopic records in naturally occurring In isotopic studies of diagenesis, 0180 values of zeolites will be discussed in another paper (FENG quartz, feldspars, carbonates, and clays are com- and SAVIN,in prep.). monly measured (ESLINGERand SAVIN,1973;YEH and SAVIN,1977; LAND,1984;DUTTONand LAND, PREVIOUS WORK 1985; LoNGSTAFFE,1986; LANDand FISHER,1987; SAVINand EpSTEIN(1970) analyzed three authigenic AVALONand LONGSTAFFE,1988; LEEet al., 1989; phillipsite samples, separated from ocean floor sediments. GIRARD et al., 1989; and many others). Zeolite Samples were treated by drying at room temperature in a minerals are common products oflow-grade meta- dry box for between 24 and 72 hours prior to loading in morphism and diagenesis, particularly in volca- the Ni reaction vessels. The results of drying aliquots of the same sample for differing lengths of time pointed out nogenic sediments. Knowledge of the conditions of the complexity of the dehydration behavior of phillipsite their formation is important not only to an under- and the difficultyof measuring its 180/160 ratio accurately. standing of problems directly related to the char- SAVINand EpSTEIN(1970) estimated a value of 1.034 for acteristics (e.g., porosity and permeability) of res- aphillipsite-water at O°C. FENG(1985) analyzed phillipsite that had been hand- ervoirs in clastic rocks, but also to an understanding picked from within a manganese nodule collected in the of high-grade diagenetic and low-grade metamor- central North Pacific Ocean. This analysis also indicated phic processes. However, stable isotope methods a fractionation factor of about 1.034 at ocean bottom have only rarely been applied to the study of zeolites. temperature. BOHLKEet al. (1984) reported 8180 values There are two main reasons for the paucity of of a phillipsite from Deep Sea Drilling Project (DSDP) Hole 396B and of a number of bulk rock samples con- isotopic analyses of zeolites. The first is the difficulty taining zeolite minerals from the same core and from DSDP Holes 417A and 417D. They studied alteration of deep-sea basalts by correlating the 8180 values of minerals * Contribution No. 183, Department of Geological Sci- and bulk rocks in those cores with the H20+ contents. ences, Case Western Reserve University. HAY et at. (1991) studied clay mineral diagenesis in core t Present address: Division of Geological and Planetary KM-3 from SearlesLake, California. They used 8180values Sciences, California Institute of Technology, Pasadena, CA of a number of minerals, including phillipsite, as the basis 91125, U.S.A. for conclusions about the water chemistry of the lake dur- 271 272 X. Feng and S. M. Savin ing diagenesis. Both BOHLKE et al. (1984) and KITA and a phase transformation during dehydration, was selected HONDA (1987) used 8180 values ofmordenite and clinop- later in the study for comparison with heulandite. Two tilolite, as well as quartz, cristobalite, kaolinite, and other clinoptilolite samples were used. One (denoted as Clinop- minerals to distinguish hydrothermally altered rocks from tilolite-l ) was donated by Giday WoldeGabriel of Los those of diagenetic origin. Alamos National Laboratory. He separated the mineral In all of the studies mentioned above except that of from a bulk rock sample (Yucca Mountain Nuclear Waste SAVIN and EpSTEIN (1970), zeolites were dehydrated in Repository Site, bore hole G 1) by particle size separation. the Ni reaction vessels of the fluorination line prior to The 1 to 3 µm fraction used in this study was more than reaction with Brf', according to the technique of CLAYTON 90% pure clinoptilolite. A second clinoptilolite (Castle and MAYEDA (1963). Because the reaction vessels become Creek, Owyhee Co, Idaho) was obtained from Miriam lined with hygroscopic fluorine compounds, the water lib- Kastner of Scripps Institution of Oceanography. XRD and erated from the zeolites may be absorbed on the sides of TGA analyses indicated that this sample was greater than the vessels. If this occurs, the oxygen of the absorbed water 95% clinoptilolite. This clinoptilolite sample will be referred will be liberated during fluorination of the minerals, re- to as Clinoptilolite-2. sulting in erroneous measurements. STALLARD and BOLES (1989) reported 8180 values of laumontite, stilbite, and heulandite from zeolite-facies Thermogravimetric analysis (TGA) metavolcanic rocks of the Hokonui Hills, New Zealand. Thermogravimetric analyses or dehydration studies of They pointed out that incompleteness of dehydration of the zeolites were undertaken using two methods. In the the zeolites significantly affects the reproducibility of the first method, a Tern-Pres Research TG-716 Thermogravi- isotopic analysis, and showed large isotopic effects of de- metric Analyzer was used to obtain curves of weight loss hydrating at different temperatures and for different lengths vs. temperature while the temperature was raised at a con- of time. In their geological interpretations, they used the trolled rate. With this instrument the sample can be heated data obtained using those dehydration conditions that gave either in the atmosphere or under vacuum (:=dO-3 torr). the best precision. Weight can be measured with a relative error of less than Recently, KARLSSON and CLAYTON (1990) reported 5%, and temperature with a relative error of 2%. 8180 values of both framework oxygen and channel water In the second method, a McBain-Bakr balance was used of analcime, chabazite, clinoptilolite, laumontite, mor- to measure the weight loss of the sample during dehydra- denite, and natrolite. They dehydrated their samples out- tion. The McBain-Bakr balance (Fig. 1) consists ofa quartz side of the fluorination line by heating under high vacuum spring with a hook at either end. The spring is suspended to 450°C prior to analysis of framework oxygen. They from o.ie hook and a sample cup hangs from a wire about qualitatively examined the possibility of isotope exchange 20 em long attached to the other. The spring stretches or between framework oxygen and channel water during de- contracts in proportion to the change in weight of the hydration by varying the conditions of dehydration and material in the sample cup. The sample and the balance by exchanging the channel water with a water of different were enclosed in a 25 mm O.D. quartz tube, sealed at the isotopic composition. They concluded that the 8180 values of dehydrated analcime and mordenite are not affected bottom and connected to a vacuum system at the top. A furnace operated by a temperature controller was placed significantly by changing the procedures of dehydration, around the quartz tube. The length of the spring (and hence and argued that this indicated that the measurements were the sample weight) is measured with a cathetometer. The accurate. However, their treatments affected the 8180 val- precision of the cathetometer is 0.005 em, corresponding ues of chabazite and laumontite by as much as 2 per mil, to a weight change of 0.05 mg. The amount of sample leading KARLSSON and CLAYTON (1990) to conclude that used for each dehydration experiment was greater than the 180/160 ratios of those minerals should be interpreted 100 mg, and thus the relative error introduced by the cath- with caution. etometer reading is smaller then 0.1 %. In this paper we show that the dehydration conditions The apparatus was designed to permit collection of the used by STALLARD and BOLES (1989) may not have been water liberated upon dehydration. In some experiments, appropriate to yield 8180 values representative of the water vapor was introduced into the system by evaporation framework oxygen of stilbite. In addition, we show that from a water reservoir held at a constant temperature.