Vol. 48, 01, 05ok.qxd 159 ANNALS OF GEOPHYSICS, VOL. 48, N. 1, February 2005 Key words carbon isotope composition – Total In- organic Carbon Dissolved in water (TDIC) – stable isotopes – continuous flow IRMS measurements – on-line extraction technique 1. Introduction The stable carbon isotope ratios of dissolved inorganic carbon (δ 13CTDIC) have been success- fully applied in studies aimed at the evaluation of sources, sinks and fluxes of carbon taking place in natural waters (Atekwana and Krisnamurthy, 1998). In active volcanic areas, on the basis of the carbon isotope balance among dissolved carbon species in solution (CO2aq, HCO3– and CO32–), the pristine carbon isotope composition of CO2 inter- acting with thermal waters has been estimated (Favara et al., 1999; Inguaggiato et al., 2000). Temporal variations observed in the isotopic composition of TDIC in some geothermal waters of Ischia Island were recognized as significant for volcanic surveillance (Caliro et al., 1999). The determination of δ 13CTDIC is based on isotope measurements of carbon dioxide ex- tracted from the samples. The most common re- actions to produce CO2 from water samples are: i) reaction of 100% H3PO4 with solid carbonate Mailing address: Dr. Giorgio Capasso, Istituto Nazionale di Geofisica e Vulcanologia, Sezione di Palermo, Via Ugo La Malfa 153, 90146 Palermo, Italy; e-mail: g.capasso@pa.ingv.it On-line technique for preparing and measuring stable carbon isotope of total dissolved inorganic carbon in water samples (δδ 13CTDIC) Giorgio Capasso, Rocco Favara, Fausto Grassa, Salvatore Inguaggiato and Manfredi Longo Istituto Nazionale di Geofisica e Vulcanologia, Sezione di Palermo, Italy Abstract A fast and completely automated procedure is proposed for the preparation and determination of δ 13C of total inorganic carbon dissolved in water (δ 13CTDIC). This method is based on the acidification of water samples trans- forming the whole dissolved inorganic carbon species into CO2. Water samples are directly injected by syringe into 5.9 ml vials with screw caps which have a pierciable rubber septum. An Analytical Precision «Carbonate Prep System» was used both to flush pure helium into the vials and to automatically dispense a fixed amount of H3PO4. Full-equilibrium conditions between produced CO2 and water are reached at a temperature of 70°C (± 0.1°C) in less than 24 h. Carbon isotope ratios (13C/ 12C) were measured on an AP 2003 continuous flow mass spectrometer, connected on-line with the injection system. The precision and reproducibility of the proposed method was tested both on aqueous standard solutions prepared using Na2 CO3 with δ 13C = −10.78 per mil ver- sus PDB (1σ = 0.08, n = 11), and at five different concentrations (2, 3, 4, 5 and 20 mmol/l) and on more than thirty natural samples. Mean δ 13CTDIC on standard solution samples is –10.89 < per mil versus PDB (1σ = 0.18, n = 50), thus revealing both a good analytical precision and reproducibility. A comparison between average δ 13CTDIC values on a quadruplicate set of natural samples and those obtained following the chemical and physi- cal stripping method highlights a good agreement between the two analytical methods. 160 Giorgio Capasso, Rocco Favara, Fausto Grassa, Salvatore Inguaggiato and Manfredi Longo precipitated from the water sample and ii) di- rect injection of phosphoric acid onto the liquid sample. Both these methods include the use of an expensive vacuum line and/or dedicated sampling and analytical devices. The first procedure implies the use of a large water sample. Furthermore, it requires several steps for sample preparation including precipitation of carbonates by using Ba(OH)2 or Sr(OH)2, filtering and drying under a controlled (carbon-free) atmosphere, homogenization of the precipitate and finally its reaction with acids to form a CO2 gas phase into a vacuum line and gas extraction. All these laborious preparation procedures comprise several potential sources of error and they lead to slow analysis time. A more rapid procedure consists of the addi- tion of phosphoric acid onto the liquid sample. The chemical equilibrium among dissolved carbon species in the solution is shifted towards CO2. The carbon dioxide produced by means of this process can be extracted into a vacuum line (Atekwana and Krisnamurthy, 1998; Favara et al., 2002) or dy- namically extracted under a re-circulating flow of high-purity inert gas (Mc Nicol et al., 1994). In the following sections, we report a rapid and simple method for collection and preparation of water samples for δ 13CTDIC analyses. The ex- traction technique discussed in this paper is based on the liberation of CO2 yielded by acidification of water samples. It requires a low amount of sample (0.5-2 ml) and an equilibrium time of about 18 h, when full-equilibrium conditions be- tween produced CO2 and water are established. The use of an autosampler both for acid dosing and for sample injection into the mass spectrom- eter reduces many potential manual errors as well as preparation and analysis time. 2. Sampling and preparation procedures All water samples were collected in the field and dispensed into glass bottles (ca. 50 cc) and quickly sealed using an aluminum crimp cap and gas-tight rubber/teflon plugs, taking care that no air bubbles are present in the samples. In the lab- oratory, about 10 cc of pure Ar, or another carbon- free host gas, are injected with a syringe into the upturned bottles (fig. 1). The volume of water si- multaneously withdrawn is collected in a syringe, fitted with a stopcock. The water sample is then injected through the rubber septum of a 5.9 ml screw-capped glass vial. The amount of water in- jected into the vials ranges from 0.2 up to 2 ml, depending on the Dissolved Inorganic Carbon (DIC) content of the sample. The lower instru- ment detection limit (see details in the next para- graph) allows the analysis of water samples with DIC contents greater than 4 mmoles per liter. Fig. 1. Extraction of sample from sampling bottles in the laboratory. Injection of about 10 cc of pure Ar in the upturned bottles and simultaneous collection of water in a syringe with a stopcock. Fig. 2a,b. a) schematic of the Analytical Precision «Carbonate Prep System»; b) schematic diagram of the an- alytical unit (Analytical Precision AP2003). a b 161 On-line technique for preparing and measuring stable carbon isotope of total dissolved inorganic carbon in water samples (δ 13CTDIC) However, reducing the headspace of the vials, it is possible to lower the detection limit further. This was done by inserting a glass cylinder of about 1 cc into the vials thus improving the minimum de- tectable DIC content to 2.0 mmol per liter. In order to reduce any air contamination, the vials were previously flushed with ultrapure He- lium (99.9996 vol%) using an Analytical Preci- sion «Carbonate Prep System» (fig. 2a). This de- vice consists of a modified Gilson 222XL au- tosampler tray holding up to 132 samples kept in a thermostatic rack and equipped with a needle having three concentric capillary tubes. The same flushing needle provides the acidification of the water samples. A fixed amount (100-200 µ l) of 100% H3PO4 is automatically dispensed into the vials by means of a valveless pump. Addition of acid decrease the pH of the sample shifting the equilibrium of dissolved carbon species towards gaseous CO2. This reaction takes place at 70° 162 Giorgio Capasso, Rocco Favara, Fausto Grassa, Salvatore Inguaggiato and Manfredi Longo ± 0.1°C. At the end of the process, the pH of wa- ter samples ranges between 0.5 and 1 pH unit. 3. Sample analysis After 18 h, the sample vessels containing the CO2 produced by acidification and the water sample are removed from the bath and manually transferred into the analytical unit where the vials are cooled for about 2 h (fig. 2b). The ana- lytical unit consists of an injection apparatus (Gilson 222XL autosampler) and a purification system (Nafion® trap and GC column) directly connected to the mass spectrometer. Carbon isotope ratios (13C/12C) were meas- ured on an AP 2003 continuous flow mass spec- trometer using He (5.6) as carrier and CO2 (4.8) as reference gas. By using this configuration, instrument sensitivity is calibrated on a carbon dioxide content in the range 3-10 vol%. In order to improve the precision of the measurements, the analysis is repeated four times for each sample, resulting in a series of CO2 peaks reaching the mass spectrometer. Within the same analysis, two pulses of refer- ence gas bracket the sample peaks. The isotopic 13C/12C ratios (R) are expressed in the conven- tional δ -notation as follows: R RC 1 10sample reference 13 3 )-=d _ i8 B . The value of δ 13C versus PDB international standard is calculated using an internal calibrat- ed standard measured within the same batch. The entire process for each sample takes less than 10 min. The precision of the measurement is better than 0.1 permil. 4. Results and discussion The precision and reproducibility of the pro- posed method was tested both on aqueous stan- dard solutions and on some natural samples. The aqueous standard solutions were prepared with Na2CO3 powder at five DIC concentrations (2, 3, 4, 5 and 20 mmol/l). The δ 13C value of the pow- der Na2CO3 was determined by reacting about 1.5 mg of solid carbonate with an excess of 100% H3PO4 using the same apparatus previously de- scribed. Analytical results on eleven replicates re- vealed a δ 13C = –10.78 ‰ versus PDB (1σ = 0.08). 4.1. Equilibration time and correction factor Tests on the evaluation of the time needed for a complete reaction and the achievement of equilibrium conditions between dissolved and gaseous carbon dioxide were carried out on the aqueous standard solution with a DIC content of 4 mmol/l. At least triplicate aqueous standard solution samples were analyzed after different equilibra- tion times from 1 to 24 h. With increasing equi- libration time, the δ 13C composition of CO2 yielded by acidification tends progressively to- wards an equilibrium value of –10.38 ‰ versus PDB (fig. 3). Therefore the carbon isotope ex- change between gaseous CO2 and carbon diox- ide dissolved in the aqueous standard solution can be considered complete after about 18 h. Due to a carbon isotope fractionation process occurring between coexisting water and CO2, the obtained δ 13CCO2 values are enriched in heavy Fig. 3. Evaluation of equilibration times on the Na2CO3 aqueous standard solution with a DIC content of 4 mmol/l. The error bar indicates the standard devi- ation of at least triplicate measurements. After about 18 h the carbon isotope exchange between gaseous CO2 and carbon dioxide dissolved in the aqueous stan- dard solution can be considered completed reaching an equilibrium value of –10.38 ‰ versus PDB. isotopes with respect to the carbonate used to make the standard solution. As a consequence, a correction factor to δ 13CTDIC measurements has to be applied. As the experiment is carried out in a closed system, the number of stable isotope carbon species remains constant. At the pH values of the solution (0.5 ÷ 1), the only carbon species present in our system are gaseous and dissolved carbon dioxide. Under these conditions, the carbon isotope balance can be expressed by the following equation: C C C C CO C C CO gas aq 13 12 13 12 2 13 12 2 TDIC gas aq # # = + + | | (4.1) where 13C/12CTDIC, 13C/12Cgas and 13C/12Caq stand for the carbon isotope ratios of total dissolved carbon species, gaseous CO2 and dissolved CO2 respectively while, χ CO2gas and χ CO2aq represent the molar fraction of gaseous and dissolved CO2. In terms of delta notation, the eq. (4.1) be- gins C CO CO C CO CO C 1 TDIC aq gas CO aq gas CO 13 2 2 13 2 2 13 aq gas2 2 # = + + d | | d | | d (4.2) where δ 13CCO2gas is the value of measured gas, and δ 13CCO2aq is calculated considering the car- 163 On-line technique for preparing and measuring stable carbon isotope of total dissolved inorganic carbon in water samples (δ 13CTDIC) bon isotope enrichment factor between dis- solved and gaseous CO2 (ε 13Caq-gas). In vials of known volume, the molar frac- tion of gaseous CO2 can be computed as a func- tion of the DIC content in the injected sample volume and the headspace volume (table I). The molar fraction of dissolved CO2 has been cal- culated taking into account the solubility of CO2 in water (866 cc/l at 25°C calculated from Capasso and Inguaggiato 1998). For δ 13CCO2aq the value of –1.04 δ per mil was subtracted from the measured δ13CCO2gas ac- cording to the enrichment factor between dis- solved and gaseous CO2 (ε 13Caq-gas) in closed systems at 25°C found by Szaran (1998). Hence, the eq. (4.2) begins . C CO CO C CO CO C 1 1 04 TDIC aq gas CO aq gas CO 13 2 2 13 2 2 13 gas gas2 2 # = + - + d | | d | | d_ i . (4.3) In a closed system, the extent of isotope fractiona- tion between gaseous and dissolved CO2 is a func- tion of the headspace-water volume ratio, thus re- sulting in a difference between the measured δ 13CCO2gas and the values. Therefore, the δ 13CTDIC values can be easily obtained from measured δ13CCO2gas introducing a Correction Factor (CF) re- ported in table I. The CF values can be computed from eq. (4.3) fixing both the δ13CCO22 gas value and the headspace-water volume ratio. Table I. Computed Correction Factor (CF) as a function of the DIC content and the volume of water injected for vials with internal volume of 5.9 ml. DIC contents of water samples ([CO2aq] + [HCO3−] + [CO3− −]) are ex- pressed in mmol/l. Sample volume is expressed in milliliter. The molar fraction of dissolved CO2 (χCO2aq) has been calculated by Henry’s law. For DIC contents between 2 and 3.5 mmol/l, CF was computed considering the reduced headspace volume due to the insertion of the glass cylinder (see text). DIC Sample vol Headspace vol χ CO2aq χ CO2gas CF 2.0-3.5 2* 2.9 0.60 0.40 0.62 4-8 2 3.9 0.44 0.56 0.46 8.5-17.5 1 4.9 0.18 0.82 0.18 20-60 0.5 5.4 0.08 0.92 0.08 60-100 0.2 5.7 0.03 0.97 0.03 2* Samples analyzed with the glass cylinder inserted into the vial. 164 Giorgio Capasso, Rocco Favara, Fausto Grassa, Salvatore Inguaggiato and Manfredi Longo 4.2. Tests on standard solutions To determine the accuracy and precision of the proposed method, ten measurements were carried out for each of the prepared aqueous standard solutions having five different DIC con- tents (2, 3, 4, 5 and 20 mmol/l). Mean δ13CCO2 measurements and corrected δ13CTDIC values are reported in table II together with their standard deviation (σ). The average value of δ13CTDIC is only 0.11 δ per mil lighter with respect to the carbon isotope composition of the Na 2 CO3 pow- der (δ 13CNa 2 CO3 = –10.78‰ versus PDB). As can be seen in fig. 4, all the δ13CTDIC values are very close to the δ13CNa 2CO3 value, the maximum dif- ference being 0.23 δ per mil for the 3 mmol/l so- lution. Furthermore, on the basis of the number of analyses (n = 50), the obtained results also highlight excellent reproducibility of the meas- urements. The standard deviation values de- crease with increasing of DIC content in the pre- pared standard solutions. Using this method, the carbon isotope reproducibility is better than ± 0.14 for DIC content higher than 3 mmol/l. 4.3. Tests on natural water samples To test the proposed method on natural waters more than thirty water samples were sampled and analyzed in replicate. The main physical-chemical features of these waters are reported in table III to- gether with the analytical results. The water sam- ples were collected from springs or domestic wells belonging to different geologic environments, from active volcanic systems (Vulcano Island, Et- na, Stromboli, Popocatepetl-Mexico) to carbonate sedimentary aquifers (Western Sicily) to cover a wide range of carbon isotope composition. For comparison on the same samples δ13CTDIC values were also determined using the methodology pro- posed by Favara et al., (2002) based on chemical and physical stripping of CO2 from water sam- ples. The δ13CTDIC values reported in table III are graphically shown in fig. 5 highlighting compara- ble results giving very small differences in the whole range of isotope values. The least-squares regression line for the en- tire data set is: Y=1.011X + 0.04, r2 = 0.998. A set of samples analyzed 30 days after sam- pling highlights the fact that the storage of water samples both in the glass sampling bottles and in the vials do not compromise the carbon isotopic ratio. However, some precautions have to be tak- Fig. 4. Comparison between isotopic composition of solid Na2CO3 and Na2 CO3 aqueous standard solution at different DIC concentration (2, 3, 4, 5 and 20 mmol/l). The error bar indicates the standard deviation of at least eight measurements for each of the aqueous solutions. Mean δ 13CTDIC values are corrected accord- ing to eq. (4.2) and table I. All the resulting δ 13CTDIC values are very close to the carbon isotope composi- tion of the Na2CO3 powder (dark area) the maximum difference being 0.23 δ per mil. Table II. Mean δ 13CCO2 measurements for the aque- ous standard solutions at five different DIC contents (2, 3, 4, 5 and 20 mmol/l). Each value represents the average of ten analyses. δ 13CTDIC values have been computed according to the correction factors report- ed in table I. All the isotope values are expressed in δ ‰ versus PDB international standard. DIC δ 13CCO2meas δ 13CTDICcorr Standard deviation (1σ ) 2 –10.34 –10.96 0.29 3 –10.39 –11.01 0.20 4 –10.47 –10.93 0.14 5 –10.40 –10.86 0.05 20 –10.63 –10.71 0.05 Average δ 13CTDICcorr –10.89 Standard deviation (1σ ) 0.18 165 On-line technique for preparing and measuring stable carbon isotope of total dissolved inorganic carbon in water samples (δ 13CTDIC) T ab le II I. T es ts o n na tu ra l w at er s am pl es t og et he r w it h th ei r m ai n ph ys ic o- ch em ic al p ar am et er s. p H is e xp re ss ed i n p H un it , D IC i n m m ol /l , T D S (T ot al D is so lv ed S ol id ) in m g/ l. S am pl e vo lu m e is r ep or te d in m il li li te r an d th e co rr es po nd in g C or re ct io n F ac to r (C F ) is t ak en f ro m t ab le I . A ll t he is ot op e va lu es a re e xp re ss ed i n δ ‰ v er su s P D B i nt er na ti on al s ta nd ar d. S ta nd ar d de vi at io n – st d. d ev . (1 σ ) is c al cu la te d, at l ea st , on f ou r re pl ic at es . S am pl e L oc at io n D at e p H D IC T D S S am pl e vo l C F A v. δ 13 C C O 2 m ea s. S td . de v. δ 13 C T D IC co rr ( 1 ) δ 13 C T D IC ( 2 ) R . C am pa na E tn a 30 .1 0. 20 02 7. 45 5. 5 64 6 2 0. 46 –3 .9 6 0. 03 –4 .4 2 –4 .8 3 G . P av on e E tn a 13 .0 2. 19 98 6. 53 10 .8 52 8 1 0. 18 2. 25 0. 10 2. 07 2. 06 P on te fe rr o E tn a 17 .0 1. 19 98 7. 49 14 .2 16 36 0. 5 0. 08 1. 04 0. 28 0. 96 1. 28 V al co rr en te E tn a 13 .0 2. 19 98 6. 92 21 .3 16 81 0. 5 0. 08 3. 92 0. 18 3. 84 4. 40 P 31 E tn a 30 .1 0. 20 02 5. 99 21 .7 80 4 0. 5 0. 08 0. 45 0. 11 0. 37 0. 21 A T P op oc at ep et l 18 .0 5. 20 02 6. 80 4. 6 17 28 2 0. 46 –7 .6 0 0. 25 –8 .0 6 –7 .9 8 A X P op oc at ep et l 18 .0 5. 20 02 5. 90 28 .7 62 0 0. 5 0. 08 –2 .3 1 0. 10 –2 .3 9 –2 .6 5 Z ur ro S tr om bo li 31 .0 1. 20 03 6. 93 5. 9 30 99 7 2 0. 46 –0 .0 8 0. 28 –0 .5 4 –0 .2 5 Z ur ro S tr om bo li 26 .0 1. 20 03 6. 93 6. 3 32 69 6 2 0. 46 0. 21 0. 13 –0 .2 5 –0 .5 8 O ss id ia na S tr om bo li 02 .0 2. 20 03 6. 93 6. 6 40 40 4 2 0. 46 –0 .3 6 0. 27 –0 .8 2 –0 .7 9 S ir en et ta S tr om bo li 31 .0 1. 20 03 7. 10 7. 5 18 88 8 1 0. 18 1. 10 0. 22 0. 92 0. 91 C us ol it o S tr om bo li 31 .0 1. 20 03 6. 77 10 .8 26 12 1 1 0. 18 1. 54 0. 25 1. 36 1. 37 C us ol it o S tr om bo li 27 .0 1. 20 03 6. 61 11 .0 24 60 5 1 0. 18 2. 07 0. 14 1. 89 1. 75 C us ol it o S tr om bo li 2. 02 .2 00 3 6. 75 11 .6 25 91 3 1 0. 18 1. 84 0. 12 1. 66 1. 74 F ul co S tr om bo li 26 .0 1. 20 03 6. 39 19 .3 24 48 4 0. 5 0. 08 1. 83 0. 18 1. 75 1. 96 F ul co S tr om bo li 31 .0 1. 20 03 6. 35 19 .5 12 32 3 0. 5 0. 08 0. 90 0. 19 0. 82 0. 99 F ul co S tr om bo li 02 .0 2. 20 03 6. 45 19 .7 12 93 6 0. 5 0. 08 0. 96 0. 12 0. 88 1. 22 C . S ic il ia V ul ca no I sl an d 13 .0 1. 20 03 7. 45 7. 2 64 67 2 0. 46 1. 12 0. 08 0. 66 0. 82 E as V ul ca no I sl an d 13 .0 1. 20 03 8. 42 19 .8 22 93 0. 5 0. 08 4. 47 0. 09 4. 39 4. 78 B am ba ra V ul ca no I sl an d 13 .0 1. 20 03 5. 73 27 .5 10 38 0. 5 0. 08 –2 .6 1 0. 26 –2 .6 9 –3 .0 0 D is ca ri ca V ul ca no I sl an d 13 .0 1. 20 03 6. 63 63 .2 47 88 0. 2 0. 03 0. 90 0. 09 0. 87 0. 51 C . B ai da W es te rn S ic il y 29 .0 1. 20 03 8. 11 2. 7 36 7 2* 0. 62 –1 4. 96 0. 16 –1 5. 58 –1 5. 45 C . B ai da ( a ) W es te rn S ic il y 29 .0 1. 20 03 8. 11 2. 7 36 7 2* 0. 62 –1 4. 53 0. 13 –1 5. 15 –1 5. 45 S co pe ll o W es te rn S ic il y 29 .0 1. 20 03 7. 87 2. 7 31 8 2 * 0. 62 –1 2. 84 0. 36 –1 3. 46 –1 3. 56 S co pe ll o (a ) W es te rn S ic il y 29 .0 1. 20 03 7. 87 2. 7 31 8 2* 0. 62 –1 2. 87 0. 26 –1 3. 49 –1 3. 56 F ra gi ne si W es te rn S ic il y 29 .0 1. 20 03 8. 16 3. 5 76 4 2* 0. 62 –9 .6 3 0. 34 –1 0. 25 –1 0. 29 F ra gi ne si ( a ) W es te rn S ic il y 29 .0 1. 20 03 8. 16 3. 5 76 4 2* 0. 62 –9 .5 4 0. 23 –1 0. 16 –1 0. 29 P io pp o W es te rn S ic il y 29 .0 1. 20 03 7. 15 4. 7 61 6 2 0. 46 –1 3. 95 0. 20 –1 4. 41 –1 4. 62 P io pp o (a ) W es te rn S ic il y 29 .0 1. 20 03 7. 15 4. 7 61 6 2 0. 46 –1 4. 06 0. 07 –1 4. 52 –1 4. 62 S . N uo va W es te rn S ic il y 29 .0 1. 20 03 6. 70 5. 9 16 95 2 0. 46 –2 .1 7 0. 23 –2 .6 3 –2 .6 8 S . N uo va ( a ) W es te rn S ic il y 29 .0 1. 20 03 6. 70 5. 9 16 95 2 0. 46 –2 .3 5 0. 09 –2 .8 1 –2 .6 8 M er la W es te rn S ic il y 29 .0 1. 20 03 7. 06 7. 4 97 2 2 0. 46 –1 4. 00 0. 21 –1 4. 46 –1 4. 17 M er la ( a ) W es te rn S ic il y 29 .0 1. 20 03 7. 06 7. 4 97 2 2 0. 46 –1 4. 01 0. 28 –1 4. 47 –1 4. 17 2* S am pl es a na ly ze d w it h th e gl as s cy li nd er i ns er te d in to t he v ia l; ( a ) s am pl es a na ly ze d af te r 30 d ay s; ( 1 ) t hi s m et ho d; ( 2 ) a na ly ze d fo ll ow in g th e st ri p- pi ng m et ho d (F av ar a et a l. , 20 02 ). 166 Giorgio Capasso, Rocco Favara, Fausto Grassa, Salvatore Inguaggiato and Manfredi Longo Fig. 5. Comparison between the δ 13CTDIC values obtained using the proposed method and those car- ried out following the chemical and physical strip- ping technique (Favara et al., 2002) for some natural water samples The least-squares regression line for all the data set (Y=1.011X + 0.04, r 2= 0.998) high- lights comparable results over the whole range of isotope values. en for waters saturated with respect to carbonate minerals and/or organic matter rich waters that could modify the pristine TDIC isotope compo- sition. In the first case this is due to the precipi- tation of a solid carbonate phase, whereas the or- ganic activity leads to enrichment of 12C in bio- logically synthesised organic compounds. 5. Conclusions An automated technique for extracting and analyzing inorganic carbon isotope ratio (δ13CTDIC) from 0.5 to 2 ml water samples has been developed. This method, based on the re- lease of CO2 yielded by acidification of water samples, is a modification of the procedure previously proposed by Favara et al. (2002). The use of an autosampler both for acid dosing and for sample injection into the mass spec- trometer not only greatly reduces a lot poten- tial manual errors as well as reducing prepara- tion and analysis time but also eliminates the use of expensive devices (vacuum preparation line, special grease, liquid nitrogen, etc.). Analytical results gave good sample repro- ducibility and precision over a wide range of isotope composition and highlighted that this method is comparable to the other analytical procedures. Acknowledgements We are indebted to Jason Newton for help- ful suggestions and reviews. We also thank two anonymous reviewers for their constructive comments. REFERENCES ATEKWANA, E.A. and R.V. KRISHNAMURTHY (1998): Season- al variations of dissolved inorganic carbon and δ 13C of surface waters: application of a modified gas evolution technique, J. Hydrol., 205, 265-278. CALIRO, S., C. PANICHI and D. STANZIONE (1999): Variation in the total dissolved carbon isotope composition of thermal waters of the Island of Ischia (Italy) and its im- plication for volcanic surveillance, J. Volcanol. Geot- herm. Res., 90, 219-240. CAPASSO, G. and S. INGUAGGIATO (1999): A simple method for the determination of dissolved gases in natural wa- ters, An application to thermal waters from Vulcano Is- land, Appl. Geochem., 13, 631-642. FAVARA, R., F. GRASSA and S. INGUAGGIATO (1999): Chemi- cal and isotopic features of dissolved gases from ther- mal springs of Sicily, in Proceedings of the Fifth Inter- national Symposium on the Geochemistry of the Earth’s Surface (GES-5), 16-10 August 1999, Reyk- javik, Iceland (H. Armannssonn Publishing), pp. 574. FAVARA, R., F. GRASSA, S. INGUAGGIATO, G. PECORAINO and G. CAPASSO (2002): A simple method to determine the δ 13C of total dissolved inorganic carbon, Geofis. Int., 41, 313-320. INGUAGGIATO, S., G. PECORAINO and F. D’AMORE (2000): Chemical and isotopical characterisation of fluid man- ifestations of Ischia Island (Italy), J. Volcanol. Geot- herm. Res., 99 (1/4), 151-178. MC NICOL, A.P., G.A. JONES, D.L. HUTTON, A.R. GAGNON and R.M. KEY (1994): The rapid preparation of sea- water ε CO2 for radiocarbon analysis at the national ocean sciences AMS facilities, Radiocarbon, 36 (2), 237-246. SZARAN, J. (1998): Carbon isotope fractionation between dissolved and gaseous carbon dioxide, Chem. Geol., 150, 331-337.