progress report 11/04/07

A portable universal water sampler for dissolved gases

Martin Stute (martins@ldeo.columbia.edu)

The objectives of this OTIC project are to develop a sampler for dissolved gases that eliminates the need to collect and transport water samples by extracting the gas from solution using membrane technology and the principles of diffusion and solubility equilibrium. We have set three main criteria for the sampler: (1) it should be able to sample as many gases as possible (e.g. CFCs, SF₆, SF₅CF₃, ³He, noble gases, N₂, and CO₂); (2) the sampler should not require the use of a pure compressed carrier gas (N2 or He), and (3) it should be portable, and have minimal power consumption. To date, we have made significant progress on (1) and (2); equilibration experiments have proven that solubility equilibrium can be achieved within the sampler for the gases N₂, Ar, SF₆, O₂and CO₂. Further work is necessary to achieve the same results with CFC’s and ³He. In this report we will first outline and discuss the advances that have been completed to date, including the limitations of the sampler, and then summarize by listing the steps that are necessary to meet the three objectives described above.

Prototype design and equilibration experiments

Figure 1. Schematic of the initial prototype developed in May, 2007.

To prove the design concept we have used an experiment that takes advantage of the gases most present in the atmosphere, N₂ Ar and O₂, while using CO₂ as a more variable gas. We equilibrate ca. 100 liters of water in a Nalgene tank with air by bubbling air through tank for at least 8 hours. This reference water volume is then pumped through the membrane sampler and a time series of gas pressure and gas concentration is collected with syringes. To compare the gas concentration yielded by the membrane sampler, syringe samples are collected from the water volume and analyzed using headspace equilibration in the syringe. Both membrane and headspace samples were analyzed on a Gas Chromatograph equipped with flame-ionization and thermal conductivity detectors. We expect the corrected concentration from the headspace samples to match the concentration measured by the membrane sampler; as a further check, both of these concentrations should reflect the concentration of the above gases in the atmosphere (this is referred to as ‘Dissolved Air’ in Table 1). CO₂ is more variable and subject to local sources, and reflects the ability of the sampler to collect trace gas concentrations.

The initial prototype (Figure 1) consists of a Liqui-cel Mini-Module membrane whose maximum water flow rate is 2.5 L/min. A pressure transducer measures the gas pressure on the gas-side of the membrane. Water pressure is regulated via a flow valve, upstream of the membrane. The plumbing consists of nylon tubing with Parker o-ring compression fittings. These fittings are leak-tight and easy to assemble/disassemble.

RESULTS:

Table 1 shows the results of one of the equilibrium experiments. The ‘Dissolved Air’ concentrations are those that we expect from equilibrium with the atmosphere. For each gas, the membrane sampler yielded results within 3% of the expected values, and appears consistently more accurate than the headspace method. The reproducibility of membrane samples is very good, usually within 5%. Both the headspace and membrane technique indicate CO₂ concentrations greater than in the atmosphere, which is not surprising as these experiments were performed in a laboratory environment. The time to achieve solubility equilibrium is well-captured by examining the time rate of change in the pressure transducer. When equilibrium is achieved, the pressure variations oscillated within several tenths of a mb of the equilibrium value. The time to equilibrium depends on the efficiency of the membrane, the dissolved gas content of the water, and the flow rate. At ca. 2 L/min, solubility equilibrium was achieved within 40-60 minutes. This implies that approximately 80-120 liters of water are required to obtain a syringe sample of ~ 50 ml. Water that has greater dissolved gas content will equilibrate more quickly; this was demonstrated this summer during sampling at the LDEO borehole, where dissolved gas concentration was much greater. We think that the equilibration time is mostly a result of diffusive fractionation when the rate of mass transfer is initially high (at low pressure). However, this should be less of a concern for trace gases at low concentrations.

The same test was performed for CFC-11, CFC-12 and CFC-113. We found that CFC-11 was in good agreement, but the membrane produced values that were 50-100% greater than the headspace method. We expect this is an effect of the plastics, nylon and PTFE in the system; these plastics are known to be a source of contamination to CFC samples. We are currently evaluating whether changing the plumbing to stainless steel can eliminate this problem; the membrane housing and lumens are plastic, which may mean that CFC’s cannot be sampled in this fashion.

Table 1. Comparison between the headspace and membrane sampling method, where both methods are expected to yield dissolved gas concentrations that reflect solubility equilibrium of the water reservoir with the atmosphere (referred to as Dissolved Air). Except for CO2, both methods produced the expected results, within 8%. The membrane samples were within 3% of expected values.

	Ar (ml/L)	N2 (ml/L)	O2 (ml/L)	CO2 (ml/L)
Dissolved Air	0.3242	12.3400	6.6150	0.3183
Headspace	0.2649	12.2367	5.8368	0.5647
Membrane	0.3238	12.6723	6.2409	0.6074
(Membrane/Air)	0.99	1.027	0.943	1.908

Remaining work

Having proven the design concept, we are refining the configuration and materials in an effort to include 3He as well as CFC’s in the list of gases which can be sampled with the membrane. As mentioned above, the membrane may prove a source of CFC’s. In the case of 3He; we hope to minimize the effect of He diffusion through the walls of the sampler by changing the nylon plumbing and fittings to stainless steel. We intend to test the sampler on water samples with dissolved gas content which are significantly different from what is found in the atmosphere. This is necessary to confirm the integrity of the system and demonstrate that the membrane will yield an accurate sample when dissolved gas pressure is high.

The final phase of the design will be to integrate the sampler into a portable suitcase-type configuration, to determine the power consumption, and to evaluate second order effects, such as the buildup of condensation on the gas side of the membrane. We anticipate that this work will be completed by the early part of 2008, at which point we hope to deploy the instrument in the long-term groundwater monitoring that is taking place in Bangladesh. We will present a poster on the results of this work at the 2007 AGU Fall Meeting in the General Hydrology poster session.

We have a wiki that we have used to aggregate our notes and to document progress on the sampler; http://duck-rabbit.ldeo.columbia.edu/PoTGas/wiki/index.php/Main_Page.