It has been known for many years that sulfur compounds are particularly aggressive in imparting unwanted changes in the flavor of the product. Historically, a "Taste Test" has been used on the final product. It can screen against the release of poor taste, but cannot guard against the use of contaminated raw materials used in the final formula. Carbon dioxide is a raw material added during the bottling process. It is required to be extremely pure. Many of the possible impurities are limited to less than 1.0ppm in total content. Sulfur components are held to an even lower specification. Typically, a specification for hydrogen sulfide or carbonyl sulfide can be as low as 50ppb maximum.

Producers of high purity carbon dioxide use extensive physical and chemical treatments to remove impurities. Sulfur removal is a critical operation monitored at each phase of the process with a final analysis conducted before entering bulk storage. The storage vessels are dedicated to the high purity product therefore any subsequent contamination occurs during the distribution phase. Consequently, sulfur analyses must be conducted at each change of custody and certainly before entering the bottler's tankage. Delivery of contaminated product to a bottler can cause the shut down of an entire plant for cleanup.

Sulfur analyses can be performed by several techniques from sophisticated laboratory instruments to visual stain assessments. The industry has tolerated almost any methodology that possesses the inherent ability to achieve the desired detection levels. Of late, combustion methods, which react the sulfur to a form consistent with the detection device, are being used for Total measurements. Total sulfur may be used as a specification test but provides no useful information as to the type or source of the sulfur. Carbon dioxide is produced from several large-scale industrial processes. Each process tends to produce certain discrete sulfur fingerprints. An analytical technique that possesses the ability to achieve both current detection levels and sulfur type information is preferred.

Today, as in most industries, beverage producers are demanding more of their suppliers. Quality is an uncompromising issue in the food marketplace. Subjective "Taste Tests" are being replaced with scientific instrumentation. Raw material suppliers are faced with delivering extremely pure product without the analytical solutions productized for plant QC operations.

Arnel Inc. of Parlin, New Jersey has been developing complete chromatographic solutions for the past two years for one of the largest carbon dioxide producers in the United States. The company has a cooperative agreement to jointly engineer and market special products with Perkin Elmer, a manufacturer of gas chromatographs. The objective of the project was to design a sulfur specific analyzer with sensitivities equivalent to or better than total sulfur analyzers. Gas chromatography (GC) was chosen to provide a convenient mechanism for separating the volatile sulfur compounds. However, a detection system superior to flame photometry, the traditional GC sulfur detector was necessary to meet the criteria of sulfur specificity and sensitivity. Additional critical issues also addressed included plant environments, on-line sampling, calibration material stability, ease of use and long term system stability.

Improved Sulfur Detection for Gas Chromatography

Chemiluminescence detection of sulfur compounds as a finish to high-resolution capillary chromatography has evolved as the detector of choice for extremely low concentrations. The sensitivity exhibited by these detectors makes it possible to use high-resolution capillary columns to separate low molecular weight sulfur compounds. These small internal diameter columns are capacity limited, and therefore limit the amount of analyte injected. Consequently, high sensitivity is required of the detector to render the columns useful to the analyst. The Sulfur Chemiluminescence Detector (SCD®) is ideally suited for this task. The salient features of the SCD® contrast with the weaknesses of the traditional Flame Photometric Detector (FPD). The SCD® possesses greater intrinsic sensitivity and lower signal to noise which when applied, yields better detection levels. Direct injections of carbon dioxide can achieve minimum detectable quantities in the order of 5ppb without increasing sample volume to column overload. The dynamic range of the SCD® is >105, the response is equimolar for all sulfur compounds. The carbon rejection is 108 and the calibration is achievable at low levels with a single permeation device. The family of analyzers described herein has been engineered to mount a chemiluminescence detector manufactured by Sievers Instruments.



System Design
Trace sulfur analysis is an analytical challenge. Many of the low boiling sulfur components are extremely reactive to other chemicals within their environment and to the surfaces they contact. Any system design must consider the reactivity of the matrix and consequently the yield of the specie as it is presented to the detector. Experience designing similar systems has indicated that passive sampling with containers can result in large percentages of the sulfurs of interest being lost on the container surface or reacted. For this reason, an on-line sampling approach was chosen. The design integrated the delivery or mixing pumps from the distribution systems with manifolding to deliver product to the gas chromatograph's inlet piping. A fast loop arrangement resulted in circulating volumes of product without waste, followed by valving to the GC and immediate analysis.

Since sulfur transport and recovery is critical to accuracy, the inlet piping to the GC sampling system is comprised of special metallurgy and/or is silica coated. All non-essential dead volume is designed out of the system. Injection ports are excluded as contributing to band broadening and sulfur loss. Valves are nickel with special composition rotors to prevent reactivity and memory effects. The column is connected directly to the sampling valves (gas or liquid permitted) with no reactive surfaces exposed to the sample. All wetted tubing is silica coated. Passivation is thus assured.

Instrument calibration must be performed on a periodic basis. The calibrants are prepared external blends from experienced specialty gas suppliers or pure sulfur compounds. Gas blends are expensive, usually contentious at ppb levels and have extremely short shelf lives. Pure sulfur compounds are available for accurate calibration and are contained in fixtures of porous polymers that diffuse the substance at predictable rates at constant mass flow and temperature. Unlike gas blends, the permeation tube is traceable to the National Bureau of Standards (NBS) and provides an unparalleled path to accuracy. This system has an engineered permeation device within the gas chromatograph's sampling path for ease of calibration. Each permeation tube shipped with a system is tested in the analyzer and an NBS traceable calibrator to ensure accuracy. The difference between the two calibration methods is typically less than 5%.

Permeation Chamber Design and Integration

Permeation devices are commercially available for this application. However, they are rather large and possess a footprint as large as the gas chromatograph. They also add substantial cost to any system, as much as 50% of the cost of the GC. We have designed a
small, all glass chamber as an integral part of the gas chromatograph.It is conveniently mounted in the space normally taken by a liquid autosampler. The chamber can contain multiple permeation tubes or wafers. Although the exact number depends upon the sulfur type and concentrations, four devices are easily placed in the chamber. The permeation tube chamber temperature is controlled by an injection port heated zone. Dilution gas (carrier gas) is controlled by an internal mass flow controller or a Programmable Pneumatic Control (PPC) zone.

A four port micro volume valve is used to pass either calibration gas from the permeation chamber or the actual sample to the gas sampling valve. A rotometer on the outfall of the gas sampling valve completes the sampling hardware. It is used to monitor correct dilution gas flow from the permeation chamber and to assure proper sample flow when the four port valve is in the sampling position.

Integrated System
The complete analyzer consists of a manifold sampling station consistent with the plant design and its mechanicals for moving carbon dioxide to storage, distribution or point of usage. The product to be tested is connected to the inlet piping of the gas chromatograph's sampling system, which contains the four port selection valve. The highly modified gas chromatograph with integral permeation chamber, gas sampling valve, high-resolution methylsilicone capillary column, and sulfur specific chemiluminescence detector completes the analytical train. The detector signal is processed by an amplifier within the gas chromatograph and sent to a computer-based integrator. The integrator stores a factory method for operation of the system. All operational parameters and calibration details are provided with the system to minimize installation and training.



Analytical Cycle
Calibration is accomplished by positioning the four port selection valve to allow gas to flow from the permeation chamber to the gas sampling valve. The calibrant is dimethysulfide for the standard system. Because the SCD® yields an equimolar response to all sulfur compounds, only one is necessary for calibration. Dimethylsulfide is used because it is less ordorous and toxic than others that might be present. It is also very likely to be present from certain manufacturing processes. After a sufficient purge of the sample loop, the calibrant is injected by pushing the "Start" button. The rest is automatic. An isothermal elution is held to separate hydrogen sulfide followed by a mild temperature ramp to elute all the sulfides of interest, typically through the carbon four mercaptans. The elution time is less than five minutes. If sulfur dioxide is required to be separated from carbonyl sulfide, a mild cryogenic start is needed. Since copious quantities of carbon dioxide are always present, this does not add to the cost of the analysis.

After the completion of the column elution, the integrator processes the recorded detector signals. Each sulfur peak exceeding the method threshold is compared to the calibration file for retention matches for peak identifications and to the intensity of the calibration component (dimethylsulfide). A report is then printed and the next cycle started.

Detection Limits
The detection limit for the SCD® is <1 pg sulfur per second. Practical detection limits are on the order of 5 ppb for gas or liquid samples when analyzed with this system. Large benefits have been realized from designing to minimum internal volumes, passivation of surfaces and the resulting signal to noise that such system improvements yield.

Conclusion
A complete solution has been designed to permit beverage grade carbon dioxide to be accurately analyzed for sulfur compounds at low ppb levels. Sulfur speciation is accomplished by the use of high-resolution capillary chromatography with specificity from a chemiluminescence detector. Design criteria have dictated the analyzer to be easy to use and calibrate. The skill level at plant sites is usually nontechnical and often scalemen and truck drivers are trained to use this equipment. The reliability and uptime approaches that of process-type equipment with minimal maintenance and lower initial capital investment. The integration of the analytical equipment is complete when on-line sampling is employed.

Although this article described only the sulfur analysis requirement, many other impurities have been identified and must be measured as part of the total purity specification. The sulfur analyzers have been extended into a family of analyzers dedicated to beverage grade carbon dioxide analyses. Additional channels can be added to measure most organic impurities present. All are on-line and completely automated for ease of use.