DOI.56/sensoren6/P3. QLAS Sensor for Purity Monitoring in Medial Gas Supply Lines Henrik Zimmermann, Mathias Wiese, Alessandro Ragnoni neoplas ontrol GmbH, Walther-Rathenau-Str. 49a, 7489 Greifswald, Germany Loioni, Via Fiume 6, 63 Angeli di Rosora, Anona, Italy Abstrat Beause of their diret impat on patiene, medial supply lines underlie strit regulations and have to be monitored in terms of purity on regular base. State-of-the-art measurement solutions do not allow ontinuous bed-side monitoring. The aim of the presented projet work is to provide a ompat multispeies monitoring system based on latest quantum-asade-laser tehnologies. Keywords: gas-sensor tehnology, optial measurement methodology, quantum-asade-laser, laser absorption spetrosopy, purity measurements Introdution The neoplas ontrol GmbH is a spin-off of the INP Greifswald (Leibniz-Institute for Plasma Researh and Tehnology). The solutions offered are ustomized devies and produt speifi servies for gas and plasma diagnostis in industrial as well as eonomi environments. While implementing this business idea, lose ooperation with both the INP Greifswald and its transfer enter, the neoplas GmbH, are maintained. The produts of the Q-MAS family (Quantum asade Laser Measurement and ontrol System) are ompat and user friendly. The Quantum asade laser measurement and ontrol systems have been developed to monitor and ontrol plasma proesses and to detet highly sensitive trae gases []. Q-MAS ombine the advantages of absorption spetrosopy in the mid infrared spetral range with the unique properties of Quantum asade Lasers (QL). In ombination with thermo-eletri ooled infrared detetors Q-MAS is well suited for industrial appliations, in partiular for on-line proess monitoring. The Q-MAS produts need almost no maintenane and an be individually adapted to ustomer requirements inside industrial proesses due to its modular design. With the produt series Q-MAS it is possible to detet lowest absolute onentrations of moleules in gaseous media up to parts per billion (ppb) in real-time. Within the sope of the EU funded projet MIRIFISENS the issues of sensitivity and seletivity, multi-gas apabilities, ompatness, effiieny and ost effetiveness as speified by a number of seleted safety and seurity appliations are addressed []. In this ontext the neoplas ontrol GmbH in ooperation with the projet partner Loioni is fousing on the on-site purity monitoring of medial gases used in hospitals. The main 3 supply lines to hek are the oxygen line, the nitrogen protoxide line and the ompressed air line. ontinuous monitoring on the distribution lines ould avoid fatal aidents that unfortunately still happen to patients, espeially after maintenane on older buildings. onentration limits are defined for every gas line by loal regulatory. The main moleules to be measured are O, O, H S, SO, NO x, H O, O, N O and Oil. Expensive instrumentation has been developed to perform periodial hek to every gas port, but these systems do not allow the bed site operation. Therewith a ompat lower ost sensing tehnology ensures higher safety potential as it ould be diretly integrated in the distribution plant. As a first step a single speies monitor fousing on the hallenging high sensitive detetion of SO is developed and forms the basis for the intended multispeies solution based on MIRIFISENS tehnology. Experimental setup As a laser absorption spetrometer based on diret absorption tehnique the Q-MAS Trae ompat SO sensor aquires the transmission signal after passing the 56 m long path ell. This signal will be analyzed with respet to well-known moleular parameters, what ensures almost alibration free 8. GMA/ITG-Fahtagung Sensoren und Messsysteme 6 566
DOI.56/sensoren6/P3. measurements. In terms of the introdued SO measurements these parameters are taken from the HITRAN moleular absorption database [3]. In Figure a a representative detetor signal aquired at the hosen Full Sale value of 7 ppm and the orresponding fit result is shown. The sheme of the experimental setup is shown in Fehler! Verweisquelle konnte niht gefunden werden.b. Figure : (a) Aquired detetor signal with SO absorption feature and fit result. (b) Experimental onfiguration. The gas dilution system onsisted of two Unit MKS MF mass flow ontrollers and a diaphragm pump Thomas 7Z D. One ontroller provides a range from to sm and another one omes with a range from to slm. This set of flow ontrollers allows reasonable auray for the dilution of gas standards over a very wide range of dilution ratios by seleting the appropriate settings on the mass flow ontrollers. The distint onentration steps, whih were alternatingly tuned as ontinuous flow through the measurement ell of.5 l volume, are refleted in Table. Table : List of dilution steps and resulting SO mixing ratios and ahievable auray Target / ppb SO / sm N / sm Auray / ppb 3 +5 / - 7 98 +3 / -9 7 5 95 +35 / -3 35 9 +55 / -47 69 8 +94 / -8 7 5 5 +9 / -8 35 +395 / -344 5 5 5 +574 / -5 7 +747 / -655 This dilution system was used to provide the test atmospheres for the analyzers under test. The pressure in the sample ell was monitored via a MKS HPS Series 9 piezo transduer, whih provides a range of appliation from to 3 mbar. Using this pressure gauge the pressure in the ell was ensured to be a fixed value of ~6 mbar during the tests. An ative pressure ontrol was not applied during this test. Statistial Methods The statistial methods used to evaluate the quantitative performane fators are presented in this hapter. Beause no alternative onept for the detetion of SO was available during the test, the evaluation of the performane parameters for the sensor has to be based on the alulated mixing ratios, what depend on the settled flow rates. This approah limits the types of statistial omparisons, whih ould be applied. Qualitative observations were also used to evaluate verifiation test data. Linearity fator is assessed by linear regression with the alibration onentration as the independent variable and the analyzer response as the dependent variable. The alibration model is given in Eq. (). h error Where hallenge onentration, h() is a linear alibration urve, and the error term is assumed to be normally distributed. Variability of the measured onentration values was modelled by the relationship expressed in Eq. (). k Where, k and are onstants to be estimated from the data. After determining the relationship between the mean and variability, appropriate weighting is determined by Eq. (3). w The form of the regression model to be fitted is expressed in Eq. (3). h ( ) onentration values were alulated from the estimated alibration urve using the formula Eq. (4). h A test for departure from linearity is arried out by omparing the residual sum of squares to a hi-square distribution with 6 - = 4 degrees of freedom, as given with Eq. (5). 8. GMA/ITG-Fahtagung Sensoren und Messsysteme 6 567
DOI.56/sensoren6/P3. 6 i i i n i w i Where n is the number of repliates at onentration. The response time of the analyzers to a step hange in analyte onentration was alulated by determining the total hange in response due to the step hange (either inrease or derease) in onentration, and then determining the point in time when 95% of that hange was ahieved. Both rise and fall times were determined. Using data taken every seond, the alulation is arried out by Eq. (6). R R R Total Where R a is the final response of the analyzer to the test gas after the step hange, and R b is the final response of the analyzer before the step hange. The analyzer response that indiates the response time then is alulated applying Eq. (7). R. 95 R Total The point in time at whih this response ours is determined by inspeting the response time data, and the response time is alulated aording to Eq. (8). Time Response Time95 % Time I Where Time 95% is the time at whih R ours, and Time I is the time at whih the step hange in onentration was imposed. Sine only one determination was made, the preision of the rise and fall time results ould not be estimated. The detetion limit (LOD) was defined as the smallest true onentration at whih the analyzers expeted response exeeded the alibration urve at zero onentration by three times the standard deviation of the analyzers zero reading. The LOD is then determined by applying Eq. (9). 3 3 LOD Here is the estimated standard deviation at zero onentration. Note that the validity of the detetion limit estimate and its standard error depends on the validity of the assumption that the fitted linear alibration model aurately represents the response down to zero onentration. Statistial proedures for assessing zero and span drift were similar to those used to assess interrupted sampling. Zero (span) drift was alulated as the arithmeti differene between zero and span values respetively obtained before and after sampling of soure emissions. During this test no estimate of the preision of the zero and span drift values was made. a b Results and Disussion A Q-MAS Trae analyzer prototype was tested for the highest sensitive online monitoring of SO traes. The laboratory tests were designed to hallenge the analyzer over its full range under a variety of onditions. These tests were performed using ertified standard gases and a gas dilution system. The gas standards were diluted with high-purity gases to produe the desired range of onentrations with known auray. Laboratory testing was onduted primarily by supplying known gas mixtures to the Q-MAS Trae analyzer from the gas delivery system. The linearity of response of the Q-MAS Trae analyzer will be tested by 3-point alibrations of the SO gas filling. Prior to this hek, the analyzer is provided with the appropriate zero gas (N ) and then with a span gas onentration of 7 ppm SO, whih is defined in this verifiation test to be the nominal range of the analyzer. After any neessary adjustments to the analyzer to math that span value, the 3-point hek proeeded without further adjustments. The 3 points onsisted of three repliates eah at 7 ppb, 7 ppb, 35 ppb, 69 ppb,.7 ppm, 3.5 ppm, 5. ppm and 7 ppm in random order, and interspersed with six repliates of zero gas. Following ompletion of all 3 points, the zero and % spans were repeated, also without adjustment of the analyzer. Zero and span drift will be evaluated using data generated in the linearity and the auray tests. The zero and span drift is determined as the differene in response on zero and span gases in these two tests. This omparison will be made for all zero and span responses, using data from the linearity and the auray tests. Figure shows the linearity results obtained from the linearity tests for the Q-MAS Trae analyzer, whih was onfigured for the SO measurements. Figure : linearity results for the Q-MAS Trae set up with seond aquisition time. In Table the linear equations for the system onfiguration developed from this data is shown. 8. GMA/ITG-Fahtagung Sensoren und Messsysteme 6 568
DOI.56/sensoren6/P3. Table : Statisti results of the linearity test. Interept / ppb Slope R -.7 ± 3.74.48 ±..83 The quite low value for the regression oeffiient in the linearity tests is aused by the rather high unertainty the hosen dilution system provides for the N dilution gas at lower flow rates. The unertainties of the dilution system are listed in Table as well as visualized in terms of error bars in Figure. For the linear fit the onentration values are weighted with the unertainties of the dilution system to respet the deviation of the finally settled onentrations from the targeted ones. Figure 3: results of linear fit for measured values at Zero and.67 % of range. Table 3: Detetion limit determined aording to SO measurements. Slope.969 ±.86 Standard Deviation 4.6 for Zero values / ppb Average Noise / ppb 7.67 Detetion Limit / ppb.99 Table 3 shows the detetion limits for eah onfiguration of the Q-MAS Trae analyzer tested, determined from the detetion limit proedure desribed in the former setion. Figure 3 visualizes the results for the linear fit valid for the range from Zero to.67 % of range. The alulated detetion limit for seond aquisition time is.99 ppb. The response time results for the sensor based on a step hange in analyte onentration was determined to about 4 ± s, when limiting to an aquisition time of seond. Figure 4: trend of SO onentration during the zero / span test. In Figure 4 the trend of the SO onentration signal is shown. Zero and span data taken at the start and end of the linearity test are shown in Table 4. The drift values observed are shown in Table 5 as differenes between the pre- and post-test onentration measurements. Furthermore, Table 5 also presents the zero and span drifts as a perent of span gas onentrations. The Zero drift for the sensor tested was less than.4 % of the respetive span gas onentration. The span drift was less than. % of the respetive span onentration. Table 4: Data used to assess Zero and Span drift of the Q-MAS Trae ompat SO analyzer. omponent Sensor SO / ppb Pre-Test Zero.64 ± 8.48 Pre-Test Span 7674.57 ± 8.3 Post-Test Zero 3.3 ± 7.4 Post-Test Span 7689.83 ± 6.6 Table 5: Results of the Zero and Span drift of the SO analyzer. omponent Differene / ppb Zero.49 Span 5.6 Drift in % of Span - Zero.4 % - Span. % The final verifiation tests were onduted with support of the Leibniz-Institute for Plasma Siene and Tehnology Greifswald. Summary and onlusions The test results, whih are summarized in Table 6, onfirm that the Q-MAS Trae analyzer provides linear response over wide operating ranges. The ompat prototype onfiguration as used in this preliminary study provides very good results with respet to the sensitivity, seletivity and stability. Table 6: Results from performane analysis for the Q-MAS Trae SO Sensor. Speifiation Value Response time 4 ± s Linearity.48 Detetion limit > 3 ppb Drift Zero <.4 % Span <. % The system is rugged and portable, and the neessary setup time is minimal. The quite fast sensor response times and measurement stability allowed verifiation testing to proeed smoothly. Its design inorporates a sample 8. GMA/ITG-Fahtagung Sensoren und Messsysteme 6 569
DOI.56/sensoren6/P3. probe and sample onditioning system, making it adaptable to a wide range of measurement appliations. Although the aim of 5 ppb and below for the limit of detetion was not ahieved with this Q-MAS Trae ompat onfiguration, the positive results show that it is possible to design a system, whih will fulfil this speifiation. Reently long path ells with more than 5 meters optial path length beame available. Unfortunately suh ell would not allow keeping or even further optimizing the ompatness of the resulting sensor system. Moreover, the muh higher volume of suh a ell would lead to a signifiant and unwanted inrease of the response time. More pratial approahes are the seletion of stronger absorption lines, what ould be possible depending on the gas matrix to be analyzed in the speifi appliation. Developments for further optimization of the gas handling, e.g. by the integration of an ative pressure ontrol, as well as the extension of the analytial methods are atually in progress. Applying these drafted optimizations in a future onfiguration will allow dereasing the ahievable detetion limit by approx. a fator of 5 and therefore to detet SO onentrations below 5 ppb by a ontinuous monitoring system. In pending development steps the single QL soure will be exhanged by a QL to address several moleular speies in parallel by maintaining the ompatness of the system. First introdued by Lee et al. in 9 [5], QL arrays will now make the step from the lab to the market. This was made possible in the rather short timeframe thanks to funded researh and development projets like MIRIFISENS. Referenes [] J. Röpke, et al., J. Phys. D: Applied Phys. 45, pp (); doi:.88/- 377/45/4/43 [] Mid InfraRed Innovative lasers For Improved SENSor of hazardous substanes, FP7-IT-.3.5 ore and disruptive photoni tehnologies, http://www.mirifisens-projet.eu. [3] L.S. Rothman, et al., Journal of Quantitative Spetrosopy and Radiative Transfer 3, 4-5 (3); doi:.6/j.jqsrt.3.7. [4] M. Fisher, et al., Optis Express, 74-77 (4); doi:.364/oe..74 [5] B. G. Lee, et al., Optis Express 7, 66-64 (9); doi:.364/oe.7.66 8. GMA/ITG-Fahtagung Sensoren und Messsysteme 6 57