244
`
`Miniaturized Total Chemical Analysis Systems: a Novel Concept for Chemical Sensing
`
`A. MANZ, N. GRABER
`and H. M. WIDMER Central Analytical Research, Ciba-Geigy AG, CH-4002 Bare1 (Switzerland)
`
`Abstract
`
`Following the trend towards smaller channel inner diameter for better separation performance and shorter channel length for shorter transport time, a modular construction of a miniaturized ‘total chemical analysis system’ is proposed. The theoretical performances of such systems based on flow injection analysis, chromatography and elec- trophoresis, are compared with those of existing chemical sensors and analysis systems.
`
`Introduction
`
`A miniaturized TAS relation to a chemical (Fig. 3). Fig. 1. Typical concentrations (moles per liter) of known chemical components (molecular weight less than 1000) in human blood Serum [l]. 0925-4005/90/$3.50 0 Elsevier Sequoia/Printed in The Netherlands An ideal sensor is specific, i.e. it transduces the concentration of the chemical compound of inter- est, to the exclusion of all others, into an electrical signal. In addition, the sensor can be immersed directly into a sample solution or stream. This concept is the goal of many sensor technologists, but so far the results (selectivity, lifetime) have been unconvincing. Analytical chemistry offers a great number of methods for the analysis of almost every com- pound in any environment. Most of these meth- ods are time consuming and require a fully- equipped chemical laboratory and a qualified technician. Some of these techniques, such as chromatography, electrophoresis or flow injection analysis can be integrated into a TAS. The detec- tor or sensor in a TAS does not need high selectiv- ity, because the sample pretreatment serves to eliminate most of the interfering chemical com- pounds. Furthermore, calibration can be incorpo- rated into the system. If a TAS performs all sample handling steps extremely close to the place of measurement, then we propose that it be called a ‘miniaturized total
`The continuous monitoring of a chemical parameter, usually the concentration of a chemical species, is gaining increasing attention in the chem- ical production, environmental and medical sci- ences. The chemical compound of interest is usually accompanied by interfering species. Figure 1 shows the average concentrations of all known compounds in human blood serum. If the com- pound of interest has a concentration of lo-’ moles per liter (several ppm), the analysis system must be sufficiently selective to reject at least 100 compounds of higher concentration. The state-of- the-art strategy for solving analytical problems like these is the introduction of a ‘total chemical analysis ystem’ (TAS), which periodically trans- forms chemical information into electronic infor- mation. Sampling, sample transport, any necessary chemical reactions, chromatographic separations as well as detection are automatically carried out (see Fig. 2, for examples see ref. 2). This approach presents a possibility for accommodating the rapidly changing composition of industrial samples (e.g., river water, chemical reaction mixtures, fer- mentation broths). In this paper we present a general concept for a miniaturized TAS. Concept must be defined both in sensor and to a TAS 0 20
`
`80
`60
`40
`100 120 140 160 180
`NIJMHKR OF’ CHEMICAI.
`I’AI<AMF:‘W?S
`
`zoo
`
`Agilent Exhibit 1287
`Page 1 of 5
`
`Sensors and Actuators, BI (1990) 244-248
`

`

`245 MASS FLOW INFORMATION FLOW SIGNAL EVALUATION INTERPRETATION TRANSPORT Fig. 2. General flow chart of a quantitative chemical analysis. IDEAL SENSOR _/ EfCTF&;;S 1 I !! I IAl L J _. ELECTRONICS RECORDER p -TAS I Carriers, Reagents, Moblle Phases Hydrauhc Control, Waste Fig. 3. Schematic diagram of an ideal chemical sensor, a ‘total chemical analysis system’ (TAS) and a miniaturized TAS (JJ-TAS). chemical analysis vstem’ (p-TAS). The interface to the control and measurement electronics could include, for instance, tubing for mass flow and optical fibers. If the analysis time of a p-TAS is comparable to ,the response time of a selective chemical sensor, both are very similar in appear- TIME -3it response time HIGHLY SELECTI\ t: CHE\lI(‘\L -‘E\.GOli -. TIME response time janalysls time SIG ‘AL p ~ FLOW INJECTIOS DETECTOR .AN\LY’lS nc K 3 analysis time K: TIME cycle time + analysis time M cycle time Fig. 4. Comparison of response time, analysis time and cycle time for an ideal chemical sensor, a flow injection analysis- based and a chromatography-based TAS. ante and use (Fig. 4). Several research groups have done basic developmental work on micro pumps and valves [3,4], small flow injection analysis systems [S] and open-tubular column chromatog- raphy [6,7]. Our mai; reason for the miniaturiza- tion of the TAS is related to an enhancement of its analytical performance, rather than a reduction of its size.
`
`General Let d be a typical length in a given system (for example the diameter of a tube). By multiplying each variable by d” and the appropriate constants, it can be reduced to a dimensionless parameter which is independent of the spatial scale of the given system (for example the flow rate and the P&let number). Similar systems of different sizes can easily be compared. If we assume that a minia- turization is a simple three-dimensional down- scale, we can easily demonstrate the behaviour of the relevant physical variables. There remains one degree of freedom for mechanical parameters: time. Time Constant System In this case, the time scale is the same for the large and for the small system. Consequently, all
`
`Theory and Discussion
`
`Agilent Exhibit 1287
`Page 2 of 5
`
`

`

`246 relevant time variables (analysis time, transport time, response time) don’t change. The conse- quences are shown in Table l(a). The D@ksion-controlled system becomes im- portant when molecular diffusion, heat diffusion or flow characteristics control the separation efficiency in the given system. In this system the time scale is treated as a surface, i.e. time is proportional to d*. This means that a down-scale to l/10 of the original size (diameter of a tube) reduces the related time variables (transport time, required response time of a detector) to l/100. The Reynolds number remains constant, but the pres- sure requirements increase by a factor of 100. In the case of some electrical parameters the time scale must be constant in order for the definitions of the electric current, the electrical capacity and Ohm’s law to be consistent. Never- theless, three possibilities are presented in Table l(b): constant charge density (system a.I), con- stant electrical field strength (system a.11) and constant voltage (system a.111). Experiments are needed to show which system is best suited to miniaturization. For any given system, it is possible to start from one point and extrapolate to get the order of magnitude of the variables of a scaled-down sys- tem. Changes in geometry only change the estima- tion by a constant factor. Although these con- siderations do not contribute to a prediction of the feasibility of a system, they can, in principle, lead to the exclusion of impossible cases and give an idea of the order of magnitude. TABLE l(a). Proportionalities of some mechanical parame- ters in relation to the characteristic length d. System a: time scale remains constant during miniaturization, system b: time scale compensates for diffusion Time constant Diffusion control system a system b Time constant
`Reynolds number constant Pressure $ d-2 Pressure drop (laminar flow) constant d-2 Pressure drop (turbulent flow) di.5 d-2 TABLE l(b). Proportionalities of some electrical parameters in relation to the characteristic length d. (a.1): charge per volume remains constant, (a.11): electrical field strength re- mains constant and (a.111): voltage remains constant Time constant system a Time Space Ohmic resistance Electrical capacity constant d a.1 a.11 a.111 Electrical charge Voltage Electrical field strength Electrical current Magnetic field strength
`
`d constant Liquid Flow Figure 5 shows the laminar flow rates required for time constant (flow injection analysis) and diffusion-controlled tubing systems (chromatogra- phy, electrophoresis). A pressure gradient yields flow rates proportional to those needed in a time constant system, regardless of the spatial scale. The electroosmotic flow generated by an electrical field remains constant as long as the electrical field is kept constant during miniaturization. In addition, the reduced production of heat might allow higher electrical field strengths with smaller tube diameters. Electroosmotic propulsion can therefore meet the demands of separation systems better than a pressure-driven flow (limited to aqueous electrolyte solutions).
`
`Force constant Energy (work) 1:
`d5
`
`d
`d-’
`
`d2
`
`d
`
`d
`
`constant constant
`
`d-’
`
`Ll\li:AH FLOW RATE
`
`d
`
`1
`
`d2
`d
`
`1
`
`Linear velocity d d-’ Volume flow rate d3 d Linear acceleration d d-’ Angular velocity constant d-2 Impulse
`
`d4
`
`r
`
`100 “In lm 14 “rn
`
`100
`TYPICAL TUBE DIAMETER
`
`pm 1 mm
`
`Fig. 5. Linear flow rate as a function of the tube inner diameter. The required flow rates for a time constant flow
`system
`(FIA:
`flow
`
`injection analysis) and for dilBrsion-con- trolled separation systems (LC/SFC/CZE: liquid chromatogra- phy, supercritical fluid chromatography, capillary zone electrophoresis) are compared with Bow rates resulting from a pressure drop of 300 bar/m and of an electric field of 10 to 1000 kV/m.
`
`Agilent Exhibit 1287
`Page 3 of 5
`
`SPa=
`Power
`

`

`241
`
`Qc_
`I
`I
`IOfiL
`I pL
`(LOGARITHMIC)
`
`1M
`
`’
`‘LC CHIP’
`OPTIMUM KT;:;;
`m
`;;
`IL 100 fL
`
`Sample Pretreatment Almost all methods of standard flow injection analysis [8] can be adopted. In a time constant flow system, the effect of turbulence decreases (Reynolds number changes, Table l(a)) whereas molecular diffusion increases. A simple ‘T’ can thus function as a good mixing chamber in a miniaturized system. Separation techniques can be considered as a special case of a sample pretreatment. In Table 2 the operational conditions needed to attain equal separation performances (number of theoretical plates) with electroosmotic chromatography [9], open-tubular liquid chromatography and super- critical fluid chromatography are compared. With increasing separation performance the open-tubu- lar column inner diameter must decrease. A sepa- ration corresponding to 100 000 theoretical plates in 1 min is, by far, better than the experimental state of the art obtained with larger capillary diameters. Detection As Table 2 indicates, the detection volume restrictions are drastic in high performance sepa- ration systems (in the order of picoliters). The relationship of the signal output to the size of a detection system is critical for the detection limits in small volumes. For example, a fluorescence detector signal is proportional to d3 and an am- perometric to d*. Refractive index and potentio- metric detectors are almost totally insensitive to vohune changes, as was experimentally proved with a Ca2+-selective electrode [lo, 111. Despite the excellent detection limits obtained with fluorescence detectors, there must exist a detection volume at which potentiometric detectors are superior (see Fig. 6). In a flow injection analysis system, the detection vohune would be compara- bly greater. Generally, a definite limit is provided by the concentration at which exactly one molecule exists in the closed detection volume (Table 3(a)). This only holds for non-buffered molecules of interest. With pH measurements, for example, the total s -11
`Fig. 6. Detection limits as a function of the detection volme for refractive index, potentiometric and fluorescence detectors. FIA: flow injection analysis, LC: conventional liquid chro- matography, SFC: capillary supercritical fluid chromatogra- phy, CZE: capillary zone electrophoresis, LC ETI-I capillary LC [12], LC CHIP HITACHI: see ref. 7. TABLE 2. Calculated parameter sets for a given separation performance obtained with capillary electroosmotic (EC), liquid (LC) and supercritical fluid chromatography (SFC). Assumed constants are: diffusion coetlicients 1.6 x 10e9 m2/s (LC, EC) and 10-s m2/s (SFC), viscosities of the mobile phase 1O-3 Ns/m2 (LC, EC) and 5 x lo-’ Ns/m2 (SFC), electrical conductivity of the mobile phase 0.3 Siemens/m (EC), electrical permittivity x zeta potential 5.6 x IO-” N/V (EC), heating power 1.1 W/m (EC) Parameter Electroosmotic Liquid chromatography chromatography No. theoretical plates Analysis time Heating power Capillary i.d. Capillary length Pressure drop Voltage Peak capacity Signal bandwidth Detection volume Response time Injection pulse Stop time N t (k’ = 5) (mitt) p/L (W/m) d Olm) L (cm) P (atm)
`1 I 1 1 2.8 0.9 6.9 2.2 8.1 26 20 64 26 2600 1.4 140 220 700 220 700 0.56 0.56 1.4 1.4 70 22 70 22 3.3 0.33 52 5.2 0.8 0.08 1.2 12 16 5 16 5 1.5 49 0.075 2.4 5.1 5.1 5.1 5.1 Supercritical fluid chromatography
`n e (mm) c (ms) e (Pl) F (Pl) t (ms) p*t (s*atm) U*r (s*kV) 1 (s) 1OOk 1M 10M 1 1 1 1.1 1.1 1.1 24 7.6 2.4 6.5 21 65 5.8 58 580 180 510 >2000 0.21 0.21 0.21 42 13 4.2 94 9.4 0.94 47 4.7 0.47 21 6.5 2.1 0.41 1.3 4.1 3.3 3.3 3.3 IOOk IM 1OOk
`
`-12
`
`G
`F
`g
`
`IO
`
`\
`j
`i K ETH
`
`/
`Cf
`
`SYC
`
`1 pL
`
`10 pL 100 pL
`
`10 nL 100 nL
`I nL
`DETECTION VOLUME
`
`u(kV)
`
`Agilent Exhibit 1287
`Page 4 of 5
`
`

`

`same
`
`conditions). A p-TAS will cause less problems than a chem- ical sensor in terms of selectivity, because well- known techniques of analytical chemistry can be applied as sample pretreatment. The integration of separation techniques enables multi-component monitoring with a single device.
`
`248 TABLE 3. Theoretical limitations of detection for non- buffered solutions of small molecules: (a) concentration and corresponding volume, which contains exactly one molecule, (b) response time of a detector and corresponding uncertainty in length or volume caused by molecular diffusion. Diffusion coefficient lo-’ m*/s (a) Closed volumes: less than 1 molecule per volume Concentration Detection Corresponding log (mol/l) volume length -3 0.002 al 12 nm -4 0.02 al 26 nm -5 0.2 al 55 nm -6 2al 120 nm -1 20 al 260 nm -8 200 al 550 nm -9 2fl 1.2pm -10 20 fl 2.6 pm -11 200fl 5.5 pm -12 2 Pl 12 pm (b) Infinite volumes: uncertainty relation (space and time) Response Detection Corresponding time volume length 10 /1s 1 al 0.1 pm loo/Q 32 al 0.32 pm 1 ms lfl lpm 10ms 32 fl 3.2 pm 1OOms 1 Pl 10 pm 1s 32 pl 32 pm 10 s 1 nl 1OOpm 1.5 min 32 nl 320 pm 15 min 14 lmm concentration of protons is, by far, larger than the concentration of free H+. An uncertainty relation caused by the diffusion of molecules exists even in large detection volumes. A given response time of the detector dictates the spatial resolution of a measurement (Table 3(b)).
`A basic theory of hydrodynamics and diffusion indicates faster and more efficient chromato- graphic separations, faster electrophoretic separa- tions and shorter transport times for a miniaturized TAS. The consumption of carrier, reagent or mobile phase is dramatically smaller. A multi-channel device would allow the simultaneous performance of a large number of measurements (under the
`Wissenschaftliche Tabellen Geigy, Vol. 2, Ciba-Geigy, Basel, 1981, p. 78. N. Graber, H. Lildi and H. M. Widmer, The use of chemical sensors in industry, Sensors and Actuators, BI (1990) 239-243. S. Shoji, M. Esashi and T. Masuo, Prototype miniature blood gas analyser fabricated on a silicon wafer, Sensors and Actuators, I4 (1988) 101-107. H. T. G. van Lintel, F. C. M. van de Pol and S. Bouwstra, A piezoelectric micropump based on micromachining of silicon, Sensors and Actuators, 15 (1988) 153-167. J. Ruzicka and E. H. Hansen, Integrated microconduits for flow injection analysis, Anal. Chim. Acfa, Ml(l984) l-10. S. C. Terry, J. H. Jerman and J. B. Angell, A gas chro- matographic air analyzer fabricated on a silicon wafer, IEEZ T&s. Electron-Devices, ED-26 (1979) 1880- 1886. A. Manz. Y. Mivahara. J. Miura. Y. Watanabe, H. Miyaai and K. Sate, Design ‘of an open-tubular column liquid chromatograph using silicon chip technology, Sensors and Actuators, 81 (1990) 249-255. 8 J. Ruzicka and E. H. Hansen, Flow Injection Analysis, Wiley, New York, 1988. 9 S. Terabe. K. Otsuka and T. Ando, Band broadening in electrokinetic chromatography with micellar solutions and open-tubular capillaries, Anal. Chetn., 61 (1989) 251-260. 10 U. Schefer, D. Ammann, E. Pretsch, U. Gesch and W. Simon, Neutral carrier based Ca*+-selective electrode with detection limit in the sub-nanomolar range, Anal. Chem., 58 (1986) 2282-2285 (Figs. 3 and 4). 11 D. Ammann, T. Biihrer, U. Schefer, M. Mtlller and W. Simon, Intracellular neutral carrier based Ca*+ microelec- trode with subnanomolar detection limit, pflligers Arch., 409 (1987) 223 (Fig. 2). 12 A. Manz and W. Simon, Potentiometric detector for fast high-performance open-tubular column liquid chroma- tography, Anal. Chem., 59 (1987) 74-79; A. Manz, Potentiometrische Flilssigmembran-Mikroelektroden als Detektoren fiir schnelle und hochau5Bsende Kapillar- Fliissigchromatographie, Thesis, Swiss Federal Institute of Technology, Zurich, 1986 (in German).
`
`References
`
`Conclusions
`
`Agilent Exhibit 1287
`Page 5 of 5
`
`