Analytica Chimica Ada, 237 (1990) 115-125
Elsevier Science Publishers B.V., Amsterdam
115
Differential ion-selective membrane electrode-based potentiometric gas-sensing cells with enhanced
gas sensitivity
Hyoung Sik Yim, Geun Sig Cha and Mark E. Meyerhoff *
Department of Chemistry University of Michigan, Ann Arbor, MI 48109 (U.S.A.)
(Received 27th December 1989)
Abstract
A novel design for devising static and automated flow-through type potentiometric gas-sensing systems with
enhanced response slopes is described. The approach involves the use of two working gas-sensing electrode half-cells in
a differential measurement arrangement. One of the half-cells employs a pH-sensitive polymeric membrane electrode to
sense pH changes from diffusing analyte gas within a suitable inner electrolyte solution housed behind an outer
gas-permeable membrane. The second working half-cell is fabricated with an anion- or cation-selective membrane
electrode that responds selectively to the conjugate acid or base ionic form of the analyte gas trapped within an inner
buffer solution housed behind a similar gas-permeable membrane. When the two internal solutions of the half-cells are
in electrolytic contact, the differential response of the resulting gas-sensing scheme is significantly enhanced compared
with the response of a conventional single working electrode/reference electrode type gas cell. For the model analyte
gas ammonia, response slopes observed for both static and flow-through measurement schemes approach the 118 mV
decade-t predicted by theory. To demonstrate its analytical utility, the flow-through arrangement was used to
determine ammonia-N concentrations in bioreactor media with good correlation with conventional electrode and
enzymatic methods. The prospects of fabricating similar differential detection arrangements for CO,, NO* and SO, are
discussed.
Keywordr: Ion-selective electrodes; Ammonia; Nitrogen
Over the past two decades, potentiomeric gas- sensing devices have proved to be attractive ana- lytical tools for the direct detection of dissolved gases (e.g., NH,, CO,, NO,) in complex samples, including whole blood [l-4]. In addition, when used in conjunction with immobilized enzymes and intact cells, these gas sensors can provide a simple indirect means of quantitating a wide range of biomolecules via modern biosensing configura- tions [5-71. Whereas commercial gas sensors often used in these applications employ glass membrane pH electrodes as inner transducers in a so-called Severinghaus detection arrangement [8], the ad-
vantages of using polymetric pH sensors in an analogous measurement mode has been demon- strated previously [9]. In addition, the use of inter- nal polymeric ion-selective membrane electrodes responsive to ionic forms of the analyte gases (e.g., NH:, SO:-, CO,‘-, and NO;) to devise both static and automated flow-through potentiometric gas detectors with improved selectivity and detec- tion limits has been described [lo-161. In more recent preliminary studies [17], it has been shown that these two different approaches to gas detec- tion can be combined in a single differential potentiometric measurement cell to yield gas
0003-2670/90/$03.50 0 1990 - Elsevier Science Publishers B.V.
116 H.S. YIM ET AL.
sensor designs with significantly enhanced re- sponse sensitivities (as defined by the change in analytical signal/change in concentration, i.e., re-
sponse slope). The purpose of this paper is to provide a more detailed description of the general operating principles of this type of differential detection scheme, and to illustrate the adaptation of this new approach to a flow-through measure- ment arrangement with an analytically useful per-
formance. As with any electrochemical detector used in a
direct potentiometric measurement mode of analy- sis (i.e., non-titration method), conventional Severinghaus-style gas sensors and the newer poly- mer membrane-based devices cited above are sub- ject to precision limitations associated with the logarithmic response of such devices. Indeed, un- certainties in measured potentials of + 1 mV will result in deviations of &- 4% for sensors with slopes
of 59 mV decade-’ and +8% for those devices
based on response to divalent ions (e.g., based on inner sulfite and carbonate electrodes). One gen- eral approach suggested for enhancing the re- sponse slopes of potentiometric sensors is to use several membrane electrode based cells in series [18]. This arrangement results in response slopes n times the Nernstian value, where n is the number of two-electrode cells (working and reference) in series. Unfortunately, this approach requires a complicated array of sensors where the number of individual electrodes needed is increased, as is the number of separate sample compartments.
Alternatively, Cha and Meyerhoff [15] de- scribed a novel two-electrode differential potentio- metric cell for enzyme electrode systems that pro- vides enhanced substrate sensitivities when com- pared with conventional cells composed of a single working enzyme electrode and reference electrode. This concept involves the use of a cell with two working enzyme electrodes, one of which responds to the analyte in the positive potential direction via detection of cations, the other responding to the same analyte but in a negative detection owing to anion detection. As described here, a similar approach can also be adapted for the design of new two-electrode gas-selective sensors with en- hanced sensitivity to gases. Specifically, this con- cept is demonstrated by describing the perfor-
mance of new static and flow-through ammonia- selective gas-sensing systems with analytical re- sponse slopes approaching 118 mV decade- ‘.
THEORY
The proposed differential gas-sensing system is composed of two working gas probes, each with a different inner polymeric ion-selective membrane electrode as the transducer. For static configura- tions, this involves the use of two different gas- sensing half-cell probes where the internal filling
solutions are connected via a salt bridge (see Fig. 1). The overall notation for this type of electro-
chemical cell is shown as follows:
(2) (1) pH electrode 1 internal electrolyte [[internal buffer 1
membrane electrode sensitive to conjugate ion
One half-cell (2) responds to the basic or acidic analyte gas by detecting an increase or decrease in the pH of a thin film of electrolyte (e.g., am- monium chloride for an ammonia gas sensor) sandwiched between a pH-sensitive membrane (prepared with tridodecylamine as membrane ac- tive species [9]) and an outer gas-permeable mem- brane. The second half-cell (1) detects the analyte gas in the sample by responding to changes in the activities of the conjugate base anion or acid cat- ion of the analyte gas in a thin layer of buffer sandwiched between an ion-selective membrane
and another outer gas-permeable film. Accordingly, the overall measured potential for
such a two-working-electrode cell is the difference
in potential between the ion-selective electrode and the pH electrode:
E cell = Econj.ion - EpH (1)
or
E ce,, = K + (0.059/zi) log uco,,j,ion - 0.059 log au+
where uconj,ion (a,uZ+ or uA-) and au+ are the activities of the conjugate ions of analyte gas and protons in the thin films of internal solutions held between the outer gas-permeable membranes and the respective ion-selective membranes, and K is the sum of all constant potentials in the cell (e.g.,
FLOW-THROUGH TYPE POTENTIOMETRIC GAS-SENSING SYSTEMS 117
C a
cor\i. ion 44ctive membrane
B+&O,-BH++cH
HA+H_P=A-+&O+ I
B+l+O,E!H++Ctl
HA+H@=A-+&O+ I
BorHA
membrane
Fig. 1. Schematic diagram of differential gas sensor fabricated with two different polymeric ion-selective membranes: (a) 0.1 M BHCl or MA + NaCl; (b) 0.1 M phosphate buffer (pH 7.0) containing 0.1 M NaCl; (c) buffered internal solution such as 0.1 M Tris-HCl
(pH 7.9) for ammonia sensor; (d) pH-sensing internal electrolyte solution (see Table 1).
junction potentials at salt bridge, inner Ag/AgCl potentials of each membrane electrode). Diffusion of the basic or acidic analyte gas into the thin films results in the equilibrium hydrolysis reaction of the analyte gas:
B + H,O = BH++ OH- (for basic gases) (3)
HA + H,O = A-+ H,O+ (for acidic gases) (4)
with equilibrium constants of
K, = aBH +aOHe/PB (5)
KHA = aH+aAm/PHA (6)
where P, and P,, are the partial pressures of the dissolved gases. Therefore, since the pH of the film in contact with the ion-selective electrode is
highly buffered, at equilibrium, Uco,,j,ion (aaH+ or aA-) in the film is directly proportional to the partial pressure of the analyte gas:
aBH+= PBKB/%H- (7)
aA-= PHAKHA/aH+ (8)
On the other hand, for the pH electrode half- cell, since aOH-= Kw/aH+, then the activity of
protons in the thin film may be given by the following expressions for the two respective cases;
aH+= aB,+Kw/PBKB (9
aH+= PHAKHA/aAm (10)
However, in the pH half-cell, aconj.ion is kept high and relatively constant by using appropriate inter- nal solutions as the thin film electrolyte (e.g.,
NH: for NH, sensing, HCO; for COz sensing). Thus, substituting Eqns. 5 and 6 and Eqns. 7 and 8 into Eqn. 2 and combining all the constant terms together yields the following expression for the overall differential cell potential:
Ecel, = K’ + (0.059/z; + 0.59/l) log P,,, (11)
where zi is the charge on the conjugate anion or cation sensed in the half-cell containing the inter-
nal buffer solution. As can be seen, such a cell should respond to the partial pressure of the acidic or basic gas with a theoretical slope of 118 mV
decade-’ if the conjugate acid or base of the detected gas has a charge of f 1. If the charge on the detected ion in half-cell 2 is k2, then the
118 H.S. YIM ET AL.
W W
i---------l
I________________.
Sample
Gas Dialyzers Pump
Fig. 2. Schematic diagram of differential flow-through ammonia-N assay system. Rec., recorder, mV, pH-mV meter; pH, tubular pH
electrode; NH:, tubular ammonium electrode; PS, pulse suppressor; W, waste.
theoretical slope would be 89 mV decade-‘. In practice, such slopes will only be observed over limited concentration ranges of the analyte gas, owing to changes in the ionic composition of the internal electrolyte layers of each half-cell at in- creasing gas levels [20].
To adapt this type of differential detection scheme to flow-through arrangements (e.g., flow- injection analysis or segmented continuous flow
methods), it is convenient to use two gas dialysis chambers in which the sample and recipient solu- tion streams are flowing continuously, and the recipient solutions are allowed to contact each other downstream of the two ion-selective elec- trode detectors (see Fig. 2). In this case, depending on the exact geometry of the dialyzers used and the flow-rates of the sample and recipient streams, the detection process is most always a non-equi-
TABLE 1
Examples of some analyte gases and the associated half-cell detection chemistries required for the design of static and flow-through
differential gas-sensing arrangements with enhanced sensitivities
Analyte Pertinent equilibrium
gas reactions
Half-cell 1
(conjugate ion sensed in
buffered internal solution)
Half-cell 2
(pH-sensing)
electrolyte solution ’
Theoretical
slope b
(mV decade-‘)
NH, co*
NO,
HF
HOAc
Cl,
NH, + H,O G= NH: + OH-
CO,+H,O+HCO; +H+
HCO; = CO;- + H+
SO2 + H,O +HSO; +H+
HSO; = SO;- + H+
2N0, + H,O f NO; + NO; + 2H+
H,S + H,O = HS-+ H+
HS-it S*-+ H+
HF + F-+ H+
HOAc G= AcO-+ H+
Cl,+H,O=Cl-+ClO-+2H+
NH: [lo] ’
HCO;
CO;- [16]
HSO;
sof- [15]
NO; [14]
NO,- [22]
HS-
S2- [23]
F- [24]
OAc-
Cl- [25]
NH&l 118
MHCO, 118
MHCO, 89
MHSO, 118
MHSO, 89
MNO, 118
MNO, 118
MHS 118
MHS 89
MF 118
MOAc 118
MC1 118
a M represents alkali metal ion. b In restricted range governed by concentration of electrolyte used in half-cell 2 and also pH and
buffer capacity of internal solution employed in half cell 1. ’ Reference describing suitable ion-selective membrane electrode to sense
conjugate ion.
FLOW-THROUGH TYPE POTENTIOMETRIC GAS-SENSING SYSTEMS 119
librium one [21]. Nonetheless, if the fraction of gas transfer across the gas permeable membrane of each dialyzer is relatively constant with dis- solved gas concentration (this is not always the case; see Results and Discussion), then the potentiometric response of the system may still be expressed by Eqn. 11. The non-equilibrium nature of such flow-through measurements will, however, significantly affect the detection limits and dy- namic measuring range of these systems when compared with the static sensor configuration.
Table 1 summarizes the internal electrolyte conditions, sensing chemistries and ion-selective membrane electrodes required to adapt the dif- ferential method to the detection of various ana- lyte gases. Also listed are the predicted slopes for the resulting sensing configurations. While the use of two polymeric-type ion-selective membrane electrodes (one always being a pH-sensitive elec- trode) is convenient from cost and fabrication standpoints, in principle other types of ion-selec- tive membranes may be used, including pressed pellet solid-state, single crystal or glass. In some instances, practical implementation of this new approach will require the development of a suita- ble membrane electrode selective for the conjugate base anion of the respective gas. In others, the
Gas Dialyzers
required membrane electrode with suitable selec- tivity has already been described in the literature and an appropriate reference is provided in Table 1. For the purpose of this paper, the ammonia- sensing arrangement serves as a convenient model, since the required NH:-selective membrane can be readily prepared by incorporating the antibiotic nonactin into a plasticized PVC membrane [lo].
EXPERIMENTAL
Apparatus All potentiometric measurements were made
with a Fisher (Romulus, MI) Accumet Model 620 pH-mV meter and recorded on a Houston Instru- ments (Houston, TX) Omni-Scribe strip-chart re- corder. A schematic diagram of the AutoAnalyzer manifold used to evaluate the flow-through dif- ferential ammonia gas-sensing system is shown in Fig. 3. The recipient buffer and electrolyte solu- tions and the sample were pumped through the system with an Ismatec (Zurich) variable-speed peristaltic pump. A Technicon (Tarrytown, NY) Sampler II served as the autosampler, and was operated at a throughput of 30 samples per hour. Two 12-in. Technicon dialyzer blocks fitted with
pH Electrode- HA + H20 G= A- + H30+ Recipient Electrolyte
B + H204= BH+ + OH- - Stream
To Waste - HAorB Sample Stream
4
A- or BH+ electrode HA + Ha0 =? A- + H30+ Recipient Buffer
- Stream B + H20= BH+ + OH-
-------- ____________________-_____ Gas Membrane@
t1 -------- -_______-______________-_-
To Waste - HAorB Sample Stream
+
Fig. 3. Expanded view of each gas dialyzer unit and the chemical processes that take place within.
120 H.S. YIM ET AL.
0.2~pm pore-size PTFE membranes (W.L. Gore, Elkton, MD) served as the gas dialyzers.
Reagents All chemicals used were of analytical-reagent
grade or better. Standards and buffer solutions were prepared with distilled, deionized water. Nonactin was obtained from Sigma Chemical (St. Louis, MO), poly(viny1 chloride) (PVC), chro- matographic grade, from Polysciences (Warring- tone, PA), dibutyl sebacate from Eastman Kodak (Rochester, NY) and tridodecylamine (TDDA) and potassium tetrakis( p-chlorophenyl)borate (KTpClPB) from Fluka (Ronkonkoma, NY). Am- monia gas electrode internal filling solution ‘from Orion (Cambridge, MA) was used as the inner electrolyte of the pH-sensing half-cell in the static sensor design.
Static differential ammonia gas sensor
The pH and ammonium ion-selective polymer membranes were prepared as described previously [9,10]. After drying, pH and ammonium ion-selec- tive electrodes were fabricated by cutting a 2-mm diameter of the appropriate membrane and attach- ing them to the end of disposable plastic pipet tips having small pieces of Tygon tubing (2-mm diam- eter) at the end. Each membrane was sealed to the Tygon tubing with a paste made of tetrahydro- furan (THF), dibutyl sebacate and PVC (that used to cast the membranes except without ionophore). For the pH electrode, the inside of the plastic tube was filled with 0.05 M phosphate buffer (pH 7.5) containing 0.01 M NaCl. The ammonium-selective membrane electrode was filled with 0.1 M NH,Cl. The final static differential ammonia gas sensor was constructed as depicted in Fig. 1. The internal pH and ammonium electrodes were placed in slightly larger plastic pipet tips which had been prefilled with internal gas-sensing solutions [Orion internal filling solution and 0.1 M Tris-HCl buffer (pH 7.9) respectively]. The end of each pipet was covered by a PTFE gas-permeable membrane (0.2 pm pore size) (W.L. Gore). These gas membranes were held in place by a plastic O-ring (see [g-11] for more detailed diagrams of each half-cell as- sembly). When each internal electrode was pressed into the outer pipet, a thin film of internal buffer
or electrolyte formed between the gas membrane and the polymeric ion-selective membrane. Final electrical connections were made via a salt bridge between the bulk internal filling solutions of the two gas-sensing probes. Lithium acetate was specifically used as the salt bridge electrolyte (over sodium and potassium salts) to avoid possible leakage of interfering cations into the internal filling solutions. The PVC-nonactin-based am- monium and the TDDA-based pH electrodes dis- play minimal response to Li+ [26,27]. The re- sponse of the two electrode static sensor config- uration was evaluated by recording equilibrium potentials of the cell after making standard ad- ditions of NH,Cl to 50 ml of 0.01 M NaOH.
Flow-through system
Figure 3 illustrates the continuous-flow mani- fold used to evaluate the differential flow-through ammonia detector. As shown, the sample stream is split equally so that exactly half of the fluid flows through each gas dialysis chamber. The required tubular flow-through polymer membrane pH and ammonium ion-selective electrodes were prepared as described previously [9,12,13]. A 0.01 M NH,Cl solution and a 0.05 M Tri-HCl buffer (pH 7.5) were used as the recipient stream solutions in the pH- and ammonium-sensing flow-through half- cells, respectively. The electrochemical circuit was completed when the flows of both recipient streams were merged after the tubular membrane electrode detectors. Standard NH,CI solutions were pre- pared in 0.01 M NaOH and placed in the Auto- Analyzer sampler immediately prior to measure- ment. Recorded peak heights (in mV) were used to construct the response curves. The response of the flow system toward various amines was evaluated in a similar manner. The sampling rate for most experiments was 30 h-r with a sample-to-wash ratio of 1 : 2.
Determination of ammonia in bioreactor media
The analytical utility of the differential flow- through system was evaluated by determining the ammonia-nitrogen content of various samples of bioreactor media used to produce monoclonal an- tibodies. These samples were obtained from the
FLOW-THROUGH TYPE POTENTIOMETRIC GAS-SENSING SYSTEMS 121
University of Michigan’s Department of Chemical Engineering. The samples were diluted 1 + 9 with 0.01 M NaOH prior to measurement with the automated system. Aliquots of the same samples were also tested for ammonia-N content using two conventional methods; an L-glutamate dehydro-
genase enzymatic assay procedure [28] and a com- mercial static Orion Model 95-12 ammonia elec-
trode method [29].
RESULTS AND DISCUSSION
Static gas sensor configuration
Figure 4 illustrates the typical equilibrium potentiometric response obtained for the differen- tial static ammonia gas sensor when various con- centrations of ammonium chloride are added to a well stirred 0.01 M NaOH sample solution (n = 3).
The response time to reach the equilibrium poten- tial varied depending on the concentration of sam-
> E
-400 - ’ ’ - ’ ’ s -7.5 -6.5 -5.5 -4.5 -3.5 -2.5 -1.5
kiWH31, M Fig. 4. Typical calibration graph for differential ammonia
sensor constructed in accordance with the design shown in Fig.
1. Each point represents an average of three determinations.
ple ammonia. At concentrations > 0.5 mM the response times were typically less than 1 min, whereas in the range l-500 PM response times of 2-5 mm were observed. These dynamic response properties are similar to those of existing commer- cial potentiometric ammonia sensors [30] and newer research prototypes previously fabricated with polymeric pH and ammonium-selective inter-
nal electrodes [9-111. For three separate experiments, the average
slope of the static sensor’s differential ammonia response was found to be 96.3 f 0.5 mV decade-’ in the range 10-6-10-3 M NH,. This is far greater than the slopes observed for conventional Severinghaus-style sensors (e.g., 55-59 mV de- cade-‘) or those based on single internal poly- meric pH and ammonium ion detectors [9,10]. This slope is, however, less than the theoretical value of 118 mV decade-’ predicted using the simplified model considered above. This dif- ference is expected in the case of the static sensor configuration for a number of reasons. First, it has been found that the polymeric pH and am- monium-selective membranes used in each half-cell typically exhibit sub-Nernstian behavior even when used as independent ion sensors under equi- librium conditions (e.g., slopes range from 54 to
58 mV decade-‘). Second, owing to geometric design limitations [30], it is impossible to isolate completely the thin film of electrolyte and buffer in each half-cell from the internal bulk of these solutions. Thus, at equilibrium, a steady-state ex- change of bulk solution with the thin film causes
the overall response (i.e., change in thin-film aH+
in half-cell 1 and thin-film a,,; in half-cell 2) to be less than expected for a truly closed thin-film reaction layer [20,30]. Finally, it has been shown previously that the response slopes for gas sensing half-cells based on conjugate ion detection, in this case ammonium ions, are highly dependent on the pH, pK, and the ionic strength of the buffer solution used as the internal solution for the sensor [20]. While the 0.1 M Tris-HCl buffer used here
should provide a reasonably strong buffer capac- ity, at sample ammonia levels > lop3 M there will nonetheless be a significant pH change within the thin-film buffer layer that will cause a dramatic decrease in the observed slope (see [20] for details).
122
In practice, this limits the range in which the observed potentiometric signal is linearly related to the logarithm of ammonia concentration in the sample (in this’case, ca. lop3 M; see Fig. 4).
The detection limits of the combined differen- tial ammonia sensor will also depend on the detec- tion limits of each individual gas sensing half-cell. Indeed, despite the significant enhancement in slope over a given concentration range of dis- solved ammonia, the absolute detection limits of the static differential design (6 x lop7 M NH,), as determined by the IUPAC recommended method [31], do not differ significantly from those observed when each half-cell is employed in a conventional single working electrode ammonia probe [9,10]. Since it has been found previously that the detection limits for ammonia are better when a single working electrode-type sensor is fabricated using the ammonium ion electrode as the inner transducer [lo], it is likely that the ammonia response in this half-cell dictates the absolute lower limit of response of the differential design. The precision of the static differential am- monia sensor at higher concentrations is better than that of a single working electrode type of sensor owing to the enhanced sensitivity (98 * 0.5 mV decade-‘, n = 3, over a range from 1 X low5 to 1 x low3 M ammonia). Indeed, variations in the absolute potentials at given analyte concentra- tions ranged from f1.7 mV at 1 x lop5 M to +l.l mV at 1 X 10e3 M (s.d. for n = 3). Given the enhanced slope, at millimolar levels of am- monia the observed reproducibility corresponds to a precision of 2.6% with respect to measured am- monia concentration values.
Flow-through differential ammonia gas-sensing
system To adapt the differential detection scheme to a
flow-through measurement arrangement, the con- cept of using gas dialysis in conjunction with flowing recipient solutions and tubular flow- through ion-selective electrode detectors was em- ployed (see Figs. 2 and 3). Each of the respective flow-through half-cell gas detectors had been evaluated independently in previous work with a single gas dialyzer unit [9,12,13]; however, as shown in Fig. 3, for differential sensing the sample
KS. YIM ET AL.
stream must be split so that equal fractions pass through two gas dialysis detection units simulta- neously. Figure 5 illustrates the typical ammonia calibration graph obtained for this flow-through arrangement. In contrast to the static system, a nearly theoretical slope value of 118 mV decade- ’ in the range 2 X 10-5-10-3 M is observed. This enhanced slope, relative to the static arrangement, may be attributed to fact that the detection pro- cess is not at equilibrium. That is, given the flow- rates and relative sample to recipient stream volumes in the dialysis chambers used here (1 : l), only a fraction of the total ammonia in the sample stream actually enters the flowing recipient buffer and electrolyte streams during the time the sample spends in two dialysis chambers. Hence the geo- metric effects, and changes in the ionic composi- tion/pH of the electrolyte and buffer solutions
350
250
& 5 5
z 150
50
-50 7.
, 5 -6.5 -5.5 -4.5 -3.5 -2.5
log[Species], M Fig. 5. Response of the differential flow-through ammonia-N
detector to (0) ammonia and various amine species: (0, large)
ethylamine; (A, small) diethylamine; (A, large) methylamine;
(A) butylamine; (+) ethanolamine; (0, small) dimethylamine.
FLOW-THROUGH TYPE POTENTIOMETRIC GAS-SENSING SYSTEMS 123
that influence the observed slopes in the equi- librium static sensing configuration, are not major factors in the response characteristics of the flow-
through design. However, given the fact that both flow-through membrane electrode detectors actu- ally exhibit slightly sub-Nemstian properties in
direct response to protons and ammonium ions (e.g., 54-58 mV decade-‘; thereby yielding a max- imum slope for differential sensing of 108-116 mV decade-‘), the observed theoretically predic- ted slope for the differential flow-through am- monia detection must be due to the fact that one or both of the half-cell detectors operates in a mode where the fraction of sample ammonia gas transferred across the gas permeable membrane increases slightly as a function of sample ammonia concentration. This behavior has been observed previously with both the pH- and ammonium ion electrode-based half-cell flow-through ammonia detectors when used independently [9,12,13] and
accounts, in part, for the observed behavior of the differential flow-through system.
As with any continuous-flow analysis system,
for low-level measurements optimum reproducibil- ity can be achieved only when sample solutions have ammonia levels within a restricted concentra- tion range (l-2 decades). For example, Fig. 6
shows the typical strip-chart recording obtained for multiple ammonia standards in the range 10-5-10-3 M. A similar response is observed when the standards are run in descending con- centrations. For measurements made in this range, peak potentials are reproducible to Q + 1 mV (s.d. for six measurements at each concentration). Considering the enhanced slope of the differential detector, this level of precision corresponds to an r.s.d. of G 2% for unknown concentration de- terminations. This precision is better than that obtained previously when each of the two half-cells was employed separately in the design of flow-
through ammonia detectors (e.g., r.s.d. 4%) [9,12]. The detection limit of the combined differential ammonia sensor arrangement is ca. 5 x lo-’ M under the conditions outlined under Experimental. Naturally, this lower limit is dependent on the flow-rates of both the sample and recipient streams. The detection limit of the differential flow-through system is governed by the gas-sens-
Fig. 6. Typical strip-chart recording obtained for differential flow-through ammonia-N system. Concentrations are in M. Conditions: sample-to-wash ratio, 1: 2; sampling rate, 30 h-‘.
ing half-cell that exhibits the lower detection limit (e.g., NH:-based half-cell in these studies).
It is known that static and flow-through am-
monia detectors that employ pH electrodes as inner sensing elements display large responses to volatile amines [32,33]. The amines diffuse through the microporous membrane of the static sensor or flow-through gas dialyzer and alter the pH of the internal or recipient stream NH,Cl solution. It is also known that both static or flow-through detec- tors that are fabricated with inner ammonium ion-selective electrodes in conjunction with inter- nal recipient buffers provide a dramatic enhance- ment in the selectivity over the amines [13,33]. Thus, for any differential arrangement using these two modes of detection simultaneously, it is likely that the selectivity of the resultant system would be somewhere in between. As shown in Fig. 5, this is in fact the case for the new differential flow- through ammonia detection configuration. Most of the potentiometric response to the various amines arises from the pH half-cell [13]. Conse- quently, enhanced gas sensitivity is obtained at the expense of selectivity, at least when comparing the new differential approach with the previous flow-
124
TABLE 2
Comparison of the results obtained for the determination of
ammonia-N in media samples from bioreactors by a manual
commercial Orion ammonia gas sensor, an enzymatic method
and the proposed differential flow-through ammonia
detector a.b
Sample Method
Manual Orion ’ Differential ’ Enzymatic d
(mM) (mM) (mM)
1 2.31 2.20 2.12 2 0.89 0.80 0.93 3 2.31 2.50 2.59 4 3.64 3.73 3.23 5 0.92 0.88 1.06 6 1.76 1.91 1.83 7 2.53 2.49 2.41 8 2.31 2.25 2.46 9 3.49 3.42 3.81
10 0.89 0.90 1.04 11 1.71 1.78 1.98 12 0.89 0.81 0.83 13 2.31 2.35 2.52 14 2.35 2.40 2.39 15 0.91 0.97 1.24 16 1.10 1.22 1.37 17 2.57 2.63 2.66 18 1.07 1.27 1.44 19 2.42 2.53 2.72 20 3.48 3.14 3.81 21 0.94 0.85 0.83 22 1.99 1.86 2.05
a Average difference (differential - manual Orion) = 0.10,
s.d. = 0.07, differences range = 0.01-0.34. Average difference
(differential-enzymatic) = 0.20, s.d. = 0.18, differences range
= 0.02-0.67. b Differential = 0.974 (Orion) + 0.056; standard
error of slope = 0.030; standard error of y (est.) = 0.126; r =
0.981. Differential = 0.945 (enzymatic) - 0.026; standard error
of slope = 0.054; standard error of y (est.) = 0.227; r = 0.938.
‘Average of three determinations. d Average of two determina-
tions.
through detector prepared with only the am- monium ion-sensing half-cell [12,33].
To demonstrate the analytical utility of the new
differential flow-through ammonia detector, the proposed system was used to determine the am- monia-N content (total dissolved ammonia gas plus ammonium ions) of media samples taken from bioreactors used to produce monoclonal an- tibodies. It is known that elevated levels of am- monia-N in such culture media can inhibit hy- bridoma cell growth [34]. Ammonia-N values ob-
H.S. YIM ET AL.
mined via the differential flow-through ammonia detector correlated well with the results given by both a manual commercial ammonia electrode procedure and an enzymatic method (see Table 2).
Conclusions
As indicated in Table 1, the concept of dif- ferential gas sensing should be applicable to the detection of analytes other than ammonia. For
example, it should be possible to design an SO,- sensitive differential cell with enhanced gas sensi- tivity by detecting hydrogensulfite/sulfite from diffusing SO, in one half cell {using the new Hg(II)-based polymer membrane [15]} and a pH change in the other. Similarly, detectors for NO, could be devised by employing either a nitrate- or a nitrite-selective membrane electrode in conjunc- tion with an appropriate internal or flowing re-
cipient buffer reagent in one half-cell and nitrite solutions with a pH sensor in the other. In all instances a significant enhancement in the re- sponse slope compared with existing gas-sensing arrangements should be realized. However, as with the ammonia configurations described above, ultimately the selectivity of these systems will probably be limited by the sensitivity of the pH sensor-based half-cell to other acidic or basic gases. Nonetheless, in instances where decreasing the uncertainty in a measured gas concentration is desirable, and where interferences from other gases would be minimal, the use of differential potentio- metric gas-sensing schemes may prove to be an attractive alternative to existing measurement
methods.
This work was supported in part by grants from the National Science Foundation (EET- 8712756) and the National Institutes for Health
(GM-28882).
REFERENCES
R.L. Solsky, Anal. Chem., 60 (1988) 106R.
M. Riley, in A.K. Covington (Ed.), Ion-Selective Electrode
Methodology, Vol. II, CRC Press, Boca Raton, 1979, Chap.
1.
Handbook of Electrode Technology, Orion Research, Cam-
bridge, MA, 1982.
FLOW-THROUGH TYPE POTENTIOMETRIC GAS-SENSING SYSTEMS 125
4 B.A. Shapiro, R.A. Harrison and J.R. Waton, Clinical
Applications of Blood Gases, Year Book Medical Pub-
lishers, Chicago, 1982.
5 G.A. Rechnitz,, Science, 214 (1981) 287.
6 G.G. Guilbault, Analytical Uses of Immobilized Enzymes,
Marcel Dekker, New York, 1984, p. 112.
7 I. Karube and S. Suzuki, Ion-Sel. Electrode Rev., 6 (1984)
15. 8 J.W. Severinghaus and A.F. Bradley, J. Appl. Physiol., 13
(1958) 515.
9 W.N. Opdycke, S.J. Parks and M.E. Meyerhoff, Anal.
Chim. Acta, 155 (1983) 11.
10 M.E. Meyerhoff, Anal. Chem., 52 (1980) 1532.
11 R.H. Robins and M.E. Meyerhoff, Anal. Chem., 52 (1980)
2383. 12 Y.M. Fraticelli and M.E. Meyerhoff, Anal. Chem., 53 (1981)
992.
13 H.L. Lee and M.E. Meyerhoff, Analyst, 110 (1985) 371.
14 G.B. Martin and M.E. Meyerhoff, Anal. Chim. Acta, 186
(1986) 71.
15 D.M. Pranitis and M.E. Meyerhoff, Anal. Chim. Acta, 217
(1989) 123.
16 J.A. Greenberg and M.E. Meyerhoff, Anal. Chim. Acta,
141 (1982) 57.
17 M.E. Meyerhoff, D.M. Pranitis, H.S. Yim, N.A. Chaniota-
kis and S.B. Park, Chemical Sensors and Microinstrumen-
tation (ACS Symp. Ser. 403), American Chemical Society,
Washington, DC, 1989, p. 26.
18 R. Stepak, Fresenius’ Z. Anal. Chem., 315 (1983) 629.
19 G.S. Cha and M.E. Meyerhoff, Electroanalysis, 1 (1989)
205.
20 M.E. Meyerhoff, Y.M. Fraticelli, W.N. Opdycke, L.G.
Bachas and A.D. Gordus, Anal. Chim. Acta, 154 (1983) 17.
21 W.E. van der Linden, Anal. Chim. Acta, 151 (1983) 359.
22 P. Schulthess, D. Ammann, B. Krautler, C. Caderas, R.
Stepanek and W. Simon, Anal. Chem., 57 (1985) 1397.
23 S.E. Schmitt and E. Pungor, Anal. Lett., 4 (1971) 641.
24 MS. Frant and J. Ross, Science, 154 (1966) 1553.
25 K. Hartman, S. Luterotti, H.F. Ossward, P.C. Meier, M.
Oehme, D. Ammann and W. Simon, Mikrochim. Acta, Part
II, (1978) 235.
26 SC. Ma, N.A. Chaniotakis and M.E. Meyerhoff, Anal.
Chem., 60 (1988) 2293.
27 U. Oesch, Z. Brzozka, A. Xu and W. Simon, Anal. Chem.,
58 (1986) 2285.
28 P.J. Buttery, Anal. B&hem., 47 (1972) 294.
29 S.S. Ozturk, M.E. Meyerhoff and P.O. Palsson, Biotechnol.
Tech., 3 (1989) 217.
30 J.W. Ross, J.H. Riseman and J.A. Krueger, Pure Appl.
Chem., 36 (1973) 473.
31 IUPAC, Pure Appl. Chem., 48 (1976) 129.
32 M.E. Lopez and G.A. Rechnitz, Anal. Chem., 54 (1982)
2085.
33 Y.M. Fraticelli and M.E. Meyerhoff, Anal. Chem., 53 (1981)
1857.
34 M.W. Glacken, R.J. Fleischaker and A.J. Sinskey, Biotech-
nol. Bioeng., 28 (1986) 1676.
