ANALYST MAY 1986 VOL. 111 511 Piezoelectric Quartz Crystal Detection of Ammonia Using Pyridoxine Hydrochloride Supported on a Polyethoxylate Matrix* Colin S. 1. Lai G. J. Moody and J. D. R. Thomas Department of Applied Chemistry Redwood Building UWIST P.O. Box 13 Cardiff CFI 3XF UK The use of a nonylphenoxypolyethoxylate (Antarox CO-880) as a support-polymer is confirmed as a means of prolonging the life (to >50 d) of pyridoxine hydrochloride as a sensitive sorbent coating during the piezoelectric crystal detection of ammonia. However the matrix system incurs possible interference from hydrogen chloride gas although except for triethylamine the other gases studied at high levels (sulphur dioxide nitrogen dioxide carbon dioxide and hydrogen sulphide) give only a small piezoelectric signal.The extreme sensitivity of the piezoelectric crystal detection of ammonia to below the pg dm-3 range is inconsistent with the Sauerbrey equation which normally applies to straightforward deposition on piezoelectric transducers. This is a consequence of the low slopes of the log(frequency decrease) versus log(concentration) graphs. Such slopes can be increased by modifying the syringe dilution procedure but there are other more intransigent factors involved. Keywords Ammonia detection; flow injection analysis; pol yethoxylate support; piezoelectric quartz crystal detection; gas sensor Suitably coated piezoelectric quartz crystal detectors form a highly sensitive technique for the detection of trace amounts of atmospheric pollutants. 172 The analyte is selectively sorbed by the coating thereby increasing the mass on the crystal and decreasing the frequency of vibration.The frequency change, AF (Hz) is linearly related to the mass sorbed according to the Sauerbrey equation,3.4 which for AT-cut crystals takes the form Am AF= -2.3 X l O 6 P -A where Fis the initial frequency of the quartz plate (MHz) Am is the mass sorbed (8) and A is the area of the coating (cm2). Thus for a particular experimental set-up the change in frequency can be expressed as AF= KC (2) where C is the analyte concentration (mg dm-3 or pg dm-3 or ng dm-3) and K is a constant that includes the basic frequency of the quartz crystal the area coated and a factor to convert the mass of analyte sorbed into its gas-phase concentration.Ucon 75-H-90000 and Ucon LB-3000X were first used for detecting ammonia in air and were found to have good sensitivity.5 These were followed by coatings of extracts of Capsicum annuurn pods and ascorbic acid with and without silver nitrate,6 and later by L-glutamic acid hydrochloride and pyridoxine hydrochloride (vitamin B6 hydrochloride) ,7 which showed exceptional sensitivity. Supporting the pyridoxine hydrochloride on a matrix of a high relative molecular mass polyethoxylate Antarox CO-880 (nonylphenoxypolyethoxy-late with 30 ethoxylate units) helps to extend considerably the useful lifetime of the detector.8 This paper describes further studies on the detection of ammonia using an AT-cut quartz crystal of 9-MHz resonant frequency coated with pyridoxine hydrochloride in a matrix of Antarox CO-880.The parameters studied include coating methods for applying the sorbent to the electrode surfaces of the crystal the effect of interferences procedural steps in the syringe dilution method for obtaining low concentrations of ammonia test gas and tests of the Sauerbrey equation. ~ ~~~~~~~ ~~~ * Presented at Analyticon 85 London UK September 17-19th, 1985. Experimental Apparatus and Detector Design A laboratory-constructed piezoelectric apparatus9 was assem-bled as shown in Fig. 1 and the operating conditions were optimised. The measuring unit consisted of a frequency oscillator with a buffered output powered by a Weir 400 power supply set at 9 V d.c. The frequency output from the oscillator was measured by a Marconi Type 2431A 200-MHz digital frequency meter.A digital to analogue converter selected the last two digits of the frequency meter output for conversion into an analogue signal to a Bryans Model 28000 chart recorder reading to +1 Hz. The AT-cut quartz crystal with gold electrode (Fig. 1 inset) (Quartz Crystal Co. Wellington Crescent New Malden, Surrey) had a resonant frequency of 9 MHz. The detector cell incorporating this crystal was based on the double impinger cell design of Karmarker and Guilbaultlo wherein the gas sample was split into two streams impinging directly on opposite faces of the coated crystal. The glass-encased cell detector was immersed in a water-bath at 25 k 0.1 "C. Pyridoxine Hydrochloride Coatings Coating solutions consisted of mixtures (1 + 1 V/V) of a 0.2% mlV solution of pyridoxine hydrochloride in ethanol and water (1 + 1 VlV) and a 0.2% mlV solution of Antarox CO-880 in acetone.For the capillary coating approach the mixture was applied to the electrodes cf the quartz crystal with a fine drawn-out capillary tube (melting-point tube). For the brush-coating approach a tiny brush was used to1 apply the mixture to the electrode surfaces in the manner described by Hlavay and Guilbault.11 In each instance the coated crystal was then placed in the oven at 80 "C to evaporate the solvent and crystals were stored overnight or between measurements at this temperature. The coating applied in each instance corresponded to a decrease of about 9.5 KHz in the basic frequency of the crystal and was readily removed with ethanol and water (1 + 1 V/V).The crystal was dried before re-coating. Ammonia Test Samples The ammonia gas test samples were obtained with a 10-cm3 gas-tight Perspex syringe from ammonia vapour over the headspace of dilute (2 M) ammonium solution equilibrated a 5 12 Digital to Recorder - analogue converter ANALYST MAY 1986 VOL. 111 Power -Frequency- Oscillator - supply meter Air - 3 I -f l Perspex block Fig. 1. Schematic diagram of piezoelectric quartz crystal detection apparatus with (inset on left) detail of quartz crystal V - 11111111111111111111llllllllllllllllllll Water-bath (25 _+ 0.1 "C) -= Charcoal = Silica gel= 25 "C. Serial dilution of the headspace gas was effected by syringe dilution12 with ambient air (dry air gave AF = 0).Successive dilutions were delayed by 30-60 s in order to allow ammonia to diffuse throughout the air in the syringe. The concentration of ammonia in the headspace was checked by titration. Thus 10-cm3 samples were slowly injected from the syringe into 20 cm3 of 0.025 M sulphuric acid. The excess of sulphuric acid was titrated with 0.1 M sodium hydroxide solution using methyl orange as an indicator. Thirty replicate samples of the headspace gas contained 32.0 mg dm-3 (s.d. 0.25 mg dm-3) of ammonia. Interfering gases were analysed in a similar way with appropriate absorbents and titrants. For the syringe dilution procedure 9 cm3 of the test gas were expelled from the syringe and air (9 cm3) was sucked into a total volume of 10 cm3 plus the volume of needle and syringe connector (0.23 cm3 s.d.= 0.01). The tip of the syringe needle was closed by piercing into a rubber bung. The mixture in the syringe was allowed to stand for 0.5-1 min in order to allow it to become homogeneous by diffusion. The original test gas was thus 10.23/1.23-times diluted. In the second dilution stage 1.23 cm3 of the first mixture was diluted to 10.23 cm3 giving a mixture of (10.23/1.23)2 times dilution over the original concentration. By repeating the procedure, mixtures of low concentrations could be obtained. Samples for Testing the Sauerbrey Equation For testing the Sauerbrey equation serially diluted samples were taken as above. Additionally sample dilutions were prepared by a procedure involving replacing the syringe needle with a clean one in between each dilution.Thus, commencing with a lo-cm3 syringe-full of headspace ammonia standard (32.0 mg dm-3) [actually 10.23 cm3 after allowing for the volumes of the needle (0.033 cm3) and the connector (0.197 cm3)] the following procedure was adopted for serial dilution: (a) expel 9 cm3 of the ammonia standard and replace the needle with a new one; (b) draw in 9 cm3 of air and allow to mix; (c) expel 9 cm3 of the diluted sample from (b) again replacing the needle with a new one; (d) repeat stage (b) and the concentration of ammonia standard should be 100 times diluted [actual concentration is 32 x (1.197/10.197)2 = 0.441 mg dm-3 = 441 pg dm-3 for a syringe needle connector volume of for example 0.197 cm3]; (e) expel 1 cm3 in order to fill the needle with sample; (f) inject 5 cm3 of the sample into the piezoelectric crystal detector; (8) expel 3 cm3 of the remaining sample from (f) then replace the needle with a new one; (h) pull in 9 cm3 of air and allow to mix; and (i) repeat stages ( 4 (f) (8) and (h).Operation of Piezoelectric Quartz Crystal Detection Apparatus The responses of coated piezoelectric quartz crystals were tested on 5-cm3 samples of appropriate dilutions of the headspace ammonia test samples and the mean decrease in frequency for replicate samples was measured. The diluted samples were injected into a carrier stream of dry (silica gel) air and passed through the quartz crystal compartment (at 25 "C) at a rate of 20 cm3 min-1 by the pump of a Pitman Instruments Model 7069 air sampler (Fig.1). As stated the power supply was set at 9 V d.c. Results and Discussion The manner of operation of the piezoelectric quartz crystal detection system is essentially a gas-phase mode of flow injection analysis. Sensitivity is helped by the highly commen-ded12.13 double impinger detector cell design. This is facili-tated by the sorption and subsequent desorption of ammonia by the pyridoxine hydrochloride: CH2OH CHqOH H+CI- H+CI-Fig. 2 illustrates recorder responses for serially diluted ammonia samples while Fig. 3 confirms the previously reported8 role of Antarox CO-880 as a matrix for prolonging the life of the piezoelectric detector for ammonia. The typical responses shown in Fig. 2 illustrate that although the response of the detecting system is fast the return to the base-line frequency takes several minutes because of the relatively slow desorption of the ammonia.However as mentioned previously,8 fresh samples may be injected before returning to the base-line frequency for it is the immediate decrease in frequency caused by the injected sample that is analytically significant ANALYST MAY 1986 VOL. 111 5 13 It has previously been shown8 that Antarox CO-880 coated on the quartz crystal without pyridoxine hydrochloride was not significantly involved in ammonia sorption but that it did sorb water. The extent of water sorption (only the carrier air stream was dried) for the larger samples used here ( 5 cm3 compared with 1 cm3 in earlier studies8) is shown in Table 1 for a pyridoxine hydrochloride - Antarox CO-880 coated crystal.As expected the larger volume samples used in this study (5 cm3) produced larger frequency changes [ca. 350 Hz for 463 pg dm-3 (Fig. 2)] than were observed in the previous study8 for pyridoxine hydrochloride in an Antarox CO-880 matrix for 1 cm3 samples (215 Hz for 30 mg dm-3). These data compare with 1190 Hz for 1 mg dm-3 and a surprising 386 Hz for 10 ng dm-3 reported by Hlavay and Guilbault4 for pyridoxine hydrochloride alone on the quartz crystal. Good linearity of calibration was obtained (Fig. 3) with correlation coefficients of 0.99 when log AF was plotted against log [NH,]. t TTT Moist air I Time -f .7 -0.012 0.096 20 min 8.8 -Fig. 2. Typical recorder trace of a calibration of ammonia gas using a cpartz crystal coated with pyridoxine hydrochloride and Antarox 0-880.Sam le size 5 cm3. Numbers on peaks are NH3 concentra-tion (pg dm-$ 2.5 N 2.0 I Q 1.5 IY Fig. 3. Calibrations over several days of a piezoelectric quartz crystal for ammonia coated with pyridoxine hydrochloride and Antarox CO-880 and illustrating long functional lifetime. Day A 1; 0 2; H, 4; '7 10; 0 15; 0 2 5 ; x 30; 0 3 6 ; V 46; A 57; + 61; and * 67 Sorbent Coating Methods Capillary-tube and brush-coating approaches were compared for the application to the quartz crystal of pyridoxine hydrochloride alone and of hydrochloride in Antarox CO-880. For the pyridoxine hydrochloride alone the capillary-tube approach gave an uneven coating as the hydrochloride was concentrated in certain areas thereby reducing the reacting surface area.The brush-coating approach gave a more even coating. Both approaches gave visually even coatings for pyridoxine hydrochloride in Antarox CO-880 although the brush-coating approach produced larger frequency changes (Table 2). The brush-coating approach was used for all other results discussed here. Tests of reproducibility of the brush-coating technique were carried out for later coatings (Table 3). The five coatings shown were made on the same quartz crystal the previous coating being removed after each set of calibrations by brushing the crystal surface gently with ethanol - water (1 1 rn/V). In each instance the crystal was oven dried at 80 "C for 30 min after coating as stated in the experimental procedure.The relative standard deviations for the frequency de-creases (Table 3) were generally low being less than 4% for the various concentrations of ammonia with 4.0% for 0.012 pg dm-3 of ammonia and 0.9% for the 463 pg dm-3 ammonia sample. The area of coating (A) was measured with a Quantitiet 800 Image Analyser (Cambridge Instruments Ltd.). The amount of coating material was calculated from Sauerbrey's equation (1). Table 1. Effect of moisture on the quartz crystal brush coated with pyridoxine hydrochloride and Antarox CO-880 Laboratory . . . . . . Dilutionstage 0 1 2 3 4 5 6 7 air AFIHz 60 57 47 35 27 23 22 19 21 . . . . . . . . . . Table 2. Comparison of the capillary tube and brush-coating methods of coating pyridoxine hydrochloride with Antarox CO-880 on quartz crystal electrodes.All results are given as AF (Hz); sample volume of ammonia standard = 5 cm3 Ammonia standard (pg dm-3) Moist Coating method 463 56 6.7 0.80 0.096 0.012 laboratoryair Day I : Capillarytube . . . . . . 278 176 110 63 37 24 28 Capillarytube . . . . . . 185 134 96 59 30 23 20 Capillarytube . . . . . . 164 117 94 56 34 22 24 Brush . . . . . . . . . . 392 200 142 101 66 41 25 Day 5: Brush . . . . . . . . . . 248 139 104 66 50 33 25 Day 10: Brush . . . . . . . . . . 278 192 130 84 52 31 2 514 ANALYST MAY 1986 VOL. 111 Table 3. Tests on the reproducibility of the brush-coating technique with relative standard deviation data for another five coatings by capillary tube coating Response (AFIHz) to NH3 with piezoelectric quartz crystal coated with pyridoxine hydrochloride and Antarox CO-880 Relative standard deviation of similar Relative data for standard capillary tube [NH,]/pg dm-3 1st coating 2nd coating 3rd coating 4th coating 5th coating Mean deviation '/o coating '/o 462.6 339 333 332 337 332 334.8 0.90 6.4 55.62 189 195 198 182 194 191.6 2.9 3.4 6.70 125 140 136 133 138 134.4 3.9 6.1 0.80 87 88 94 90 90 89.8 2.7 5.5 0.096 60 62 65 58 61 61.2 4.0 9.9 0.012 35 33 36 37 34 35.0 4.0 9.9 Area of coating/cm2 0.56 0.51 0.58 0.53 0.51 0.533 5.6 11.7 Amount deposited/pg 28.0 24.3 35.9 25.7 22.8 27.3 19.0 21.4 A F due to coating/Hz 9334 8905 11576 9042 8334 9438.2 12.0 10.7 ~ ~~ ~~~ ~ ~ Table 4.Interferences in the piezoelectric crystal detection of ammonia AFIHz NH3 .. NH3 . . so . . NO2 . . HC1 . . HCl . . coz . . HZS . . TEA* . . Concentration/ Gas mg dm-3 . . . . . . . . 3.85 . . . . . . . . 0.46 . . . . . . . . 101 . . . . . . . . 75 . . . . . . . . 109 . . . . . . . . 13.2 . . . . . . . . 1477 . . . . . . . . 116 . . . . . . . . 5.3 Dry lab. air . . . . . . . . Moist lab. air . . . . . . * TEA = triethylamine. Vitamin B6 coating 408 26 1 21 47 45 19 27 21 1 0 10 -Antarox CO-880 coating 37 23 34 32 2674 255 30 41 43 0 11 Vitamin B6 + Antarox CO-880 coating 320 44 35 1496 143 32 43 383 0 16 -The mass of coating material deposited on the crystal surface varied from 22.8 to 35.9 pg.As the coating material was a 1 + 1 mixture of pyridoxine hydrochloride and Antarox CO-880 the amount of pyridoxine hydrochloride was deemed to be half of the mean value of 27.3 pg that is 13.7 pg with a relative standard deviation of 19%. The frequency decrease due to the coating was 9.44 kHz that is slightly more than the 9.24 kHz for a similar analysis of the capillary tube coating method for which relative standard deviation data are presented in Table 3. The two sets of relative standard deviation data show that the brush-coating approach is more reproducible. Chemical Interferences Interferences in the piezoelectric crystal assay of ammonia from other gases are listed in Table 4 for the pyridoxine hydrochloride - Antarox CO-880 coating and for the separate materials.With a few exceptions these confirm previous observations.7 The generally low interferences from acidic gases on pyridoxine hydrochloride alone are expected as the concentra-tions of the interferents are much higher than the concentra-tion of ammonia gas injected and the observed frequency decreases are much less than those for 3.85 mg dm-3 of ammonia (Table 4). However 5.3 mg dm-3 of triethylamine caused a frequency change of 211 Hz. This is not unexpected, because amines have similar structures and properties to ammonia though Hlavay and Guilbault7 found that trimethyl-amine had no effect. Interference profiles for the mixture of pyridoxine hydro-chloride and Antarox CO-880 as coating material are generally similar to responses recorded for pyridoxine hydrochloride alone except for hydrogen chloride gas (109 mg dm-3) with a frequency decrease of 1496 Hz.This response is due to the reaction between hydrogen chloride and Antarox CO-880 and is even more marked when the crystal is coated with Antarox CO-880 alone. Such sensitivity can be attributed to hydrogen bonding between the hydrogen chloride and the ethoxylate oxygen as confirmed by broadening of the ethoxylate infrared absorption band at 3500 cm-1 (vSH stretching) for Antarox CO-880 in the presence of hydrogen chloride. The piezoeIec-tric interference reaction is reversible as shown by the fast return (ca. 5 s) of the frequency decrease to the base line. This is now being evaluated for hydrogen chloride sensing14 as the frequency decrease for 109 mg dm-3 of ammonia is a much greater response than 400 Hz for 100 mg dm-3 using trimethylamine hydrochloride as substrate coating.l1 Application of the Sauerbrey Equation The analytical utilisation of coated piezoelectric quartz crystal detectors has been based on the assumption that Sauerbrey's equation1 is valid i.e. that the mass increase caused by sorption is directly related to the concentration of the sample in the flowing gas stream and is proportional to the decrease of the resonance frequency. 1.2 However all analysis of previous published data by Beitnes and Schrcbder15 shows that the sensitivity of piezoelectric crystal detectors for flowing gas streams does not obey the Sauerbrey equation. Thus as previously indicated,16 despite the poorer sensitivities that would be expected from incomplete sorption on the crystal coating mixing effects with carrier gas etc.the observed decreases in frequency are often greater than the valu ANALYST MAY 1986 VOL. 111 515 Table 5. Comparison of ammonia present in 5 cm3 of sample according to dilution calculations (for 3rd coating data of Table 3) and ammonia calculated [from the frequency change by the Sauerbrey equation (l)] to be sorbed on the piezoelectric crystal coating Concentration of Ammonia present Ammonia calculated to NH3 in in 5 cm3 of be sorbed on coating sample/pg dm-3 AFIHz samplehg (for A = 0.58 cmZ)/ng 463 332 2320 56 198 280 6.7 136 34 0.80 94 4 0.096 65 0.48 0.012 36 0.06 1030 620 420 290 202 112 2.5 2.0 2 Q Q ol -J 1.5 1 .o -2 -1 0 1 2 Log([ NHsIIw d r r 3 ) Fig.4. Ammonia content of standards deduced by syringe dilution procedures related to frequency decrease (A for Table 5 data and C) compared with ammonia content (assuming 100% absorption) calcu-lated from Sauerbrey equation (1) using observed frequency decrease for the corresponding syringe diluted standards (B for Table 5 data and D). Lines A and B are for the ordinary syringe dilution method (with allowance for needle and connector volume) and lines C and D are for the alternative dilution method with needle replacement (with allowance for connector volume) deduced from Sauerbrey’s equation. Conversely observed decreases in frequency relate according to the Sauerbrey equation to mass changes that are greater than the amount of the sought component actually present in the sample (see Table 5).A graph of the data of columns 1 and 2 of Table 5 (A of Fig. 4) has a much lower slope than that of unity expected for log AF versus log C according to the logarithmic form of equation (2). By changing the dilution procedure from the ordinary serial dilution with allowance for needle and needle connector volumes to one where the needle is changed for each dilution stage (detailed in the Experimental section relating to testing the Sauerbrey equation) the slope of log (sample ammonia concentration) versus log AF is steeper (Fig. 4 C) and nearer to the expectation of the Sauerbrey equation (Fig. 4 D). Even greater anomalies than exist in the above data occur in other reported work e.g.for the 0.01 pg dm-3 ammonia sample of Hlavay and Guilbault7 with a AF of 386 Hz the ammonia calculated to be sorbed by the pyridoxine hydro-chloride coating is about 1 pg compared with the 0.00005 pg deemed to be present in the 5-cm3 sample used. The anomalies are related to the gentle slopes observed for the log AF versus log Cgraphs and some previously reported7 slopes are very gentle e.g. 0.0615 for the coating of L-glutamic hydrochloride and 0.0978 for a coating of pyridox-ine hydrochloride. In this study the slopes of the graphs of Fig. 3 are between ca. 0.15 and ca. 0.2 while that of C in Fig. 4 improves to >0.4. Table 6. Frequency decrease (AF) data for the removal of 463 pg dm-3 of ammonia from a 10-cm3 syringe by successive evacuation and re-filling with dry air after replacing the needle at the end of each evacuation.(The 5-cm3 samples examined for piezoelectric frequency changes corresponded to the volume fraction between 1 and 6 cm3) AF/Hz for different tests ~~~ ~ Re-fill No. 1 2 3 4 0 323 316 320 320 1 28 25 16 28 2 7 7 6 6 3 0 0 0 0 Data such as the above have led Beitnes and Schroderls to investigate systematic errors in the syringe dilution method, but the likelihood of sorptions on syringe walls between dilutions is demonstrated (Table 6) to be an incomplete explanation of the anomalies and suggest that other factors are involved. The problem revolves around the better t’han predicted sensitivities and Beitnes and Schroder15 found that alterna-tive dilution methods such as bottle dilution also give sensitivities that are better than predicted.Although not a complete solution a change in the syringe dilution procedure, as discussed above brings the experimental response nearer to the predictions of the Sauerbrey equation (Fig. 4 C) but there could be other factors not recognised here. 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. References Guilbault G. G. Ion-Sel. Electrode Rev. 1980 2 3. Alder J. F. and McCallum J. J. Analyst 1983 108 1169. Sauerbrey G. Z . 2. Phys. 1959 155 206. Sauerbrey G . Z. 2. Phys. 1964 178 457. Karmarkar K. H. and Guilbault G . G. Anal. Chim. Acfa, 1975 75 111. Webber. L. M. and Guilbault G. G . Anal. Chem. 1976,48, 2244. Hlavay J. and Guilbault G. G. Anal. Chem. 1978,50,1044. Moody G. J. Thomas J. D. R . and Yarmo M. A . Anal. Chim. Acta 1983 155 225. Cannard A. J. Moody G. J. Thomas J. D. R. and Yarmo, M. A. unpublished work. Karmarkar K. H. and Guilbault G. G. Anal. Chim. Acta, 1974 71 419. Hlavay J. and Guilbault G. G. Anal. Chem. 1978,50 965. Karasek F. W. and Tienay J. W. J. Chromatogr. 1974 89, 31. Cooke S . West T. S. and Watts P.,Anal. Proc. 1980,17,2. Lai C. S. I. Moody G. J. and Thomas J. D. R. to be published. Beitnes H . and Schrcbder K. Anal. Chim. Acta 1984 158, 57. Lai C. S. I . Moody G. J. and Thomas J. D. R . Anal. Proc., 1985 22 10. Paper A5f386 Received October 28th I985 Accepted November 25th 198