Long-term stability of glass electrodes in aqueous media

Tdmta,Vol 22,pp 102>1027PergamonPress, 1975PnntedtnGreatBntam

LONG-TERM

STABILITY OF GLASS ELECTRODES IN AQUEOUS MEDIA

Bo KARLBERG* Department of AnalytIcal Chemistry, Umverslty of Umei, S-901 87 Umei, Sweden (Recewed 4 November 1974 Accepted 17 January 1975)

Summary-The stability of the potential of glass electrodes has been studled The potential changes only shghtly durmg the hydration of freshly etched electrodes With glass electrodes previously used m alkaline solutions, structural transformations wlthm the gel-layer give rise to large potential drifts m neutral or acldlc test solutions In alkaline solutions all glass electrodes are attacked, especially the low-temperature type, and the potential changes with time Drymg hydrated electrodes affects the stability only slightly Alternating transfers between acidic and basic solutions decrease the stability Recommendations for precise measurements with glass electrodes are given

There are several recommendations m the hterature for the care of the glass electrode 111order to reduce drift of the potential These recommendations, which Involve treatment such as presoaking, condrtlonmg, and brmgmg the glass electrode to constant temperature, are often contradictory The recommended length of the soaking period, for instance, varies from a few hours to several days The hterature has been summarized m treatises on the subject IV2 The electrochemical properties of a glass electrode are related to the state of the gel-layer 3 An alteration m the structure of the gel-layer causes an lmmedlate change m potential It 1s therefore necessary to know the history of a glass electrode exactly, so that the state of the gel-layer can be established Processes which alter the structure of the gel-layer include, among others, hydration and chermcal degradation, which can occur slmuitaneously and which determine the thickness of the gel-layer Hydration contmuously renews the gel-layer, the external part of which 1s subJect to some chemical degradation m any solution The degradation 1s especially rapid m strongly basic solutions In this work, the factors affecting the drift m potential of some commercial glass electrodes have been studied m the light of more recent knowledge about the gel-layer properties 3-g

very seldom larger than 0 1mV throughout an observation penod of 12 hr At least two reference electrodes were used m each measuring series and these were allowed to eqmhbrate m the test solution for at least 6 hr before the measurements with the glass electrodes were started All test solutions contamed chloride ions, allowing a chffuslonfree arrangement for the Ag/AgCl electrode The glass electrodes were always brought to constant temperature before use since even a small temperature &fference between the test solution and the glass electrodes resulted m an emf drift of several mV Solutions

All solutions were prepared from reagent grade chermcals The buffer compositions were as follows 0 1M hydrochloric acid-“pH 1” 005M potassium hydrogen phthalate + 0 1OM so&urn chloride-“pH 4” 0 02M “tris” + OOlM hydrochloric acid + 0 1OM so&urn chloride-“pH 8” 0 OlM borax + 0 02M so&urn chloride-“pH 9” cn 0 0025M calcium hydroxide + 0 02M sochum chlorIde”pH 12” A little silver nitrate solution was added to each buffer and to each of the other solutions to prevent dlssolutlon of silver chioride from the reference electrode and this addition altered the chloride ion content only shghdy Nitrogen was used to expel carbon dloxlde from the most alkaline solutions &cult

and emf recordmg

The electrode vessel was a double-walled tltratlon vessel, Metrohm EA 880, kept at 25 0” and shielded m an earthed EXPERIMENTAL metal cage Not more than three glass electrodes were mElectrodes vestigated at a time A programmable umt selected and Only hthla glass electrodes were studled Several specsconnected the electrodes m selected order to an HP 3460 mens of each of the followmg electrode types were used B voltmeter vu2 a follower (Analog Devices Model 302), Ingold LOT low-temperature electrodes, Ingold 201 and and the ernf was recorded by an IBM typewrIter m direct Metrohm UX general purpose electrodes, and Ingold HA connectlon The time and order of connection of the elechigh-temperature electrodes Etchmg was performed m 2% trodes were programmed m advance as were the number hydrofluorlc acid for l-3 mm The Ag/AgCl electrode, preof recordmgs and the time intervals between recordmgs pared by Brown’s method,“’ was used as reference The The typewriter prmted both the time and the emf value potentials of freshly prepared and aged reference electrodes A reading for each electrode system could be taken about m 0 1M hydrochloric acid were compared devlatlons were every third tnmute The electrode was connected to the voltmeter for about 1 mm before a reading was made The long-term stab&y of the voltmeter and follower was mves* Present address Astra Pharmaceuticals AB, AnalytIcal tigated separately and the drift was found to be less than 0 1 mV over a period of 24 hr Control, S-151 85 Sodertalje (Sweden) 1023

1,024

Bo

“pH II1 ____________________....

‘bH 4”

z L

I

12mV “pH 8”

‘) ,@--7-7_.

.._. _________ _.__.._.-.. 1

“pH 9”

____ _._______.-..._ _.- ____

“pH 12” -..

__ __.._ .._.__.

. ---.-

- --- --------

KARLBERG

chemical attack by water, require more time to reach a steady value, m accordance with slower formation of the external layer The thickness of the total gellayer 1s known to mcrease for all electrodes even after they have reached a steady potential, but this increase apparently occurs wlthm the internal part of the layer At higher pH values the probable mechanism of electrode glass attack 1s neutrallzatlon of slhclc acid groups llpl* The hydration process will be retarded by such a reaction and the growth of the gel-layer will consequently be reduced. A different drift pattern 1s thus expected m alkaline solutions With the HA and 201 electrodes, the drift from lower to higher emf observed at pH l-9 1s compensated at pH 12 With the LOT electrode, the shift towards higher emf values at the very begmnmg of the curve 1s compensated to such an extent even, at pH 8 and 9 that a drift from higher to lower emf values results Drift pattern of dried (previously hydrated) electrodes

The three types of glass electrode (LOT, 201, and HA) were hydrated for about one month m dlsttlled Hydration time / hr water, dried for 2 hr m a warm air stream at 70-80” Fig, I Emf drift patterns with freshly etched glass electrodes m different buffers Ingold HA (-), Ingold 201 and allowed to cool m a desiccator They were then brought carefully to constant temperature and im(-----), Ingold LOT ( ) mersed m the pH-4 buffer Typical emf values observed at different times are given m Table 1 The tmmedmte conclusion drawn 1s that drying does not RESULTS AND DISCUSSION influence the long-term stability The LoT electrode Lhji. pattern offreshly etched electrodes 1s especially msenatlve to drying. It has previously The potential drifts of three freshly etched elec- been shown that a slmllar drying procedure affects trodes were studied simultaneously m aqueous solu- the response behavtour m non-aqueous solutions “,” ttons of various pH The results are summarized m Lack of water m the gel-layer hinders proton transfer, which occurs via water molecules, and consequently Fig 1 Several interesting conclusions can be drawn from the results The total drift of the electrodes dur- the response becomes slow However, lmmerston of mg the gel-layer formation 1s rather small but 1s a glass electrode m an aqueous solution for only dependent on the glass composltlon This finding con5 mm 1s sufficient to restore the rapid response The forms with current knowledge of the rate of gel-layer redlstrlbutlon of the water m the gel-layer IS so fast formation of the different glasses, this rate increases that the long-term stablhty 1s not affected m the order HA < 201 < LOT A fully-developed gelInfluence of raped pH changes on the long-term stabzhty layer consists of two distinct regions, the external region bemg thinner than the internal one The LoT In Table 2, cell emf values are gtven for the three glass 1s readily attacked by water, resulting m rapid different Ingold electrodes m going from pH 4 to pH 1 and from pH 4 to pH 9, respectively The change growth of the gel-layer The external region 1s therefrom pH 4 to pH 1 does not seem to have any mfore formed rather rapidly Since the response properfluence on the long-term stablltty while, for the Ingold ties of the glass are located m the external part, only LoT only, the change from pH 4 to pH 9 causes small changes of the cell emf are expected once this layer has been established. In fact, the drift curves a small but continuous increase of the emf values for the LOT electrode m pH 1, 4, 8, and 9 buffers These results are expected smce a soft glass like the LOT glass 1s more easily attacked m a basic solution are almost parallel with the time axis after one hour of hydration On the other hand, the curves for the than a hard glass, ze , the HA glass The degradation 201 and HA glasses, which are more resistant to of the outermost part of the gel-layer results in a 0

2

4

6

8

IO

12

Table 1 Potentials of “dried” electrodes m pH-4 buffer, mY

Electrode

01

03

05

10

Time, hr 20

45

100

180

240

LOT 201 HA

1168 1073 106 4

1166 1069 1060

1164 1066 105 5

1167 1062 1054

1167 1059 105 1

1169 105 6 105 2

1167 1060 1050

1168 1065 1054

1168 1068 1056

Stablhty of glass electrodes

1025

Table 2 Emf varlatlon with time followmg transfer to pH-1 and pH-9 buffers after 24 hr in pH-4 buffer

Electrode

4

LOT 201

308 4 297 8

HA

296 4

LOT 201 HA

-2068 -218 7 -2202

10

Tome, mm 20

From pH 4 to pH 1 308 4 308 3 298 0 298 0 296 5 296.4 From pH 4 to pH 9 -2072 -2074 -2186 -2185 -220 1 -220 1

changing potential If the electrodes are subsequently re-immersed m the pH-4 buffer, a similar drift pattern is expected The gel-layer of the LOT electrode, which has been degraded m the basic solution, starts to regenerate and this process causes a drift (Table 3) Stable potentials are obtained for the 201 and the HA electrodes throughout, but slightly different potentials are obtamed m the pH-4 buffer, depending on the pH of the solution from which the electrodes have been transferred The shift may be due to changes m the proton actrvmes wrthm the gel-layer This proton activity change may m turn be caused by an altered standard state within the gel-layer, the number of protons may m fact remam constant The results m Tables 2 and 3 indicate that rapid changes within the pH-region l-9 do not drmmrsh the longterm stability of the 201 and the HA glass electrodes, only the LOT electrode exhibits deviations during and after contact with the pH-9 buffer The fact that small potential shifts may appear depending on the direction of the pH change m the test solutron immediately raises the question of whether an electrode has a Nernstian response over a broad range of pH or not, irrespective of the duectron of pH-change Proton activity changes m the gellayer 1111occur m basic solutions when protons are exchanged for cations such as sodmm, resulting in the well-known alkali error For low-temperature electrodes this deviation occurs even at pH 9-10 However, it 1s known that by no means all glass electrodes obey Nemst’s law in the “ideal” range (pH l-8) I5 Test procedures with the hydrogen electrode as a reference have been described I6

40

100

308 2 2919

308 1 2919

296 4

296 4

-2076 -218 5 -220 1

-208 1 -2183 -220 1

hr

Fig 2 Varlatlons m potential of two hydrated glass electrodes, Ingold LOT (-) and Metrohm UX ( ) m alternatmg test solutions Ektslc test solution OOIM NaOH + 099M NaCl, acldlc test solutlon 0 IM HCl + 09M NaCl The broken vertical lmes denote the time of the electrode transfers

Large pH changes (from acidic to basic solutrons and vrce versa) resulted m unexpected behaviour of the low-temperature Ingold LOT electrode, and the general purpose Metrohm UX electrode (Fig 2) In the basic solution (O*OlMNaOH + 0 99M NaCl) the LOT electrode exhrbrts a strong and regular drift m potential, whrle the potential of the Metrohm electrode varies rather irregularly After the first transfer to the acidic solution (0 1M HCl + 0 9M NaCl), both curves show distmct peaks

Table 3 Emf varlatlon with time followmg transfer to pH-4 buffer after 24 hr m pH-1 or pH-9 buffer Electrode

4

LOT 201 HA

1202 1109 108 3

LOT 201 HA

1177 1108 1094

10

Time, mln 20

From pH 1 to pH 4 1201 1200 1107 1106 1083 1083 From pH 9 to pH 4 1184 1189 1110 1110 1095 1094

40

100

1200 1106 1084

1200 1105 1084

1193 1111 1094

1198 1109 1094

1026

J30 KARLBERG

Several factors must be taken mto account m the interpretation of the drift patterns First, we have the previously mentloned attack on the glass m alkaline solutions Secondly, hydrogen Ions m the gel-layer are exchanged for sodmm Ions m the basic solution, and reexchange takes place m the acidic solution This ion-exchange process strongly affects the emf m nonaqueous solutlons’3 and ought also to be of great importance m aqueous solutions Thirdly, reduction of the gel-layer thickness resultmg from the storage m basic solutions causes rehydratlon to occur m the acldlc solution Furthermore, breakdown of the gellayer structure may occur when traces of acid or base m the surface region of the glass are rapidly neutrahzed during the transfer of the electrode Other processes may also be involved m determmmg the emf changes The slow and contmuous emf drift observed when the LOT electrode 1s m basic solution 1s certainly not due to the ion-exchange process since exchange of hydrogen ions m the gel-layer 1s fast It 1s probably the alkaline attack on the electrode glass which causes the emf drift since the LOT glass has a low durability 3 On the other hand, the slow attamment of the peak emf value m acidic solution may be partly caused by the ion-exchange process, since sodmm ions within the gel-layer require a conaderable time to reach exchange equlhbrmm The results presented in Fig 3 support the mterpretatlon that the neutrahzatlon reactlon may damage the gel-layer An Ingold 201 electrode was transferred to a OOlM sodium hydroxide solution after different pretreatments Curve a, showmg the emf varlatlon after a direct transfer from 0 1M hydrochloric acid, should be compared with curve c, which was obtained after electrode transfer ura short periods of lmmerslon m a series of buffers with successively increasing pHvalues The ~rlatlons m curve c are much less than those m curve a Even the neutrahzatlon of traces of the pH-4 buffer causes a drift in potential (curve b), while storage m distilled water seems to be more satisfactory (curve d)

t 5mv

I d

IO

5

I5

hr

Fig 4 Emf drrft with an Ingold LOT electrode (hydrated

for several months before the start of the experiment) m 0 IM HCI Pretreatments were accordmg to the scheme Absolute @me, Time m OOIM KOH, Curve days kr 1 20 a h 3 7 25 20 d 31 23 The electrode was stored m dlstrlled water between the base-acrd test cycles C

The “peak phenomenon” appearing m acidic soiutlon (Fig 2) was studied more closely for an Ingold LOT electrode (Fig 4) The alkaline storage solution was OOlM potassium hydroxide and the acidic solution 0 1M hydrochloric acid (both aqueous) The electrode had been hydrated for several months before the experiment was started The curves m Fig 4 have been separated from each other for the sake of clarity, m reality, the equhbrmm emf values were all very similar It 1s obvious that the character of the peak changes when the experiment 1s repeated Addltlonally, attainment of the peak value IS successively slower This last finding can readdy be explained Treatment m the alkalme solution causes the externd part of the gel-layer to be reduced m thickness,’ and a further reduction 1sprobably caused by the neutrahzatlon The structural rigidity of the gel-layer mcreases towards the unhydrated glass core, and consequently the d&islon coefficients for the mobile tons m the gel-layer decrease m the same dire&on Hence, the ion-exchange process slows when the gel-layer structure increases m rigidity owmg to the reduction m thickness The hydration process re-opens the structure to some extent, but three weeks of further hydration after test b m basic solution 1s obviously not enough to form a layer similar to that formed after several months of hydration (compare curves a and c m Fig 4) CONCLUSIONS

5

IO

hr

Fig 3 Varlatlon m potential of an Ingold 201 In OOlM NaOH Hlstorv of the electrode (a), 48 hr III 0 Ii14 HCl. direct transfer; (h), 19 hr m pH-4‘ duffer, direct transfer, (c), 49 hr m 0 1M HCI, transfer ura lO-15-set dlppmgs in buffers at pH 3. 5, 7 and 9, (d), 24 hr m dIstilled water, direct transfer

Conclusions and reco~endations m thus section are based partly on the results presented m this paper and partly on the results of earlier studies Temperature

For precise measurements, the glass electrode must be brought carefully to constant temperature when

Stablhty of glass electrodes

it is m use, since a difference of as httle as 2-3” between the electrode and the test solutton causes an emf drift of several mV durmg the first hour It is self-evidenf that the temperature of the test solution must be thermostatically controlled for precise work Hydratwn

Well-hydrated electrodes should be selected for precise measurements in aqueous solutions The hydratron process causes the largest mrttal emf changes when a glass electrode without a gel-layer is immersed in water Electrodes used m solutions with etching properties are subject both to hydration and to other structural transformations of the gellayer, and these processes together cause a pronounced drift m emf Low-temperature electrodes are especially unstable m basic solutions, errors resulting even when the electrode is used agam at a lower pH Storage

Storage m distilled water, recommended by many manufacturers, IS quote adequate Storage m air for short periods is also possible if coatings can be prevented and if the electrode 1s soaked m water lust before use, but it must be emphasized that this is valid only for electrodes that have been completely hydrated New electrodes should be kept m water The pH of the aqueous storage solution 1s not crmcal provided too basic or too acidic solutions are avoided Effect of pH changes

Since small changes m pH do not alter the electrode stability, it is prudent to store the electrode m a solution of similar pH to the test solution If an electrode 1s used alternatively m actdrc and basic solutions, the stability of its potential m any solution decreases rapidly Electrodes used for end-pomt deter-

1027

mmation in actd-base tttrations should not, therefore, be used for precise pH measurements Bu&r capacity Emf changes may appear m gomg from strongly to weakly buffered solutions, even if the soluttons have the same pH The opposite transfer IS less detrimental Thus, electrodes should be kept m an unbuffered solution before use m a test solution with low buffer capacity The expected emf stabihty m such test solutions is, however, necessarily lower than m well-buffered solutions Acknowledgement-The author IS indebted to Prof G Johansson for valuable dIscussIons and to Dr M Sharp for lmgmstx revlslon of the manuscript submltted

REFERENCES

I G Mattock, pH Measurements and Tltratlon, Heywood, London, 1961 2 R G Bates, Determrnatlon of pH, Wdey, New York, 1973 3 A Wlkby and B Karlberg, Electrochlm

Acta, 1974,

19, 323

A Wlkby, J Electroanal Chem, 1972, 38, 429 Idem, lbrd, 1972, 39, 103 B Karlberg, lbzd, 1973, 45, 127 Idem, lbzd, 1974, 49,

I

F G K Baucke, J Non-CrystalIme Sohds, 1974, 14, 13

A Wlkby, Elechochlm Acta, 1974, 19, 329 A S Brown, J Am Chem Sot, 1934, 56, 646 D Hubbard, E H Hamdton and A N Fmn, J Res Nat1 Bur Std. 1939. 22. 339

R W Douglas and T G M El-Shamy, J Am Ceram Sot,

1967, SO, 1

B Karlberg, Anal Chum Acta, 1973, 66, 93 14 B Karlberg and A Wlkby, Acta Chem Stand, 1973, 27, 1855 15 I Hansson, rbzd, 1973, 27, 931 16 B( Olm, Ibtd, 1960, 14, 126