SABLE SYSTEMS INTERNATIONAL

At the end of this page is a general section on common electrode problems. For best results, read the whole of this page first before you read about them for the first time! This page will be updated as time permits; if you have specific comments or wishes, mail us! 

Ideally, a polarographic oxygen sensor (oxygen electrode) delivers a stable current proportional to the partial pressure of oxygen contacting its membrane, independent of time and across a wide temperature range. If you need to use a oxygen electrode, you need stability. Stability may be measured over minutes, or hours; long-term monitoring applications demand stability over periods up to a year.  

Given this ideal, most oxygen electrodes fall far short of it. It's likely that among the scientists who use them, oxygen electrodes have contributed to more signs of premature aging, and a greater range of pernicious psychological disorders, than almost any other single factor apart from department chairs and other senior administrators. Understanding why oxygen electrodes behave as they do may help to mitigate this suffering, and can even lead to better oxygen electrode performance in the instances where problems can be both identified and cured. 

This page presents a systematic examination of factors that influence the steady-state current output of oxygen electrodes. We will also weigh the relative importance of these factors, list the symptoms they produce, and - where possible - suggest cures.  

First, we must distinguish between the drifting of steady-state sensitivity, and the short-term transient changes followed by a slow approach to stabilization, that follow any major change in operating conditions. A major change might include: a change of membrane, of temperature, of applied voltage, or of oxygen partial pressure. Such a transient change of sensitivity is easily recognizable and easy to deal with (not pleasant, but easy). It is usually sufficient to wait for a certain predetermined time, the stabilization time, before resuming measurements. Typically, the stabilization time lasts for several minutes.  Expect it when you first turn on an oxygen analyzer and it polarizes the oxygen electrode, and expect it whenever you change operating conditions thereafter.  

For best stability, we recommend leaving your oxygen electrode polarized - which is to say, don't turn off your oxygen analyzer unless you're not going to use it for at least a day!

After this initial stabilization time has elapsed, the output from the oxygen electrode settles down to a level which can appear to be constant over a period of several hours, but which nevertheless can drift noticeably, at constant oxygen partial pressure, over a period of months. These transient changes and long-term drifts are normal, and need not concern you unduly. Of more concern are short-term drift and short-term noise. 

Before dealing with short-term drift and noise rigorously, we need to understand how oxygen electrodes work! 

An oxygen electrode consists of an anode, typically made of chlorided silver, and a cathode, typically made from thin platinum wire. Only the very tip of the platinum cathode is exposed; the rest of the platinum wire forming the cathode is sealed in glass or epoxy. A membrane covers the tip of the platinum wire cathode, and beneath the membrane a thin layer of electrolyte allows electrical contact between the anode and the cathode. 

The reaction occurring in the electrode is: 

 O₂ + 4H⁺ + 4e^-  --> 2 H₂O 

I = JnF 

J = I / (nF) 

Do a thought experiment. A perfectly permeable membrane will overwhelm the cathode's ability to reduce the oxygen at any but the most unthinkably low oxygen partial pressures, thus producing an unusably sensitive oxygen electrode. A perfectly impermeable membrane won't allow any oxygen to reach the cathode at all, leading to an unusable insensitive oxygen electrode. Between these extremes lies the territory of more-or-less usable oxygen electrode membranes, which allow enough, but not too much, oxygen to reach the cathode. And, obviously, the bigger the cathode, the greater the amount of oxygen that can be reduced, and the greater the current (I) that the oxygen electrode will produce at a given oxygen partial pressure. 

The more permeable the membrane, 

• the greater the speed with which oxygen reaches the cathode and is consumed
• the more oxygen is consumed
• the faster the response of the oxygen electrode
• the more stirring is required.

• the slower the rate of oxygen consumption at the cathode
• the less oxygen is consumed
• the slower the response of the oxygen electrode
• the less stirring is required.

Let's quantify this talk about membrane permeability! Can't you feel an equation coming on? First let's calculate the oxygen flux rate that's possible at the cathode: 

J =  (PO2) / ((Zs/DsSs) + (Zm/DmSm) + (Ze/DeSe)) 

OK, what does this mean? Well, we showed in our very first equation that the probe current is determined by J, and now we've just shown that J is determined not only by the partial pressure of oxygen, but also by the characteristics of the boundary layer, the membrane, and the electrolyte! This means that if you want stable readings, you need a stable electrolyte; you need a stable membrane; and you need a constant boundary layer, which translates to a constant rate of stirring or sample movement. 

Of the various resistances to diffusion listed above, by far the biggest (we stress again) is usually the membrane. And, of course, this means that as the membrane becomes more permeable, the other resistances become more and more significant. So, for the best stability from the point of view of the above equation, you should use a low-permeability (usually meaning thicker) membrane and put up with the lower resulting J, meaning, in practice, slower response times. 

Needless to say, if the characteristics of the membrane change with time, or the characteristics of the electrolyte change with time, you'll get a change in J, which means - Drift. Short term changes mean - Noise. These can both come from other sources too, as we'll see. 

Here are the common membrane materials, and their solubility and diffusion parameters, assuming a membrane thickness of 0.001 inch (25.4 microns): 
 
 
 

Let's sum this all up in terms of electrode current, which after all is what we're measuring! Neglecting diffusion resistance posed by the boundary layer and the electrolyte (translation: assuming the membrane way outweighs them), we have 

I = (nFA DmSmPO₂) / Zm 

Let's reiterate - As we've just seen, we can cut down on the stirring requirement by decreasing the permeability of the membrane. We can do this by specifying a less permeable membrane material, or a thicker membrane, at a cost in terms of speed of response. In all but a very few highly specialized applications this is not a problem. What we lose in terms of fast response, we gain in stability. 

Speaking of stability - How do these equations relate to stability, the overriding concern (and frustration) of oxygen electrode users? 

The most important single factor causing drift is membrane tension. Because the membranes used for oxygen electrodes are normally thin and partially elastic, their thickness, and hence the resistance they offer to oxygen diffusion, can be substantially altered if they become stretched when being mounted. One of two deleterious effects may then occur. Either the membrane may gradually slip under the restraining means (usually an O-ring), to return to its original thickness, or it may suffer "cold flow" as a result of the constant stress, with a consequent change of thickness. These thickness changes often continue during a period of months following mounting and are observed as a gradual drift of output signal. True stability can only be achieved if membrane stretching is avoided during mounting. At the same time you must not allow a thick electrolyte layer to remain between the membrane and the sensing cathode. This follows from the rule that the diffusion resistance of the electrolyte layer should be as small as possible relative to that of the membrane.  

Here's a subtlety. We've just shown that the thickness of the electrolyte layer between the membrane and the cathode determines, in part, the sensitivity of an oxygen electrode. Hence physical phenomena that directly influence the thickness of the electrolyte layer can give rise to instability. The membrane that encloses an oxygen electrode is semipermeable in the sense employed in discussions of osmotic effects, that is, it allows the passage of water vapor but blocks the passage of ions. When an oxygen electrode is immersed in pure water, therefore, water vapor enters the sensor, diluting the electrolyte! If the walls of the electrolyte chamber were completely rigid, the internal pressure would rise until the water vapor pressure in the electrolytic solution equaled the water vapor pressure in the water outside the sensor. Equilibrium would then be attained. In practice, however, the hydrostatic pressure inside the sensor at equilibrium, known as the osmotic pressure, can be as high as 104 kPa at equilibrium (!), and can't be restrained by the tension in the membrane. So movement of water into the sensor continues, with the membrane bulging outward to accommodate the increased volume of electrolyte. This entry of water into the sensor causes an increase in the electrolyte layer thickness, Ze, and a consequent decrease in the output from the sensor. 

In order to function properly, an oxygen electrode must be filled with an electrolyte having at least a minimum conductivity for current flow. This conductivity is decreased if the sensor dries out, or if the electrolyte becomes diluted by sample water entering through a "leak" in the membrane sealing system. The symptom manifested by the sensor in this case is a gradual decrease of signal at constant oxygen concentration. To cure the problem, it is only necessary to renew the electrolyte, and to ensure adequate sealing of the membrane to the sensor body. 

Gravitational effects on the electrolyte layer can also occur. The membrane may sag to some extent if it supports the whole weight of the electrolyte. The effect is most significant if the electrolyte volume and the membrane area are large, and if the membrane is poorly tensioned and supported. Gravitational effects can be minimized by clamping the sensor in a fixed orientation. 

Vibrations of the sensor or turbulence in the sample sometimes causes the sensor to deliver a noisy signal. This problem is particularly acute if the membrane is poorly tensioned, or if air bubbles are enclosed in the sensor. The explanation is thought to be that convection currents are caused in the electrolyte, thus transporting oxygen from the electrolyte reservoir to the cathode. 

That about covers membranes, if you'll excuse the pun. Let's move on to other sources of strange oxygen electrode behaviors that will drive you crazy or make you grow as a person, depending on your point of view. 

 First, let's re-visit stirring sensitivity, and the speed of response trade-off that comes with it. Well, what about changing the cathode area rather than (or in addition to) membrane permeability? An alternative way to decrease the stirring requirement, after all, is to reduce the cathode diameter, resulting, in extreme cases, in the so-called microcathode electrode. Such electrodes consume very little oxygen (and, as a result, produce very low currents). A couple of decades ago scientists fell in love with microcathode electrodes because of their supposed freedom from boundary layer effects. Most scientists realize now that what they thought was love was actually a meaningless and expensive infatuation. Why? Well: 

• Because of their small cathode area, microcathode electrodes are plagued by fouling, poisoning, drift, noise and other problems.
• Microcathode electrodes still require some stirring, contrary to what their manufacturers would have you believe.
• So if you're gonna have to stir anyway, you might as well stir anyway, as Sam Goldwyn might have said.
• So why put up with microcathode electrodes in the first place?
• Masochism. Fashion. Who knows.

But, however good the electronics, the ultimate determinant of performance is going to be the electrical characteristics of the oxygen electrode itself. Oxygen electrodes are usually operated with a constant total applied voltage of about 0.65V, shared between the anode/electrolyte interface, the cathode/electrolyte interface, and the electrolyte conductor. Ideally, the electrolyte solution should be an almost perfect conductor, and its share of the applied voltage should therefore be virtually zero. Also ideally, the anode should be operated under equilibrium conditions so that its interfacial voltage is constant. Ideally (again!), therefore, the cathode's interfacial voltage should remain constant, giving us a stable current and thus a stable PO₂ reading, the Holy Grail of oxygen electrode users.  

However, the electrolyte conductor may take an increasing share of the applied voltage with the passage of time, either because it becomes progressively depleted of salt due to the anode reaction, or because it becomes gradually diluted with sample water due to faulty sealing of the membrane to the sensor body. To make things worse, the anode interface may take a variable (rather than constant) fraction of the applied voltage, either because it becomes increasingly blocked by the deposition of nonporous insulating products, or because electroactive vapors from the sample interfere with the anode reaction. In either case the cathode voltage is also variable, usually in the sense that it decreases with time, causing a downward drift in sensor output. 

The oxygen electrode reaction is also subject to interference by deposits on the cathode surface, either of electroplated metals (the anode metal is frequently plated onto the cathode during operation), or of organic substances adsorbed from the electrolyte, or of products of side reactions at the cathode caused by interfering vapors from the sample. The smaller the cathode, in general, the worse this problem becomes! Again, a downward drift of sensor output at constant oxygen partial pressure is the usual consequence.  

Ideally, the electrochemical reaction at the anode of an oxygen electrode should result in the creation of an insoluble porous solid phase which adheres firmly to the anode metal. In practice, the products generated at most of the commonly chosen anode metals, such as silver, lead, cadmium, thallium, or zinc, are soluble in the electrolyte to some extent because of the formation of complexes between the metal ions and the anions of the electrolyte. Diffusion of these ions and subsequent plating of the anode metal on the cathode is therefore a common phenomenon.  

The plating is less rapid when the membrane is well tensioned, and when the zone surrounding the cathode, through which diffusion of the metal ions occurs, is long and narrow. Also the useful life of the sensor is longest, for any particular sample, for membranes of low permeability, since the initial rate of formation of the metal ions is proportional to the membrane's permeability. 

Sometimes the plating occurs over the whole of the area of the cathode, an effect which is most easily noticeable at gold cathodes because of the change of coloration. This is not too troublesome if the anode is silver, but can lead to a decrease in cathode activity when the anode metal is more basic. The cathode should be polished in these circumstances to restore its activity. This is quite easy to do. 

In other circumstances the plating occurs at the periphery of the cathode, as tree-like dendritic formations, thus extending the sensing area and leading to an increase in displayed concentration at constant sample concentration. Again, the cathode must be polished at regular intervals if this becomes a problem. 

More serious is anode blocking. The continuous operating lifetime of an oxygen electrode is often limited by the blocking of its anode by nonporous reaction products. Anode blocking usually occurs when a particular quantity of anode product has been deposited on each unit area of the anode, and therefore the larger the area of the anode, the lower the permeability of the membrane, and the lower the concentration of oxygen to which the sensor is exposed, the longer will be the life of the sensor. After anode blocking has occurred, activity is easily restored by stripping the product off the anode with a suitable cleaning solution. A silver anode, for example, may be cleaned with a commercial detarnishing fluid, or with an ammoniacal solution. It will usually need re-chloriding after this treatment. 

Temperature effects are important too. The current delivered by an oxygen electrode changes by about 3% to 4% per ^oC, at constant oxygen partial pressure. This temperature coefficient is greatest for the least permeable membranes. When temperature changes occur during the course of measurements, as they inevitably do during field work for example, you can use this temperature coefficient to interpret the measured current in terms of oxygen partial pressure. Hence one is concerned not only with the stability of sensor output at constant temperature, but also with the stability of the temperature coefficient of oxygen solubility and diffusivity in the membrane. 

Extra care is necessary when an oxygen electrode must be operated well above room temperature, to avoid enclosing an air bubble in the sensor. The reason is that this bubble expands more than the displaced electrolyte would have done, and the additional volume might be created by "inflation" of the membrane. This, of course, leads to a diminished output from the sensor, and sometimes to an irreversible stretching of the membrane.  

The zero current of an oxygen electrode is the current produced by the sensor when exposed to a medium containing no oxygen. It is never zero, although the sensor should be selected so that it is negligible in comparison with the signals of interest in any particular study. Commercial sensors are available having zero currents in the range from 0.01% to 1% of the signal generated in air-saturated water. Hence the stability of zero, or residual, currents is of little interest when measurements are made in well-aerated clean water, but becomes crucially important when measurements are made in anaerobic media. 

Even a cursory study of residual currents reveals that they are rarely constant in time. For example, a current spike followed by an exponentially decaying current tail is observed whenever a sensor is switched on after being exposed to air while off circuit. The time constant for this decay can be minutes or even hours (!), depending upon the particular sensor, and upon the state of cleanliness of the cathode. Beware, microcathode electrode users... 

Because of the transient nature of the residual current, the practice of electronically adjusting the zero point of a sensor by subtraction of a constant current is of limited usefulness. The residual current is completely internal in origin, and so becomes less important as the oxygen permeability of the membrane increases, because the signal level increases in comparison with the constant error level.  

Finally, let's look at some of the major compromises between the stability and speed of response of oxygen electrodes. 

We've mentioned membrane permeability in several places, in the context of oxygen electrode stability. This is because the oxygen must pass through the diffusion boundary layer and the electrolyte layers as well as the membrane, and these two additional resistances are likely to vary during operation; therefore stability can be expected only when the membrane impedance exceeds those of the other two by a safe margin. This reasoning alone dictates a low permeability membrane for an oxygen electrode, whenever stable, long-term monitoring of oxygen is important.  

Another advantage accrues from the use of a low permeability membrane, namely that only a low rate of stirring is required to reduce the impedance of the diffusion boundary layer to a negligible value in comparison with that of the membrane. Indeed, very high stirring rates are needed to supply oxygen at the rate required by thin PTFE, PEA, or FEE membranes. Why, then, are highly permeable membranes most commonly chosen for oxygen electrodes? The reason is that, in some applications, speed of response and the ability to measure traces of dissolved oxygen are of predominating importance. Plus, the residual current is independent of the membrane chosen, while the sensitivity of the oxygen electrode  is proportional to the oxygen permeability of the membrane. Hence, any sensor is more capable of measuring to lower concentration limits, when it is fitted with a permeable membrane, than when fitted with an impermeable one. 

 A compromise between speed of response and high sensitivity, on the one hand, and stability and low stirring requirement on the other, is therefore necessary when selecting a sensor for any particular application. When rapid response is indispensable, then frequent recalibration is inevitable, and you'll just have to live with some instability. When stability is of prime importance, a sluggish response is inevitable. 

Here's a useful catalog of oxygen electrode drift and noise problems. Most of these problems are addressed in our main page on how oxygen electrodes operate, and the appropriate section can be pulled up by clicking on the appropriate link in a future edition of this page.  

General drift 

• Osmotic effects (downward)
• Release of membrane stress (downward)
• Leakage or evaporation of a gross excess of electrolyte (upward)
• Deposition of anode metal at edge of cathode causing increase of cathode area (upward)
• Membrane erosion (upward)
• Membrane clogging (downward)
• Anode blocking (downward)
• Dilution of electrolyte with sample water (downward)
• Cathode deactivation by anode metal or poisons

• Poorly stabilized anode supply voltage
• Poorly compensated temperature variations
• Flow rate (stirring) variations of sample
• Gravitational (orientational) effects
• Applied pressure effects
• Vibrational effects

• Barometric pressure effects
• Zero (residual) current variations

Sable Systems is a company founded by scientists for scientists. We are certain that you will enjoy doing business with us. 

Sable Systems Home Page

Sable Systems Tech Update Index