Dissolved Oxygen Circuit

This page describes aspects of measuring Dissolved Oxygen (DO2) , focussing mostly on the Clark type of sensor. Marine life depends on the small amount of oxygen dissolved in water. Productivity can be affected by having insufficient oxygen. Oxygen levels can vary widely in rivers, lakes and the ocean, so monitoring oxygen levels can be important. There are three 'families' of sensor commonly used:

The Clark sensor was developed the earliest of these three types, and despite its problems can provide satisfactory results if it is well understood. It still appears to be the most popular type, probably because of availability, and despite the advantages of the galvanic (self polarising) type. The sensors are all expensive, though the relatively new fibre optic type could eventually become lower cost.

Operation

The Clark polarographic sensor has a pair of electrodes, usually gold (Au) and silver (Ag) operating in an electrolyte of potassium chloride (KCl). The electrodes are separated from the sample solution (water and dissolved oxygen) by a membrane, which allows oxygen to pass, while keeping the sample solution out. The sensor cathode (Au) is held at a constant potential (~ -0.8V) reference the anode (Ag/AgCl). This polarisation makes the cell selective to oxygen mostly. The electrical current that flows is then proportional to the oxygen that passes through the membrane.

The galvanic cell is related to the Clark cell, being a 'self biasing' cell. It can be seen as a cell generating an output current proportional to oxygen. It seems less troublesome in some applications. They do wear out, but they can last several years.

Considerations

These apply especially when using continuous rather than pulsed polarisation:

• The barometric pressure needs to be considered during the calibration procedures.
• The sensitivity and response time are affected by the membrane, which controls the flow of oxygen.
• The silver electrode is the anode, and the gold electrode is the cathode.
• The sensor consumes oxygen, the rate being dependant on the membrane and electrode physics.
• The sensor takes some time to stabilise after new membranes and electrolyte are installed, and after startup.
• The membrane needs to be cared for by keeping it moist when not in liquid.
• The electrolyte lasts about 2 weeks before needing replacement.
• The anode (silver) needs cleaning regularly as it becomes contaminated by brown AgCl and black silver sulphide.
• The affects of salinity can be corrected by calibrating with saturated solutions of oxygen and salt water. The temperature is controlled to obtain a desired mg/l of dissolved oxygen at saturation and the measured salinity. Water held at a constant temperature will settle to a saturated value for the salinity and temperature involved. The settling process can be speeded up by stirring and bubbling air.
• Note that changing temperature changes the salinity.
• The sensor is sensitive to temperature, typically 4% or 5% per degree. It seems pointless to measure the oxygen without measuring temperature too. The sensor and the measured solution should all be settled at the same temperature. The calibration has to be done the same way, so the sensor body must be considered.
• A calibration chamber or bath may be required so temperature can be controlled.
• The flow rate past the probe effects the reading. Stable stirring is required for accurate measurement.
• Air contains about 21% oxygen. Therefore when it is bubbled through water only 21% as much oxygen dissolves as would be dissolved by using oxygen alone.
• Moist air is used as 100% reference by having a moist sponge in a flask, with the sensor held above it.
• Nitrogen gas (heavier than air) can be used as a zero reference.
• About a teaspoon of sodium sulphide (photographic fixer) per 100 ml of water can be used to 'kill' the oxygen for a zero reading.
• The membrane can be made from ordinary cling wrap film (Kitchen wrap or Glad wrap). Commercial membranes may be similar, or use a Teflon film similar to plumbers tape. Teflon film has a slower response, as do thicker films.

Problems

These problems are mostly not present with a galvanic cell, which is generally thought to be superior, but the problems can also be minimised by using the pulsed method.

• Isolation of the Anode. The chemical reaction produces AgCl. Over time this builds up on the anode, inhibiting operation. It has to be cleaned.
• Zero Shift. The internal electrolyte (KCl) becomes more alkaline because of the OH produced. This causes a zero shift. Over time, the electrolyte needs to be changed.
• Depletion of Chloride. Chloride ions are consumed and the electrolyte needs to be replenished.
• Warm-Up Time. When the probe is disconnected, the power supply is cut off. On connecting the probe again, the user must wait for the probe to be polarised. This causes a drift with time, typically reading up to 20% high for the first 10-50 minutes.
• Hydrogen sulphide is often present in waters. The silver anode is attacked, forming a silver-sulphide coating which inhibits operation over time.
• Interference (spurious) responses. The sensor also responds to chlorine and hydrogen sulphide gases. Thus the salinity of the test solution has an effect on the reading, and sulphides can cause false results.

Measurement Units

• Saturation percent is often used as it relates directly to the calibration, which is set to read 100%. The test reading is the percentage of the reading at calibration (typically for a saturated solution). Thus it gives a ready comparison with the calibration (assuming the other conditions like temperature and salinity are constant). Typical instruments read from 0% to 200%, as supersaturated solutions can exist.
• If the amount of oxygen is required for further calculations, it can be determined in terms of the value of the calibration concentration. This is measured in parts per million (ppm) or mg/l, which is the same thing. Typical instruments read up to 20 mg/l. The instrument is set to show the calibration value in mg/l, at calibration time.
• It helps for the calibration and test waters to be at the same salinity and temperature.
• For calibration using water saturated with air, use the table below to determine the oxygen concentration. Seawater has a salinity of 35 - 36 ppt, so use the table value of 6.72 mg/l at 25 degrees C and 1013 hPa .
• For calibration using air at 100% humidity, use the table value of 8.26 mg/l at 25 degrees C and 1013 hPa.
• The measurement in mg/l can be converted to moles of oxygen molecules per litre for chemical calculations. The mole is a large number (6.02 x 10²³). A mole of oxygen molecules weighs 32 grams, from the atomic number and the number of atoms per molecule. Note that it also occupies 24.5 litres at standard temperature and pressure, like any gas, even when mixed.

Calibration Tables

The amount of oxygen dissolved in the water when saturated depends on the temperature and salinity. This table gives a feel for the dissolved oxygen relationship in mg/l. Further corrections are required for barometric pressure. The amount of oxygen absorbed depends on atmospheric pressure. Note that 1013 hPa = 760 mm of Hg. The two references at the top of the page have formulae, and tables that can be printed.

Dissolved oxygen (mg/L)

Concentration (mg/l)

Circuit Operation

Circuit suitable for micro-electrodes

Referring to the diagram above, the sensor details shows one form of Clark sensor. The gold cathode is connected to the circuit so it is biased to -830 mV. Any current which flows is converted to voltage. This circuit is suitable for microelectrodes where the sensor current is very low. Values of a few picoamps imply a feedback resistor of hundreds of megohms, so a very low bias current opamp is needed in this case. The capacitor across the feedback resistor is required for stability. It may be very small (30 pF) with the very high value feedback resistors, and needs suitable insulation. The current input node is at the bias potential. The output is 830 mV with no sensor current, so additional components may be required (not shown). Alternatively the polarising voltage can be applied directly to the sensor electrode, and the + input of the current to voltage converter grounded.

Circuit suitable for multiple sensors, and pulsed operation

Typical commercial oxygen sensors have an output current of microamps. The current can be measured by using an instrumentation amplifier across the series load resistor. In this case the instrumentation amplifier suppresses electrical noise and interference, allowing the sensor to work with others in the test solution. The sensor and its polarising supply are floating from ground, except for the small current through the instrumentation amplifier's bias return resistor.

The load resistor is sized so that there is 10-20 millivolts across it at full scale oxygen. It represents a short circuit. Set the gain resistor for an appropriate output voltage.

The polarising supply is floating (insulated) from ground. It can be powered by a battery or by an isolated power supply (small transformer inverter with electrostatic screen between primary and secondary). The inverter has separate windings for a multi-channel system. This inverter can be switched on and off to use the pulsed mode. The output of the opamp does not load the sensor when it is powered down, due to its internal diodes (using an LT1077 etc.).

Pulsed Mode

The pulsed mode described was developed for oceanography by Dr. Chris Langdon, and he has published several papers about it. The polarising potential is pulsed every few minutes. It is off during the rest of the time, so the oxygen levels inside the membrane can recover. This makes the method more or less flow independant, and reduces most of the effects that are membrane dependant. There are many benefits, such as longer period between calibration, less oxygen consumed, flow rate dependency removed. The repeatability of measurements improves by an order of magnitude. The measurement is taken after the pulse has 'developed', and before it approaches the steady state condition. For a typical YSI style sensor the measurement is taken 1.5 seconds after the pulse is applied, and lasts for less than one second. If the measurement is taken too soon the oxygen linearity suffers. If it is taken too late it approaches the steady state performance, and there is no advantage. The disadvantage of this system is that the response to oxygen change is slower.

Pulsed Operation

To use the pulsed mode the following ADC needs to have 16 bits resolution for sufficient dynamic range. Alternately, the signal can be clipped as the initial peak following switch on is not needed. The polarising voltage should be adjusted to one volt for the pulsed mode because the sensor's internal voltage drop is increased during the stronger part of the current pulse.

Galvanic Sensor

Galvanic Sensor Circuit

This circuit is suitable for the galvanic type of electrode, which does not require an external polarising potential. The load resistor is sized to about 2 x 47 ohms, representing a short circuit. There is about 3 mV full scale across the load resistors. The amplifier bias current return resistor can be 1 megohm. The sensor floats (from ground), while the amplifier bias current from the inputs can return to the power supply common (ground).