Powered by boiling mercury!

Gallery opened August 2005

Updated: 17 Aug 2017

Eyewitness report from South Meadow station
Back to Home PageBack to The Museum

The choice of mercury for a working fluid may seem bizarre- after all, it is both expensive and horribly poisonous. However, there were what appeared to be sound thermodynamic reasons for grappling with the challenges involved in its use. After these challenges proved too great, mercury had a bit of a renaissance in projects for power plants in space. Here is the story...

There are some other working fluids which have their own galleries in the Museum:



Mercury has got to be the ultimate dodgy working fluid mercury: as remarked elsewhere on this site, in the Steamwheel gallery, (where mercury is used as a weight and sealant rather than a working fluid) it is an insidious poison of a most unpleasant kind. It has a much higher boiling point than water, at 357 degC. Experimental power station installations were tested in the USA but ultimately nothing came of them, the cost, complexity, and potential danger of the plant. Probably a good thing.

In a Rankine cycle, as used in steam power plants, most of the heat is supplied to boil water into steam; a major increase in cycle efficiency can only be achieved by doing the evaporation at a higher temperature. Steam has the relatively low critical temperature of 374 degC (705 degF), and a very high critical pressure at 3206 psi, so it is desirable to find a working fluid that has a critical temperature well above the metallurgical limit, and with a moderate vapour pressure at this temperature. Mercury fits the bill; it has a critical temperature of 1477 degC, (2690 degF) which is only 61 degC below the melting point of iron, and at a vapour temperature of 588 degC (1091 degF) its saturation pressure is only 300 psi.

However, the use of mercury as a single working fluid presents formidable problems. At the usual temperature of heat rejection, the vapour pressure is exceedingly low, and so the specific volume is enormous. This mean that it would be necessary to maintain an almost impossibly high vacuum in a condenser the size of a cathedral. Another snag is that mercury has a low latent heat (about one-eighth that of steam) implying that you will need a lot of it to transfer the heat in a cycle. And it is not cheap. But it is poisonous.

Therefore, to use the whole temperature range from the metallurgical limit down to the temperature of the nearest river, mercury was always used as the high-temperature section of a two-stage or binary cycle, in which condensing the mercury vapour boiled water into steam.

(Thermodynamic info from Engineering Thermodynamics: Work & Heat Transfer Rogers & Mayhew, Longmans 1957)


This is part of a British Association Paper called "Some of the Developments of Mechanical Engineering during the Last Half Century" By Sir Frederick Bramwell. It appeared in The Scientific American Supplement No. 312, on 24th December, 1881. While the mercury is not being used here as the working fluid (steam is) it's an interesting account of a bizarre idea.

"There was another kind of marine engine that I think should not be passed over without notice; I allude to Howard's quicksilver engine. The experiments with this engine were persevered in for some considerable time, and it was actually used for practical purposes in propelling a passenger steam-vessel called the Vesta, and running between London and Ramsgate. In that engine the boiler had a double bottom, containing an amalgam of quicksilver and lead. This amalgam served as a reservoir of heat, which it took up from the fire below the double-bottom, and gave forth at intervals to the water above it. There was no water in the boiler, in the ordinary sense of the term, but when steam was wanted to start the engine, a small quantity of water was injected by means of a hand-pump, and after the engine was started, there was pumped by it into the boiler, at each half revolution, as much water as would make the steam needed. This water was flashed on the top surface of the reservoir in which the amalgam was confined, and was entirely turned into steam, the object of the engineers in charge being to send in so much water as would just generate the steam, but so as not to leave any water in the boiler. The engines of the Vesta were made by Mr Penn, for Mr Howard, of the King and Queen Ironworks, Rotherhithe. Mr Howard was, I fear, a considerable loser by his meritorious efforts to improve the steam-engine."

"There was used, with this engine, an almost unknown mode of obtaining fresh water for the boiler. Fresh water, it will be seen was a necessity in this mode of evaporation. The presence of salt, or of any other impurity, when the whole of the water was flashed into steam, must have caused a deposit on the top of the amalgam chamber at each operation. Fresh water, therefore, was needed; the problem arose how to get it; and that problem was solved, not by the use of surface condensation, but by the employment of reinjection, that is to say, the water delivered from the hot well was passed into pipes external to the vessel; after traversing them, it came back into the injection tank sufficiently cooled to be used again. The boilers were worked by coke fires, urged by a fan blast in their ashpits, but I am not aware that this mode of firing was a needful part of the system."

The idea of using an amalgam of mercury and lead as a reservoir of heat was not a good one, and one regrets that Mr Howard was a considerable loser. The liquid with the highest specific heat, and so greatest storage capacity, is not some exotic chemical. Remarkably, it is plain water. Much cheaper and much safer!


The man behind the use of mercury vapour turbines in electric power stations was William Le Roy Emmet of the General Electric Company. Emmet devoted a great deal of time and energy to their development and promotion, "as a more efficient power generation system than steam turbines." The difficulties and risks in using the new technology led to termination of the development after his retirement. He died in Erie, Pennsylvania, on September 26, 1941, at the age of 82, so he had presumably managed to avoid mercury poisoning.

Mercury has a much higher boiling point than water, so the Carnot efficiency of a thermodynamic cycle that uses it is higher, in theory at least. The plants were dual-cycle, (also known as binary cycle) the hot metallic exhaust from the mercury turbines being used to produce steam for a second stage of steam turbines.

1914 Plans are made to build an experimental mercury boiler and power plant. (see below)

1923 The Hartford Electric Light Company (Helco) installs an experimental General Electric mercury-steam generating unit at Hartford, Conn.

1928 The first commercial mercury cycle generating unit in the world is installed at Helco's South Meadow Station in Hartford. Power output from the mercury turbine generator was 10 MW, plus the production of 125,000 lb per hour of steam at 250psi and 700 degF. According one of my sources (possibly not that reliable) the Hartford plant used 13 pounds of mercury per kilowatt generated; that would be 130,000 pounds or 58 tons, which seems like an awful lot. See Oliver Lyle's comments below.

1931 A General Electric mercury plant is installed at Schenectady, New York State, for the New York Power and Light Company. The mercury turbine output was 20 MW, plus the production of 325,000 lb per hour of steam at 400psi and 750 degF.

1933 A mercury topping cycle added to existing steam plant at South Kearney, NJ. The mercury turbine output was 20 MW, plus the production of 290,000 lb per hour of steam at 400psi and 750 degF.

1947 Construction begins on the last mercury/steam plant; the Schiller station at Portsmouth, NH.

1949 A new mercury plant at South Meadow, Hartford, begins operation on January the 2nd.

1950 The 40 MW Schiller station goes on load.

Left: A mercury/steam power plant.

This is a mercury topping cycle.

The mercury is vapourised in the boiler, drives the mercury turbine, and is condensed by boiling water to steam. This steam is superheated in the mercury boiler and then drives a conventional steam turbine. The "bled steam heaters" refer to the practice of heating the feed water by bleeding steam from various points along the turbine; this increases cycle efficiency and had been a standard feature of steam power plants for many years. Trust me, it works.

From "Electric Power Stations" Vol 1, by T H Carr, 4th edition, 1954

Left: A mercury/steam power plant.

On close inspection this diagram shows exactly the same plumbing as the one above. I have however included it here, as diagrams of mercury/steam plants are, to put it mildly, rather rare.

According to this book, the mercury circuit operated at 200psi and 538 degC. Electrically welded joints were used in the pipework because the poisonous nature of mercury was recognised, and leaks were not acceptable. Even so, it is hard to believe that a big installation could have been completely free from leaks, and I for one would not have cared to work there, or indeed within a hundred miles of it.

From "Electric Power Stations" Vol 1, by T H Carr, 4th edition, 1954.

Left: Another mercury/steam power plant diagram

This diagram shows a slightly more sophisticated plant with a mercury economiser as well as a water economiser as part of the mercury boiler. The conditions of the two working fluids are shown on the temperature/entropy graph at the bottom. (entropy = s) Note that the mercury is much hotter than the water when it is at less than half the pressure.

The great merit of a T/s plot is that the areas on it are equal to the energy transferred.

from Engineering Thermodynamics: Work & Heat Transfer Rogers & Mayhew, Longmans 1957

In one American plant (it is not clear which) the mercury circuit operated at 113 psi and 507 degC, driving a 15 MW turbine-alternator set at 720 rpm. The mercury condenser produced steam at 400 psi and 371 degC, at the rate of 200,000 lb/hour; this drove conventional steam turbines and alternators.

The extract below is taken from a journal that appears to have been called simply "The Tech" published by students of the Massachusetts Institute of Technology. This issue is dated Saturday, April 25, 1914.

Engineering and Scientific Notes: Mercury Turbines
"In a paper recently read before the American Institute of Electrical Engineers, there were described some interesting experiments to test the feasibility of employing mercury as the working agent in heat engines. Mercury boils under one atmosphere at a temperature of 677 degF (358 degC) and in a 28 inch vacuum it will condense at 455 degF (235 deg C). It is proposed to use the mercury vapor to drive a turbine and to utilize the heat from the exhaust in generating steam to drive a steam turbine. Despite the poisonous character of the vapor and the comparatively high cost of mercury, the liquid has several physical properties admirably suited for this purpose. Its high temperature is associated with very moderate pressure, it is clean in use and does not wet nor corrode the blades, while its high density makes possible a very moderate blade speed, an important factor. As to the question of cost, it has been estimated that the mercury would not cost more than a maximum of ten dollars per kilowatt output and, of course, the same mercury could be used repeatedly. It is claimed that a tandem mercury and steam plant would show a gain of 44 per cent on output per pound of fuel. A boiler of considerable size is now under construction in order to test such a plant on a large scale."

Left: A later article on mercury power.

From "The Tech" MIT, for 1st March 1 1926.


"The Efficient Use of Steam" by Oliver Lyle, of the Tate & Lyle sugar-refining family, was the bible of steam usage in Britain during the second world war and after. It was aimed more towards process steam rather than power generation, but the sixth edition (1958) contained the following information on mercury/water binary cycles: "Mercury boils at 900F under the very modest pressure of 80psi. At 28.5 in vacuum mercury vapour has a temperature of 437 degF (225 degC). If therefore, mercury is boiled in a boiler, and the saturated mercury vapour passed through a mercury turbine the mercury vapour can be exhausted into a condenser which can be a water boiler, and can raise steam to 250psi.This steam can be given some superheat up to say 600 degF (316 deg C) from the flue gases of the mercury boiler. Owing to the low latent heat of mercury it is necessary to use about 10 pounds of mercury for every 1 pound of water."

and later in the text:

"The disadvantages are somewhat formidable. Mercury vapour is extremely poisonous. Mercury does not wet metal surfaces.* The plant is complicated and costly. But several large mercury-steam stations have been working for some years in the USA and have shown very high sustained overall thermal efficiencies. other fluids might give better results. Diphenyl oxide has been suggested as being more suitable than mercury."

* If the surfaces are not wetted by the working fluid, then heat transfer would be greatly impaired. But you will see below that mercury does wet tantalum.

Now that's an interesting point above about the large amount of mercury required. And it's not cheap stuff.


In September 1914 the British Thomson Houston was granted British patent GB21689 for 'Improvements in and relating to power systems' which is an obfuscatory patent title if I ever saw one. Clearly BTH thought they has something worth hiding, insofar as you can hide a patent. The patent probably roused little interest in Europe, as there other matters going on at the time.

The British Thomson-Houston Company was the British arm of the General Electric Company of the USA.

Left: The BTH patent of 1914

The text below is taken from the patent:

A plant comprises a mercury-vapour turbine and a steam turbine, the exhaust from the mercury turbine being utilized to generate steam for the steam turbine.

A mercury-vapour generator 8, provided with tubes 6, is heated by a furnace 5. The mercury vapour passes by way of a pipe 20 to a turbine 15, whence it is exhausted by way of a pipe 22 to a condenser 21, which also functions as a steam-generator. The condensed mercury returns to the generator 8 by way of a pipe 25 and economizer 27, situated in the furnace flue. The water for the steam turbine is heated in a coil 40 situated in the furnace flue, and afterwards passes through the generator 21 where it is formed into steam by the heat given up by the mercury. The steam is led by a pipe 45 to a steam turbine 18, which exhausts by way of a conduit 46 to a condenser 47.

A by-pass valve 41 is provided to divert mercury vapour from the turbine when the pressure rises above a predetermined limit. A cooler 48, divided into two compartments by a perforated partition 52, is connected as shown to the mercury and steam condenser systems, and to an air-pump 49 so that one pump serves to extract air from the mercury and water. The action of the cooler prevents any mercury vapour from being withdrawn by the pump. The valves of the two turbines are under the control of a single governor, but are so set that the mercury valve is first opened and consequently the regulation due to variations of load is mainly effected through the steam turbine. The turbines may be arranged upon a common shaft or upon separate shafts.

This is most interesting because it shows that even at this early date the technology for later plants was in place, notably the use of Field tubes in the mercury boiler, and a turbine bypass valve rather than a conventional safety valve venting to atmosphere. (The latter not being a good idea in a mercury plant)


Emett first proposed the use of mercury vapour in 1914, and got the backing of T H Soren, the vice-president of the Hartford Electric Light Company. (HELCO) Following a deal with the General Electric Company, a 5000 HP experimental system was built at the company's Dutch Point generating station in 1923. The system contained a hefty 20 tons of mercury, and used mercury vapor at 35 psi, (gauge) and 812 degF to drive a single-wheel mercury turbine. The mercury condensed at 1.5 psi, (absolute) and 485 degF, and in doing so generated steam in excess of 200 psig to drive a steam turbine. To improve efficiency, the steam was reheated in the mercury boiler to about 700F.

(Info from Connecticut, the Industrial Incubator 1982, published by the History and Heritage Committee of the Hartford section of the Americian Society of Mechanical Engineers.

There were some problems; it was reported in Popular Science (March 1931) that at one point a boiler head blew out with the escape of thousands of dollars worth of mercury, only part of which was recovered. Also a turbine disc disintegrated and put the unit out of action for several months. It was stated that "workers were overcome by the fumes of the poisonous mercury. There were, however, no fatalities, nor was anyone permanently injured, because of the unusual precautions taken by the company to protect the men who had charge of the work." That does not sound like mercury poisoning as we know it, Jim. One wonders about the long-term health of the workers who were "overcome".

After this plant had been fully studied a second experimental plant was built at South Meadow, giving improved results but also suffering some problems.

Info from Popular Science Sept 1929 and March 1931


In November 1928 the first commercial mercury cycle generating unit in the world began operation at Helco's South Meadow Station in Hartford.

Left: The first commercial mercury plant installed at South Meadow in 1929

The mercury turbine ran on mercury vapour at 884 degF. The power output was 13,000 HP. More than a thousand gallons of mercury, weighing nearly 70 tons, were in use in the cycle. The current price of mercury was about $200 per gallon; the working fluid represented an investment of more than $200,000. Leaks were discouraged.

According to Popular Science for September 1929, the plant had recently been dismantled for engineering improvements to be made.

From Popular Science September 1929

Left: The two mercury-condenser/steam-boilers at the plant at South Meadow in 1931

The snail-shaped thing between them is the mercury turbine.

From Popular Science, March 1931

Left: The mercury part of the system at South Meadow in 1931

It appears from the picture that the mercury vapour safety valve does not exhaust to atmosphere, (which would be a really bad idea) but just bypasses the turbine. This might prevent turbine overspeed, though presumably there were conventional safety devices that measured the rotor speed directly, but it would not stop the boiler exploding. (which would be an even worse idea) I suspect that in reality the safety valve was connected to a big empty tank.

This magnificent illustration is taken from Popular Science, March 1931

Left: The mercury part of the plant at South Meadow as at 1949

The mercury vapour leaves the mercury boiler drum, and drives the turbine at the top of the drawing; it then exhausts to the two adjacent mercury-condenser/steam-boilers; these are as close as possible because of the difficulties in handling large volumes of mercury vapour. Feed water passes through the economiser, leaving at 410 degF, and is converted to steam at 450 degF and 410 psi by the condensing mercury. This steam passes through a superheater in the top of the mercury boiler and then goes to the station steam header at 700 degF and 385 psi, which powers conventional steam turbines. The condensed mercury goes back to the mercury boiler drum down a spiral path to prevent the heavy mercury building up excessive momentum and wrecking the pipe-work. Note the use of air pre-heating; air was supplied to the burners at 523 degF.

The South Meadow plant was reported in Popular Science March 1931 as producing 143 kiloWatt-hours of electricity per 100 pounds of coal burned, while the best conventional steam plant produced only 112 kiloWatt-hours. The average efficiency acrosss the whole country was only 59 kW-h per 100 lb of coal.

The legacy of these experiments is still with us. Mercury contamination is in sediment of the the Connecticut River. Wethersfield Cove just south of Dutch Point is a former bend in the Connecticut River (it is an oxbow lake) that has up to 3 ppm Hg in its sediment and is stated to contain more than half a ton of mercury

From Power Generation March 1950

The following was sent to me by Rich Colopoulos in August 2017:

"I would like to pass on some anecdotes about South Meadow Station. I was hired by Northeast Nuclear Energy Company in December of 1980 as a Mechanics Helper on Unit 2 of the Millstone Nuclear Power Station Complex in Waterford Connecticut. This is about 50 miles southeast of the South Meadow Power Station. South Meadow is located on the Connecticut River which empties into Long Island Sound which is were Millstone is located on. Some of the mechanics I worked with came from South Meadow Station after it was closed (I don't know the closing date but Unit 2 came online in the mid seventies)."

"I was assigned to help an older gentleman named Jimmy Preste who was in general just a great guy and loved to tell stories to pass the time while we were supposed to be working! He told me that they used to put wooden boards down on the mercury in the main condenser and that they would provide enough buoyancy so that could walk on them to service the condenser! I don't know if this is true but it sure is a great story and I don't know why he would have made it up, I never asked about the mercury he just volunteered the information. He and other mechanics also volunteered the ominous information that "there was a lot of mercury down by the river". I sure hope that this is not so true or that if so it has been cleaned up."

Far be it from me to question, but I would not have thought that an ordinary wooden board floating in mercury would have borne a man's weight; it should be easy enough to calculate. Any volunteers?


The Schiller power station at Portsmouth, New Hampshire, USA, was the last binary mercury/steam plant built. The site has two decommissioned GE Mercury Turbines and two B&W Mercury/Steam boilers. It currently uses three 50 megawatt coal-fired steam boilers built in the 1950s. One of these is being replaced with a fluidized-bed boiler to burn whole-tree wood chips and other clean low-grade wood materials. Read about The Northern Wood Power Project (external link)

The Schiller station consisted of two 7500 kW mercury units and one 25,000 kW steam turbine set. The rated capacity was 40,000 kW. It was equipped to run on either Bunker "C" fuel oil or pulverised bituminous coal.

I am deeply grateful to David E Corey for providing these photographs of the plant, taken in 2006.

Left: Mercury turbine-alternator in the centre, with mercury/steam boilers either side

Looking at the unit from the Alternator end. The two units on the sides are the mercury/steam boilers; from these the steam traveled back to the top of the mercury boiler, was superheated and traveled out to a steam turbine that has since been sold to Mexico...

Once again the mercury turbine has been installed between the two condenser/boilers to minimise the path length for the large volumes of mercury vapour.

Picture and info courtesy of David E Corey

Left: Mercury Vapor Lines

The top pipe in the foreground is a steam line from one of the two mercury-condenser/steam-boilers, returning to the main mercury boiler for superheating; the 3 pipes in the background are the three separate mercury lines that go to the turbine nozzles. The mercury lines each have their own manual throttles and isolation valves.

Picture and info courtesy of David E Corey

Left: Mercury Drum at the mercury Boiler

Picture courtesy of David E Corey

Left: Upper Level area in front of Mercury Boiler

The vertical pipes are the mercury feed to drum lines. Sootblowers can be seen between them.

Picture courtesy of David E Corey

Left: Underneath of mercury/steam boiler

In the mercury/steam boiler the mercury condensed as it generated steam. The big pipe is the condensed mercury drain line.

Picture & info courtesy of David E Corey

Left: Mercury storage tank

Picture courtesy of David E Corey

Left: One of the instrument panels for the power-plant

The top left gauge reads mercury pressure at Boiler #1 stop-valve. The centre gauge reads steam pressure at the turbine stop valve. The missing gauge at top right was for mercury pressure at Boiler #2 stop-valve; presumably Boiler #2 was decommissioned before #1.

In the original photograph the black labels, each with two indicator lamps, running down each side of the panel are legible. From the top the left column labels are 'Cooling water pump No 1', 'Cooling water pump No 2', 'Cooling water sump pump No 1' and so on.

On the right hand column the remaining labels are 'Condensate return pump No 1', 'Condensate return pump No 2', 'Nitrogen cooler pump No 1','Nitrogen cooler pump No 2', 'Air compressor No 1', and 'Air compressor No 2'.

Picture & info courtesy of David E Corey


I have unearthed a NASA report from 1969 that describes a nuclear-heated mercury power plant intended to generate large amounts of electricity in space. It is called :"DESIGN AND FABRICATION OF A COUNTERFLOW DOUBLE-CONTAINMENT TANTALUM-STAINLESS STEEL MERCURY BOILER" from the Lewis Research Center, Cleveland, Ohio. It was part of the SNAP-8 program to develop a Rankine cycle power system for space applications.

The reactor is cooled with a mixture of sodium and potassium, (NaK) which heats a mercury boiler. The mercury vapour drives a turbo-alternator and is then condensed and subcooled by a secondary NaK heat rejection loop which transfers the waste heat to a radiator for rejection to space. Why is a second NaK loop used? I have no idea, but this is NASA we're talking about so I bet there was a good reason.
The report concentrates on the design of the potassium/mercury boiler, which is an exotic device indeed. It is a spiral stucture designed to fit in a small toroidal space in a cylindrical spacecraft; the heat exchanger tubes are of tantalum, which is readily wetted by mercury, making for good heat transfer. The net power output was 37 kW.

Left: The NASA nuclear mercury turbine system.

Note the extra cooling and lubricating loop using polyphenyl ether.

Left: The performance data.

The extraordinary thing is that the overall efficiency seems horribly low at 6.9%. A conventional terrestrial steam power station would give some where around 40%.

So why is it so bad? The Carnot efficiency E = (T1 - T2 ) / T1 where T1 is the heat input (top) temperature and T2 the heat rejection temperature. Looking at the diagram above, the mercury leaves the boiler at 670 degC. The effective rejection temperature in the condenser is a bit less clear but the NaK coolant leaves it at 335 degC. This gives a Carnot efficiency of 35%. That's obviously a theoretical maximum, but a long step from 6.9%; there must be some serious losses somewhere.

The rejection temperature has to be high because in space there are no rivers flowing by to cool your condensers- the only way to lose heat is to radiate it away from a flat plate, which needs to be pretty hot to radiate effectively. Hence the radiators in the flow diagram; these are nothing like central-heating "radiators" which work mostly by convection.
This is one reason- possibly the most important- for the choice of mercury as a working fluid; it permits a high heat rejection temperature so the radiators are efficient, and therefore relatively small and light.

This project was only one of several nuclear-powered sytems that were intended to provide large quantities of electricity for long-duration space missions. One of the reasons that they have not so far been used is that no-one is very comfortable with the notion of launching a nuclear reactor on top of a rocket.

You can read the full NASA report on the potassium/mercury boiler here (External link)

A word of explanation about the use of sodium and potassium alloys. (NaK) When they are between about 40% and 90% potassium by weight they are liquid at room temperature, which is very handy for a cooling fluid. The mixture with the lowest melting point (the eutectic mix), consists of 78% potassium and 22% sodium, and is liquid over the enormous range of -12.6 to 785 C.

Another exotic heat-transfer/lubricating fluid is used in this system- polyphenyl ether, which is interesting stuff in itself, consisting of rotatable benzene rings joined by ether (-O-) links. You can read more about it here (External link)

Back to Home PageBack to The Museum EntranceTop of this page