Full Text
THE ATOMIC HYDROGEN FLAME¹
Irving Langmuir.
If a tungsten wire is heated by an electric current in a vacuum, the heat is dissipated almost entirely by radiation. The radiated energy increases in proportion to the absolute temperature raised to the power 4.7. If an inert gas—say nitrogen or argon—is introduced into the vessel at atmospheric pressure, then a considerable part of the heat will be dissipated by convection and thermal conduction. These losses due to the dissipation of energy by convection and thermal conduction can be determined by subtracting from the total expenditure of energy the energy which, at the same temperature, is radiated in a vacuum. In this way it was found² that the losses due to convection and thermal conduction increase in proportion to the temperature raised to the power 1.9, up to temperatures as high as \(3660^\circ K\) (the melting temperature of tungsten).
This result is in excellent agreement with the theory of heat convection, based on the assumption that in the immediate vicinity of the filament the heat is transferred by thermal conduction.
In the case of heating a tungsten filament in hydrogen³ the heat losses increase in proportion to the temperature raised to the power 1.9 only up to \(1700^\circ K\), and thereafter grow much more rapidly. For example, between \(2600^\circ\) and \(3400^\circ K\) the heat losses in hydrogen at atmospheric pressure increase in proportion to the 5th power of the temperature, and at a pressure of \(55\) mm—even to the 6th power.
The abnormal behavior of hydrogen is explained by the fact that at high temperatures hydrogen molecules dissociate into atoms according to the equation:
\[ \mathrm{H}_2 = 2\mathrm{H}, \tag{1} \]
absorbing in the process a large quantity of energy. The enormous thermal conductivity of hydrogen at high temperatures (at \(3400^\circ K\), 23 times
¹ General Electric Review, March, 1926. Translated by S. N. Rzhevkin.
² I. Langmuir, Phys. Rev., 34, 401, 1912.
³ I. Langmuir, Trans. Amer. Electrochem. Soc., 20, 225, 1911; Journ. Amer. Soc., 34, 860, 1912.
greater than in nitrogen) is caused by the absorption of energy from the incandescent wire by dissociating hydrogen molecules, and then by the liberation of this energy in the colder gas upon recombination of the diffused hydrogen atoms into molecules.
Confirmation of the view that this effect is due precisely to the formation of monatomic hydrogen, and not of some other endothermic polymorphic form of hydrogen, for example, \(H_3\) (corresponding to ozone \(O_3\)), may also be seen in the fact that the loss of heat by the incandescent filament at high temperatures is considerably greater at low hydrogen pressures (50 mm) than at atmospheric pressure. According to the law of mass action, the degree of dissociation of a gas into atoms must be greater at low pressures than at high; whereas in the formation of molecules containing more than two atoms the opposite phenomenon would be observed.
Experimental facts were soon obtained \(^{1}\), showing that hydrogen at low pressures, when brought into contact with tungsten or platinum at \(1300^\circ K\) or higher, acquires entirely new chemical properties, which very much resemble the properties of hydrogen in the atomic form.
It was found that metallic oxides \(^{2}\) such as \(WO_3\), \(CuO\), \(Fe_2O_3\), \(ZnO\), or \(PbO_2\), placed in a vessel filled with hydrogen at low pressure, are rapidly reduced to the metallic state if a tungsten filament located in the same vessel (or in another vessel connected by a glass tube) is heated to a temperature above \(1500^\circ K\). At the same time the hydrogen in the vessel gradually disappears. Thus atomic hydrogen, produced by an incandescent filament, can react with certain metallic oxides at room temperature, whereas molecular hydrogen does not possess this property. Atomic hydrogen can also react at room temperature with oxygen or phosphorus (forming \(PH_3\)). The dissociation of hydrogen by an incandescent tungsten filament can, however, be prevented \(^{3}\) by a negligible quantity of oxygen or water vapor, if it is in contact with the wire.
Atomic hydrogen also has the property of dissolving in platinum at room temperature \(^{4}\), and causes a considerable increase in its electrical resistance. By bringing the platinum wire into contact with oxygen, its resistance can be returned to the normal value. A quantitative study of the heat loss by a tungsten wire at various temperatures in hydrogen, especially—
\(^{1}\) I. Langmuir. Journ. Amer. Chem. Soc., 34, 13 ff., 1913.
\(^{2}\) Langmuir. Trans. Amer. Electrochem. Soc., 29, 254—5, 1916.
\(^{3}\) I. Langmuir. Journ. Amer. Chem. Soc., 35, 2251, 1916; Gen. El. Rev. 25, 445, 1922.
\(^{4}\) Freeman. Journ. Amer. Chem. Soc., 35, 927, 1913.
at low pressures, makes it possible to determine the degree of dissociation1 and the heat of formation of the hydrogen molecule from atoms. Since these experiments were carried out more carefully, the data concerning the temperature of the tungsten filament came into general use. Lewis and Randall2, as well as other authors, showed that the third law of thermodynamics gives a relation between the heat of dissociation and the degree of dissociation.
By means of this relation and the new temperature scale of Forsythe and Worthing3, the experimental data of 1914 were recalculated, and the degree of dissociation of hydrogen at various temperatures was found. The results are expressed by the formula:
\[ \log_{10} k = - \frac{21,200}{T} + 1.765 \log_{10} T - 9.85 \cdot 10^{-5} T - 0.256, \tag{2} \]
where \(k\) is the equilibrium constant, defined by the relation
\[ k = \frac{p_1^2}{p_2}, \tag{3} \]
where \(p_1\) is the partial pressure of atomic hydrogen, and \(p_2\) is the pressure of molecular hydrogen in atmospheres. Let \(x\) be the degree of dissociation of hydrogen, i.e., the fraction of hydrogen molecules that have decomposed into atoms. Then, if \(P\) is the total pressure, we have:
\[ p_1 = \frac{2Px}{1+x}; \quad p_2 = \frac{(1-x)P}{1+x}, \tag{4} \]
and the equilibrium constant \(k\) in equation (2) will be determined by the relation:
\[ k = 4P^2 \frac{x^2}{(1-x)^2}. \tag{5} \]
Table 1 gives the equilibrium constant and the degree of dissociation at various temperatures, calculated on the basis of equations (2) and (5).
From equation (2), applying the Clapeyron equation, we can calculate \(H\)—the heat absorbed in the dissociation of molecular hydrogen (at constant pressure)
\[ H = 97,000 + 3.5T - 0.00045T^2. \tag{6} \]
This quantity is expressed in small calories per gram-molecule of hydrogen (2.016 g).
The heat of reaction at constant volume will be:
\[ Q = 97,000 + 1.5T - 0.00045T^2. \tag{7} \]
IRVING LANGMUIR
TABLE 1.
| Temp. T° | K atmospheres | \(x\) at 1 atm. | Temp. T° | K atmospheres | \(x\) at 1 atm. |
|---|---|---|---|---|---|
| 300 | \(2{,}63 \cdot 10^{-64}\) | \(2{,}56 \cdot 10^{-34}\) | 3 200 | \(9{,}78 \cdot 10^{-2}\) | 0,154 |
| 1 000 | \(5{,}50 \cdot 10^{-17}\) | \(3{,}71 \cdot 10^{-9}\) | 3 400 | 0,256 | 0,245 |
| 1 200 | \(2{,}48 \cdot 10^{-13}\) | \(2{,}48 \cdot 10^{-7}\) | 3 600 | 0,593 | 0,361 |
| 1 400 | \(1{,}04 \cdot 10^{-10}\) | \(5{,}08 \cdot 10^{-6}\) | 3 800 | 1,24 | 0,438 |
| 1 600 | \(9{,}81 \cdot 10^{-9}\) | \(4{,}95 \cdot 10^{-5}\) | 4 000 | 2,56 | 0,625 |
| 1 800 | \(3{,}42 \cdot 10^{-7}\) | \(2{,}92 \cdot 10^{-4}\) | 4 500 | 10,9 | 0,855 |
| 2 000 | \(5{,}93 \cdot 10^{-6}\) | \(1{,}22 \cdot 10^{-3}\) | 5 000 | 34,7 | 0,9469 |
| 2 200 | \(6{,}18 \cdot 10^{-5}\) | \(3{,}92 \cdot 10^{-3}\) | 6 000 | 169,0 | 0,9884 |
| 2 400 | \(4{,}36 \cdot 10^{-4}\) | \(1{,}04 \cdot 10^{-2}\) | 7 000 | 649 | 0,9969 |
| 2 600 | \(2{,}30 \cdot 10^{-3}\) | \(2{,}40 \cdot 10^{-2}\) | 8 000 | 1570 | 0,4987 |
| 2 800 | \(9{,}58 \cdot 10^{-3}\) | \(4{,}83 \cdot 10^{-2}\) | 9 000 | 3030 | 0,9993 |
| 3 000 | \(3{,}30 \cdot 10^{-2}\) | \(9{,}03 \cdot 10^{-2}\) | 10 000 | 5000 | 0,9996 |
Thermal Conductivity of Hydrogen.
Upon dissociation of hydrogen its thermal conductivity increases greatly. Let us consider the law determining the loss of heat in hydrogen with slight dissociation. The author has shown that in this case the loss of heat through thermal conductivity and convection by a body of temperature \(T_2\), in a gas of temperature \(T_1\), will be
\[ W_c = S \cdot (\Phi_2 - \Phi_1). \tag{8} \]
Here \(W_c\) is expressed in watts; \(S\) is a factor depending only on the shape and dimensions of the cooling body and on the relative thickness of the surrounding layer of gas, but not depending on temperature. The quantities \(\Phi_2\) and \(\Phi_1\) are expressed as follows:
\[ \Phi = \int_{0}^{T} k\,dT; \tag{9} \]
where \(k\) is the coefficient of thermal conductivity.
For high temperatures in hydrogen we have approximately
\[ \Phi_2 = 1{,}06 \cdot 10^{-4} T^{3/2} \]
and \(\Phi_1 = 0{,}3\) watts per cm at 300°.
FLAME OF ATOMIC HYDROGEN
The dissociation of hydrogen increases the amount of heat removed from the heated body by the amount
\[ W_D = S D Q_1 c, \tag{10} \]
where \(S\) is the same shape factor, \(D\) is the coefficient of diffusion of hydrogen atoms through molecular hydrogen, \(Q_1\) is the heat evolved (in watt-seconds) when \(1\ \mathrm{g}\) of hydrogen atoms combines into molecules; \(c\) is the concentration (expressed in grams per \(\mathrm{cm}^3\)) of atomic hydrogen at the temperature \(T_2\).
It was found that
\[ D = 0.00214\, T_2^{\frac{3}{2}} . \tag{11} \]
From equation (7) we see that, at temperatures of \(2000\)–\(3000^\circ K\), \(Q_1\) is approximately equal to \(49{,}000\) cal. or \(205{,}000\) watt sec.
\[ c=\frac{0.0244}{T_2}\cdot \frac{Px}{1+x}. \tag{12} \]
Substituting these quantities into (10), we obtain:
\[ W_D = 10.7\, S \sqrt{T_2}\, P \cdot \frac{x}{1+x} \tag{13} \]
(\(W_D\) in watts, \(P\) in atmospheres, and \(S\) in \(\mathrm{cm}\)).
At \(T_2 = 3600\) and one atmosphere, \(x = 0.361\) and \(W_D = 170\,S\), whereas from equation (8) for this temperature we have \(W_c = 22.4\,S\). Thus, the heat transfer in hydrogen between 3600 and 300 increases by a factor of 8.6 owing to the presence of dissociation. At 5000 the increase in heat transfer will be approximately elevenfold.
Arc in hydrogen at low pressure.
In attempting to obtain the Balmer spectrum of hydrogen without the superposition of the secondary spectrum, R. W. Wood1 constructed a very long discharge tube of small diameter, through which he passed, in an atmosphere of moist hydrogen at a pressure of \(0.5\ \mathrm{mm}\), a current of 20 amperes. In doing so he observed a number of remarkable phenomena. Short pieces of tungsten wire introduced into the region of the discharge became strongly incandescent, whereas a thin glass thread, or a platinum wire, in the same position was not heated to any appreciable extent. When the hydrogen was dried with phosphorus pentoxide, a strong secondary spectrum appeared (belonging to molecular hydrogen), while the Balmer spectrum (belonging to atomic hydrogen) almost disappeared. Incandescence of the tungsten wire was not observed in dry hydrogen.
In agreement with Prof. Wood, the author assumed that moisture paralyzes the catalytic action of dry glass walls, which accelerate the reverse transformation of atomic hydrogen into molecular hydrogen. Thus the tube with moist hydrogen was filled with almost pure atomic hydrogen and, owing to its diffusion to the surface of the catalytically active tungsten wire, the latter was heated. A calculation based on a formula analogous to (13) showed that at an atomic-hydrogen pressure of only 0.16 mm and at 500° it would be sufficient to maintain the tungsten wire at 2400° K.
These conclusions are confirmed by Wood’s observations: the walls of the tube are only slightly heated in moist hydrogen, whereas in dry hydrogen the heating is very strong. The tungsten wire becomes red-hot even in a side tube (5 mm in diameter) at a distance of 4 cm from the discharge tube, which indicates the ability of hydrogen atoms to diffuse in comparatively large quantities from the discharge region.
On the basis of what has been set forth, the author conceived the possibility of obtaining higher concentrations of atomic hydrogen by producing a powerful voltaic arc between tungsten electrodes in hydrogen at atmospheric pressure and blowing the atomic hydrogen out of the arc by means of a stream of hydrogen.
The arc in hydrogen at atmospheric pressure.
The study of arcs in various gases between tungsten electrodes was begun in the General Electric Co. laboratory several years ago1. The arc in hydrogen stands out because of its large voltage drop and small cross-section. A direct-current arc of 10 amperes between massive tungsten electrodes at a distance of 7 mm in hydrogen at atmospheric pressure has the appearance of a sharply defined bright red line, about 0.5 mm in diameter; the potential gradient is 150 volts per cm, which is approximately 15 times greater than in nitrogen or argon. This abnormal behavior of hydrogen was attributed to dissociation, which is the cause of the enormous consumption of energy in the arc.
A simple calculation shows that the ordinary release of heat by thermal conduction and convection, without the participation of dissociation, can in no case explain so large an energy consumption as 1500 watts per cm of arc length. The calculation also shows that the loss of heat by convection cannot be so great. Furthermore, from calculations not given here, it follows that between cylinders of 0.5 mm and 1.5 mm diameter the heat transfer observed
of magnitude is possible by thermal conduction in pure atomic hydrogen when the temperature falls from 10,000° to 5,000°. Between the 1.5 mm and 6 mm cylinders, it is possible to explain the transfer of heat by diffusion of atomic hydrogen through molecular hydrogen. It proves impossible to explain the transfer of heat from the surface of a cylinder 6 mm in diameter to the walls of the vessel either by thermal conduction of molecular hydrogen or by convection. Therefore one has to conclude that atomic hydrogen transfers heat directly to the massive tungsten electrodes, which is quite probable, since in the experiments of Mackay and Ferguson the length of the arc was 7 mm, and one of the electrodes was a massive piece of tungsten about 15 mm in diameter. We thus arrive at the conclusion that the removal of the large quantity of heat developed in the hydrogen arc can be explained only by the proximity of massive tungsten electrodes.
Preliminary experiments with a flame of atomic hydrogen
In order to test the possibility of blowing atomic hydrogen out of the arc, an alternating-current arc of 20 amp. was constructed, burning between two tungsten wires 6 mm in diameter, placed across a horizontal alundum tube, 10 cm in diameter, through which a stream of hydrogen was passed. At voltages from 300 to 800 volts it was possible to maintain an arc up to 2 cm long. The magnetic field of the arc blew it out, so that it assumed a fan-shaped form. Iron wires 2–3 mm in diameter melt within one or two seconds if they are held at a distance of 3–5 cm above the arc.
By directing a stream of hydrogen from a thin tube into the arc, it is possible to blow atomic hydrogen out of the arc and form an extremely hot flame. To maintain the arc in a stable condition, the electrodes have to be brought together to a distance of 1–3 mm; the arc is then not held wholly between the electrodes, but extends in the form of a fan to a distance of 5–8 mm above the electrodes. The flame of atomic hydrogen extends far beyond the limits of the arc. At a distance of 1 or 2 cm from the arc, molybdenum melts readily (melting point 2900°). At the edge of the arc one can melt a tungsten wire (melting point 3660° K). Quartz melts much more reluctantly than molybdenum, whence it follows that the cause of the rapid heating of metals in the flame of atomic hydrogen is due, in part, to their catalytic action, which accelerates the combination of hydrogen atoms into molecules on the surface of the metal.
The use of hydrogen under these conditions for melting metals has many advantages. Iron can be fused without contamination by carbon, nitrogen, and oxygen. Owing to the strong reducing—
to the action of atomic hydrogen, alloys containing chromium, aluminum, silicon, or manganese can be melted without the use of fluxes and without surface oxidation. In an atomic-hydrogen flame, rather large pieces of alumina, $\mathrm{Al_2O_3}$, manganese oxide, $\mathrm{MgO}$, or thorium oxide, $\mathrm{ThO_2}$, can readily be melted. Oxides melted in this way show no appreciable reduction to metals, which probably occurs for the reason that the metal formed at such a high temperature immediately volatilizes.
The technical development of the method of utilizing the atomic-hydrogen flame is the fruit of the work of many persons, among whom it is necessary to mention R. Palmer and R. Weiman.
The Temperature of the Atomic-Hydrogen Flame in Comparison with the Temperature of Other Flames.
Let us suppose that we could obtain atomic hydrogen in a vessel at atmospheric pressure and room temperature, and that we could “burn” it, converting it into the molecular form in the flame. What would be the temperature of this flame, and how is it to be compared with the temperature of other flames? Taking the heat of reaction (per $2\ \mathrm{g}$ of hydrogen) as equal to 98,000 calories and the specific heat of molecular hydrogen (per $2\ \mathrm{g}$) as equal to $6.5 + 0.0009\,T$, we find that the heat of reaction is sufficient to heat the hydrogen to $9200^\circ K$.
A similar calculation for an oxyhydrogen flame gives a temperature of $3700^\circ$, and for an oxyacetylene flame—$7000^\circ K$. Of course, at such high temperatures the combustion products will be strongly dissociated, so that the full heat of reaction cannot be liberated, as a result of which the flame temperatures obtained are much lower. For the atomic-hydrogen flame we shall find:
\[ (1-x)H_0 = [2xC_1 + (1-x)C_2]T, \]
where $T$ is the flame temperature, $x$ is the degree of dissociation at this temperature, and $C_1$ and $C_2$ are the mean specific heats of the gram-atom and gram-molecule. By trial, using the values of $x$ from Table 1, we find that this equation is satisfied if we put $T = 3990^\circ K$ and the degree of dissociation $x = 0.62$. Similar calculations for other flames, taking into account the degree of dissociation of water vapor and carbon dioxide, give for the oxyhydrogen flame $T = 3370^\circ$ and $x = 0.20$, and for the oxyacetylene flame $T = 3750^\circ K$, at which temperature the water vapor proves to be dissociated by 38%, and the carbon dioxide by 89%.
These temperatures of the oxyhydrogen and oxyacetylene flames are undoubtedly too high, since hydrogen (and,
possibly oxygen) in the partially dissociated products of combustion at these high temperatures must be strongly dissociated into atoms. As a consequence of this, the degree of dissociation must also be put higher than was assumed in the preceding calculations. An error of this kind does not enter into the calculation of the temperature of the atomic-hydrogen flame. Thus we arrive at the conclusion that the atomic-hydrogen flame is considerably hotter than both the oxyhydrogen and the oxyacetylene flames.
There is still another factor that compels us to regard the temperature of the atomic-hydrogen flame as considerably higher than the \(4000^\circ\) obtained from calculation. Namely, at the moment of leaving the arc, atomic hydrogen is not at room temperature, as the calculation assumes, but at a much higher temperature. The conditions are partly analogous to those in the oxyhydrogen flame, where both gases are preheated beforehand.
Fig. 1. Burner for the atomic-hydrogen flame
Thus, the upper limit of the temperature is determined only by the degree of dissociation of hydrogen and by the rate at which heat is lost through radiation and contact with bodies of low temperature.
Fig. 2. Appearance of the atomic-hydrogen flame.
The circumstance that the degree of dissociation of hydrogen into atoms at a given temperature is considerably less than the degree of dissociation of carbon dioxide into oxygen and carbon monoxide is an important factor making the atomic-hydrogen flame much hotter than the oxyacetylene flame.
The amount of energy transferred by the atomic-hydrogen flame to some surface can be easily calculated by formula (13).
Suppose that a flame or jet of atomic hydrogen is directed onto a flat metallic surface whose temperature is below 2,000° and which catalyzes the combination of hydrogen atoms. Thus, on the surface of the metal the concentration of atoms will be practically zero. At a distance of 5 mm from the surface let us assume a degree of dissociation of 0.50 and a temperature of 4,000°. The form factor in equation (15) will be equal to the area of the given surface divided by the thickness through which diffusion occurs. From these considerations we find that the energy imparted to the surface is equal to 450 watts per cm² per second, which is 25 percent greater than the amount of energy radiated
Fig. 3. Steel tubes one inch in diameter, welded along their length in an atomic-hydrogen flame.
Fig. 4.
Fig. 5.
Welding a chromed-steel tube to glass. Welding in an atomic-hydrogen flame gives an extremely strong joint between dissimilar metals. The place of welding may be stretched, bent, twisted, or subjected to any other deformation without fracture.
from the surface of tungsten at the melting temperature (3,660°). The rate of diffusion of atomic hydrogen thus proves to be so great that, for rapid melting of the metal, it is not necessary for the gas jet to bring atomic hydrogen closer than 5 mm to the metal surface.
A jet of molecular hydrogen of 100 cm³ per second (at room temperature), when converted into atomic hydrogen at 5,000° ($x = 0.95$), is capable of liberating a power of 1.6 kW. This figure may give a clearer idea of the applications that an atomic-hydrogen flame can have for the melting and soldering of metals.
Apparatus of the Arc for Obtaining Atomic Hydrogen.1
A burner for soldering and welding by means of an atomic-hydrogen flame is shown in Fig. 1 in two different forms. The electrodes, between which the arc is produced, consist of two tungsten rods set at an acute angle; they are insulated
Fig. 6. Steel tube, flattened after welding without the slightest kink or crack along the line of the weld. The atomic-hydrogen flame makes it possible to weld other metals as well, such as, for example, copper and steel, copper and German silver, or to weld different grades of steel.
from each other [[unclear: continuation cut off on this page]].
Fig. 7. Steel plate, \(1/8\) inch thick, welded lengthwise from two parts and then bent through an angle of \(90^\circ\).
by means of lava insulators. The hydrogen jet is directed precisely at the arc through a small opening. The figure shows an arrangement that makes it possible to surround the welding site with an atmosphere of hydrogen, which emerges in a slow stream from a whole series of openings. The arc is supplied with alternating current from a special transformer at a voltage of 400 volts. A direct-current arc at a voltage of 250 volts also gives a good result. The current in the arc is from 20 to 70 amperes, depending on the thickness and massiveness of the objects being welded.
A flame analogous to the flame of atomic hydrogen, but less energetic, can also be obtained with illuminating gas or with ammonia decomposed at a temperature of \(650^\circ\mathrm{C}\) into its constituent parts \((N+H_3)\).