THEORY OF THE CAKING PROCESS OF GRANULAR AMMONIUM NITRATE AND POSSIBLE WAYS TO COMBAT CAKING

N. P. KURIN

Introduction

Pure granular ammonium nitrate, during storage, cakes into very strong monolithic lumps and, as a result, loses its free-flowing character, which is extremely necessary for the uniform application of this fertilizer to the soil.

Organizing the process of crushing caked nitrate at the places where it is consumed is associated with a number of serious difficulties. Therefore, the problem of obtaining non-caking \(\mathrm{NH_4NO_3}\) has acquired primary importance.

In the USSR and abroad, a number of studies were carried out on the process of caking and on finding ways to combat it. However, in all these studies the mechanism of formation, the structure of the granules, and the mechanism of caking of granular nitrate in a layer were not given proper elucidation. As a result, it was not possible to assess the reliability of individual methods of combating caking or to correctly determine the direction of further investigations in this field.

The purpose of the present work is to partially fill this gap.

Mechanism of Formation and Structure of Granules

Granular ammonium nitrate is obtained by spraying a melt and cooling the droplets as they fall in a hollow tower in a stream of rising air.

The melt fed to granulation at \(160\text{--}165^\circ\) contains \(\sim 98.5\%\) \(\mathrm{NH_4NO_3}\).

The temperature of the granules when they arrive on the conveyor, depending on the temperature of the atmospheric air supplied for cooling, is \(80\text{--}98^\circ\) in summer and \(70\text{--}80^\circ\) in winter. Owing to cooling on the belt conveyor, the temperature of the granules before loading into paper bags decreases in summer to \(60\text{--}70^\circ\) and in winter to \(55\text{--}60^\circ\). Granular nitrate contains \(0.8\text{--}1.0\%\) moisture. The granules have a more or less regular spherical shape. The granulometric composition is very nonuniform; along with large granules (\(1.5\text{--}3\ \mathrm{mm}\)) there is a significant quantity (\(\sim 26\%\)) of small granules (less than \(1.5\ \mathrm{mm}\)) [1].

When individual granules are examined under a microscope, cracks and large and small pores are found. Often large pores and voids are visible to the naked eye. Of course, the fine internal structure of the granules cannot be examined under a microscope with glass lenses; however, the granules must contain a large number of fine pores in the form of voids in intercrystalline spaces.

Schemes of solidification of a rotating and a nonrotating granule during free fall in a stream of rising air are shown in Fig. 1.

The nonuniformity of the thickness of the solid shell of a nonrotating granule may be caused by differences in the rate of heat transfer. Physical heat and the heat of crystallization are transferred from the center of the liquid melt droplet to its periphery by thermal conduction, and from the periphery are removed by convection and radiation. In this case, in the peripheral region of the droplet the temperature

Fig. 1. Schemes of solidification of granules during fall in a stream of rising air: a) rotating granule, b) nonrotating granule

falls below the saturation temperature, a supersaturated solution is formed, and solidification begins with the crystallization front directed from the periphery toward the center.

In the initial stage of cooling and crystallization, when the highest cooling rate occurs and, consequently, the greatest degree of supersaturation of the solution, a thin solid shell of the finest crystals should form on the surface of the droplet. As a result of less intensive cooling, and consequently of a lower supersaturation of the solution, the inner regions of the granule should be filled with coarser crystals, with the crystallization front directed counter to the heat flow, i.e., from the periphery toward the center.

Degree of filling of the granule volume with crystals, mother liquor, and air. Porosity of the granule

Without departing substantially from the truth, it may be assumed that the volume of the thin shell of the granule will be close to, or in the limiting case equal to, the volume of the droplet at the initial crystallization temperature. The subsequent crystallization process, associated with the appearance of solid phases having a considerably greater specific gravity than that of the melt, must lead to the formation, within the volume bounded by the thin shell of the granule, of intercrystalline void spaces filled with mother liquor and air (Fig. 2).

Let us consider the quantitative aspect of the volumetric changes occurring during the formation of a granule from a droplet. As initial assumptions we adopt the following.

1. The weight of the droplet is equal to the weight of the granule, since the change in weight, due to slight drying during the fall of the granule, may be neglected.

2. The total volume of the hot granule is equal to the volume of the thin shell of the granule and, consequently, is equal to the volume of the droplet at the crystallization temperature.

Let us introduce the following notation:

\(g\) — weight of the droplet and of the granule,

\(v_k\) — volume of the droplet at the crystallization temperature,

\(v_1\) — volume of the hot granule at the crystallization temperature,

\(v_2\) — volume of the cooled granule at the final temperature,

\(\gamma_k\) — specific weight of the liquid of the droplet at the crystallization temperature,

\(\gamma_1\) — specific weight of the crystals of ammonium nitrate in the hot granule,

\(\gamma_2\) — specific weight of the crystals of ammonium nitrate in the cooled granule.

Fig. 2. Diagram of the normal structure of a granule

\(v_{kr}, v_{rastv}, v_{vozd}\) — volumes occupied in the cooled granule, respectively, by nitrate crystals, mother liquor, and air,

\(x_{H_2O}\) — water content in the granule in weight percent,

\(y_{soli}\) — nitrate content in the mother liquor of the cooled granule in weight percent.

Taking into account the initial assumptions and the notation introduced, it may be written that the total volume of the hot granule is equal to the volume of the droplet, i.e.

\[ v_1 = v_k . \tag{1} \]

The shell of the hot granule, like all its solid contents, during cooling, owing to the change in the specific weight of the nitrate crystals, will change the volume enclosed in it, equal to the volume of the hot granule, until it reaches the final temperature. The resulting total volume of the cooled granule is related to the total volume of the hot granule by the relation

\[ v_2 = v_1 \frac{\gamma_1}{\gamma_2} \tag{2} \]

or, taking equation (1) into account,

\[ v_2 = v_1 \frac{\gamma_1}{\gamma_2} = v_k \frac{\gamma_1}{\gamma_2}. \tag{3} \]

The total volume of the cooled granule will consist of the volumes of the nitrate crystals, the mother liquor, and the air,

\[ \text{i.e.}\qquad v_2 = v_{kr} + v_{rastv} + v_{vozd}. \tag{4} \]

Starting from equation (4), the degrees of filling \((\alpha)\), \((\beta)\), and \((\gamma)\) of the space of the cooled granule with nitrate crystals, mother liquor, and air can be expressed, respectively, by equations (5), (6), and (7):

\[ \alpha = \frac{v_{kr}}{v_2}\cdot 100 \ \text{in } \%, \tag{5} \]

\[ \beta = \frac{v_{rastv}}{v_2}\cdot 100 \ \text{in } \%, \tag{6} \]

\[ \gamma = \frac{v_{vozd}}{v_2}\cdot 100 \ \text{in } \%. \tag{7} \]

Let us first determine the quantity \(\beta\). It is obvious that

\[ v_{\text{soln}}=\frac{g'_{\text{soln}}}{\gamma_{\text{soln}}} \tag{8} \]

\[ g_{\text{soln}}=g_{H_2O}+g_{\text{salt}} \tag{9} \]

\[ g_{H_2O}=\frac{g\cdot x_{H_2O}}{100} \]

and from the proportion

\[ \begin{array}{rcl} g_{\text{salt}} & \ldots & y\% \\ \dfrac{g\cdot x_{H_2O}}{100} & \ldots & (100-y)\% \end{array} \]

we find

\[ g_{\text{salt}}=\frac{g\cdot x_{H_2O}\cdot y}{100(100-y)}. \]

Substituting the values of \(g_{H_2O}\) and \(g_{\text{salt}}\) into equation (9), we obtain

\[ g_{\text{soln}}=\frac{g x_{H_2O}}{100-y}. \tag{10} \]

From equations (8) and (10) we find the volume of mother liquor in the granule:

\[ v_{\text{soln}}=\frac{g\cdot x_{H_2O}}{\gamma_{\text{soln}}(100-y)}. \tag{11} \]

But the weight of the droplet and of the granule obtained from it are equal to one another and, consequently,

\[ g=v_k\cdot \gamma_k, \tag{12} \]

whence

\[ v_{\text{soln}}=\frac{v_k\cdot \gamma_k\cdot x_{H_2O}}{\gamma_{\text{soln}}(100-y)}. \tag{13} \]

In equations 8–13 the symbols denote:

\(g_{\text{soln}}\) and \(\gamma_{\text{soln}}\) — respectively, the weight and specific gravity of the mother liquor in the cooled granule,

\(g_{H_2O}\) and \(g_{\text{salt}}\) — respectively, the weights of water and saltpeter in the mother liquor of the cooled granule.

Substituting into equation (6) the values of \(v_{\text{soln}}\) from equation (13) and \(v_2\) from equation (3), we obtain the final equation for the degree of filling of the volume of the cooled granule with mother liquor, in percent:

\[ \beta=\frac{\gamma_2\cdot \gamma_k\cdot x_{H_2O}\cdot 100}{\gamma_1\cdot \gamma_{\text{soln}}(100-y)}. \tag{14} \]

In the granulation of dry saltpeter, \(\beta=0\).

The degree of filling of the volume of the cooled granule with saltpeter crystals \((\alpha)\) can be determined from the following relations:

\[ v_{\text{cr}}=\frac{g_{\text{cr}}}{\gamma_2}, \]

where \(g_{\text{cr}}\) is the weight of saltpeter crystals in the volume of the cooled granule; but

\[ g_{\text{cr}}=g-g_{\text{soln}}, \]

or, substituting here the value of \(g_{\text{soln}}\) from equation (10), we find

\[ g_{\text{cr}}=\frac{g(100-y-x_{H_2O})}{100-y}, \]

whence, taking into account equations (5), (3), and (12), we find

\[ \alpha=\frac{v_{\text{cr}}}{v_2} =\frac{\gamma_k(100-y-x_{H_2O})\cdot 100}{\gamma_1(100-y)}. \tag{15} \]

Vertical axis: Degree of filling of the granule volume with crystals or mother liquor, vol.%; degree of filling of the granule volume with air, %

Horizontal axis: Temperature to which the granule was cooled, °

Fig. 3. Effect of the final cooling temperature on the degree of filling of the granule volume with crystals (curve 1), with mother liquor (curve 2), and with air (curve 3), and on the porosity of the granule (curve 4).

In the granulation of dry saltpeter, \(x_{H_2O}=0\), \(y\) is also equal to zero, since the solution is absent, and therefore

\[ \alpha=\frac{\gamma_k}{\gamma_1}\cdot 100. \tag{16} \]

The degree to which the volume of the cooled granule is filled with air, \(\gamma\), can be determined from equation (17)

\[ \alpha+\beta+\gamma=100\%, \tag{17} \]

whence

\[ \gamma=100-\alpha-\beta. \tag{18} \]

In the granulation of dry saltpeter \(\beta=0\), and

\[ \gamma=100-\alpha. \tag{19} \]

Equation 18 can also be expressed in terms of the initial quantities. Using for this purpose the values of \(\alpha\) and \(\beta\) from equations (14) and (15), we obtain:

\[ \gamma=100-\frac{\gamma_k\cdot(100-y-x_{H_2O})\cdot100}{\gamma_1(100-y)} -\frac{\gamma_2\cdot\gamma_k\cdot x_{H_2O}\cdot100}{\gamma_1\gamma_{\text{soln}}(100-y)}. \tag{20} \]

The sum of the degrees to which the granule volume is filled by mother liquor and by air gives the porosity \((A)\) of the wet granule, in percent,

\[ A=\beta+\gamma. \tag{21} \]

Applying equations (15), (16), (14), (18), (19), and (21), and using literature data on the specific gravity [2] and concentration [1] of saturated aqueous solutions of ammonium nitrate at various temperatures, the degrees to which the granule volume is filled by saltpeter crystals, mother liquor, and air, as well as its porosity, were calculated as functions of the final cooling temperature at a moisture content of 1% in the saltpeter. The specific gravities of ammonium nitrate crystals at various temperatures required for these calculations were determined from the diagram of L. I. Kuznetsov-Fetisov [1], with the use of certain other data [3]. It was also assumed that a solid film on the surface of the granule, at a water content in the saltpeter of 1%, begins to form at a temperature of 155°, and accordingly the specific gravities of the crystals \((\gamma_1)\) of cubic form I \((\varepsilon)\) and of the saturated solution \((\gamma_k)\) of saltpeter were taken as equal, respectively, to 1.570 and 1.435.

The results of the calculations are given in Table 1 and shown in Fig. 3. As is seen from Table 1 and Fig. 3, the degree to which the granule volume is filled by crystals is, at a moisture content of 1% and a temperature of 150°, 31.66%; with further lowering of the temperature, in accordance with the decrease in the solubility of saltpeter in water, \(\alpha\) increases and reaches only 88.8% at 20°. It should be noted that the phase changes occurring at temperatures of 125.6, 85.1, and 32.3° (as shown by S. I. Volfkovich and G. V. Glazova [4], in addition to the known five stable modifications of ammonium nitrate, there is also one metastable modification with a transition point in the temperature range 44–57°), which are associated with changes in the specific volumes of saltpeter crystals, do not affect the degree to which the granule volume is filled by crystals.

The greater part of the granule volume at 150° is filled with mother liquor (65.6%); as the temperature to which the granule is cooled is lowered, \(\beta\) decreases, reaching a value of 3.45% at 20°. In addition to temperature, the magnitude of \(\beta\) is affected by phase transitions of the saltpeter crystals. Thus, in the transition of cubic form I \((\varepsilon)\) to tetragonal II \((\delta)\) at 125.6°, \(\beta\) increases from 26.03 to 26.58, i.e., by 2.11%. Phase transitions of stable

Table 1

Degrees of filling of the granule volume by crystals of ammonium nitrate \(\alpha\), mother liquor \(\beta\), and air \(\gamma\). Porosity and apparent specific gravity of the moist and dried granule
(Granule moisture content equals 10% by weight)

Temperature to which the granule was cooled, °C	Crystalline forms of nitrate	Specific gravity: nitrate crystals	Specific gravity: saturated nitrate solution	Concentration of saturated aqueous nitrate solution, wt. %	Degree of filling of granule volume: crystals \(\alpha\), %	Degree of filling of granule volume: mother liquor \(\beta\), %	Degree of filling of granule volume: air \(\gamma\), %	Granule porosity: moist, %	Granule porosity: dried, %	Apparent specific gravity of granule: moist	Apparent specific gravity of granule: dried
150	Cubic form I \((\varepsilon)\)	1,575	1,434	98,47	31,66	65,61	2,73	68,34	9,52	1,440	1,425
140	Cubic form I \((\varepsilon)\)	1,584	1,432	97,59	53,47	41,95	4,58	46,53	9,52	1,448	1,433
130	Cubic form I \((\varepsilon)\)	1,594	1,429	96,61	64,44	30,07	5,49	35,56	9,52	1,457	1,442
125,6	Cubic form I \((\varepsilon)\)	1,598	1,428	96,07	68,14	26,03	5,83	31,86	9,52	1,461	1,446
125,6	Tetragonal form II \((\delta)\)	1,632	1,428	96,07	68,14	26,58	5,28	31,86	9,52	1,492	1,447
120	Tetragonal form II \((\delta)\)	1,635	1,426	94,97	73,23	20,83	5,94	26,77	9,52	1,494	1,479
100	Tetragonal form II \((\delta)\)	1,646	1,4145	90,86	81,40	11,64	6,94	18,68	9,52	1,505	1,489
90	Tetragonal form II \((\delta)\)	1,652	1,405	88,53	83,43	9,37	7,20	16,57	9,52	1,510	1,495
85,1	Tetragonal form II \((\delta)\)	1,655	1,401	87,34	84,18	8,53	7,29	15,82	9,52	1,513	1,498
85,1	Rhombic form III \((\gamma_1)\)	1,633	1,401	87,34	84,18	8,42	7,40	15,82	9,52	1,493	1,478
80	Rhombic form III \((\gamma_1)\)	1,635	1,3940	86,08	84,83	7,71	7,46	15,17	9,52	1,495	1,479
60	Rhombic form III \((\gamma_1)\)	1,642	1,3519	80,42	86,73	5,67	7,60	13,27	9,52	1,501	1,486
40	Rhombic form III \((\gamma_1)\)	1,649	1,3415	73,70	87,93	4,27	7,80	12,07	9,52	1,507	1,492
32,3	Rhombic form III \((\gamma_1)\)	1,652	1,333	70,96	88,25	3,90	7,85	11,75	9,52	1,510	1,495
32,3	Rhombic-bipyramidal form IV \((\beta)\)	1,716	1,333	70,96	88,25	4,05	7,70	11,75	9,52	1,568	1,553
20,0	Rhombic-bipyramidal form IV \((\beta)\)	1,725	1,3115	65,18	88,78	3,45	7,77	11,22	9,52	1,576	1,560

of crystalline modifications at 85.1° decreases, and at 32.3° increases the degree of filling of the granule volume with mother liquor by 1.29 and 3.85%, respectively.

However, as our calculations have shown, even at such a high temperature as 150°, the volume of the granule is filled not only with crystals and mother liquor, but also with air. At the indicated moisture content of the nitrate and 150°, the degree of filling of the granule volume with air, \(\gamma\), is equal to 2.73%. A further lowering of the granule temperature increases \(\gamma\); at 20° the latter reaches a value of 7.77%. The phase transitions of some crystalline forms of nitrate into others also have a noticeable influence on the value of \(\gamma\). Thus, the phase transition at 125.1° decreases \(\gamma\) from 5.83 to 5.28%, i.e., by 9.43%; at 85.1° it increases it, and at 32.3° it decreases \(\gamma\), respectively, by 1.51 and 1.91%. The jumps characterizing the indicated phase transitions are clearly visible in Fig. 3, curve 3.

From all the above it is clear that, during solidification of a droplet of molten ammonium nitrate containing water, 1) one cannot speak of any constant crystallization temperature at which all the material of the liquid phase will pass into the solid phase; on the contrary, as cooling proceeds, owing to the decrease in the solubility of nitrate in water, there will be a gradual filling of the granule volume with nitrate crystals precipitating from the mother liquor; 2) one cannot expect continuous filling of the granule volume by the crystalline phase; in reality, instead of a monolithic structure, a porous structure must be obtained, with the granule volume filled, in addition to crystals, also with mother liquor and air.

As is seen from Table 1, the porosity of a granule with 1% moisture at 150° is 68.34% of the total volume of the granule; of this, 65.61% falls on the mother liquor and 2.73% on air. As the temperature to which the granule is cooled is lowered, its porosity decreases, reaching 11.22% at 20°. The porosity of the wet granule does not depend on the phase transitions of the crystalline forms of nitrate.

Turning to a discussion of the porosity of the dried granule, it may be assumed that during drying the external dimensions of the wet granule will be preserved, since precipitation of crystals from the mother liquor will occur mainly in the pores of the granule.

On this basis, the porosity of a completely dried granule \((A')\) can be expressed by equation (22)

\[ A' = A - \frac{\beta \cdot \gamma_{\text{soln}} \cdot y}{100 \cdot \gamma_2}, \tag{22} \]

where

\[ \frac{\beta \cdot \gamma_{\text{soln}} \cdot y}{100 \cdot \gamma_2} \]

is the degree of filling of the granule volume by crystals of ammonium nitrate precipitated upon complete evaporation of moisture from the mother liquor.

Using equation (22) and taking into account equations (21), (18), (14), and (15), we obtain:

\[ A' = 100 - \frac{\gamma^{k}}{\gamma_1}\,(100 - x_{H_2O}). \tag{23} \]

It follows from equation (23) that the porosity of the completely dried granule is determined by the initial moisture content and by the ratio of the specific gravities of the droplet and of the crystals of the granule shell at the temperature of its formation, and does not depend on the final temperature to which the granule is cooled. As is seen from the data of Table 1, the porosity of the dried granule calculated from equations (22) or (23), at its initial moisture content of 1%, is 9.52%.

Apparent Specific Weight of a Granule

The apparent specific weight of a granule \((\gamma_{\text{app}})\) is the ratio of the weight of the granule \((g_{gr})\) to its volume \((v_{gr})\), including the volume of the pores, i.e.

\[ \gamma_{\text{app}}=\frac{g_{gr}}{v_{gr}}. \tag{24} \]

For a moist and cooled granule

\[ g_{gr}=\frac{v_{gr}\cdot \alpha}{100}\gamma_2+\frac{v_{gr}\cdot \beta}{100}\cdot \gamma_{\text{soln}}. \]

Substituting the expressions obtained for \(g_{gr}\) into equation (24), we obtain

\[ \gamma_{\text{app}}=\frac{\alpha\cdot \gamma_2+\beta\cdot \gamma_{\text{soln}}}{100}. \tag{25} \]

Replacing \(\alpha\) and \(\beta\) in equation (25) by their expressions from equations (14) and (15), we shall have

\[ \gamma_{\text{app}}=\frac{\gamma_k\cdot \gamma_2}{\gamma_1}. \tag{26} \]

In an analogous manner it can be shown that the apparent specific weight \((\gamma'_{\text{app}})\) of a granule obtained by granulating a moist melt but then dried will be expressed as

\[ \gamma'_{\text{app}}= \frac{\alpha\cdot \gamma_2+\beta\cdot \gamma_{\text{soln}}\cdot \dfrac{y}{100}}{100} \tag{27} \]

or, taking equations (14) and (15) into account,

\[ \gamma'_{\text{app}}= \frac{\gamma_k\cdot \gamma_2}{\gamma_1}\cdot \left(\frac{100-x_{H_2O}}{100}\right). \tag{28} \]

As the data of Table 1 show, the apparent specific weights of moist and dried granules increase as the final temperature is lowered within the range of existence of the individual crystalline modifications of ammonium nitrate.

Phase transitions from one crystalline form to another are also associated with noticeable changes in the apparent specific weights of moist and dried granules. As was to be expected, the apparent specific weight of the dried granule proved to be less than the analogous value for the moist granule. Thus, when a granule with 1% moisture is cooled to \(20^\circ\), its apparent specific weight should be equal to 1.577; the analogous value for the dried granule is 1.56.

Bulk Volumetric Weight of Granulated Ammonium Nitrate

As is known, by bulk volumetric weight is meant the weight of a loose material per unit volume. This quantity depends on the apparent specific weight of the individual granules and on the density of their packing.

If the granules of nitrate are likened to spheres of identical size, then the diameter of the spheres will have no effect on the free space between them, and it will be determined only by the nature of the packing [5].

The bulk density of moist and dried granulated ammonium nitrate may be expressed, respectively, by equations (29) and (30)

\[ \gamma_{\text{bulk}}=\gamma_{\text{app}}\cdot K =\frac{(\alpha\cdot\gamma_{2}+\beta\cdot\gamma_{\text{soln}})K}{100} =\frac{\gamma_{k}\cdot\gamma_{2}}{\gamma_{1}}K . \tag{29} \]

\[ \gamma'_{\text{bulk}}=\gamma'_{\text{app}}\cdot K = \frac{\left(\alpha\cdot\gamma_{2}+\beta\cdot\gamma_{\text{soln}}\cdot\frac{y}{100}\right)\cdot K}{100} = \frac{\gamma_{k}\cdot\gamma_{2}}{\gamma_{1}} \left(\frac{100-x_{\mathrm{H_2O}}}{100}\right)K , \tag{30} \]

where \(K\) is a coefficient characterizing the packing density.

With dense packing, the free space between the spherical granules will be 26.2%, and with loose packing, 47.6%; hence \(K\), characterizing the degree to which the space is filled by the spherical granules, is, respectively, equal to 73.8 and 52.4%, or, as fractions of unity, 0.738 and 0.524. Substituting the obtained values of \(K\) into equations (29) and (30), we obtain the value of the bulk specific gravity. For granules containing 1% moisture and cooled to 20°, the bulk density with dense packing proved to be 1.1636, and with loose packing 0.8262. If the granules are cooled to the same temperature and dried, then the bulk density will be somewhat lower, namely—for the two indicated packings, respectively—1.1519 and 0.8179.

Thus, for granules of one diameter the bulk density will lie within the indicated limits, and its actual value will be determined by the nature of the packing, while the latter depends on the quality and duration of shaking. The independence of the packing density, and consequently also of the bulk density, from the grain sizes over wide limits has also been observed for free-flowing materials whose particles are not spherical; therefore, certain deviations from the regularity of the spherical shape of ammonium nitrate granules should not substantially affect the cited limits of the values of \(\gamma_{\text{bulk}}\). The latter quantity may increase somewhat only as a result of the filling of intergranular spaces by finer granules, which are always present in industrial granulated ammonium nitrate.

Thus, proceeding from the most general data on the specific gravities of the melt and of the individual crystalline modifications of ammonium nitrate, and making highly probable assumptions concerning the nature of the volume changes during solidification of a drop into a granule, we have succeeded in deriving equations and, using them, in calculating numerical values of such important quantities as the degree of filling of the granule volume by crystals, mother liquor, and air; the porosity and apparent specific gravity of a moist and dried granule; and the bulk density of granulated ammonium nitrate.

The correctness of our calculations is confirmed by certain experimental data. Thus, according to the investigations of Ross, Yee, and Hendricks [6], the porosity of granules at room temperature is about 10%. According to our calculations, at 20° the porosity of moist (1% moisture) and dried granules, as indicated above, is respectively 11 and 9.5%, i.e., the agreement is good. According to the experimental data of the same investigators, the apparent specific gravity of individual granules at room temperature is 1.56, while according to our calculations at 20° for moist and dried granules it is, respectively, 1.57 and 1.56. The bulk density of granulated ammonium nitrate is 0.859–0.883, which is close to the bulk density of ammonium nitrate with loose packing (0.826).

The water vapor and gaseous decomposition products of ammonium nitrate released during partial drying, penetrating the melt of a droplet that has not yet solidified, may further increase the porosity of the granules, up to the formation of large internal voids (Fig. 4) and wide channels communicating with the surface (Fig. 5). However, a significant development of this phenomenon will indicate an irregularity in the technological process of granulation.

Fig. 4

Mechanical strength of the granule

The greatest mechanical strength would be possessed by a monolithic granule, and its strength would be determined by the temporary resistance characteristic of homogeneous polycrystalline anhydrous ammonium nitrate. In this case the nitrate granule would behave like a steel ball, i.e., when a load is applied, up to a certain limit it would deform elastically, and then would rupture along its diameter. A certain mechanical rigidity of real ammonium nitrate granules arises as a result of the action of molecular forces manifested when individual crystals and their intergrowths come into contact with one another through their edges, faces, and corners, and also owing to the effect of wedging and arching. The liquid phase present in the granule in the form of mother liquor serves as a kind of lubricant, by virtue of which the sliding of crystals and their intergrowths over one another is facilitated when an external load is applied to the granule. Under the influence of the latter, the granule, undergoing plastic deformation, should be transformed from a sphere into a flattened spheroid with a more complete en—

Fig. 5

by the filling of internal voids with crystals, and only after this, with a further increase in load, should destruction of the granule occur. As our observations have shown, this is indeed the case.

Since the mechanical strength of any openwork system is directly proportional to the number and strength of the individual bonds, the strength of an ammonium nitrate granule must be the greater, the more points of contact arise between the separate crystalline formations inside the granule during its solidification. The number of individual bonds in a granule, and consequently also its mechanical strength, should increase as the size of the crystals decreases. The sizes of the crystals and of their druses within the granule will depend on the crystallization conditions. Specifically, the higher the concentration of the melt fed to granulation and the more intensively the granule falling through the air is cooled, the greater the supersaturation of the solution that will arise, and the finer the crystals and their druses in the granule will be.

In accordance with the foregoing, this should lead to the formation of a large number of individual bonds and to an increase in the mechanical strength of the granules. It should be expected that, with an increase in moisture content and a rise in temperature, the plastic properties of the ammonium nitrate granule will increase, since under these conditions the amount of lubricating mother liquor will increase and the cohesive strength of the individual crystals and their druses will decrease owing to partial dissolution of the bridges in the granule. It should be expected that drying of the granule should reduce plasticity and increase the mechanical strength of the granules. As experiments have shown, the resistance to crushing of completely dry granules 2–2.38 mm in diameter, of American manufacture, is equal to 0.63–1.12 kg, which amounts to 14 to 35 kg/cm².

Mechanism of Caking of Granulated Ammonium Nitrate

As indicated above, under production conditions granulated ammonium nitrate is loaded into bags at a temperature of 60–70° in summer and 55–60° in winter and is stacked either in piles in the warehouse or in railway cars; in the process it cools, cakes, and loses its flowability.

Let us consider the mechanism of this process, taking into account the porous structure of the granules described above, the presence in them of mother liquor, and their ability to undergo plastic deformation under the influence of an external load.

If several granules are brought into contact with one another, as occurs in a layer, then a certain number of individual bonds arises between them, equal to the number of points of contact of the granules with one another. The higher the strength of the individual bonds, the greater the mechanical strength of the layer and, consequently, the more pronounced its tendency to cake.

The strength of an individual bond \((F)\) will be equal to

\[ F = S \cdot f, \tag{31} \]

where \(S\) is the contact surface between two granules,

\(f\) is the molecular forces acting per unit contact surface.

However, as experience shows, if dry granules are poured in a thin layer, then the strength of the individual bonds arising in this case is so small that no appreciable strength of the layer appears, and it crumbles when the confining walls are removed.

This, according to equation (31), can be explained by the slight development of the contact surface \((S)\) (Fig. 6a) between individual granules, since

the quantity $f$ depends only on the physicochemical properties of ammonium nitrate and, under given conditions, is constant.

The creation of significant mechanical strength in a layer of granulated ammonium nitrate, at a constant number of unit bonds, requires a considerable increase in the strength of the latter. In accordance with the preceding discussion, this may occur when the contact surface $(S)$ between individual granules is increased, which can be achieved: 1) by plastic deformation of granules under the influence of an external load (Fig. 6b), 2) by

Fig. 6. Nature of the contact between two touching granules in: a) absence of mechanical deformation of the granules, b) mechanical deformation of the granules, c) presence of mother liquor at the point of contact, d) mechanical deformation and presence of mother liquor at the point of contact

the creation of crystalline bridges between granules through the formation of crystals from the mother liquor, if the latter has in some way entered the intergranular space (Fig. 6c), and 3) by the simultaneous action of plastic deformation of the granules and the formation of crystalline bridges between granules (Fig. 6d).

Deformation of Granules in a Layer under the Influence of Mechanical Load

If a layer of granules in a cylinder is subjected to compression by means of an external load transmitted through a piston (Fig. 7), at constant temperature and moisture content, then each granule will deform and, ultimately, in the case of loose packing, will assume the shape of a cube or a parallelepiped slightly elongated in the direction perpendicular to the compressive force (Fig. 7a), or, in the case of dense initial packing of the granules, a shape close to a twelve-faced polyhedron (rhombic dodecahedron) (Fig. 7b). The greatest deforming force on an individual granule, for the same total external load, will be observed in the case of loose packing. The higher the strength of the granules, the greater the force that must be applied to produce plastic deformation. Deformation of granules under compression will lead to compaction of the layer; in particular, this will cause a decrease in the free volume between the granules and within the granules. Under considerable loads one may expect displacement of mother liquors from the internal volume of the granules into the intergranular space and even complete destruction of the granules with the formation of a more or less monolithic mass.

The more strongly the original spherical shape of the granules is distorted under the influence of a load, the greater the contact surface that develops between individual granules, the stronger the individual bonds between them will be, and the more strongly the layer as a whole will be cemented.

Fig. 7. Deformation of granules in a layer under the influence of an external load;
a and b—the shapes of deformed granules under loose and dense packing, respectively.

Thus, one of the conditions for reducing the caking of granular ammonium nitrate is a reduction of the external load during storage.

Influence of the initial moisture content on the caking of a layer of granular ammonium nitrate at constant temperature

As was shown above, the higher the moisture content of the nitrate, the greater the amount of mother liquor contained inside the granules, the lower the packing density of crystals in the internal space of the granules, the lower their rigidity and the greater their plasticity. Consequently, at constant temperature and pressure, with increasing moisture content the deformability of the granules and the contact surface between them will increase, which will inevitably lead to strengthening of the individual bonds and to an increase in the degree of caking of the layer as a whole.

Influence of the nature of packing, granule diameter, and particle-size composition on the caking of a layer of granular ammonium nitrate

Along with the strength of the individual bonds between separate granules, the number of these bonds will also affect the mechanical strength of the nitrate layer. The greater this quantity, the greater the mechanical strength of the layer and the higher its caking; and, under appropriate conditions, the cakedness of the nitrate as well.

The number of individual bonds will depend on the nature of the packing, the diameter of the granules, and their particle-size uniformity.

If one takes a sufficiently large cube (Fig. 8) with edge \(l\) and places in it spherical granules of identical diameter \(d\) according to the principle of loose packing, in which the free intergranular space

is equal to 47.6%, and each sphere, except for those located at the faces, will have 6 points of contact with other sphere-granules; then [5] the number of granules in the cube \((n)\) will be equal to

\[ n=\left(\frac{l}{d}\right)^3 . \tag{32} \]

Fig. 8

Taking into account that one bond is formed upon contact of two granules, the number of unit bonds \((K_1)\) in the given volume will be

\[ K_1=3n=3\left(\frac{l}{d}\right)^3 . \tag{33} \]

However, the actual number of unit bonds resisting tensile or compressive deformations along a plane parallel to one of the faces of the cube will be smaller and will be

\[ K_2=\left(\frac{l}{d}\right)^2 . \tag{34} \]

With close packing, when each sphere has 12 points of contact with other spheres along the planes of a rhombic dodecahedron and the free space between granules is 26.2%, the number of granules in a cube with edge \(l\) will be 1.41 times greater, i.e.,

\[ n'=1.41\left(\frac{l}{d}\right)^3 . \tag{35} \]

The total number of unit bonds in the same volume between individual granules will be

\[ K_1' = 6 \cdot n' = 8.46 \left(\frac{l}{d}\right)^3 . \tag{36} \]

The minimum number of unit bonds resisting compressive or tensile deformations along a plane parallel to one of the faces of the indicated cube will be equal to

\[ K_2' = 6 \cdot 1.41 \left(\frac{l}{d}\right)^2 = 8.46 \left(\frac{l}{d}\right)^2 , \tag{37} \]

where 6 is the number of unit bonds resisting compression or rupture for one granule.

Thus, from a comparison of equations (33), (34), (36), and (37), it is seen that, with dense packing, characteristic of the rhombic dodecahedron, the number of unit bonds between granules in the volume of a cube with edge \(l\), and the number of these bonds resisting rupture or compression deformations, are respectively \(\frac{16.9}{6} = 2.82\) and \(\frac{8.46}{1} = 8.46\) times greater than is the case with loose packing. It should be expected that the caking capacity of a layer of saltpeter with dense packing will be approximately 2.8 times higher than the caking capacity with loose packing, if the strengths of the unit bonds in the one and the other cases are identical. Practically, dense packing of the distribution of granules in bags may arise as a result of shaking; therefore, to reduce caking, shaking of the bags during filling and transport should be limited.

Fig. 9

From the equations given, it is also seen that the number of unit bonds between granules per unit volume is inversely proportional to the cube of the granule diameter [7], while the number of bonds resisting rupture and compression deformations along one of the planes of the cube is inversely proportional to the square of the granule diameter.

Hence, in order to reduce the number of unit bonds between granules in

per unit bulk volume, and, in order to reduce caking, it is desirable to manufacture the saltpeter in the form of larger granules.

The situation with caking becomes still more complicated when the granulometric composition of the granules is nonuniform. In this case (Fig. 9), owing to the filling of the free spaces between large granules by smaller ones, the packing density will be higher; the same will occur with the number of individual bonds and with the strength of the bed of granulated saltpeter. Thus, to reduce caking, it is desirable to obtain saltpeter in the form of granules of a single size.

Influence of Temperature on the Caking of a Bed of Granulated Ammonium Nitrate

The strength of the individual bonds between granules of completely dry ammonium nitrate will be determined by the degree of development of the surface of physical contact between separate touching granules (strengthening of the bonds through the formation of bridges arising upon precipitation of crystals from the mother liquor is impossible here because the latter is absent). In view of the comparatively high mechanical rigidity of hot and cold dry granules, at least in the temperature range (70–20°) and at low external loads, one cannot expect any significant development of the contact surface and of the bond strength between individual granules as the temperature is lowered. For the same reason, phase changes associated with the transition of some crystalline forms of the saltpeter into others cannot substantially increase the strength of the individual bonds and cause caking of the saltpeter bed when the temperature is lowered, as was observed by V. A. Klevke and N. V. Tsel’m [8], despite the facilitated conditions for transfer of material from one granule to another and for the formation of crystalline sinters during the destruction of some crystalline forms and the emergence of others.

The picture changes sharply if we are dealing with moist granulated ammonium nitrate.

As is known, ammonium nitrate has exceptionally high solubility in water and a sharply expressed temperature dependence of solubility. Thus, when saltpeter is loaded into sacks at a temperature of 60° and a moisture content of 1%, one ton of it will contain 10 kg of water, in which 41.1 kg of saltpeter will be dissolved. If the saltpeter is cooled to 20°, the content of $\mathrm{NH_4NO_3}$ in the solution will decrease to 18.7 kg, and 22.4 kg will pass into the solid phase, giving a significant quantity of new crystals. As was indicated above, at a moisture content in the saltpeter of 0.8–1% the mother liquor is located in the internal pores of the granules, and their external surface remains dry. Therefore, in the absence of an external mechanical load and with constant pressure of the surrounding gas, cooling will lead to the separation of crystals from the mother liquor in the internal space of the granules, which will cause an increase in their mechanical strength, a decrease in plasticity, and a reduction of caking in the initial period of storage, even when an external load is subsequently applied.

However, if a bed of moist granulated saltpeter is cooled in the presence of an external load, then, as a result of their higher plasticity, the hot moist granules will be substantially deformed, with an increase in the contact surface and a reduction of the internal free volume and, at sufficient pressure, with the displacement of part of the mother liquor into the intergranular space (Fig. 10). In this case, as a result of the decrease in the solubility of the saltpeter upon cooling, new crystals will precipitate on the external surface of the granules at the places where the mother liquor is concentrated, and the initially smooth surface of the granules will become covered with irregularities from newly formed crystals. The greatest concentration

...of mother liquor, owing to surface migration of the liquid, will occur at the points where the granules touch one another. Here, as a result of the crystallization process caused by cooling, more or less strong bridges will arise, which will produce a certain mechanical strength of the layer and the loss of its flowability.

The higher the moisture content of the nitrate and the external pressure under which cooling of the granule layer takes place, the greater their plasticity and deformation, the larger the contact surface and intergranular crystal formation, and the stronger the bridges arising between the granules and the greater the caking.

Fig. 10. Deformation of moist granules \(R\) under the influence of an external load and wetting of the surface of contacts by mother liquor \(C\).

Taking into account the porous structure of the granules and the considerable filling of the pores with air, a change in barometric pressure must have a significant effect on the caking of the layer during its cooling. When barometric pressure falls, the gas enclosed inside the granule, expanding, can expel through the pores part of the mother liquor into the intergranular space, which will lead to wetting of the contacts between granules and to the subsequent formation of bridges and caking. Thus, a fall in barometric pressure will intensify the adverse consequences of mechanical deformation of the granules. An increase in barometric pressure will act in the opposite direction.

The appearance of mother liquor on the outer surface of granules during their cooling is also possible as a result of phase transitions of some crystalline forms of nitrate into others.

Phase transitions associated with the radical breaking of some crystal lattices and the formation of others destroy the walls of the tiniest reservoirs located near the surface of the granules, formed as a result of the capture of liquid during crystallization, and thereby facilitate the migration of mother liquor into the zone of contacts between granules.

In addition, the phase transitions at temperatures of 125.6 and 32.3° are accompanied by a decrease in the volume of the granules (~3.5%), which, while the porosity remains constant (see Table 1), can also cause the expulsion, together with air, of part of the mother liquor onto the outer surface of the granules.

Thus, phase transitions at the indicated temperatures can cause penetration of mother liquor into the contact zone with subsequent—

…resulting in the formation of more or less strong crystalline bridges between granules as they cool.

Influence of Hygroscopicity on the Caking of a Layer of Granulated Ammonium Nitrate

As is known, saturated aqueous solutions of ammonium nitrate have substantially lower water-vapor pressures than pure water at the same temperatures. Therefore, during the transport and storage of granulated ammonium nitrate in non-hermetic containers, it can both absorb moisture from and give moisture to the air. The driving force for the sorption or desorption of moisture will be equal to the difference between the pressure of water vapor in the air and the vapor pressure of water over the granules at a given moisture content and temperature. Numerically, this quantity is equal to \((P_{\mathrm{H_2O}} - P_{\mathrm{H_2O\,sorp}})\). Theoretically, an arbitrarily small excess of \(P_{\mathrm{H_2O}}\) over \(P_{\mathrm{H_2O\,sorp}}\) is sufficient for all the solid ammonium nitrate to pass into solution, given the corresponding amount of air.

Expressing the quantities \(P_{\mathrm{H_2O}}\) and \(P_{\mathrm{H_2O\,sorp}}\) through the relative humidity of the air and the humidity characteristic of the hygroscopic point, and also considering the absorption of moisture from air by ammonium nitrate as a special case of absorption of a gas by a liquid from an inert-gas medium, N. E. Pestov [9] showed that the rate of absorption of water vapor from the air is determined by the equation

\[ Q = K(h_u - h), \tag{38} \]

where \(h_u\) and \(h\) are the relative humidity of the air, in percent, and the humidity of the air characteristic of the hygroscopic point,

\(K\) is the coefficient of the rate of moisture absorption,

\(Q\) is the amount of moisture, in grams, absorbed by \(1\ \mathrm{cm}^2\) of the surface of the nitrate in 1 hour.

The partial pressure of water vapor in the air and its relative humidity vary over wide limits and depend on geographic location, season, time of day, and other meteorological conditions [9]. Therefore, when ammonium nitrate is stored, even at constant temperature, in non-hermetic containers, processes of moistening and drying of the nitrate are always possible.

When the driving force of sorption \((P_{\mathrm{H_2O}} - P_{\mathrm{H_2O\,sorp}})\) is positive, moisture penetrates into the intergranular space and then into the pores of the granules and dissolves part of the solid nitrate. In this process, as a result of dissolution, the intercrystalline bonds are weakened and the amount of mother liquor increases; the granules become more plastic and therefore are more strongly deformed under the influence of an external load. As was shown above, under the influence of mechanical deformation of the granules, a decrease in barometric pressure, an increase in the volume of air in the pores with a certain rise in temperature, and the phase transitions of certain crystalline forms into others, the mother liquor may enter the intergranular spaces and wet the contact surfaces between individual granules. During the subsequent drying process, or when the temperature decreases, or under the simultaneous action of both of these factors, supersaturated solutions are formed, followed by separation of the crystalline phase, formation of strong bridges between granules, and loss of flowability of the layer. Repeated repetition of the process of dissolution and crystallization, caused in particular by hygroscopicity in the presence of the deforming influence of external loads and other effects, can lead to complete destruction of the granules and the formation of a completely monolithic layer of ammonium nitrate with high mechanical strength and complete loss of flowability.

The unit bonds between individual granules that arise as a result of the action of the above-listed causes may produce a certain caking of the layer. Some practical methods for measuring caking have been developed by N. E. Pestov [7], [9]. The measure of caking of granular nitrate, as also of other loose materials, should be taken to be the energy of destruction \((U)\) of all unit bonds formed between individual granules in a unit bulk volume; it may be expressed by the following equation:

\[ U=\frac{1}{2}N \cdot n \cdot a, \tag{39} \]

where \(a\) is the energy required to destroy one unit bond,
\(N\) is the total number of granules in a unit bulk volume,
\(n\) is the number of unit bonds per granule for the given type of packing,
\(\frac{1}{2}\) is a coefficient allowing for the fact that one unit bond is formed when two granules come into contact.

If the energy required to destroy all unit bonds is supplied in the form of the energy of free fall, and if the expenditure of this energy on overcoming air resistance, producing plastic deformations, and other secondary processes is neglected, then

\[ U=\frac{1}{2}N \cdot n \cdot a=\gamma_{\text{bulk}}\cdot h, \tag{40} \]

where \(\gamma_{\text{bulk}}\) is the bulk volumetric weight of the granular nitrate,
\(h\) is the fall height of the caked nitrate.

Solving equation (40) for \(h\), taking into account that the weight of one granule is

\[ g_{\text{gr}}=\frac{\gamma_{\text{bulk}}}{N}=\frac{1}{6}\pi d^{3}\gamma_{\text{part}}, \]

we obtain

\[ h=\frac{3\cdot n \cdot a}{\pi d^{3}\cdot \gamma_{\text{part}}}. \tag{41} \]

It follows from equation (41) that the fall height of the caked nitrate \((h)\) required to destroy all the bonds formed between the granules of a unit bulk volume is inversely proportional to the cube of the granule diameter \((d)\). Having experimentally measured the value of \((h)\), it is easy, from equation (41), to calculate the energy required to destroy one bond.

As an indicator of caking, one may use the fall height \((h)\) of a nitrate sample at which all unit bonds are destroyed. If the fall height of a caked nitrate sample is limited to some practically convenient value \(h_{1}\), then, by measuring by means of sieving the degree of crushing of the nitrate to granules of a specified diameter, one may calculate the height required to destroy all bonds between granules from the following equation:

\[ h=\frac{100\cdot h_{1}}{c}, \tag{42} \]

where \(c\) is the degree of crushing of the nitrate sample, in percent, to granules of the specified diameter when falling from the height \(h_{1}\).

Methods for Combating the Caking of Granular Ammonium Nitrate

The regularities of the caking process of granular ammonium nitrate set forth above make it possible to consider the degree of perfection of individual methods for combating this phenomenon and to outline a program for further research in this direction. In doing so, it is also necessary to take into account questions of the explosion safety of ammonium nitrate, which have acquired great importance in connection with a number of explosions of ammonium nitrate abroad [10].

The explosion hazard and fire hazard of ammonium nitrate increase when organic compounds, and in particular nitro compounds, are present in it. In this case, under appropriate conditions, spontaneous explosions are possible during the production, transportation, and storage of nitrate.

Method of Individual Isolation of Granules

The method consists in the individual isolation of each granule by means of a film of hydrophobic organic substances. This method was proposed in 1933 by I. R. Krichevsky, L. Kantorovich, and M. Bergauz [11]. Paraffin was used as the hydrophobic substance in an amount of 0.2% of the weight of the nitrate, which reduced the rate of absorption of moisture from the air by 3–5 times compared with pure nitrate.

However, it was subsequently established that a film of pure paraffin, owing to its low elasticity, does not possess sufficient strength; therefore, other organic hydrophobic substances, both in pure form and in the form of mixtures with paraffin, were later studied as coatings (vaseline, rosin, bitumens, resins, semi-gudron, etc.) [12, 13, 1].

Good results were obtained by A. M. Dubovitsky and F. G. Margolis [14], who used paraffinic fuel oil with an admixture of 25% pure paraffin as the hydrophobic substance. To eliminate the sticking together of granules coated with paraffinic fuel oil, dusting with finely ground water-insoluble inorganic additives (kaolin, ash, apatite flour, limestone, etc.) was used in an amount of 2–3%. The resulting elastic, strong film of paraffinic fuel oil reduced the hygroscopicity of granular nitrate by 2–2.5 times and made the nitrate practically non-caking.

As the investigations of Zabolotsky [15] showed, paraffinated anthracene oil can be successfully used as a hydrophobic coating.

The coating methods used in practice do not provide complete isolation of the granule by a hydrophobic film; evidence of this is the comparatively rapid dissolution of such granules when they are immersed in water, as well as their certain ability to sorb moisture from the air. In this connection, the mechanism of the protective action of hydrophobic coatings consists not only in the fact that they separate direct contact between the nitrate of the granules and reduce the rate of sorption of vapor from the air, thereby preventing the appearance in the granules of large amounts of mother liquor; what is very important, these hydrophobic coatings also prevent the migration of the mother liquor from the pore openings \(a\) (Fig. 11) (if, under the influence of the processes described above, it has entered there from the internal space of the granules or has formed as a result of absorption of moisture from the air) over the non-wettable surface \((S)\) to the contact points \((K)\). As a result, strong single bonds cannot arise between individual granules, nor can significant caking occur as a result of crystallization during subsequent lowering of temperature and drying.

The method described, which has found wide application at plants in the USSR and abroad, makes it possible to obtain ammonium nitrate that practically does not cake when it contains 0.4–1% of an organic additive. However, the introduction of organic impurities, owing to the possibility of a number of reactions occurring between them, ammonium nitrate, and the products of its decomposition (for example, HNO₃), with the formation of new explosive substances and substances that initiate explosion, may under appropriate conditions increase the explosion hazard and fire hazard of ammonium nitrate; therefore this method of combating caking cannot be regarded as a solution to the problem.

Fig. 11. Diagram of a section of a granule covered with a hydrophobic film: R — granule, S — shell of hydrophobic material, a — pores, K — place of contact of two granules.

The presence of about 1% water in nitrate should reduce the danger of explosion, since thin films of water will damp the incipient explosive wave by shielding individual crystals of ammonium nitrate and by absorbing thermal energy for evaporation. However, this protection cannot be considered very reliable, and therefore it is necessary to develop other methods of combating caking without the use of organic additives.

The method of individual isolation would give especially good results if it were possible to develop reliable hydrophobic coatings for granules from inorganic material.

Method of separating granules by means of loose materials

The method consists in mixing the granules with a certain amount (2–5%) of finely divided materials insoluble in water. It was assumed that the powdery material, being uniformly distributed in the intergranular spaces, would eliminate contact between individual granules and thereby prevent caking of granulated ammonium nitrate. Limestone, kaolin, kieselguhr, ash, phosphorite flour, and many other substances having lower hygroscopicity than ammonium nitrate were tested as such dusting materials.

However, as shown by the experiments of S. I. Volfkovich, R. Remen, and others [16, 1, 2], in general the results were negative, and in the presence of such dusting materials the hygroscopicity and caking tendency did not decrease, but in many cases increased. From the standpoint of the porous structure of the granules considered by us, with their usual content of about 1% water, even with complete exclusion of moisture ingress from the atmosphere, this was to be expected, since under the influence of various fac-

the mother liquor could pass from the internal spaces of the granules to the periphery and wet not only the outer surface of the granules but also the dusting layer, especially at the points of contact between granules (Fig. 12). During storage of saltpeter in nonhermetic containers, wetting of the dusting layer may occur owing to newly formed mother liquor during sorption of moisture from the air. Upon subsequent lowering of the temperature, or drying, or both together, crystals will precipitate from the liquid phase and will firmly cement the granules and the loose interlayer into a single conglomerate, with loss of the layer’s flowability.

Fig. 12. Schematic section through granules covered with a layer of dusting additive. \(R\)—granules, \(S\)—dusting layer.

Thus, this method cannot achieve its purpose if one is dealing with moist ammonium nitrate and limits oneself to the use of small amounts of less hygroscopic, but readily wettable, dusting inorganic additives.

This method may have some prospects when finely ground nonwettable materials are used.

Method of Group Isolation

This method consists in eliminating the effect of atmospheric moisture by packing the granulated saltpeter in moisture-impermeable containers. It might be expected that it would give positive results in preventing the caking only of dry ammonium nitrate, since, as was shown above, the absence of mother liquor in the granules and their high hardness will prevent the formation of strong individual bonds owing to the small area of physical contact and the impossibility of forming crystalline bridges because of the absence of mother liquor.

In the USSR and abroad, multilayer paper bags with impregnation of individual layers with bitumen have come to be used as such containers.

However, even if one assumes complete hermeticity of such bags (which is doubtful), then when hot (~60°) and moist saltpeter (~1%) is loaded into them, the caking process will nevertheless occur, since this container does not eliminate the influence of compression, temperature, barometric pressure, and phase transitions. Under the influence of the indicated factors, mother liquor from the internal volume of the granules can enter the intergranular space and, migrating over the surface of the granules, will accumulate in the contact zones between them. Mother liquor on the surface of the granules may also appear as a result of condensation of water vapor from the gas phase surrounding the granules when the temperature is lowered. Upon subsequent cooling, the crystals precipitating from the solution will create strong bridges between granules, with loss of flowability of the layer.

Practice has confirmed the correctness of the foregoing. Saltpeter containing 0.8–1% moisture, loaded into multilayer paper bags, cakes strongly within several days.

Drying Method

Drying of ammonium nitrate obtained by granulation of a moist melt will reduce the amount of liquid in the pores of the granules and will cause the precipitation of addi-

of a considerable quantity of crystals in the internal space of the granules and thereby increase their mechanical hardness and reduce their plasticity. Completely dry nitrate, loaded into airtight containers, should retain its free-flowing properties, since, with a reasonable limitation of the external load, the formation of strong individual bonds between rigid separate granules is precluded owing to the small surface of physical contacts and the absence of mother liquor.

For the same reasons, hot granulated nitrate loaded into non-airtight containers will also, during the initial storage period, practically not cake.

However, drying does not eliminate the pores themselves, which constitute the internal free volume of the granules, nor does it eliminate the hygroscopicity of the nitrate; therefore, during prolonged storage in non-airtight containers, when under the influence of atmospheric moisture mother liquor appears in the granules, conditions will arise for the formation of strong individual bonds between granules, which must inevitably lead to loss of the free-flowing character of the layer—its caking.

Combination of methods

From all that has been set forth above it is clear that all the methods listed above have substantial shortcomings, and each of them separately does not provide a completely reliable means for combating caking. However, much better results can be obtained if several methods are used simultaneously.

Thus, with the simultaneous use of the methods of individual isolation, group protection, and separation of granules by means of free-flowing materials insoluble in water, even when pores are present in the hydrophobic films on the granules and the container is not fully airtight (paper bituminized sacks), the formation of strong individual bonds between granules can be made so difficult that the nitrate practically does not cake, even if before loading it had 0.8–1% moisture and was loaded into the container at a temperature of about 60°.

This has now been confirmed quite reliably in the Soviet Union as a result of scientific-research work and plant practice [1].

Unfortunately, at present a major shortcoming of this combined method is the introduction into the nitrate, for the formation of hydrophobic films on the granules, of about 0.5% of an organic substance—paraffinic fuel oil—which, as was indicated above, must inevitably increase the combustibility and the possibility of explosions of ammonium nitrate. In our opinion, the increase in combustibility and explosion hazard, in the presence of organic substances in the nitrate, will be especially promoted by drying the nitrate, since in this case, owing to the absence of water, the processes of nitration of the organic substances by the more concentrated nitric acid formed as a result of decomposition of the nitrate will be facilitated.

In light of the foregoing, there arises a need to develop reliable hydrophobic coatings for granules and airtight containers made of inorganic material.

Positive results with respect to preventing caking may be obtained by the simultaneous application of drying, group isolation, and cooling methods. As was indicated above, completely dry granulated nitrate, when loaded into airtight containers, should not cake because of the impossibility of forming strong individual bonds. In practice it is difficult to obtain nitrate without moisture; however, reducing its moisture content to 0.1–0.15% is entirely possible, and therefore, if such predried nitrate is cooled before loading into containers, then during storage it should not cake. However, in view of the high rate of moisture sorption from the air—

therefore, especially stringent requirements must be imposed on the airtightness of the container for such saltpeter, since otherwise the moisture absorbed from the air, under the influence of other factors, will inevitably cause caking. The multilayer paper bags with bituminized layers now used in practice can hardly meet the indicated requirements, and therefore the method can be applied if it proves possible to develop a cheap but fully airtight container.

Method of Water-Removing Additives

To prevent the formation of appreciable quantities of liquid in the granules—which, as was shown above, owing to the low plasticity of the granules and the impossibility of a mother liquor appearing on their surface, will make it impossible for strong individual bonds to arise between them—it is advisable to introduce water-removing inorganic additives into the saltpeter melt before its granulation.

As such additives one may use: 1) solid substances insoluble in the mother liquor and 2) soluble inorganic salts capable of forming crystal hydrates.

The first group of substances may include, for example, calcined gypsum, capable of combining with 2 molecules of water to form the solid crystal hydrate $\mathrm{CaSO_4\cdot 2H_2O}$. One of the conditions for reliable action of such solid additives is a high degree of their dispersion (close to the analogous value for cement), ensuring uniform distribution of the additive in the granule, a high completeness of utilization of the additive, and a high rate of absorption by it of moisture from the mother liquor.

The finely ground additive, heated to the melt temperature, should be introduced into the melt before granulation. Technically, a uniform feed of solid additives can readily be effected, for example, by means of a small screw feeder connected to a section of the pipe supplying the melt to the spraying basket. Uniform distribution of the additive in the melt is achieved by intensive mixing of the melt in the basket itself during its rotation, or by the simultaneous action of the basket and a simple propeller-type mixer.

The second group of water-removing additives should also include soluble inorganic salts capable of forming crystal hydrates, such as, for example, $\mathrm{Mg(NO_3)_2}$, $\mathrm{Ca(NO_3)_2}$, $\mathrm{Fe(NO_3)_3}$, $\mathrm{Cu(NO_3)_2}$, $\mathrm{Zn(NO_3)_2}$, $\mathrm{Al(NO_3)_3}$, and inorganic salts with other anions and cations (including mixtures of salts) capable of binding considerable quantities of water of crystallization.

The indicated substances may be introduced either in finely divided form directly into the melt after evaporation, or in the form of a solution at various points of the production process flowsheet, beginning with the section in which nitric acid is neutralized with ammonia.

A soluble salt or mixture of salts introduced into the ammonium nitrate melt will, upon cooling of the granules obtained, pass from solution into the solid phase with the formation of solid crystal hydrates, through removal from the mother liquor of the moisture present in it, with favorable consequences in the sense of reducing caking.

Both water-insoluble and water-soluble additives must possess:

1) the ability to form crystal hydrates by removing water from the mother liquor of ammonium nitrate under the temperature conditions of production;

2) a comparatively high rate of formation of crystal hydrates, which will make it possible to eliminate the presence of liquid in hot granules and their caking;

3) relatively high moisture absorption per unit weight of the additive;

4) low hygroscopicity, in order to avoid rapid absorption of moisture from the atmosphere during storage of the nitrate;

5) low corrosive action on the equipment;

6) low cost and availability.

They must not impair the agronomic properties or increase the explosion hazard and fire hazard of ammonium nitrate.

The stoichiometric content of additive required to absorb all the water in the nitrate may be calculated from the following equation:

\[ a=\frac{100}{1+100\,\frac{n \cdot M_{\mathrm{H_2O}}}{M_{\mathrm{add.}}\cdot x}}, \tag{43} \]

where \(a\) is the content of dry additive in percent by weight;

\(n\) is the number of moles of water absorbed by one mole of additive to form crystalline hydrates;

\(M_{\mathrm{H_2O}}\) and \(M_{\mathrm{add}}\) are the molecular weights of water and of the additive, respectively;

\(x\) is the water content in the granules in percent by weight.

The amount of additive calculated from equation (43), if the mechanism of its action is limited only to the phenomenon of crystal formation, should be increased in order to bind the moisture that will be absorbed as a result of hygroscopicity from the air during storage of the nitrate.

This method may give especially good results if it is combined with the method of group isolation.

Rigid-framework method

The method consists in creating a rigid framework in the granules by forming a cellular internal structure through the introduction into the nitrate melt of certain solid, powdery inorganic additives and salts. In the ideal case, each crystal or group of crystals in an ammonium nitrate granule will be enclosed in a cell made of substances insoluble, or less soluble than ammonium nitrate (Fig. 13). The presence of an internal cellular structure, the presence of a framework, will create greater mechanical strength and lower plasticity of individual granules and will substantially slow the migration of moisture from the internal regions of the granules into the intergranular spaces and moisture exchange with the atmosphere.

Fig. 13. Schematic section of a granule having an internal framework: \(a\)—crystals or groups of crystals of ammonium nitrate, \(b\)—framework walls.

The impossibility of creating strong individual bonds between separate granules, owing to low plasticity and the slight development of the surface of physical contact under the influence of external mechanical loads, and the absence of mother liquor in the intergranular spaces, should lead to a sharp reduction in caking or, perhaps, to its complete absence even when hot nitrate is packed in bags.

To create a framework and a cellular structure, it is possible to use both solid inorganic substances in a pulverized state and salts in dissolved form. The greatest results should be expected from substances capable, as a result of interaction with aqueous solutions of ammonium nitrate, of giving highly developed colloidal systems. Among such substances should be included compounds of aluminum, iron, silicon, and magnesium, as well as a number of pulverized products such as cements, phosphorite flour, apatites, pyrite-furnace cinders, ground dunites, etc.

Method of Reducing the Solubility of Ammonium Nitrate

As was shown above, one of the most important properties of ammonium nitrate that determines caking is the high solubility of nitrate in water, which, when mother liquor enters the intergranular spaces and the temperature decreases, leads to the formation of strong crystalline bridges between individual granules and to the loss of the free-flowing character of the layer.

If it were possible by some means to reduce the solubility of nitrate in water, this would lead to the impossibility of forming strong crystalline bridges between granules and to the impossibility of high caking of the nitrate.

It is known from the theory of solutions [17] that the introduction into a solution of the principal salt of other salts can change its solubility. The introduction into a solution of salts with common ions lowers the solubility of the principal salt. Proceeding from this, one may expect a favorable effect on reducing the solubility and caking of ammonium nitrate primarily from nitrate salts.

As shown by F. G. Margolis, A. M. Dubovitskii, and G. V. Glazova [18], the introduction into ammonium nitrate containing 1% moisture of 1% \(Mg(NO_3)_2\) reduces by a factor of 2.7 the separation of crystals from the mother liquor, while simultaneously sharply reducing the caking of the nitrate. A similar action should also be expected from a number of other water-soluble salts and their mixtures.

The method described may give especially positive results if it is combined with the method of group isolation, which limits or completely excludes moisture exchange with the atmosphere.

The methods of water-repellent additives, of a rigid framework, and of reducing solubility presuppose the introduction into the nitrate melt of solid and dissolved inorganic additives, closely connected with one another, since sometimes one or another inorganic additive may cause a reduction in the caking of granulated nitrate by simultaneous action along two or three of the described pathways. This is especially to be expected when an additive consisting of several inorganic salts or complex substances is introduced into the nitrate (for example, phosphorite flour, dolomites, apatite, dunites, and a number of other substances), which, as a result of chemical action with nitric acid or with nitrate, are capable of giving a number of new chemical compounds (double salts, complex compounds) with different mechanisms of action on caking.

Nevertheless, despite the possibility of additives acting by various mechanisms, in developing practical methods for combating caking it is advisable to select additives guided by the requirements of the methods described, since this will shorten the laborious path of experimental searches.

Conclusions

1. The problem of obtaining noncaking and explosion-safe granular ammonium nitrate is of great importance. Solving the problem will increase the effectiveness of nitrogen fertilizers and will make the work of collective and state farms easier.

1. In the present work it has been shown that the granules formed when drops of melt fall through air must necessarily possess a porous structure.

2. For the first time, equations have been derived by means of which one can calculate the degree of filling of granules with crystals, mother liquor, and air over a wide temperature interval. Equations have been derived for calculating the porosity of granules, their apparent specific gravity, and their bulk volumetric weight. The results of the calculations agree with experimental data.

3. The work considers the mechanism of caking of granules in a layer and elucidates the influence of external load, initial moisture content, the nature of the packaging, granule diameter, granulometric composition, temperature, barometric pressure, phase transitions, and hygroscopicity.

4. A classification of methods for combating caking is given, and the degree of their effectiveness is assessed.

5. New methods are proposed for combating caking by stabilizing the internal structure of the granules through the introduction of water-repellent substances, by creating a rigid framework with the formation of a cellular structure; the positive qualities of the method of reducing caking by decreasing the solubility of saltpeter are also evaluated.

The work described indicates a path for experimental searches in the field of obtaining non-caking ammonium nitrate without the use of organic compounds.

LITERATURE

1. A. M. Dubovitskii and Ya. I. Kil’man. Technology of Ammonium Nitrate, Goskhimizdat, 1949.

2. S. I. Vol’fkovich, A. M. Dubovitskii, N. I. Kryuchkov, V. I. Orlov, and N. E. Pestov. Technology of Nitrogen Fertilizers, ONTI, 1935.

3. Gmelin-Kraut’s. Handbuch der anorg. Chem., Vol. Ammonium, 103.

4. S. I. Vol’fkovich and T. V. Glazova. Izv. AN SSSR. Otdel khim. nauk, 4, 314, 1943.

5. M. M. Dubinin. Physicochemical Foundations of Sorption Technology, ONTI, 293, 1935.

6. W. Ross, J. Jee, S. Hendrieks. Ind. Eng. Chem., 37, 11, 1079, 1945.

7. N. E. Pestov. Zhurn. Khim. prom., 11, 33, 1940.

8. V. A. Klevke and N. K. Tsel’m. Khim. prom., 5, 11, 1947.

9. N. E. Pestov. Physicochemical Properties of Granular and Powdered Chemical Products. Publishing House of the Academy of Sciences of the USSR, Moscow–Leningrad, 1947.

10. V. A. Klevke and D. Yu. Gamburg. Khim. prom., 7, 22, 1950.

11. I. R. Krichevskii, L. Kantarovich, and M. Bergauz. Zhurn. khim. prom., 7, 41, 1933.

12. W. Ross, D. Adams, J. Jee, and C. Whittaker. Ind. Eng. Chem., 36, 12, 1088, 1944.

13. P. Miller, G. Lenaeus, W. Saeman, and M. Dokken. Ind. Eng. Chem., 38, 7, 709, 1946.

14. A. M. Dubovitskii and F. G. Margolis. Khim. prom., 5, 8, 1947.

15. T. V. Zabolotskii. Zhurn. prikl. khim., 23, 11, 1127, 1950.

16. S. I. Vol’fkovich and R. Remen. Hygroscopicity of Ammonium Nitrate and Its Mixtures, 1927.

17. A. V. Rakovskii. Introduction to Physical Chemistry, GONTI, 1938.

18. F. G. Margolis, A. M. Dubovitskii, and G. P. Glazova. Khim. prom., 5, 138, 1951.