Int J Thermophys (2012) 33:91–104DOI 10.1007/s10765-011-1113-9

Characterization of Sodium Nitrate as Phase ChangeMaterial

Thomas Bauer · Doerte Laing · Rainer Tamme

Received: 25 August 2010 / Accepted: 24 October 2011 / Published online: 15 November 2011© Springer Science+Business Media, LLC 2011

Abstract In this article the results of material investigations of sodium nitrate(NaNO3) with a melting temperature of 306 ◦C as a phase change material (PCM)are presented. The thermal stability was examined by kinetic experiments and long-duration oven tests. In these experiments the nitrite formation was monitored. Althoughsome nitrite formation in the melt was detected, results show that the thermal stabilityof NaNO3 is sufficient for PCM applications. Various measurements of thermophysicalproperties of NaNO3 are reported. These properties include the thermal diffusivity bythe laser-flash, the thermal conductivity by the transient hot wire, and the heat capac-ity by the differential scanning calorimeter method. The current measurements andliterature values are compared. In this article comprehensive temperature-dependentthermophysical values of the density, heat capacity, thermal diffusivity, and thermalconductivity in the liquid and solid phases are reported.

Keywords Hot wire · Laser-flash · Phase change material · Sodium nitrate ·Thermal energy storage

1 Introduction

This paper considers the latent heat storage concept using phase change materials(PCMs). PCMs are particularly advantageous for two-phase heat transfer fluids. Inthis concept, both the heat carrier and the PCM undergo a phase change at approxi-mately the same temperature.

T. Bauer (B) · D. Laing · R. TammeInstitute of Technical Thermodynamics, German Aerospace Center (DLR),Pfaffenwaldring 38-40, 70569 Stuttgart, Germanye-mail: thomas.bauer@dlr.de

123

92 Int J Thermophys (2012) 33:91–104

Our ongoing research focuses on high-temperature latent heat storage designs forthe two-phase heat carrier water-steam in the temperature range from 120 ◦C to 350 ◦C.The condensation and evaporation temperatures of the carrier are pressure-dependentand occur between 120 ◦C (2 bar) and 350 ◦C (165 bar). Depending on the water-steam pressure, a suitable PCM needs to be identified [1]. Appropriate PCMs for thistemperature range are anhydrous alkali nitrate/nitrites and their mixtures, if require-ments, such as handling, thermal stability, steel compatibility, and economic aspectsare taken into account [2]. The research focuses on latent heat storage systems forindustrial process heat utilization in the low-temperature range (120 ◦C to 250 ◦C)and solar thermal power generation in the high-temperature range (250 ◦C to 350 ◦C).Solar thermal power technologies using optical systems for concentration of sunlightcan be classified into dish, tower, and through systems. Direct steam generating par-abolic troughs are an example where sodium nitrate (NaNO3) could be utilized asa PCM for thermal energy storage. These systems aim for a pressure of around 100bar. At this pressure the water-steam phase change occurs at a temperature of around310 ◦C. This temperature suits the melting temperature of NaNO3 [1].

NaNO3 is found in naturally occurring deposits. These deposits contain variousother salts and require processing to obtain pure NaNO3. On a large scale, NaNO3-rich deposits in Chile are exploited. Smaller but also significant quantities are producedsynthetically. At the time of writing, the world production volume (without China)of NaNO3 is relatively small with a value of a few hundred thousand tons per year.The world KNO3 market is a few times larger than the NaNO3 market. The nitrateproduction is driven by demand. For example, in 1928 the annual world productionvolume of NaNO3 was considerably higher with a value of about 3.2 million tons [3,4].Constraints in the fertilizer market may or may not occur due to the large quantities ofNaNO3 required for storage systems in solar thermal power plants. It can be arguedthat NaNO3 is not consumed or destroyed in the storage application. Utilization ofNaNO3 can be seen as a long-term investment in the security of electricity supply.

The colorless or white crystalline salt has a molar mass of 84.99 g · mol−1. NaNO3is very soluble in water. It is hygroscopic, but does not form hydrated solid phases.The two major grades of NaNO3 are technical and agricultural. NaNO3 is used as afertilizer as well as in a number of industrial processes. The industries include glass,enamel, porcelain, explosives, and charcoal briquette manufacturing. Some gradesof NaNO3 contain an anti-caking agent. Molten salt mixtures containing NaNO3 areemployed as heat-treatment baths, as heat transfer fluids, and as sensible heat storagemedia [3,4].

2 Melting Enthalpy and Melting Temperature of NaNO3

Lumdsen and Jriri [5,6] reviewed experimental data of the melting temperature andthe melting enthalpy of pure NaNO3. Both authors report the average melting temper-ature with a value of 306 ◦C. The reported enthalpy data vary stronger in a range from172 J ·g−1 to 187 J ·g−1 [5,6]. The average values are 177 J ·g−1 [5] and 178 J ·g−1 [6].NaNO3 has been also considered for differential scanning calorimeter (DSC) calibra-tion with a recommended value of 178 J · g−1 [7]. There is some larger discrepancy in

123

Int J Thermophys (2012) 33:91–104 93

terms of the second-order transition enthalpy at around 275 ◦C. Values from 12 J · g−1

(drop calorimeter) to 45 J · g−1 (DSC) are reported [6].An experimental difficulty with molten NaNO3, in particular at elevated tempera-

tures, is the creeping tendency. Here, a thin film of molten salt forms on the surface.The salt wets and spreads spontaneously on many crucible surfaces [8,9].

3 Thermal Stability of NaNO3 and Nitrite Formation

The thermal decomposition of NaNO3 can be divided into three characteristic regions[10–16]:

1. Little nitrite formation: above the melting point (Tm) to about 450 ◦C2. Nitrate–nitrite equilibrium: about 450 ◦C to 550 ◦C3. Decomposition of the nitrite with release of nitrogen oxides above about 550 ◦C

3.1 Low-Temperature Region (Tm to 450 ◦C)

Literature about the low-temperature region (Tm to 450 ◦C) is generally limited. Kustreported about experiments of the eutectic KNO3–NaNO3 with Tm = 222 ◦C. Theyfound small amounts of nitrites at temperatures as low as 295 ◦C in the melt. Due tothe experimental procedure, it was assured that this nitrite was not due to impuritiesin the nitrate. The nitrite ion concentration was reported to reach almost equilibriumwithin (7 to 16) h at temperatures between 295 ◦C and 340 ◦C. Although there is someuncertainty about the exact measurement conditions, these data have been convertedto the molar ratio NO2

−/NO3 at a partial pressure of po2 = 0.21 (Fig. 4) [13].In order to improve the data basis for the thermal stability of NaNO3 in the low-tem-

perature region, long-duration tests of NaNO3 at 350 ◦C were performed. The NaNO3was purchased from BASF. The technical non-food grade salt has a purity of min.99 %. The major impurities are NaNO2 (50), Na2CO3 (400), NaCl (250), and Na2SO4(300) with maximum values in mg/kg. The salt was heated in a borosilicate glass bea-ker which was placed in a batch furnace (air at atmospheric pressure). Mass losses ofthe melt were monitored with an analytical balance at room temperature. The nitriteconcentration was monitored by taking samples of the melt periodically. For nitriteanalysis, the solidified salt was dissolved in deionized water. The nitrite (NO2

−) con-centration was determined by a photospectrometric standard method (Analytik JenaAG, type Spekol 1100, 4500-NO2

− colorimetric method). The crucible material wasborosilicate glass, and the sample mass was 100 g. The results show that about 0.04mass% NO2

− forms within a few hours in the melt. This value remains constant within24 h. The value increases to about 0.07 mass% NO2

− after about 2600 h.After the first melting, a mass loss from (0.1 to 0.3) mass% was detected. This loss is

attributed to moisture in the salt. Subsequently, mass variations during the 2600 h testwere monitored four times and were found to be smaller than ±0.05 mass%. Overall,the measurements showed that NaNO3 is thermally stable at 350 ◦C and that a smallamount of nitrite is formed.

123

94 Int J Thermophys (2012) 33:91–104

3.2 Intermediate Temperature Region (450 ◦C to 550 ◦C)

It is well known that molten nitrates decompose to nitrites in the intermediate temper-ature region from about 450 ◦C to 550 ◦C. The thermal dissociation is reversible, andthe equilibrium reaction can be written for NaNO3 as [8,11–16]

NaNO3(l) ↔ NaNO2(l) + 1/2 O2(g) (1)

Kinetic data of the oxidation and decomposition of different alkali metal nitrateshave been reported. They include KNO3/KNO2 [10–12,15], KNO3–NaNO3/KNO2–NaNO2 [8,12], and NaNO3/NaNO2 [12,15]. The time to reach equilibrium dependson parameters, such as the reaction direction (decomposition or oxidation), the atmo-sphere, and the experimental setup. A typical experimental setup reported in the lit-erature is bubbling of gas through the melt. At higher temperatures the equilibrium isgenerally reached faster compared to lower temperatures. The reported periods rangefrom days at a temperature of around 300 ◦C [13] to tens of minutes at a temperatureof about 700 ◦C [11].

We examined the decomposition of NaNO3 in NaNO2 in the temperature range from450 ◦C to 550 ◦C. For all experiments the sample was technical NaNO3 with a mass of110 g to 170 g (BASF type U, min. 99.0 %). Open crucibles made of glazed ceramics(450 ◦C) and dense aluminum oxide (500 ◦C and 550 ◦C) were used. The temperatureof the salt was monitored by a mineral insulated thermocouple (Inconel 600, Class1) within the melt. Synthetic air with rates of 100 mL · min−1 and 600 mL · min−1

was purged through the melt via a stainless steel tube. This gas supply increased thedecomposition rate and ensured a uniform distribution of the melt.

Figure 1 plots nitrite results of measurements at three temperatures (450 ◦C, 500 ◦C,and 550 ◦C). Additional experiments with increased air flow from 100 mL · min−1 to600 mL ·min−1 at 500 ◦C are also shown. For the increased gas flow in the first exper-iment, at the beginning only a small increase of the decomposition rate was observed.

0.000

0.005

0.010

0.015

0.020

0.025

0.030

0.035

0.040

0.045

0 100 200 300 400 500 600 700 800 900 1000

Time t, h

Mol

ar r

atio

NO

2- /NO

3- ,pO

2=0.

21(a

ir) [1

]

550 °C 100 ml min-1500 °C 600 ml min-1 (Experiment 1)500 °C 100 ml min-1 (Experiment 1)500 °C 100 ml min-1 (Experiment 2)450 °C 100 ml min-1

( )t

NO

NO 157.0

3

2 e10395.0 −−=

( )t

NO

NO 0484.0

3

2 e10179.0 −−=

( )t

NO

NO 00467.0

3

2 e100620.0 −−=

550 °C 100 ml·min-1

500 °C 600 ml·min-1 (Experiment 1)500 °C 100 ml·min-1 (Experiment 1)500 °C 100 ml·min-1 (Experiment 2)450 °C 100 ml·min-1

Fig. 1 Kinetics of nitrite formation in NaNO3 in contact with synthetic air

123

Int J Thermophys (2012) 33:91–104 95

0

0.05

0.1

0.15

0.2

0.25

0.3

260 270 280 290 300 310 320

Temperature, °C

DS

C S

igna

l, μV

·mg-1 NaNO3

NaNO3 (96 mol%) -NaNO2 (4 mol%)

NaNO3

NaNO3 (96 mol%) - NaNO2 (4 mol%)

Fig. 2 DSC-measurement of NaNO3 with and without NaNO2 impurity

Both experiments at 500 ◦C (100 and 600) mL · min−1 indicate a very similar molarNO2

−/NO3− ratio in equilibrium. A simple empirical exponential growth model was

fitted to the measurements with 100 mL · min−1 using non-linear regression tech-niques. The model uses two parameters and can approximately describe the reactionkinetics.

The three model curves show that the temperature level not only affects the amountof NO2

− in equilibrium but also the decomposition rate of NO3− to NO2

−. The equi-librium at 550 ◦C was quickly reached after several tens of hours. On the other hand,at 450 ◦C the time constant was much longer with several hundreds of hours.

In the following, we describe examinations of the effect of nitrite impurities on themelting behavior of NaNO3. Figure 2 shows two DSC measurements carried out bythe authors (Netzsch-Gerätebau GmbH, DSC404). A DSC enthalpy calibration wasnot performed. Hence, values are given in the units of µV rather than mW. Comparedto pure NaNO3, it can be seen that due to the addition of NaNO2 the melting rangebroadens and the melting temperature is reduced. In order to achieve a high resolution,a low heating rate of 0.2 K · min−1 was used. The crucible material was platinum, andthe atmosphere was argon.

A known method for purity determination is the use of the van’t Hoff law(Eq. 2). This law describes the melting-temperature reduction due to impurities, whereR = 8.31441 J · mol−1 · K−1 is the molar gas constant, H2 is the melting enthalpy(J · mol−1), and T2 is the melting temperature (K) of the pure material. T is the tem-perature where the impure material is completely liquid (K) and x1(T ) is the molarfraction of the impurity;

x1(T ) = H2

R

T2 − T

T 22

(2)

Bader [17] presents a method to calculate the DSC heat flux depending on the impurityfraction. Figure 3 shows the calculation for NaNO3. This calculation (Fig. 3) and theDSC-measurement (Fig. 2) show good agreement considering that no desmearing ofthe DSC signal was applied. Both figures show the effect of peak broading and loweringof the liquidus temperature due to the nitrite impurity. It can be seen that an impurity of

123

96 Int J Thermophys (2012) 33:91–104

0

2

4

6

8

10

12

14

260 270 280 290 300 310 320

Temperature, °C

Nor

mal

ized

hea

t flu

x

5 %3 %2 %

1 %

0.5 %

0.2 %

Molar fraction of the impurity

Fig. 3 Calculated DSC heat flux of NaNO3 with different quantities of NaNO2 as an impurity

4 mol% sodium nitrite in sodium nitrate results in a reduction of the liquidus temper-ature of about 6 K. Such a reduction would not be acceptable for PCM applications.

3.3 High-Temperature Region

At high temperatures (above about 550 ◦C for NaNO3 in air atmosphere), furtherdecomposition takes place with the evolution of the oxides of nitrogen in the gasphase. Here, KNO3 has been found slightly more stable compared to NaNO3 and theeutectic KNO3–NaNO3 (Fig. 4) [8,10,15]. Experimental difficulties with salt creep-ing have been reported. At temperatures higher than 500 ◦C, creeping salt in contactwith materials, such as stainless steel or quartz leads also to the liberation of nitrogenoxides [8].

200 300 400 500 600 700 800

Temperature, °C

Mol

ar r

atio

NO

2- /NO

3- ,pO

2=0.

21(a

ir) [1

]

NaNO3 (kintetics, DLR this work) NaNO3 (oven 2600 h, this work) NaNO3 Bradshaw [12]NaNO3 Sirotkin [15]KNO3-NaNO3 Bradshaw [12]KNO3-NaNO3 Nissen [8]KNO3-NaNO3 Kust [13]KNO3 Bradshaw [12]KNO3 Freeman [10]KNO3 Sirotkin [15]

Decomposition withnitrogen release and oxide formationLittle nitrite formation

KNO3

Tm = 337 °C

KNO3-NaNO3

Tm = 222 °C

NaNO3

Tm = 306 °C

1 x 10-5

1 x 10-4

1 x 10-3

1 x 10-2

1 x 10-1

1 x 100MNO3(l) ↔ MNO2(l) + 1/2 O2(g)

Fig. 4 Literature and DLR measurement data of the nitrite concentration in alkali metal nitrates. The figurealso shows the melting temperature of KNO3, NaNO3, and KNO3–NaNO3

123

Int J Thermophys (2012) 33:91–104 97

3.4 Summary of the Three Temperature Regions

Figure 4 shows the nitrite concentration in different alkali metal nitrates depending onthe temperature in the three characteristic regions. It can be concluded that heating ofthe PCM far above the melting temperature may lead to some constraints, since theNO2

− results in a change in the NaNO3 melting behavior. Here, the kinetics of theequilibrium reaction may or may not allow full oxidation of the nitrite to nitrate oncooling. Although data around 350 ◦C are limited, the current measurements and workby Kust show that the nitrite formation at this temperature is below 0.5 mass% NO2

−.As can be seen in Fig. 3, there are small changes in the melting behavior in terms ofthe formation of a melting range and the reduction of the liquidus temperature at thisimpurity level. Hence, for the PCM application with a maximum temperature limit of350 ◦C, changes in the melting behavior of NaNO3 can be neglected.

It can be argued that the compatibility of molten NaNO3 and graphite is a criticalaspect. Abald reports about the incompatibility of molten NaNO3 with graphite at310 ◦C [18]. Also, it was found that nitrate salts react with graphite seals at tempera-tures above about 350 ◦C [19].

4 Thermophysical Properties of NaNO3

4.1 Density and Thermal Expansion

The density of NaNO3 at room temperature is well known with a value of 2.260 g ·cm−3 at 25 ◦C [20]. Literature data of the linear expansion coefficient in the solidstate are also available. For polycrystalline NaNO3, the room temperature value is40 × 10−6 K−1 [21]. The linear expansion coefficient increases steadily up to a max-imum at the solid–solid conversion of NaNO3 at around 275 ◦C. For higher tempera-tures, up to the melting temperature, the coefficient drops again [20,22]. Using theseliterature expansion data, we calculated the density in the solid state from room tem-perature up to the melting temperature (Fig. 5). Petitet et al. [23] reviewed data aboutthe volume change on melting �V related to the volume in the solid state at the melt-ing temperature at atmospheric pressure Vs and reported the following �V/Vs valuesin percent: 8.3, 9.6, 9.7, 10.0, and 10.7. In this work a value of 9.7 % from Schinkeand Sauerwald [24] was assumed (Fig. 5). This value is also the average of the fivereported numbers. Figure 5 shows that the approaches to calculate the liquid den-sity at the melting temperature from the room-temperature density, the linear thermalexpansion coefficient, and the volume change on melting are consistent with availableliterature data in the liquid range [25–29].

4.2 Heat Capacity

Heat capacity data of NaNO3 are available from various authors [21,30–35] and havebeen reviewed by Jriri et al. [6] (Fig. 5). There is some discrepancy between oldervalues mainly measured by drop calorimeters and newer values measured by DSC.

123

98 Int J Thermophys (2012) 33:91–104

1.75

1.85

1.95

2.05

2.15

2.25

Landolt-Börnstein 2004 [20]Schinke 1960 [24]Touloukian 1977 [21], Schinke (calc. from thermal expansion)Kracek 1931 [22], Touloukian (calc. from thermal expansion)Tufeu 1985 [25]Polyakov 1955 [26]Janz 1968 [27] (calc. from several sources)Janz 1972 [28] (new recommendations)Murgulescu 1969 [29]

Schinke

ρ liquid,306°C=1.908 g·cm-3, g·c

m-3

11.11.21.31.41.51.61.71.81.9 DLR (this work)

Goodwin 1909 [31]Mustajoki 1957 [32]Ichikawa 1983 [33]Rogers 1982 [30]Takahashi 1988 [34]Carling 1983 [35]Jriri 1995 [6]Touloukian 1970 [21]Average liquid value

c p ,liquid,306°C=1.655 J·g-1·K-1

average does not include

older values around 1.8 J·g-1·K-1

c p, J

·g-1

·K-1

0.10.150.2

0.250.3

0.350.4

0.450.5

DLR (this work)Kobayasi 1982 [42]Kato 1975 [41]Zhang 2000 [37]Araki 1984 [40]Odawara 1977 [39]Gustafsson 1968 [38]Ohta 1990 [44]Knothe 1985 [43]

calculated from a = k /(ρ ·c p )

a liquid,306°C=0.163 mm2·s-1

a, m

m2 ·s

-1

0.5

0.6

0.7

0.8

0.9

1

0 50 100 150 200 250 300 350 400 450 500

Temperature, °C

DLR (this work) calc. Kobayasi 1982 [41]DLR (this work) measured White 1967 [50]Nagasaka 1991 recom. [45] Santini 1984 [51]Bloom 1965 [46] McDonald 1970 [52]Omotani 1984 [47] Zhang 2000 [36]Turnbull 1961 [48] Gustafsson 1968 [37]Tufeu 1985 [25] McLaughlin 1964 [53]Omotani 1982 [49] Kitade 1989 [54]

Nagasaka recommended

k liquid,306°C=0.514 W·m-1·K-1

k, W

·m-1

·K-1

ρ

Fig. 5 Thermophysical properties of solid and liquid NaNO3; the melting temperature of NaNO3 is 306 ◦C

Older values, in particular in the liquid range, are higher. Nguyen-Duy [36] used adrop calorimeter, and these exceptionally high values are not included here.

Our measurements of the heat capacity were made with a heat flux type DSC(Netzsch DSC404) in an argon flow (100 mL · min−1) with a heating rate of10 K · min−1. NaNO3 was purchased from Merck (purity 99.99 %) and used withoutfurther purification. The sample mass of about 20 mg was checked before and after theDSC measurement with a microbalance. Aluminum crucibles with a lid were used.Sapphire was utilized as a heat capacity reference material with heat capacity datarecommended by Netzsch-Gerätebau GmbH. Figure 5 shows the average values fromsix newly prepared measurements. The maximum deviation from all measurementswas ±4 % of these average values. For temperature ranges from 80 ◦C to 190 ◦C andfrom 350 ◦C to 380 ◦C, our measurements agree to within ±3 % with literature valuesfrom Rogers and Janz [30], Takahashi et al. [34], Carling [35], and Jriri et al. [6]. Theheat capacity in the liquid range is approximately constant. Table 1 shows tabulatedmeasurement results taking three times the standard deviation in the measurements

123

Int J Thermophys (2012) 33:91–104 99

Table 1 Temperature-dependent average heatcapacity values of NaNO3measured by DSC

Temperature (◦C) Heat capacity (J · g−1 · K−1)

100 1.269 ± 0.086

110 1.289 ± 0.085

120 1.312 ± 0.092

130 1.334 ± 0.087

140 1.359 ± 0.087

150 1.384 ± 0.085

360 1.658 ± 0.115

380 1.658 ± 0.115

as the uncertainty. Measurements of molten NaNO3 at high temperatures are difficultfor two reasons. First, NaNO3 in the liquid state strongly wets common containermaterials. As a result, the salt creeps out of the container [9,30]. Second, there isnitrite formation in the melt, so that measurements refer to the binary NaNO2–NaNO3system, rather than the single NaNO3 salt.

4.3 Thermal Diffusivity

The thermal diffusivity was measured by a variety of methods. They include the hot-wire method using an electrical insulating coating [37], optical techniques [38,39], andthe stepwise heating method [40–42], as well as laser-flash methods using two-layer[43] and three-layer arrangements [44]. In the liquid range at the melting temperature,the average of all these values is 0.163 mm2 ·s−1. This value agrees with the calculatedvalue using the relation a = k/(ρcp) as shown in Fig. 5 by the framed text boxes.With the exception of measurements by Kobayasi et al. [42], no work in the solidrange could be identified.

Our thermal-diffusivity measurements of NaNO3 in the liquid and solid states werecarried out by the laser-flash method using a three-layer arrangement with a platinumcrucible and a commercial apparatus (Netzsch LFA457). A single laser pulse beamilluminates the bottom area of a platinum crucible (Fig. 6). The heat penetrates throughthree layers: the platinum crucible (thickness of 0.30 mm), either a liquid or solid saltsample (0.66 mm, diameter of 12 mm), and the platinum lid (0.31 mm). The bottomand top of this crucible were coated with a thin graphite layer to enhance the absorptionand emission of radiation. The temperature rise versus time at the rear surface of the lid

Pt - layer 1

Laser beam

SiC lid with thread

Pt lid with expansion holes Solid or liquid salt samplePt crucible

SiC holding plate

Optics and IR-detector(liquid N2 cooled)

SiC holder with threadGraphite coating

Graphite coating

Salt - layer 2

l1

l2

l3 Pt - layer 3Steel distance ring

Fig. 6 Schematic cross section of the measuring assembly with three layers to measure the thermaldiffusivity

123

100 Int J Thermophys (2012) 33:91–104

Table 2 Temperature-dependent averagethermal-diffusivity values ofNaNO3 measured by thethree-layer laser-flash method

Temperature (◦C) Thermal diffusivity Number of newly prepared(mm2 · s−1) measurements

46.8 0.6202 ± 0.0382 2

99.3 0.5232 ± 0.0950 2

199.3 0.3350 ± 0.0102 2

289.5 0.2350 ± 0.0232 3

319.5 0.1786 ± 0.0130 5

349.5 0.1843 ± 0.0135 5

400.8 0.1895 ± 0.0143 5

is measured by a liquid nitrogen cooled InSb photocell. The atmosphere was nitrogen(100 mL · min−1). A mathematical model describes the temperature rise versus timesignal and was fitted to the experimental data using a non-linear regression algorithm.This three-layer model includes the thicknesses of the three layers and thermophysicalproperties of the layers, as well as a correction of the heat losses and the laser pulselength (Netzsch Cape-Lehmann model).

NaNO3 was purchased from Merck (purity of 99.99 %) and used without furtherpurification. First, the compressed salt powder was melted in the crucible within theapparatus. Subsequently, the thermal diffusivities in the liquid and then in the solid statewere measured isothermally. Two to Five measurement series with a newly preparedcrucible fill and three shots per temperature were carried out. The reproducibility inthe liquid state was generally high. Virtually all single shots had a deviation of lessthan ±5 % compared to the overall average. The salt-layer thickness is the major causeof measurement uncertainty. For example, an error of 0.01 mm in thickness leads toan error of about 3 % in the diffusivity value. The reproducibility in the solid state wasrather poor, so that the measurements can only give an indication of the diffusivity.Before and after the measurement series, the accuracy of the layer arrangement wasverified by pure water measurements at room temperature using literature values from[21]. Table 2 lists measurement results for the thermal diffusivity using three timesthe standard deviation in the measurements as the uncertainty.

4.4 Thermal Conductivity

A comprehensive review and a recommendation of the thermal conductivity of moltenNaNO3 has been given by Nagasaka and Nagashima [45]. They reviewed data fromvarious authors [46–54]. The recommended value is rather low, since many otherhigher values are thought to include systematic errors due to radiation and convection(Fig. 5) [45]. The few measurement values in the solid state differ widely, so thatno consistent data could be identified. Kobayasi et al. [42] reported values as low as0.2 W · m−1 · K−1 measured by the square-wave technique. Figure 5 does not showthese exceptionally low values fully. One possible reason for this discrepancy may bedifferences in the crystal structure in solid NaNO3.

Our measurements of the thermal conductivity in the solid phase on polycrystallineNaNO3 were carried out by the transient hot-wire method. The platinum resistance

123

Int J Thermophys (2012) 33:91–104 101

Table 3 Temperature-dependent thermal-conductivityvalues of NaNO3 measuredby the hot-wire method

Temperature (◦C) Thermal conductivity(W · m−1 · K−1)

47.3 0.91 ± 0.06

144.9 0.82 ± 0.06

245.8 0.73 ± 0.05

thermometer technique based on one wire was used (Netzsch TCT426) [55]. The wirematerial was Pt90Rh10 with a length of 163 mm and a diameter of 0.3 mm. First, NaNO3powder with a minimum purity of 99 % (BASF Type U) was liquefied. Compared toother measurements, in this thermal-conductivity measurement, technical grade ratherthan laboratory grade salt was used. The reason was the large salt block required forthis type of measurement. The molten NaNO3 in the crucible solidified from the bot-tom in order to account for the large volume change on solidification. Two blockswith dimensions of approximately 180 mm × 110 mm × 70 mm were cast. Subse-quently, plane surfaces were machined. One plane surface had slots that enclosed thehot wire and the voltage taps. The slots had a depth and width of approximately 0.5mm. At each temperature level, five measurements with a duration of 900 s and a wirepower of 1 W (6.1 W · m−1) were carried out. In order to start the measurement, atemperature stability criterion was defined. The deviation of the temperature betweena reference thermocouple and the measurement wire, determined via the resistancetable (supplied by Netzsch), was smaller than 0.01 K/10 min. For the measurement,the second heating cycle was used. Table 3 lists thermal-conductivity results with anuncertainty of ±7 % of the measurement values. Figure 5 shows that in the solid phasenear the melting temperature good agreement between the measured conductivity bythe transient hot-wire method and the calculated value from thermal-diffusivity mea-surements by the three-layer laser-flash method is obtained. In general, more researchis required in order to establish a better understanding of the thermal-diffusivity andthermal-conductivity data in the solid phase of NaNO3.

5 Summary and Conclusions

This article focused on fundamental material aspects of sodium nitrate (NaNO3) asa PCM for latent heat storage. The impact of nitrite formation in the NaNO3 melton PCM utilization has been examined. Literature values show that NaNO3 meltsform considerable amounts of nitrite in the temperature range from about 470 ◦C to700 ◦C. Our kinetic measurements at 550 ◦C, 500 ◦C, and 450 ◦C and at a 2 600 hoven test at 350 ◦C extended the range to lower temperatures. In the low-temperaturerange (≤450 ◦C) the nitrite formation kinetics were found to be slow with hundreds ofhours to reach equilibrium. If high-temperature NO2

−/NO3− data (450 ◦C to 700 ◦C)

are extrapolated to 350 ◦C, they agree with our oven long-term tests at 350 ◦C. Ourresults indicate that previously published nitrite concentrations in KNO3–NaNO3 from295 ◦C to 340 ◦C may have been reported too low. Our examinations show that nitritein molten NaNO3 results in an undesirable reduction of the NaNO3 melting tem-perature and the formation of a melting range. Hence, the maximum temperature of

123

102 Int J Thermophys (2012) 33:91–104

NaNO3 utilized as a PCM can be limited by this mechanism. Although there is alsonitrite formation at low temperatures in alkali metal nitrates, the absolute values aresmall. Hence, with a temperature limit of 350 ◦C, no significant change in the meltingbehavior of the PCM due to nitrite formation is expected.

Thermophysical properties of NaNO3 in the liquid and solid ranges have beenreviewed. Our measurements on the heat capacity, the thermal diffusivity, and thethermal conductivity have been presented. In the solid range, reliable density and heatcapacity data from the literature are available, but there is a lack of consistent thermal-diffusivity and thermal-conductivity values. In general, there is very limited work onthe characterization of thermal conduction properties of alkali metal nitrate salts fromroom temperature to the melting temperature. In this article, the first measurements inthe solid range by two methods have been presented. Both measurements show that thethermal diffusivity and thermal conductivity drop significantly from room temperatureto the melting temperature. Further work is required to clarify the larger discrepancyof the available thermal-conductivity and thermal-diffusivity values (e.g., impact ofcrystal structure on the thermal conductivity). In the liquid range near the meltingtemperature of 306 ◦C, previous authors focused on single thermophysical propertyvalues. The presented work used the relation k = aρcp , where k is the thermal con-ductivity, a is the thermal diffusivity, ρ is the density, and cp is the heat capacity. Withthe help of this equation, consistent thermophysical data in the liquid range near themelting temperature were identified for NaNO3.

This article contributes to a compilation of thermophysical property data of NaNO3from room temperature to the decomposition temperature. It includes solid-phase,phase-change, and liquid-phase properties. The knowledge of nitrite formation inmolten NaNO3 was extended to lower temperatures. Besides the considered PCMapplication, this information can also be used in other fields.

Acknowledgments The authors thank the German Federal Ministry for the Environment, NatureConservation and Nuclear Safety for the financial support given to the ITES project (Contract No.03UM0064).

References

1. R. Tamme, T. Bauer, J. Buschle, D. Laing, H. Müller-Steinhagen, W.-D. Steinmann, Int. J. EnergyRes. 32, 264 (2008). doi:10.1002/er.1346

2. T. Bauer, D. Laing, W.-D. Steinmann, U. Kröner, R. Tamme, in Proceedings Eurosun 2008, Lisbon,Portugal, 2008. Paper ID 163

3. W. Laue, M. Thiemann, E. Scheibler, K.W. Wiegand, Ullmann’s Encycl. Ind. Chem. (2000). doi:10.1002/14356007.a17_265

4. L. Pokorny, I. Maturana, W.H. Bortle Kirk-Othmer Encycl. Chem. Technol. (2006). doi:10.1002/0471238961.1915040916151115.a01.pub2

5. J. Lumdsen, in Thermodynamics of Molten Salt Mixtures, ed. by J. Lumdsen (Academic Press, NewYork, 1966), p. 109

6. T. Jriri, J. Rogez, C. Bergman, J.C. Mathieu, Thermochim. Acta 266, 147 (1995). doi:10.1016/0040-6031(95)02337-2

7. R. Sabbah, A. Xu-wu, J.S. Chickos, M.L. Planas Leitão, M.V. Roux, L.A. Torres, Thermochim.Acta 331, 93 (1999). doi:10.1016/S0040-6031(99)00009-X

8. D.A. Nissen, D.E. Meeker, Inorg. Chem. 22, 716 (1983). doi:10.1021/ic00147a004

123

Int J Thermophys (2012) 33:91–104 103

9. C.M. Kramer, Z.A. Munir, J.V. Volponi, Thermochim. Acta 55, 11 (1982). doi:10.1016/0040-6031(82)87002-0

10. E.S. Freeman, J. Am. Chem. Soc. 79, 838 (1957). doi:10.1021/ja01561a01511. R.F. Bartholomew, J. Phys. Chem. 70, 3442 (1966). doi:10.1021/j100883a01212. R.W. Bradshaw, R.W. Carling, Sandia National Laboratories Report SAND87-80055 (Sandia National

Labs, Livermore, 1987)13. R.N. Kust, J.D. Burke, Inorg. Nucl. Chem. Lett. 6, 333 (1970). doi:10.1016/0020-1650(70)80243-414. F. Paniccia, P.G. Zambonin, J. Phys. Chem. 77, 1810 (1973). doi:10.1021/j100633a01815. G.D. Sirotkin, Russ. J. Inorg. Chem. 4, 1180 (1959)16. J.A. Plambeck, “Fused Salt Systems”, in Encyclopedia of Electrochemistry of the Elements, vol. 10,

ed. by A.J. Bard (Marcel Dekker, New York, 1976), p. 18917. R.G. Bader, J.E.K. Schawe, G.W.H. Höhne, Thermochim. Acta 229, 85 (1993). doi:10.1016/

0040-6031(93)80316-318. E.R. Abald, Corrosion Guide (Elsevier, Amsterdam, 1968)19. J. Moreno, D. Kearney, presented at NREL Trough Thermal Storage Workshop, Golden, CO, 200320. K. Gesi, 30 KNO3 Family (Landolt-Börnstein, Hellwege, 2004). doi:10.1007/b6804521. Y.S. Touloukian, C.Y. Ho, R.W. Powell, M.C. Nicolaou, in Thermophysical Properties of Matter, vols.

5, 10, 13 (Plenum, New York, 1973, 1977, 1979)22. F.C. Kracek, J. Am. Chem. Soc. 53, 2609 (1931). doi:10.1021/ja01358a02223. J.P. Petitet, M. Fraiha, R. Tufeu, B. Le Neindre, Int. J. Thermophys. 3, 137 (1982). doi:10.1007/

BF0050363624. H. Schinke, F. Sauerwald, Zeitschrift für anorganische Chemie 287, 313 (1960). doi:10.1002/zaac.

19562870417 [in German]25. R. Tufeu, J.P. Petitet, L. Denielou, B. Le Neindre, Int. J. Thermophys. 6, 315 (1985). doi:10.1007/

BF0050026626. V.D. Polyakov, S.H. Beruli, Izv. Siktora, Fiz.-Khim. Anal. 26, 164 (1955) [in Russian]27. G.J. Janz, F.W. Dampier, G.R. Lakshminarayanan, P.K. Lorenz, R.P.T. Tomkins, National Standard

Reference Data Series (NSRDS), vol. 15 (National Bureau of Standards, Washington, DC, 1968)28. G.J. Janz, U. Krebs, H.F. Siegenthaler, R.P.T. Tomkins, J. Phys. Chem. Ref. Data 1, 581 (1972). doi:10.

1063/1.325310329. I.G. Murgulescu, S. Zuca, Electrochim. Acta 14, 519 (1969). doi:10.1016/0013-4686(69)87037-430. D.J. Rogers, G.J. Janz, J. Chem. Eng. Data 27, 424 (1982). doi:10.1021/je00030a01731. H.M. Goodwin, H.T. Kalmus, Phys. Rev. Ser. I 28, 1 (1909). doi:0.1103/PhysRevSeriesI.28.132. A. Mustajoki, Ann. Acad. Sci. Fenn. A6 Physica 5, 3 (1957) [in German]33. K. Ichikawa, T. Matsumoto, Bull. Chem. Soc. Jpn. 56, 2093 (1983). doi:10.1246/bcsj.56.209334. Y. Takahashi, R. Sakamoto, M. Kamimoto, Int. J. Thermophys. 9, 1081 (1988). doi:10.1007/

BF0113327535. R.W. Carling, Thermochim. Acta 60, 265 (1983). doi:10.1016/0040-6031(83)80248-236. P. Nguyen-Duy, E.A. Dancy, Thermochim. Acta 39, 95 (1980). doi:10.1016/0040-6031(80)80002-537. X. Zhang, M. Fujii, Int. J. Thermophys. 21, 71 (2000). doi:10.1023/A:100660482075538. S.E. Gustafsson, N.-O. Halling, R.A.E.Z. Kjellander, Z. Naturforsch A Phys. Sci. 23, 44 (1968)39. O. Odawara, I. Okada, K. Kawamura, J. Chem. Eng. Data 22, 222 (1977). doi:10.1021/je60073a02540. N. Araki, Int. J. Thermophys. 5, 53 (1984). doi:10.1007/BF0050207941. Y. Kato, K. Kobayasi, N. Araki, K. Furukawa, J. Phys. E Sci. Instrum. 8, 461 (1975). doi:10.1088/

0022-3735/8/6/01142. K. Kobayasi, N. Araki, Y. Iida, in Proceedings 7th International Heat Transfer Conference, vol. 6,

München, Germany, 1982, pp. 467–47243. W. Knothe, Doctoral Thesis, RWTH-Aachen, 1985 [in German]44. H. Ohta, G. Ogura, Y. Waseda, M. Suzuki, Rev. Sci. Instrum. 61, 2645 (1990). doi:10.1063/1.114185345. Y. Nagasaka, A. Nagashima, Int. J. Thermophys. 12, 769 (1991). doi:10.1007/BF0050240446. H. Bloom, A. Doroszkowski, S.B. Tricklebank, Aust. J. Chem. 18, 1171 (1965)47. T. Omotani, A. Nagashima, J. Chem. Eng. Data 29, 1 (1984). doi:10.1021/je00035a00148. A.G. Turnbull, Aust. J. Appl. Sci. 12, 324 (1961)49. T. Omotani, Y. Nagasaka, A. Nagashima, Int. J. Thermophys. 3, 17 (1982). doi:10.1007/BF0050395550. L.R. White, H.T. Davis, J. Chem. Phys. 47, 5433 (1967). doi:10.1063/1.170181151. R. Santini, L. Tadrist, J. Pantaloni, P. Cerisier, Int. J. Heat Mass Transf. 27, 623 (1984). doi:10.1016/

0017-9310(84)90034-6

123

104 Int J Thermophys (2012) 33:91–104

52. J. McDonald, H.D. Davis, J. Phys. Chem. 74, 725 (1970). doi:10.1021/j100699a00753. E. McLaughlin, Chem. Rev. 64, 389 (1964). doi:10.1021/cr60230a00354. S. Kitade, Y. Kobayashi, Y. Nagasaka, A. Nagashima, High Temp. High Press. 21, 219 (1989)55. Standard C1113-99 (ASTM International, West Conshohocken, 2004)

123