Solar Energy Materials & Solar Cells 95 (2011) 1831–1837

Contents lists available at ScienceDirect

Solar Energy Materials & Solar Cells

0927-02

doi:10.1

n Corr

Villeurb

E-m

journal homepage: www.elsevier.com/locate/solmat

Development and characterisation of a new MgSO4�zeolite composite forlong-term thermal energy storage

Stephanie Hongois a,b,c, Frederic Kuznik a,b,n, Philippe Stevens c, Jean-Jacques Roux a,b

a INSA de Lyon, CETHIL, CNRS UMR5008, F-69621 Villeurbanne Cedex, Franceb Universite Lyon 1, F 69622, Francec Electricite de France (EDF), R&D division, Les Renardi�eres 77818 Moret sur Loing, France

a r t i c l e i n f o

Article history:

Received 14 September 2010

Received in revised form

21 December 2010

Accepted 31 January 2011Available online 10 March 2011

Keywords:

Thermal energy storage

Thermochemical process

Zeolites

Magnesium sulphate

Seasonal storage

Building application

48/$ - see front matter & 2011 Elsevier B.V. A

016/j.solmat.2011.01.050

esponding author at: INSA de Lyon, CETHI

anne Cedex, France. Tel.: +33 472 438 461; f

ail address: frederic.kuznik@insa-lyon.fr (F. K

a b s t r a c t

A seasonal chemical heat store based on the hydration/dehydration cycle of a magnesium sulphate

(MgSO4) composite material has been developed. During summer, the material stores heat by an

endothermic dehydration reaction and heat used for space heating is released in winter by rehydrating

the material. For this specific purpose, a new thermal energy storage composite material has been

developed, which is the subject of the present paper.

The active material needs to be dispersed to optimise the gas–solid reaction kinetics and the

thermal power released/absorbed. Zeolites have proved to be a favourable porous expanded structure

for MgSO4 with energy densities of 150–400 kWh/m3 at a storage temperature compatible with solar-

thermal collectors. Characterisation by thermogravimetric analysis (TGA) and differential scanning

calorimetry (DSC) on 10 mg samples shows that almost 80% of the total energy density can be stored at

150 1C although the material is not fully dehydrated. At a bigger scale, during hydration tests on 200 g

sample, an energy density of 0.18 Wh/g is achieved, i.e., 45% of the theoretical energy density of the

form-stable zeolite–MgSO4 composite sorbent. Micro-calorimetry measurements have confirmed that

this energy density value remains at this level over 3 charge/discharge cycles.

& 2011 Elsevier B.V. All rights reserved.

1. Introduction

Thermal energy storage systems could make an importantcontribution in reducing our dependency on fossil fuels, but alsoin contributing to a more efficient and environmentally benignenergy use [1,2]. As the demand in thermal comfort of buildingsrises, the energy consumption correspondingly increases. For exam-ple, in France, the energy consumption of buildings has increasedby 30% in the last 30 years. Housing and tertiary buildings areresponsible for the consumption of approximately 46% of allenergies and approximately 19% of the total CO2 emissions [3].

The annual solar energy incident on building surfaces is fargreater than the building needs for space heating and hot water,especially during the summer season. It is obvious that a long-term thermal energy storage system is a means to rationalise theuse of this renewable energy. Thermal energy storage can beaccomplished using either sensible heat storage, latent heatstorage, physical sorption heat storage or chemical heat storage.Chemical heat storage has the highest potential for seasonal heat

ll rights reserved.

L, CNRS UMR5008, F-69621

ax: +33 472 438 522.

uznik).

storage (Fig. 1): the storage density is high and there is no heatloss to the environment during storage.

Potential candidates for chemical heat storage are numerousbut some of them have been identified in [4]:

–

MgSO4 and H2O with an energy storage density of 2.8 GJ/m3, – Si and O2 with an energy storage density of 37.9 GJ/m3, – FeO and CO2 with an energy storage density of 2.6 GJ/m3, – FeO and H2O with energy storage density of 2.2 GJ/m3, – CaSO4 and H2O with energy storage density of 1.4 GJ/m3.

Among the potential candidates, magnesium sulphate is themost appropriate because of its high energy density, its storagetemperature compatible with solar collectors, the availability ofthe chemical reaction compounds and its non-toxicity. The sub-ject of this study is to develop and characterise a magnesiumsulphate based composite material for long-term thermal energystorage.

2. Presentation of the thermal energy storage concept

The main principle of the storage system is based on an opensystem that uses the moisture of the ventilation exhaust air from

Fig. 1. Storage density versus physical phenomena involved.

Fig. 2. Presentation of the storage concept: heat release from the store (left) and

heat storage (right).

Fig. 3. Temperature lift versus composite material—hydration (SM16 is a silica

gel/MgSO4 composite material and ZM15 a zeolite/MgSO4 composite material).

S. Hongois et al. / Solar Energy Materials & Solar Cells 95 (2011) 1831–18371832

the building as a source of water for the exothermal hydrationreaction. The heat generated by the reaction is used to heatincoming fresh air via a mechanical heat recovery heat exchanger(balanced ventilation system with heat recovery), which providesheating to the building. A schematic diagram of the heat releaseprocess, typically in winter, is shown in Fig. 2.

The storage of heat in the material, typically in summer, requireshot air between 80 and 150 1C. The higher the temperature, thegreater the amount of heat stored. Air solar collectors with evac-uated tubes are used in order to reach high temperatures and toobtain an optimised operation time with good performance. The hotair from the collectors is used to transfer heat and recharge thethermal energy storage system. The material developed for thestorage system must be compatible with this specific feature.

3. Composite material development

3.1. Heuristic arguments

The thermochemical storage working pair is magnesiumsulphate (MgSO4)/water vapour (H2O). Heat is released/storedby the hydration/dehydration of magnesium sulphate by thefollowing reaction:

MgSO4:7H2O3MgSO4þ7H2O

Magnesium sulphate as fine powder is difficult to use in astorage reactor because it rapidly forms a skin of the hydrated salton the surface of the bulk powder when it reacts with humid air.This skin then forms a barrier to further solid/gas reaction betweenthe moist air and the surface of the solid, which considerably slowsdown the reaction and limits its power. A composite material wastherefore developed using a microporous matrix to offer a high

specific surface for reaction and a high magnesium sulphateimpregnation. A zeolite based matrix in the form of pellets wasused to take advantage of the additional storage capability of thehost matrix by sorption of water.

The temperature lift obtained for two composites tested,zeolite+MgSO4 and silica gel+MgSO4, is shown in Fig. 3. Theobjective of a storage system for usage in buildings is to have atemperature lift of at least 20 1C during the heat release process.The selected candidate for our studies is a mesoporous zeolitematrix with magnesium sulphate.

3.2. Preparation of the composite material

The composite material is prepared by impregnation of mag-nesium sulphate in a zeolite matrix using an aqueous solution ofMgSO4. A 13X zeolite molecular sieve from MERCK is used. Thismaterial is pelletized in beads of 2 mm diameter. The aqueoussolution of MgSO4 is gradually concentrated by successive addi-tions of appropriate quantities of salt until full incorporation ofmagnesium sulphate into the zeolite matrix, or the solution isfiltered after a period of at least 2 h to obtain the desired levelof salt.

The impregnated zeolite is then dried in a furnace at 150 1C inair. At this temperature, the material is only partially dehydrated,since the complete zeolite dehydration occurs above 200 1C.However, this is sufficient for the release of 6 moles of waterfrom MgSO4 �7 H2O. The set point of 150 1C is chosen as thetemperature, which can easily be reached with evacuated tubesolar-thermal collectors. The dehydration step is allowed tocontinue in the oven until the sample mass becomes stable. Bythis method, 50 mg samples of zeolite–MgSO4 composite materi-als have been prepared with MgSO4 content ranging from 10 to25 wt%.

4. Characterisation of the storage composite material

4.1. SEM characterisation of the zeolite/MgSO4 composite material

Scanning electron microscopy (SEM) pictures of zeolite,zeolite–15 wt% MgSO4 (ZM15) and zeolite–25 wt% MgSO4 com-posite (ZM25) are shown in Fig. 4. While observing the imagesfrom (a) to (c), it can be clearly seen that the microporosity of thezeolite pellet is gradually occupied by the MgSO4 salt as the massfraction of MgSO4 increases.

4.2. Thermogravimetric analysis of form-stable zeolite/MgSO4

Thermogravimetric analyses (TGA) were performed on thesamples to determine loss of water with increasing temperature.

Fig. 4. Pictures of (a) zeolite, (b) form-stable zeolite/15 wt% MgSO4 composite and (c) zeolite-20%wtMgSO4 composite.

Fig. 5. Thermogravimetric analysis of pure MgSO4.

Fig. 6. Thermogravimetric analysis of the zeolite/MgSO4 composites.

Fig. 7. Principle of the heat flow calorimetry.

S. Hongois et al. / Solar Energy Materials & Solar Cells 95 (2011) 1831–1837 1833

The TGA of 10 mg of pure magnesium sulphate is shown in Fig. 5.There are three stages of dehydration depending on the tempera-ture: 22, 71 and 261 1C. On comparing the TGA curve with thetemperatures available from the evacuated tube collectors itappears that the optimum temperature of dehydration is about150 1C, which corresponds to a dehydration level of about 89%.

The same TG analysis performed on 10 mg of the zeolite–MgSO4 composite materials shows a less resolved curve with amore continuous dehydration phase. The tested materials arezeolite–MgSO4 with 10, 15 and 20 wt% of MgSO4 (ZM10, ZM15and ZM20, respectively). The successive dehydration steps are nolonger discernable (Fig. 6). In terms of mass percent of waterdesorbed, the behaviours of the composites compared to thezeolite are difficult to separate.

The similarity between the zeolite–MgSO4 materials and the purezeolite TGA curves seems to show that MgSO4 does not behave like asalt hydrate anymore when it is in the zeolite pores. Many hypoth-eses are under investigation, such as a replacement of the charge

balancing cations of the 13� zeolite skeleton (mainly Na+ in the13X type) by Mg2+ ions from magnesium sulphate. This assump-tion needs to be confirmed or refuted by complementary analyses.

A second series of tests with isothermal steps at 150 1C havebeen performed to simulate the (re)charge dehydration phase at aconstant temperature as it takes place in the storage system. Wethus assume the ideal case of a summer day when the inlet airtemperature, which corresponds to the outlet air collector tem-perature, rapidly increases to the level required for the materialdehydration. These dehydration tests at the mg scale with anisothermal step at 150 1C corroborates the TGA results in Fig. 5:a large part of the total energy density can be used even if thematerial is not fully dehydrated, which requires higher tempera-tures (�270 1C). Indeed, 90% of the initial water content ofhydrated MgSO4 and 80% of the initial water content of thehydrated zeolite–MgSO4 materials are lost at Tr150 1C after 2 h.

4.3. Differential scanning calorimeter analysis of form-stable zeolite/

MgSO4

The apparatus is a heat flow calorimeter (Fig. 7). The principleis based on the difference of specific heat capacities between areference material and the sample, which creates a temperaturedifference. The heat flux difference is then calculated using thistemperature difference [5].

The study of the material (re)charge dehydration phase has beenpursued with differential scanning calorimetry (DSC) tests with anisothermal step at 150 1C. These measures were performed onzeolite, MgSO4 and zeolite–MgSO4 materials: ZM10, ZM15 andZM20 to make a comparative study. The deepest endothermic peakcorresponds to the zeolite–15 wt% MgSO4 dehydration (Fig. 8), but

S. Hongois et al. / Solar Energy Materials & Solar Cells 95 (2011) 1831–18371834

the results show low differences between the samples. A quantitativestudy has then been launched with micro-calorimetry experimentsto actually measure the amount of energy released. The study hasfocused on the form-stable composite zeolite–15 wt% MgSO4

(see Section 4.6).

4.4. Porosity measurements

Porosity measurements have been carried out on pure zeolite13X pellets as well as zeolite�25% MgSO4 by analysing theadsorption isotherms of two different gases within the pores ofthe materials:

�

3

TabPor

C

Z

Z

Z

adsorption of nitrogen N2 at 77 K, in order to study the wholeporosity of the samples and

Fig. 8. DSC analysis of the zeolite/MgSO4 composites.

00

50

100

150

200

250

300

P/P0

Vads

(cm

/g S

TP)

0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1

Fig. 9. N2 adsorption isotherm on pure zeolite (a) and ZM25 (b) at 77 K.

le 1osity characteristics of zeolite and ZM25.

haracteristics Isotherm type

N2 isotherm at 77 K

Model

Langmuir

St (m2/g) Vt (cm3/g)a

eolite (reference: sample mass) 957 0.389

M25 (reference: sample mass) 413 0.182

M25 (reference: zeolite mass) 475 0.209

a At P/P0¼0.98.

�

adsorption of carbon dioxide CO2 at 293 K, in order tospecifically study the microporosity of the material.

The composite with 25% MgSO4 was chosen because testsshowed that this concentration is not optimal and the material isoverloaded with the salt hydrate. Both tests have been carried outafter a preparation at 200 1C for 13 h under vacuum. The N2

adsorption isotherms belong to type I according to BET classifica-tion, which accounts for a mainly microporous material, i.e., withpore diameter o2 nm (Fig. 9) [6].

Because of the great affinity of CO2 versus N2 to thesematerials under these experimental conditions and its higherdiffusivity at a temperature level of 293 K, the adsorbed CO2

quantities can be determined with an excellent precision at lowrelative pressure.

The analysis of the N2 and CO2 isotherms can be used toevaluate the specific area, the pore volume and the pore sizedistribution. Total specific area St and pore volume Vt are given bythe Langmuir model, whereas the Dubinin–Asthakov model onlyaccounts for the microporous specific area Sm and pore volume Vm(Table 1). The porosity measurements comparison between zeoliteand ZM25 is relative to the same zeolite mass, since zeolite is thehost material. The analysis of N2 isotherms based on the Langmuirmethod shows that the impregnation of zeolite by magnesiumsulphate causes a decrease of the pore volume of about 46%. Thiscould be caused either by penetration of MgSO4 into the micro-pores, excess MgSO4 blocking the surface of some of the pores orby a breakdown/enlargement of part of the micropores induced bythe impregnation of the salt. The latter could explain why largerpores appear in the magnesium sulphate treated sample. Inaddition 96–97% of the zeolite total specific area is obtained frommicropores in both the pure zeolite and the composite material,but this area has been reduced to a half in the composite material.

The Dubinin–Asthakhov analysis of the CO2 isotherm leads toSm and Vm values superior to St and Vt values evaluated by theLangmuir method, because CO2 is a far more appropriate probethan N2 for microposrosity, specifically if the porosity is nothomogeneous. The porosity measurements can still be comparedwith the same evaluation method. As seen with N2 isotherms, theresults of CO2 adsorption Fig. 10 show a dramatic decrease of themicroporous specific area and porous volume (�33%). Since apart of the microporosity cannot be reached by the N2 molecule,the porosity measurements relative to CO2 adsorption prove to bemore reliable.

The pore size distribution measured with the DFT techniqueconfirmed that the samples are essentially microporous with anaverage pore diameter between 0.8 and 0.9 nm (Fig. 11). For moreaccuracy, the results comparison referred to the pore volume atconstant zeolite mass. The insertion of MgSO4 within the zeolitepores gives rise to a light mesoporosity with a large distribution

CO2 isotherm at 293 K

Dubinin–Asthakhov Dubinin–Asthakhov

Sm (m2/g) Vm (cm3/g)a Sm (m2/g) Vm (cm3/g)a

921 0.374 934 0.374

400 0.142 546 0.219

460 0.163 627 0.252

Pore diameter [nm]

Por

e vo

lum

e [c

m3 .

g zeo

lite]

10 20 30 40

0.01

0.02

0.03

0.04

51

Pore diameter [nm]

Por

e vo

lum

e [c

m3 .

g zeo

lite]

10 20 30 40

0.01

0.02

0.03

51

-1-1

Fig. 11. Pore size distribution of zeolite (a) and ZM25 (b).

Fig. 12. Experimental setup for hydration tests.

0

20

40

60

80

100

120

0.004P/P0

Vads

(cm

3 /g S

TP)

0 0.002 0.0120.010.0080.006

Fig. 10. CO2 adsorption isotherm on pure zeolite (a) and ZM25 (b) at 293 K.

S. Hongois et al. / Solar Energy Materials & Solar Cells 95 (2011) 1831–1837 1835

within the range of 4–40 nm. However, this phenomenon is minorin comparison with the predominance of the microporosity.

4.5. Macro-reactor experiment of form-stable zeolite/MgSO4

Sorption tests have been carried out in a macro-reactor describedin Fig. 12. This reactor is not perfectly insulated, which means that

the temperature rise measured is lower than what could be reachedin a system with better insulation. It is divided into two zones: thefirst contains the sample (200 g of zeolite–MgSO4) whereas thesecond is dedicated to humidity measurements.

The influence of two main controlling parameters was studied:the air mass flow and air relative humidity (at fixed temperatureof 25 1C). At the building scale, the storage system will beintegrated using the humidity in the house as a water vapoursource for the hydration (discharge) reaction. The temperaturelifts and the corresponding release of thermal power as a functionof time are presented in Fig. 13 and 14, respectively.

The maximum temperature lifts and power values are mea-sured for the highest airflow rate available with this setup(8 L/min) and the highest relative humidity (80%). During thesetests, the energy density achieves the maximum value of0.18 Wh/g with an airflow rate of 8 L/min and inlet air at 50%RH. This value corresponds to 45% of the theoretical energydensity of zeolite–MgSO4. The results reveal that the system ishighly sensitive to the humidity of the inlet air. Indeed, thehydration curve of ZM15 with humid air at 20% RH and a flowrate of 8 L/min shows a flat pattern. This behaviour tends toindicate that the inlet air relative humidity must reach a thresh-old value of about 50% in order to get DT¼15 1C in the reactor.

4.6. Micro-calorimetry experiments

Micro-calorimetry analyses were performed using a Calvetapparatus (Fig. 15). The calorimeter is made up of two vessels: aninternal vessel, where the sample is placed, located inside anexternal vessel, and the calorimetric block, maintained at constanttemperature. These two vessels are linked by multiple thermocou-ples, interconnected in series, which guarantee a good thermalcontact and a high precision. When the adsorption occurs in theexperimental cell, the heat released is quickly transferred via thethermocouples to the calorimetric block. The internal vessel tem-perature is not affected by this heat transfer due to the high thermalconductivity and the large size of the calorimetric block. Therefore,the reaction takes place in isothermal conditions [7].

Water vapour adsorption on a 100 mg sample of form-stablecomposite pellets (ZM15) was studied by micro-calorimetry at30 1C. This sample had previously undergone 3 sorption/deso-rption cycles. The sorption occurs in a macro-reactor as presentedin Fig. 12, whereas desorption is made in an oven at 150 1C. Theexperimental adsorption isotherm as a function of the equili-brium vapour pressure is shown in Fig. 16. The curve is typical ofthe adsorption in a microporous material (type I isotherm in theBET classification). The adsorbed mass fraction at equilibrium isclose to 0.15 g/g.

Fig. 13. Temperature lifts in the macro-reactor—influence of the airflow (graph A) and humidity (graph B).

Fig. 14. Power released in the macro-reactor—influence of the airflow (graph C) and humidity (graph D).

Fig. 15. Calvet calorimetric cell (cross-section and top view). Fig. 16. Water vapour isotherm on ZM15 composite material at 30 1C.

S. Hongois et al. / Solar Energy Materials & Solar Cells 95 (2011) 1831–18371836

Other characteristics can be determined by micro-calorimetry,such as differential and integral adsorption heats. An isothermal(30 1C) adsorption integral heat of 4532 J/g of water vapour isachieved, which is nearly twice as high as the latent heat ofvaporisation of water. The resultant energy density reaches0.18 Wh/g of ZM15. This value is very close to the thermal energyreleased during the first hydration at the macro-scale (see Section

4.5.). This result is particularly interesting, since it proves that theheat storage material ZM15 is able to keep high energy densityafter at least 3 charge/discharge cycles.

The volumetric energy density of material is a key character-istic for the designing of a compact thermal energy storagesystem. The energy density of the composite material ZM15 wasmeasured by micro-calorimetry at 166 kWh/m3, which indicates

MgSO4 powder476 kWh.m-3

Zeolite 13X131 kWh.m-3

ZM15257 kWh.m-3

ZM15166 kWh.m-3

Theoretical valuesMeasurement

Total hydration reactionMgSO4 + 7H2O MgSO4.7H2O

Water vapouradsorption on zeolite+ total hydrationof MgSO4

Water vapouradsorption on zeolite

Fig. 17. Energy densities comparison.

S. Hongois et al. / Solar Energy Materials & Solar Cells 95 (2011) 1831–1837 1837

an increase of 27% in comparison with the theoretical energydensity of pure zeolite 13X (Fig. 17).

5. Conclusion

The development and characterisation of a novel zeolite/MgSO4 form-stable composite for low temperature heat storageapplications in buildings were presented in this paper. The MgSO4

component was confined in the microporous structure of thezeolite pellets with an optimum percentage of 15 wt%. With anairflow rate of 8 L/min and inlet air at 50% RH, an energy densityof 0.18 Wh/g (166 kWh/m3) is obtained, i.e. 45% of the theoreticalenergy density of ZM. Micro-calorimetry experiments reveal thatthis energy density level is maintained over three charge/dis-charge cycles. Compared to a water-based storage with a tem-perature lift of 50 1C, which has an energy density of 55 kWh/m3,a ZM15-based system would thus lead to a three times morecompact storage tank. Based on all these results, the form-stablecomposite proves to be a promising thermal energy storagematerial suitable for long-term storage purposes, thanks to itsgood thermal properties, and thermal and chemical reliability.Further research will focus on the optimisation of the hostmaterial properties and the evaluation of the storage system

energy density in real operating conditions in large-scaleexperiments.

Acknowledgements

The authors thank Dr. H. Groult and Dr. E. Briot of the PECSAlaboratory in Jussieu (Paris) for the porosity measurements, aswell as Prof. A. Auroux and Dr. S. Benicci of the IRCE laboratory(Villeurbanne) for the micro-calorimetry measurements.

References

[1] I. Dincer, M.A. Rosen, Thermal Energy Storage—Systems and Applications,John Wiley and Sons, 2002.

[2] Ph Stevens, Le stockage de chaleur in: P. Odru (Ed.), Le stockage de l’energieDunod, ISBN 978-2-10-054805-7, 2010.

[3] French Ministry of Ecology and Sustainable Development, Climate plan 2004:let’s act together to challenge of climate change, Report, 2004.

[4] Thermal energy storage for solar and low energy buildings. Jean-ChristopheHadorn (Ed.), IEA solar heating and cooling Task 32, 2005.

[5] International Organisation for Standardization, Plastics—differential scanningcalorimetry (DSC). part 1: general principles, norm, ISO 11357-1, 2009.

[6] S. Brunauer, P.H. Emmet, E. Teller, Adsorption of gazes in multimolecularlayers, J.Am .Chem. Soc. 60 (1938) 309–319.

[7] Groupe Franc-ais des Zeolites, dir. Frederic Thibault-Starzyk, Les Materiauxmicro et mesoporeux—Caracterisation. EDP Sciences, 2004.