1
2
Nanocomposites derived from silica and carbon for low temperature
3
photothermal conversion
4
5
A. Charlot, O. Bruguier, G. Toquer, A. Grandjean, X. Deschanels
6
ICSM, UMR 5257, BP 17171, 30207 Bagnols-sur-Cèze, France,
7
8
Corresponding author: X. Deschanels, xavier.deschanels@cea.fr, +33 4 66 79 60 87
9
10
Abstract
11
12
Carbon-oxide nanocomposite films have been fabricated using a sol-gel synthesis of
13
hybrid precursors i.e. tetraethyl ortosilicate (TEOS) + methyl-cyclodextrin (MCD) or
14
colloidal zirconia + MCD, followed by carbonization in an inert atmosphere.
15
Reflectance measurements were performed on these samples to determine solar
16
absorptance () and thermal emittance (). The ratio  was used to quantify the
17
selectivity of the samples, the bigger this value the better the selectivity. The
18
selectivity of the films was optimized according to various parameters such as the
19
amount of MCD in the precursor, the thickness of the film or the surface roughness.
20
The best value of was obtained for a TEOS + MCD film coated on a copper
21
substrate.
22
23
Keywords: Selective solar absorber, Nanocomposites, Sol-gel preparation, Solar
24
absorptance, IR emissivity
25
1
26
1. Introduction
27
28
The basic principle of photothermal conversion is to transfer the collected incident
29
solar energy to a heat-transfer fluid through an absorber. However, part of the
30
incident radiation is lost by reflection at the surface of the absorber, which constitutes
31
the optical losses. In an effort to minimize the optical losses, spectrally selective
32
absorber surfaces are widely used for the photothermal conversion of solar energy.
33
These surfaces are characterized by high solar absorptance (low reflectance) in the
34
visible and near-infrared regions (0.3-2.5µm), and a low emittance (high reflectance)
35
in the long wavelength (infrared) region (2.5-20µm). No intrinsic material is capable of
36
this selectivity. It can however be achieved by the use of a combination of various
37
compounds which tailor the optical properties of the surface [1-4]. A solution
38
implemented to achieve this goal is to make thin solar absorbing coatings on a
39
reflecting metallic substrate. Many examples [5] deal with this topic, leading to an
40
important set of solution. Sputtering and electrolysis of environmentally hazardous
41
elements such as Cr or Ni were used for the fabrication of these selective films. Many
42
studies [6-12] have been carried out to try to obtain selective coatings by green
43
chemistry processes (sol-gel) using eco-friendly and low cost compounds (silica,
44
carbon). To our knowledge, such work has not lead to any industrial applications.
45
In this present work, selective surfaces were obtained by a process inspired by the
46
work of Mastai [11]. It is based on coating a metallic substrate (Cu, Al) with a film
47
containing “nano carbon” grains embedded in an oxide matrix. Our purpose was to
48
modify the nature of the oxide matrix to improve the selectivity of the coatings. For
49
that, silica matrix was partially or completely substituted by zirconia. This film is
50
obtained by the carbonization of a hybrid precursor (oxide + methyl-cyclodextrin)
2
51
obtained using a sol-gel process. In order to optimize the selectivity of the coating,
52
various parameters such as the carbohydrate concentration, the thickness of the
53
deposited layer and the nature of the oxides used (i.e. silica, zirconia or a mixture
54
silica + zirconia) were studied. The evolution of the optical properties of the coatings
55
on the efficiency of the solar collector is also discussed.
56
57
2. Experimental and measurements
58
59
2.1. Substrate Preparation
60
61
Smooth aluminum or copper plates of 25 x 50 mm2 were used as substrates. These
62
substrates were cleaned by electrolysis treatment at 5V for 1min in a commercial
63
solution of Uniclean® 248 concentrated at 10% wt. A platinum electrode was used as
64
the cathode. In a second step, the substrate is dipped in a sulphuric acid bath (10
65
%wt.) for 2 minutes, to remove the oxide layer. Lastly, the substrate is rinsed
66
thoroughly in distilled water and then dried by blowing with nitrogen gas.
67
68
2.2. Nanocoatings preparation
69
70
Different hybrid precursors (oxide + methylcyclodextrin) based on silica oxide,
71
zirconia oxide or a mixture of silica + zirconia oxide, were obtained by a sol-gel
72
process. Briefly, a stock solution was first prepared using 8,815 g TEOS, 5 g EtOH
73
and 6,285 g H2O pH 2 (HCl). This solution was then mechanically shaken for 1 hour.
74
Simultaneously, methylβ-cyclodextrin was dissolved in 25 g EtOH, and to which 3,63
75
g of the solution stock was added. The resulting solution (S1) was stirred
3
76
continuously at room temperature for 2 hours. The metallic substrate was dip-coated
77
onto the solution S1 to obtain the MCD + TEOS precursor samples.
78
Acetate stabilized colloidal zirconia was purchased from Nyacol® (particle size 10
79
nm). The addition of these colloidal particles to the solution S1 was done to obtain
80
the slurry S2. Acid acetic (pH = 3) was added to the slurry S2 to correct its viscosity.
81
The slurry S2 was sonicated to prevent the agglomeration of the particles. The
82
metallic substrate was dip-coated onto the slurry S2 to obtain the MCD + colloidal
83
ZrO2 and MCD + a mixture of TEOS and colloidal ZrO2 samples.
84
The dip-coating was performed initially at rates of 1, 2, 6, mm/s. An optimum speed
85
of 2 ± 0.3 mm/s was used for the rest of the experiment. The coating was performed
86
in a glove box at 50 % humidity. The samples were dried under vacuum for 12 hours
87
and were placed in a furnace to carbonize the methyl-cyclodextrin. The
88
carbonization process was carried out under flowing argon at temperatures ranging
89
from 400 to 540 °C for 1 hour. The furnace was raised to this temperature at
90
1°C.min−1, and then cooled to 30 °C at 20 °C min−1.
91
In the work presented below, the samples referred as C/SiO2, C/ZrO2, C/SiO2 + ZrO2
92
were respectively prepared from MCD + TEOS precursor, MCD + colloidal ZrO2
93
and MCD + a mixture of TEOS and colloidal zirconia and carbonized at 500 °C
94
under argon atmosphere.
95
96
2.3. Characterization methods
97
98
The hemispherical reflectance R of the samples was measured in the wavelength
99
range of 0.28 –16 µm. A Shimadzu UV-3600 UV-VIS-NIR spectrophotometer
100
equipped with an integrating sphere of 50 mm diameter, circular beam entrance and
4
101
sample ports of 20 mm diameter, was used in the wavelength range of 0.28 – 2,5
102
µm. The infrared wavelength range of 2.5 – 16 µm was covered with a Perkin Elmer
103
Spectrum 100 FTIR spectrophotometer with an integrating sphere of 80 mm diameter
104
and a circular beam entrance and sample port of 20 mm diameter.
105

α
2.5
0.28
106
Isol    1  R  dλ
2.5

0.28
Isol   dλ
(1)
107


16
2.5
108
IP   1  R  d

16
2.5
Ip   d
(2)
109
110
The measurements were combined to create one spectrum and the absorptance 
111
and emittance  were calculated using formulas Eq. (1) and Eq. (2), where R is the
112
total reflectance, Isol the solar spectrum (A.M1.5) and Ip is the Planck black body
113
power system. The comparison of the selectivity of the surface is based on the 
114
ratio (: the solar absorptance,  the thermal emittance), the higher the value of 
115
the better the selectivity. Typical spectra of C/SiO2 or C/ZrO2 films deposited on
116
aluminum substrate are presented on Fig.1. The comparison of these measurements
117
with those obtained on a metallic substrate of aluminum highlights the selectivity
118
achieved by the deposition of a carbon + silica or carbon + zirconia layer.
119
Characteristic chemical bonds of silica were identified from the reflectance spectrum
120
(see Fig. 1).
5
121
SEM analyses were obtained with a FEI Quanta FEG 200 ESEM environmental
122
scanning electron microscope coupled with EDX Bruker XFlash 4010 SDD
123
microanalyser. Thicknesses were evaluated with X-film ® software provided by
124
Bruker Company. It enables to measure the thickness of a film from EDX analyses.
125
For three primary energies E0 (5, 10 and 15kV), the k-ratio of the elements were
126
measured with the same current beam and acquisition time. The values of k-ratios
127
and the composition of the analyzed layer were used as input for the X-film ®
128
software to calculate the thickness.
129
The conversion of the MCD + TEOS or MCD + colloidal ZrO2 precursors was
130
carried out in a tubular furnace and in the TGA apparatus. A flow of argon over 30
131
L/h was used in the tubular furnace to evacuate the gaseous species produced
132
during the thermal treatment. The thermo gravimetric analysis (TGA) was performed
133
using a Setaram Setsys analyser. Approximately, 50 mg of MCD + TEOS precursor
134
was placed in an open 120 µl alumina pan and heated at a heating rate of 10 ° C.min-
135
1
to 850 ° C under argon flowing gas (20 mL.min-1).
136
137
Insert Figure 1
138
139
Insert table 1
140
141
4. Results and discussion
142
143
Fig. 2 shows a typical TGA curve of TEOS + MCD precursor. The weight loss
144
observed on this curve can clearly be attributed to the decomposition of the MCD
145
precursor into carbon composite particles. This decomposition is achieved at a
6
146
temperature of about 500°C. Preliminary measurements of the absorptance indicate
147
that this property increases with the temperature of carbonization of the precursor.
148
However, for temperatures around 500 °C, the aluminum substrate becomes soft
149
(Tmelting = 660°C). In order to optimize the ideal temperature of the heat treatment,
150
absorptance measurements have been performed for samples which have been heat
151
treated at temperatures ranging from 400 to 540 °C. Therefore, a compromise
152
between these two factors was obtained at 500 ° C. The same heat treatment was
153
chosen for the carbonization of MCD + colloidal ZrO2 precursor.
154
155
Insert Figure 2
156
157
Fig. 3 displays the evolution of the ratio for the various samples versus the
158
amount of the MCD in the precursors. The comparison between the ratio, for
159
metallic substrates (Al or Cu) and the value obtained after the film deposition,
160
demonstrates clearly the efficiency of the implemented process. The ratio
161
decreases with the amount of MCD in the precursor; this is due to an increase in the
162
emittance of the samples. Increasing the MCD concentration in the precursor led to
163
an increased amount of carbon nanoparticles in the film. EDX microanalysis shown in
164
Fig. 5e confirms this result. In addition, SEM observations of the samples with the
165
highest MCD content (Fig. 5c and 5d) clearly show the presence of carbon
166
nanoclusters, with a characteristic size range between 20 and 150 nm. Also, the
167
thickness of the layer deposited is increased; as shown in Fig.4. Therefore,
168
unfortunately, the emissivity of the film increases because the reflectance of the
169
metallic substrate is lower. EDX microanalysis also revealed the presence of sodium
170
as impurity which arises from the substrate cleaning by the commercial solution
7
171
Uniclean®. A possible route for the optimization of this ratio would be to further
172
reduce the concentration of MCD in the precursors. Work is in progress with this
173
purpose in mind.
174
175
Insert Figure 3
176
177
Multilayered samples prepared by successive dipping of the metallic substrate in the
178
hybrid precursor (S1 or S2) were also assessed. For the same reason discussed
179
previously this method leads to a decrease in the ratio with the number of layers
180
deposited (Fig. 4). Finally, the highest value of the  ratio is obtained for samples
181
originating from TEOS + MCD precursor deposited onto copper substrate.
182
183
Insert Figure 4
184
185
From a microstructural point of view, the presence of carbon nanoclusters in the
186
samples with the highest MCD content has been previously discussed. For the
187
samples with low carbon content, the surface of the silica based sample (Fig. 5a) was
188
smooth whereas samples which were prepared with zirconia were rough (Fig. 5d).
189
Moreover, the C/ZrO2 sample had numerous cracks while the C/SiO2 samples were
190
crack-free. The average grain size observed on the C/ZrO2 samples is around 40 nm.
191
This value is higher than the initial grain size of the colloidal zirconia (10 nm). The
192
crystallite growth observed can be induced by the heat treatment used for the
193
carbonization precursor. A slight increase in the ratio was observed for the
194
samples based on C/SiO2 + ZrO2 compared to C/SiO2 (Fig. 3). This result could be a
8
195
consequence of their roughness; several authors [2, 5] reported that roughened
196
surfaces are favorable for obtaining good selectivity.
197
198
Insert Figure 5
199
200
As it has been mentioned before, the surface selectivity leads to minimization of the
201
optical losses. The simplified heat balance of a solar collector is therefore given by
202
the following expression Eq. (3) :
203
204


Qu  Sc    ES  Sa      Ta4  T04  Sa  H  Ta  T0 
(3)
205
206
Where QU is the useful energy, Sc the collector surface, Sa the absorber surface, H
207
the coefficient of thermal losses, Es the solar radiation flux, Ta the temperature of the
208
absorber, T0 the temperature of the environment,  the constant of Stefan-
209
Boltzmann,  the absorptance and  the emittance. If one considers only the optical
210
losses, i.e. the heat losses by conduction and convection are neglected, the
211
efficiency  of the collector has the following expression Eq. (4) :
212
213


Sc
 ES
Sa

   Ta4  T04

(4)
214
215
The conversion efficiency () of the collector for T0 = 20 ° C and Ta = 100 and 200 °
216
C is shown as a function of the thickness of the coating on Fig. 6. The evolution of 
9
217
indicates that it is desirable to use a selective surface when the absorber temperature
218
is sufficiently high, typically 200 ° C. Under these conditions, the best selective
219
surface is obtained for a coating of C/SiO2 on a copper substrate, corresponding to a
220
thickness of about 700 nm. It is interesting to note that this value is not the highest
221
value obtained for the  ratio. This means that the effect of the absorptance is more
222
important than the emittance on the efficiency of the collector.
223
224
Insert Figure 6
225
226
5. Conclusion
227
228
Thin films based on carbon-oxide nanocomposites with good optical properties for
229
selective solar absorbers were fabricated by a sol-gel process involving low cost
230
compounds like TEOS, colloidal zirconia and methyl-cyclodextrin. The optimum
231
selectivity = 32 was achieved in a C/SiO2 sample deposited onto a copper
232
substrate. A possible route for the improvement of the properties of these coatings
233
would be to reduce the methyl-cyclodextrin content in the precursor. The optical
234
characteristics of the samples were mainly depend on the thickness of the films and
235
were not influenced by the nature of the deposited layer or C/ZrO 2 C/SiO2 although
236
this parameter strongly modifies their surface. Further studies on durability in
237
dependence on environmental conditions have to be performed to evaluate precisely
238
the potential of such composite coatings for industrial applications.
239
240
241
10
242
243
6. Acknowledgments
244
245
The authors thank the CEA “DSM Energie” project for funding this study and Dr.
246
Micheal Moloney for his help during the preparation of this paper.
247
248
7. References
249
250
[1] C.G. Granqvist, Physics Teacher 22/6 (1984) 372.
251
[2] C.G. Granqvist, Adv. Mater. 15/21 (2003) 1789.
252
[3] C.G. Granqvist, Solar Energy Mater. & Solar Cells 91/17 (2007) 1529.
253
[4] J. Spitz, D. Maziere-Bezes, J. Optics 15/5 (1984) 325.
254
[5] W.F. Bogaerts, C.M. Lampert, J. Mater. Sc. 18/10 (1983) 2847.
255
[6] T. Bostrom, G. Westin, E. Wackelgard, Solar Energy Mater. & Solar Cells 91/1
(2007) 38.
256
257
[7] L. Kaluza, B. Orel, G. Drazic, M. Kohl, Solar Energy Mater. & Solar Cells 70/2
(2001) 187.
258
259
[8] G. Katumba, J. Lu, L. Olumekor, G. Westin, E. Wäckelgard, Jour. Sol-Gel Sci.
Technol. 36 (2005) 33.
260
261
[9] G. Katumba, L. Olumekor, A. Forbes, G. Makiwa, B. Mw<akikunga, J. Lu, E.
Wäckelgard, Solar Energy Mater. & Solar Cells 92/10 (2008) 1285.
262
263
[10] D. Katzen, E. Levy, Y. Mastai, Appl. Surf. Sc. 248/1-4 (2005) 514.
264
[11] Y. Mastai, S. Polarz, M. Antonietti, Adv. Funct. Mater. 12/3 (2002) 197.
265
[12] K.T. Roro, N. Tile, A. Forbes, Appl. Surf. Sc. 258/18 (2012) 7174.
266
267
11
268
List of figures
269
270
Figure 1: Hemispherical reflectance spectrum of C/SiO2, C/ZrO2 nanocomposite
271
deposited on aluminum substrates and C/SiO2 deposited on copper substrate
272
compared to the values obtained for aluminum and copper. The chemical bonds of
273
silica are identified in the absorber composite layer Al-C/SiO2. ML refers to the
274
deposition of multilayers onto the sample.
275
276
Figure 2: TGA curve in argon flow of TEOS + MCD precursor. The parameter mf/mi
277
represents the ratio between the final mass of the sample divided to its initial mass.
278
279
Figure 3: Graph showing the evolution of the quality ratio  versus the amount of
280
the MCD in the precursor for C/SiO2, C/ZrO2, C/SiO2+ZrO2 deposited on Al
281
substrate and C/SiO2 deposited on Cu substrate. The value of the ratio  for Al and
282
Cu metallic substrate is indicated for comparison.
283
284
Figure 4: Evolution of the ratio  versus thickness of the deposit film onto the
285
metallic substrate. ML refers to the deposition of multilayers onto the sample.
286
287
Figure 5: SEM images and EDX microanalysis of the various coatings deposited on
288
aluminum substrates (a) C / SiO2 (14 wt % MCD), (b) C / SiO2 (32.9 wt % MCD),
289
(c) C / SiO2 + ZrO2 (38.7 wt % MCD), (d) C / ZrO2 (12.5 wt % MCD), (e) EDX
290
microanalysis of the samples (a) (b) (c). The spectra are normalized to the aluminum
291
peak.
292
12
293
Figure 6: Efficiency of a photothermal collector based on C/SiO 2 deposited on Cu
294
substrate versus the thickness of the coating at Ta = 100 ° C and 200 ° C.
295
296
297
298
List of Table
299
300
Table 1: Characteristics of the samples investigated in this work.
301
13