Journal of Nanoparticle Research 3: 385–400, 2001. © 2001 Kluwer Academic Publishers. Printed in the Netherlands.
Copper and copper oxide nanoparticle formation by chemical vapor nucleation from copper (II) acetylacetonate Albert G. Nasibulin1 , P. Petri Ahonen1 , Olivier Richard1 , Esko I. Kauppinen1,∗ and Igor S. Altman2 VTT Chemical Technology, Aerosol Technology Group, P.O. Box 1401, FIN-02044 VTT, Finland; 2 Institute of Combustion & Advanced Technologies, Odessa National University, Dvoryanskaya 2, Odessa, 65026, Ukraine; ∗ Author for correspondence (Tel.: +358 9 456 6165; Fax: +358 9 456 7021; E-mail:
[email protected])
1
Received 22 January 2001; accepted in revised form 20 June 2001
Key words: Cu, Cu2 O, nanoparticle, copper acetylacetonate, thermal vapor decomposition, chemical nucleation
Abstract Crystalline nanometer-size copper and copper (I) oxide particle formation was studied by thermal decomposition of copper acetylacetonate Cu(acac)2 vapor using a vertical flow reactor at ambient nitrogen pressure. The experiments were performed in the precursor vapor pressure range of Pprec = 0.06 to 44 Pa at furnace temperatures of 431.5◦ C, 596.0◦ C, and 705.0◦ C. Agglomerates of primary particles were formed at Pprec > 0.1 Pa at all temperatures. At 431.5◦ C the number mean size of the primary particles increased from Dp = 3.7 nm (with geometric standard deviation σg = 1.42) to Dp = 7.2 nm (σg = 1.33) with the increasing precursor vapor particle pressure from 1.8 to 16 Pa. At 705.0◦ C the primary particle size decreased from Dp = 24.0 nm (σg = 1.57) to Dp = 7.6 nm (σg = 1.54), respectively. At furnace temperatures of 431.5◦ C and 596.0◦ C only crystalline copper particles were produced. At 705.0◦ C the crystalline product of the decomposition depended on the precursor vapor pressure: copper particles were formed at Pprec > 10 Pa, copper (I) oxide at Pprec ≤ 1 Pa, and a mixture of the metal and its oxide at intermediate vapor pressures. A kinetic restriction on copper particle growth was shown, which leads to the main role of Cu2 molecule participation in the particle formation. The formation of copper (I) oxide particles occurs due to the surface reaction of the decomposition products (mainly carbon dioxide). For the explanation of the experimental results, a model is proposed to build a semiempirical phase diagram of the precursor decomposition products.
Introduction Copper and copper oxide particles are of significant technological interest. Applications for copper powder include bronze self-lubricating bearings, conductive epoxys, metal-bonded abrasive wheels and cutting tools, and braking systems. Ultra-fine copper particles are a base for developing technologies such as metal injection molding as well as for electronics, ceramics and for thick/thin-film applications. Copper oxides have applications in thin-film oxygen pressure sensors, as a binder in pastes for thick-film microelectronic
circuits, as a p-type semiconductor and they exhibit luminescence (Majumdar et al., 1996; Holzschuh & Suhr, 1990). The importance of producing copper and copper oxide particles is exemplified by applications such as high surface area catalysts that are used in diverse chemical processes, for example, the water-gas shift reaction (Campbell et al., 1987), the butanol dehydrogenation reaction (Shiau & Tsai, 1997) and the carbon monoxide oxidation (Van der Meijden, 1981; Du et al., 1997). Copper-based catalysts are used as a key intermediate in the industrial synthesis of methanol
386 (Klier, 1982; Campbell et al., 1987; Yurieva et al., 1993; Klenov et al., 1998), which is promising as an environmentally friendly fuel for the power industry. In addition, copper and copper-based materials have applications as catalysts in traditional and new organic syntheses, for example, the condensation of aromatic halides, known as Ullmann reaction (Dhas et al., 1998), synthesis of cyclic amines from aminoalcohols and their alkyl derivatives (Hammerschmidt et al., 1986; Kijenski et al., 1989), synthesis of methylamines (Gredig et al., 1997), thermal cracking of plastics and many others (Kijenski et al., 1984; Runeberg et al., 1985; Vultier et al., 1987; Pereia et al., 1994; Shannon et al., 1996). Copper has been also identified as a good catalyst for the combustion of methane (Tijburg, 1989) and selective oxidation of hydrocarbons (Adams & Jennings, 1964; Voge & Adams, 1967). Thus, studies of the formation of copper and copper oxide particles are important. Accordingly, it has been a subject of much research during the last decades. Copper and copper oxide particle formation have been studied by solution reaction: reduction of copper (II) acetate in ethanol (Ayaappan et al., 1997) and in water and 2-ethoxyethanol using hydrazine (Huang et al., 1997), the reaction of copper (II) chloride with organolithium compounds (Takahashi et al., 1988), reduction of copper dodecylsulfate by sodium borohydrate (Lisieski et al., 1996), by using reverse micelles (Lisieski & Pileni, 1993; 1995), thermal and sonochemical reductions of copper (II) hydrazine carboxylate (Dhas et al., 1998), chemical deposition in two-phase system octane–water (Vorobyova et al., 1997), by means of electrolysis (Folmanis & Uglov, 1991; Kirchheim et al., 1991; Pietrikova & Kapusanska, 1991) and others (Herley et al., 1989). Ding et al. (1996) prepared copper nanoparticles by a mechanochemical process and studied the influence of milling conditions on particle structure and size. Nanocrystalline copper was prepared by consolidation of mechanically milled powder by Weins et al. (1997). Copper oxide formation was studied via oxidation of copper particles (Kaito et al., 1973; 1993; Kellerson et al., 1995). Much work has been devoted to the investigation of aerosol formation of copper or its oxide particles by physical methods, including molecular beams (Bowles et al., 1981), direct laser vaporization (Moini and Eyler, 1988), by using gas evaporation (Kashu et al., 1974;
Peoples et al., 1988; Xu et al., 1992), sputtering (Haas & Birringer, 1992), melting in a cryogenic liquid (Champion & Bigot, 1996) and others (Long & Petford-Long, 1986; Girardin & Maurer, 1990; Bouland et al., 1992). Another and a very attractive way to obtain aerosol particles is via the chemical route. This method might be the least expensive for aerosol particle formation under controlled conditions. Little work has been devoted to copper and its oxide formation by chemical vapor nucleation. Majumdar et al. (1996) generated CuO powder by spray-pyrolysis from Cu(NO3 )2 solution. Daroczi et al. (1998) studied the production of copper and iron nanocomposites by thermal decomposition of copper–ferrocyanide in an open vertical tube. The obvious advantage of the chemical vapor nucleation method is the possibility to produce nanosized particles at relatively low temperatures and ambient pressure. The current work is devoted to the investigation of copper and copper oxide particle formation from metal–organic compound, copper (II) acetylacetonate (Cu(acac)2 ). A suitable equilibrium vapor pressure (P = 13.1 Pa at t = 150.0◦ C) and a convenient decomposition temperature (tdec = 286◦ C) were the reasons for the choice of this precursor. The selection of this metal–organic substance was also based on its popularity as a precursor for chemical vapor deposition processes (e.g., Pelletier et al., 1991; Pauleau & Fasasi, 1991; Gerfin et al., 1993; Marzouk et al., 1994; Hammadi et al., 1995; Maruyama & Shirai, 1995) and on the knowledge of decomposition reaction kinetics (Tsyganova et al., 1992). The goals of the investigation are to produce copper and/or copper oxide nanoparticles at ambient pressure and at a temperature as low as possible, to characterize the obtained nanoparticles synthesized with various reactor conditions, and to discuss nanoparticle formation based on experimental results.
Experimental Materials For this study copper (II) acetylacetonate, Cu(acac)2 , (Aldrich Chemical Company, 97%) has been used as a precursor. The decomposition of Cu(acac)2 vapor leads to the formation of copper vapor that is supersaturated at the experimental conditions. The Cu(acac)2
387 decomposition reaction can be presented as CH3
CH3 O
HC
O Cu
O CH3
∆T
CH O CH3
O Cu + CuO + H2C CH2(CO) + CO2 + H3C C CH3 36.0 22.3 38.0 O + H2O + H3C C C2H5 , 3.2 0.3
(1)
where numbers below the decomposition reaction products indicate molar percentage fraction of the gaseous reaction products which were measured using mass-spectrometry analysis by Tsyganova et al. (1992). Such reaction products indicate significant destruction of the ligand, acetylacetone, which is possibly formed on the initial stage of the thermal decomposition. The Cu(acac)2 vapor decomposition was studied using the manometric method in static conditions in clear ampoules at 290–335◦ C and at initial pressures of 98–173 hPa (Tsyganova et al., 1992). In our study two sources of nitrogen (AGA, 99.9 vol.% and 99.999 vol.%) have been used as a carrier gases. Inert nitrogen was used in order to prevent the additional reoxidation of the formed copper vapor by atmospheric oxygen. The compaction of Cu(acac)2 powder in a saturator and hence, blocking the flow were prevented by mixing the powder with inert chromatographic carrier, silicon dioxide, SiO2 (Balzers Materials, 99.9%) with a grain size of 0.2–0.7 mm. Experimental methods A vertical laminar flow reactor for experimental investigation of Cu(acac)2 decomposition under controlled conditions was designed and constructed (Figure 1). The experimental device consisted of a saturator, a laminator, and a furnace. The saturator and the laminator consisted of a stainless steel tube with an internal diameter of 22 mm. A removable cartridge, to hold the mixture of Cu(acac)2 powder and the chromatographic carrier, was inserted inside the tube. An absolute filter (Munktell, MK 360) was used to clean the vapor–gas flow downstream of the saturator. The saturation mixture and the filter were retained on a stainless steel net. In order to laminarize the flow and to avoid turbulence
of the flow proceeding from small diameter to large one after the saturator a laminator has been used. It was constructed as a cylindrical cone in the junction part. Between the laminator and the furnace, a Teflon thermoinsulator was used. A ceramic tube, with external and internal diameters of 28 and 22 mm, respectively, inserted inside the furnace (Entech, Sweden) has been used as a reactor. The flow of pure filtered nitrogen carrier gas was supplied from a high-pressure cylinder to the saturator. Then the gas passed through the heated Cu(acac)2 powder and the vapor saturation by the precursor was reached. Inside the laminator, a steady state laminar flow was established. Then the vapor–gas mixture entered to the furnace where the temperature is maintained higher than the Cu(acac)2 decomposition temperature. The formation of supersaturated copper vapor, as a result of the precursor decomposition reaction, led to the nucleation process and further growth of particles via condensation, coalescence, and agglomeration processes. The flow rate of the gas-carrier was measured by a flow meter (DC-2, BIOS) and was referred to the standard condition (t = 25◦ C, P = 101325 Pa). Temperatures were measured by nichrome–nickel thermocouples (K-type) which had been calibrated with an accuracy of 0.1◦ C by using an oil bath and thermoresistors calibrated against the Finnish National Standard. The aerosol number size distributions in the range of 3–200 nm were measured by a differential mobility analyzer (DMA) system consisting of a charger, a classifier (Winklmayr et al., 1991, modified Hauke, length of 11 cm), a condensation particle counter (CPC, TSI 3027), and a supporting software. A sheath flow rate for DMA system was maintained at Qsh = 14.5 lpm. The morphology, the primary particle size, and the crystallinity of the particles were investigated with a field emission scanning electron microscope (Leo Gemini DSM982) and a field emission transmission electron microscope (Philips CM200 FEG), respectively. An electrostatic precipitator (Combination electrostatic precipitator, InTox Products, Albuquerque, NM, USA) was used to collect the aerosol particles on a carbon-coated copper grid (SPI Holey Carbon Grid). Electron diffraction patterns of the particles were used for determination of the crystalline phase. The samples for X-ray diffraction (XRD, Philips MPD 1880 powder X-ray diffractometer) spectrometry were collected on silver filter disks (Millpore AG4502500, 45 µm pore) and studied with Cu Kα (λ = 0.154 nm)
388
Figure 1. Schematic presentation of the experimental setup.
radiation. A Mettler Toledo TA8000 system equipped with TGA850 thermobalance was used for thermogravimetric analysis (TGA) of the samples under flowing nitrogen atmosphere with the heating rate of 10◦ C/min for sample sizes of approximately 3 mg. Inside the reactor a known temperature gradient was maintained, which gives the possibility to determine the location where the decomposition of the precursor occurs. Tsyganova et al. (1992) reported that the Cu(acac)2 vapor decomposition was a first-order rate reaction. The rate constant of this reaction can be calculated by using the Arrhenius equation k = k0 exp (−Ea /RT ), where the pre-exponential factor is k0 = 3.02×107 s−1 , the activation energy is Ea = 115.4 kJ/mole, and R is the universal gas constant. It is obvious that most of the copper acetylacetonate vapor
thermolysis occurs in the vicinity of the highest temperature zone in the furnace. In Figure 2 the dependence of temperature and the rate constant of Cu(acac)2 decomposition reaction inside the reactor are shown. On the basis of data presented in Figure 2, the decomposition of the metal–organic precursor occurs at location x = 255 ± 12 mm, and at the temperature of t = 431.5 ± 0.5◦ C. Determination of Cu(acac)2 vapor concentration It is known that the vaporization rate of the solid precursor changes with time due to decreasing surface area of the powder (Chou & Tsai, 1994;
389
Figure 2. The dependence of temperature and the rate constant of Cu(acac)2 decomposition reaction inside the reactor.
Figure 3. The dependence of the evaporation rate at different conditions.
Kodas & Hampden-Smith, 1999). Certainly, the vapor pressure of the precursor is a crucial factor in the decomposition reaction. It can be a reason for the irreproducibility of the experimental results and, as it will be found out later, it is even the reason for a change of decomposition reaction products. Therefore, studies of gas flow saturation by Cu(acac)2 vapor are necessary. The usage of the removable cartridge (Figure 1) allows determining of the saturation degree of the carrier gas flow by Cu(acac)2 vapor. If the flow rate of the gas and the mass difference of the cartridge with
Cu(acac)2 powder in a certain time interval are known, it is possible to calculate the evaporation rate and hence, the vapor pressure of the metal–organic compound at the entrance of the reactor. From Figure 3 it can be seen that the evaporation rate decreases with experimental time. The gas saturation by Cu(acac)2 vapor decreases very rapidly at a saturator temperature of tsat = 190.0◦ C and at a flow rate of Q = 2000 cm3 /min. Because of high gas velocity, the saturation time is too short to heat the carrier gas in the upper part of the saturator. The observation of saturator mixture color showed that the
390 precursor was consumed in the lower part where gas temperature reached the saturator temperature. At the flow rate of Q = 330 cm3 /min and at tsat = 170◦ C, the evaporation decreases less rapidly and depends on the decrease of the powder surface area. Decreasing the saturator temperature down to 150◦ C leads to the significant decrease of the evaporation rate dependence on time. Four days of continuous operation using the flow rate of Q = 330 cm3 /min revealed that a significant change in the results (concentration and number size distribution of agglomerated particles) was found only after the first 24 h period. The effect of the composition of the copper acetylacetonate and silicon dioxide mixture on the evaporation rate was also investigated. Variation of the mass ratio of Cu(acac)2 and SiO2 from 1 : 3 to 1 : 150 did not significantly affect on the precursor vapor pressure at the entrance to the reactor. The influence of using a small precursor fraction (1 : 150 relation) was found only after about 7 h because of the exhaustion of Cu(acac)2 material, while duration of an experiment was only about 30–40 min. During the following experiments only newly prepared powder mixtures consisting of 4 g (1.53×10−2 mole) of copper (II) acetylacetonate and 16 g (0.27 mole) of silicon dioxide have been used as the saturator mixture. The dependence of vapor pressure, as determined from the cartridge mass difference, on the saturator temperature at Q = 330 cm3 /min is presented in Figure 4. Also, a fitted curve and literature data of
Cu(acac)2 equilibrium vapor pressure (Teghil et al., 1981; Tonneau et al., 1995) are shown. From the figure, a difference between the literature data and our results can be seen. It is necessary to note that the measured saturator temperature does not coincide with the powder temperature inside the saturator and has been used only as a reference temperature during the experiments. The main crucial experimental parameter is the precursor concentration, which has been determined after each experiment. Thus, during our experiments, wellcontrolled parameters of the setup regarding the contents of vapor–gas phase have been maintained. Results and discussion Experimental results Experiments on particle formation were carried out at two fixed temperature profiles inside the furnace. One of these profiles is shown in Figure 2. Experiments were also performed with a similar temperature profile with a temperature maximum at t = 705.0 ± 0.5◦ C, where this condition was valid at x = 255 ± 15 mm. In the following section, we refer to those temperature profiles as 431.5◦ C and 705.0◦ C. First, experiments on the influence of residence time (flow rate) on particle number size distributions were carried out. At the furnace temperature of tfurn = 431.5◦ C, at flow rates higher than 400 cm3 /min,
Figure 4. The dependence of the precursor vapor pressure on the saturator temperature.
391
Figure 5. Number particle size distributions at the different flow rates.
a bimodal particle size distribution was measured (Figure 5). SEM and TEM micrographs showed two kinds of particles – small agglomerated copper particles and needle-like particles of Cu(acac)2 . Thus, the mode with the smaller mean size is connected to copper particle formation and the second one is due to nucleation of undecomposed Cu(acac)2 vapors downstream of the furnace. The variation of the flow rates at the furnace temperature of 705.0◦ C showed that the optimum flow rate was about 330 cm3 /min. The choice of this flow rate was determined by the concentration range of the condensation particle counter (1–105 particles/cm3 ) and by reasonable time of a sample collection for XRD analysis. Hereafter, only experimental results obtained at the flow rate of Q = 330 cm3 /min are presented. TEM micrographs of the nanoparticles synthesized at reactor temperature of 431.5◦ C and at the precursor vapor pressures of Pprec = 1.8 and Pprec = 16 Pa are presented in Figure 6. As one can see, the primary aerosol particles are fairly monodisperse and the size of the particles increases with increasing saturator temperature. In the same figure, the electron diffraction patterns of the agglomerated particles are presented. The electron diffraction ring pattern simulations performed for copper fits with the experimental results. These results have been confirmed by XRD analysis. TGA showed that at the furnace temperature of 431.5◦ C the decomposition of the precursor was not complete.
At the furnace temperature of 431.5◦ C only 20% of Cu(acac)2 was thermolyzed. The remaining undecomposed Cu(acac)2 forms a layer around copper particles that can be seen on the TEM images (Figure 6). It is worth noting that the electron diffraction ring pattern enclosed in Figure 6(a) is not so clear as the others due to the large amount of amorphous unreacted precursor. Microdiffraction patterns for this experimental condition have been also performed in order to determine unambiguously that the particles presented in Figure 6(a) are crystalline copper. TEM results for the nanoparticles synthesized at 705.0◦ C and at the precursor vapor pressures of Pprec = 1.8 and Pprec = 16 Pa are presented in Figure 7. With these conditions, the size of the primary particles decreases with increasing the vapor pressure of the precursor, that is, there is an inverse situation compared to the preceding samples synthesized at the furnace temperature of tfurn = 431.5◦ C. Electron diffraction patterns of the agglomerated particles produced at the furnace temperature of tfurn = 705.0◦ C indicate the presence of crystalline copper particles at the vapor pressure of Pprec = 16 Pa and copper oxide (Cu2 O, cuprite) particles at the vapor pressure of Pprec = 1.8 Pa. In Figure 8 XRD diffractograms of the particles collected on a silver filter are shown. The variation of the crystalline phase with changing the precursor vapor pressure can be seen. At the vapor
392
Figure 6. Transmission electron microscopy images and electron diffraction patterns of particles produced at the precursor vapor pressure of Pprec = 1.8 Pa (a) and Pprec = 16 Pa (b). The furnace temperature is 431.5◦ C.
pressure of Pprec = 1.8 Pa and Pprec = 6 Pa, a mixture of crystalline copper and copper (I) oxide particles was synthesized and at higher saturator temperatures only copper particles were formed. The additional experiments of XRD phase identification were carried out at the intermediate furnace temperature of tfurn = 596.0◦ C. At the precursor vapor pressures of Pprec = 6 and 44 Pa only copper crystalline product was found. The effect of the experimental conditions on product compositions is presented in Table 1. As one can see from Table 1, the dependence of the decomposition
Figure 7. Transmission electron microscopy images and electron diffraction patterns of particles produced at the precursor vapor pressure of Pprec = 1.8 Pa (a) and Pprec = 16 Pa (b). The furnace temperature is 705.0◦ C.
product on the precursor vapor pressure is revealed only at highest experimental temperature of tfurn = 705.0◦ C. The possible reasons of this phenomenon are examined in the discussion part of the article. In an attempt to reduce agglomeration, the aerosol flow was diluted by excess nitrogen flow immediately downstream of the furnace. The particle number size distribution was not significantly changed, indicating that agglomeration of the particles occurred inside the furnace. Another way to avoid the formation of the agglomerated particles is to decrease the number concentration of the primary particles by
393
Figure 8. XRD spectra of particles synthesized at the furnace temperature of 705.0◦ C and collected on the silver filter. Table 1. List of experimental conditions, corresponding crystalline product compositions, and methods used for the phase identification Furnace temperature, ◦ C
Vapor pressure of precursor, Pa
Crystalline products
Method
431.5 596.0 705.0 705.0 705.0
0.15–44 6–44 0.06–1.0 1.8–10 16–44
Cu Cu Cu2 O Cu/Cu2 O Cu
XRD, ED XRD ED XRD, ED XRD, ED
resolution picture shows the possibility to produce single crystalline individual nanoparticles.
Discussion
decreasing the Cu(acac)2 vapor pressure. In Figure 9 and Figure 10 the number size distribution of the produced particles at different saturator temperatures of tfurn = 431.5◦ C and tfurn = 705.0◦ C are presented. It is known that the area limited by the curve determines the particle number concentration. At tfurn = 431.5◦ C the concentration grows smoothly when the saturator temperature increases, but at tfurn = 705.0◦ C the change of the concentration is not smooth. This is probably in connection with the formation of different decomposition products. At precursor vapor pressures below 0.1 Pa, single crystalline individual particles are formed, the formation of agglomerated particles is observed at higher vapor pressures. In Figure 11, TEM micrograph of copper (I) oxide nanoparticle that was produced at the furnace temperature of 705.0◦ C and at the vapor pressure of Pprec = 0.04 Pa is presented. This high
The size distribution of primary particles obtained from TEM micrographs is presented in Figure 12. At 431.5◦ C the number mean size of the primary particles is increased from Dp = 3.7 (with geometric standard deviation of σg = 1.42) to Dp = 7.2 nm (σg = 1.33) with increasing precursor vapor pressure from 1.8 to 16 Pa. At tfurn = 705.0◦ C, the primary particle size is decreased from Dp = 24.0 nm (σg = 1.57) to Dp = 7.6 nm (σg = 1.54), respectively. At the furnace temperature of tfurn = 431.5◦ C, increasing the vapor pressure increased the size of the primary particles, but at tfurn = 705.0◦ C an inverse situation was observed. Apparently, it is connected with the different products of the precursor decomposition. The diameter of the primary copper (I) oxide particles (tfurn = 705.0◦ C, Pprec = 1.8 Pa) is about three times larger than that of the copper particles (tfurn = 705.0◦ C, Pprec = 16 Pa). The qualitative explanation of the primary size dependence on the crystalline products can be obtained from the consideration of physical properties of these compounds. It is known that the sintering rate, which controls the size of primary particles, is dependent on the self-diffusion coefficient of the elements contained in the substance: with the larger
394
Figure 9. Number size distributions of agglomerated particles at the furnace temperature of tfurn = 431.5◦ C.
Figure 10. Number size distributions of agglomerated particles at the furnace temperature of tfurn = 705.0◦ C.
coefficient the sintering rate is enhanced and the size of primary particles becomes larger. According to the literature data (Smithells & Brandes, 1983; Kofstad, 1972), the self-diffusion coefficients of copper atoms are 1.9×10−12 cm2 /s in pure crystalline copper and 1.5×10−10 cm2 /s in crystalline copper (I) oxide at temperature t = 705◦ C. The almost two orders of magnitude of difference in the diffusion coefficients is most likely the reason for the formation of such a
different size of copper and copper (I) oxide primary particles. As it was found, the product of the decomposition reaction depended on the precursor vapor pressure only at tfurn = 705.0◦ C: copper particles were formed at the higher vapor pressures (Pprec > 10 Pa), copper (I) oxide particles at the pressures of Pprec < 1 Pa, and a mixture of Cu and Cu2 O was formed at the intermediate vapor pressures. At the furnace temperatures of
395 431.5◦ C and 596◦ C, only copper particles were formed (Table 1). At the first glance, the revealed dependence of oxygen content in particles upon the precursor pressure seems to be anomalous. Indeed the formation of
Figure 11. High resolution transmission electron micrograph of Cu2 O particle produced at the furnace temperature of 705.0◦ C and at the vapor pressure of Pprec = 0.04 Pa.
Figure 12. Particle number size distributions of primary particles.
primary particles can be presented by two sequential steps. At the first stage, small clusters are formed by means of homogeneous nucleation process, and secondly, the particles grow because of a coalescence of the clusters and vapor condensation on the particles. Increasing the precursor pressure leads to the increase of the oxygen-containing gas pressure in the system and its participation in these two stages. The presence of oxygen in the particles with increasing the precursor pressure at least cannot vanish. Let us demonstrate that this prima facie consideration is incorrect for the explanation of the experimental results. A few possible reasons for the change of decomposition products were examined. The first reason could be the presence of impurities in the carrier gas. The appearance of the copper (I) oxide in the crystalline products occurs at the precursor vapor pressure of 10 Pa, which corresponds to a concentration of 2.6×1015 molecules/cm3 . Meanwhile, the maximum concentration of admixtures in the nitrogen carrier gas sample is about 2.5×1016 molecules/cm3 , that is, the concentration of the main component in the reaction is of the same order or less than the impurity concentration. In order to check this hypothesis, the nitrogen gas cylinder was changed to a liquid nitrogen tank to obtain ultra-pure carrier gas (99.999 vol.% with maximum oxygen concentration of 8.1×1013 molecules/cm3 ). Decreasing the content of admixtures by two orders of magnitude did not change
396 the product of the Cu(acac)2 decomposition. Moreover, the results by Tonneau et al. (1995) and Hammadi et al. (1995) on chemical vapor deposition are also contrary to this assumption. They showed that when the vapor mole fractions of oxygen and Cu(acac)2 were about the same, only crystalline copper was formed. Accordingly, the presence of impurities in the gas-carrier is not likely to be the reason for the different products of the reaction. Another possible reason is connected with the kinetics of the decomposition. It is well known that changing the gas phase concentration of the reactant can lead to the change of the decomposition mechanism (Kondrat’ev, 1964). For example, at lower precursor concentrations a unimolecular decomposition can occur and as a result of the decomposition Cu2 O is formed. Increasing the vapor pressure leads to the increase in the probability of collisions of two or more molecules during the Cu(acac)2 thermolysis reaction, which leads to the decomposition product of pure copper. However, in order to form Cu2 O molecule it is necessary to have two precursor molecules, but for copper formation only one precursor molecule is needed. Thus, this explanation is correct for the inverse behavior. Moreover, it is hard to believe that switching the mechanism from a unimolecular to a collision reaction can occur in a such small vapor pressure range. Therefore, the explanation of switching the decomposition mechanism does not seem to be justified for the elucidation of the variation of the reaction products. Also, a possible reason of the condensed product variation may be related to the kinetics of particle formation. Then the explanation of the phenomena is possible only in the case of any second-order reaction. Indeed, the ratio of partial pressures of all species formed in the first-order reactions is the same at all precursor pressures. Thereby, the relative role of different gaseous species in the condensed particle growth cannot depend on the precursor pressure. The possible reason for the crystalline product change is the existence of a secondary reaction. We assume that this reaction is a formation of copper dimers that are quite stable at the experimental conditions (Petrov, 1986): 2Cu(g) ⇔ Cu2 (g).
(2)
In order to explain the revealed transition between the crystalline products let us consider the mechanism of the formation of different condensed substances from gas phase in detail.
It is worth noting that there is a restriction prohibiting the growth of copper particles from gas via single Cu atom adsorption on the cluster surface. The physical fundamentals of such restrictions for different materials are discussed by Altman et al. (2001). They are based on the mechanism of energy transfer during a phase transition. The energy release during Cu atom condensation is about 3.5 eV, and this energy has to be dissipated directly during the adsorption. There are two possible ways for this process: an electron excitation and a phonon creation. Since this energy value is large compared to a typical Debye energy value (0.05 eV), only the electron excitation for energy dissipation is left. However, the energy band structure of copper contains the sp-band gap (Knoesel et al., 1998). The value of the sp-band gap energy at the experimental conditions is about 4.6–4.7 eV. The existence of this gap leads to impossibility of the direct energy dissipation from an adsorbing copper atom and to a prohibition of the atom condensation as a consequence. The energy release for Cu2 molecule condensation is about 5 eV, which is larger than the copper energy gap. Therefore, this process becomes feasible and the growth of the particles is made possible by condensation of Cu2 molecules. Then the reason of the crystalline product change can be understood. Indeed, due to the second-order reaction (2), the Cu2 partial pressure increases with the precursor pressure as its second power instead of all other gas partial pressures. Thus, since the condensation of Cu2 molecules determines the copper particle formation, the relative role of this process increases with increasing precursor pressure faster when compared to other processes. The second question to be discussed is the possibility of copper (I) oxide formation. Because of low reagent concentration, the variety of possible pathways can be limited by bimolecular reactions. In this case, the only way of copper (I) oxide formation is the reaction on the surface of a growing oxide particle: Cu2 + CO2 ⇔ Cu2 O(s) + CO.
(3)
It is obvious that this reaction should be activated. Let us propose the physical model of the transition. The formation of copper (I) oxide particle occurs if the flux of CO2 molecules (with an energy larger than the barrier Eo ) jCO2 on a particle surface is larger than the flux of Cu2 molecules jCu2 , in the opposite case copper particle formation occurs. Thereby, transition between the two products might occur due to the variation of these two fluxes. At a given temperature of T
397 ∗ (the transition pressure of Cu2 O product pressure Pprec identification) can be written as
the fluxes may be defined as jCu2 = jCO2
PCu2
, 2πmCu2 kB T PCO2 Eo = , exp − kB T 2πmCO2 kB T
(4)
where mCu2 and mCO2 are masses of Cu2 and CO2 molecules, respectively, and kB is Boltzmann’s constant. It is obvious that the copper particle growth leads to the exhaustion of the copper vapor and to the decrease of relative value of jCu2 flow. The initial ratio of jCu2 and jCO2 flows is more than unity (in case of pure copper formation), but at the end of the precursor decomposition this ratio becomes smaller (in case of copper (I) oxide formation). That is why there is a transition region instead of the sharp border between the different crystalline products. The absence of copper (I) oxide (at Pprec > 10 Pa) most likely means that the ratio of amount of the oxide to pure copper is less than the relative sensitivity f of XRD phase identification. In this case, at the precursor pressure of Pprec = 16 Pa, the copper (I) oxide particles are formed only at the end of the precursor decomposition, where the jCu2 /jCO2 flux ratio becomes smaller than unity due to the exhaustion of copper vapor and an accumulation of the decomposition products (carbon dioxide). In order to obtain the characteristic value of the precursor pressure where the transition between products occurs, let us consider the final location of the decomposition. The partial pressure of carbon dioxide can be written as PCO2 = 2Pprec ,
(5)
∗ 2 ∗ Kp fPprec 2Pprec Eo =√ exp − . √ mCu2 mCO2 kB T
(8)
Based on the experimental data of the disappearance ∗ = 16 Pa at tfurn = of copper (I) oxide phase (Pprec ◦ 705.0 C) and thermodynamic data (Kp = 1.84 Pa−1 ), at the sensitivity of f = 0.01, the value of the energy barrier for reaction (3) is Eo = 0.595 eV (57.3 kJ/mole). This value seems to be realistic. Then, it is easy to show that the boundary pressure of copper (I) oxide formation with mole fraction of less than 0.01 may be calculated as const Eo ∗ Pprec = , (9) exp − Kp kB T where const = 3.4 × 104 . The semiempirical temperature dependence of the boundary pressure obtained on the basis of Eq. (9) is presented in Figure 13. The area on the diagram above the boundary pressure line is related to the region of the absence of copper oxide products. As one can see the proposed semiempirical phase diagram is in agreement with the experimental results of the current work. The boundary pressure ∗ Pprec decreases by decreasing the furnace temperature. That is why at lower temperatures (tfurn = 431.5◦ C at Pprec = 0.06 to 44 Pa and tfurn = 596◦ C at Pprec = 16–44 Pa) only pure copper particles in the final product of Cu(acac)2 decomposition were obtained. It is worth noting that the diagram could be useful for the
where the factor 2 is obtained on the basis of mass-spectrometry results of gaseous decomposition products (Tsyganova et al., 1992). The copper vapor pressure at the end of the decomposition, when copper (I) oxide starts to be formed, can be written by taking into account the copper exhaustion and phase identification sensitivity f : PCu = fPprec .
(6)
The partial vapor pressure PCu2 can be found using equilibrium constant Kp of the reaction (2): 2 PCu2 = Kp PCu .
(7)
Combining Eqs. (4)–(7) with the condition jCu2 = jCO2 , the expression for the boundary precursor
Figure 13. The phase diagram of Cu(acac)2 decomposition crystalline products. Marks ‘▼’ and ‘▲’ correspond to the experimentally determined copper and copper (I) oxide products, respectively.
398 the precursor decomposition and copper and copper (I) oxide particle formation. After the Cu(acac)2 vapor evaporation and heating the vapor up to a high enough temperature, the formation of copper vapor and decomposition products, as a result of Cu(acac)2 decomposition reaction, occurs inside the furnace. The next important stage is the formation of gaseous copper dimers that participate in nucleation and condensation processes. Formation of copper (I) oxide particles occurs at low precursor vapor pressures due to the surface reaction of Cu2 vapor and the products of the decomposition (mainly, carbon dioxide). The last stage is the agglomeration process of the formed primary particles, which exist at the precursor vapor pressure of Pprec ≥ 0.1 Pa. Conclusion
Figure 14. The schematic presentation of copper and copper (I) oxide particle formation by the Cu(acac)2 vapor decomposition.
prediction of the product of the precursor vapor decomposition at other experimental conditions. The proposed approach for the building of the decomposition product phase diagram can be used for other experimental systems with chemical reactions. Figure 14 summarizes our viewpoint of the main stages occurring inside the furnace at 705.0◦ C during
For this study, a vertical laminar flow reactor has been constructed and tested for the investigation of nanoparticle formation via chemical vapor nucleation. It has been shown that crystalline nanometer-size copper and copper oxide particles can be produced by thermal vapor decomposition of a metal–organic precursor, Cu(acac)2 , at relatively low temperatures. Individual primary particles are formed at the precursor vapor pressure of Pprec < 0.1 Pa. At higher vapor pressures, particles form aggregates. At tfurn = 431.5◦ C, the number mean size of the primary particles increased from Dp = 3.7 nm (with geometric standard deviation of σg = 1.42) to Dp = 7.2 nm (σg = 1.33) with increasing precursor vapor particle pressure from 1.8 to 16 Pa. At tfurn = 705.0◦ C, the primary particle size decreased from Dp = 24.0 nm (σg = 1.57) to Dp = 7.6 nm (σg = 1.54), respectively. At the furnace temperatures of 431.5◦ C and 596.0◦ C, only crystalline copper particles were produced. At the furnace temperature of tfurn = 705.0◦ C, the product of the decomposition reaction depended on the precursor vapor pressure: copper particles were formed at vapor pressures higher than 10 Pa, copper (I) oxide at pressures lower than 1 Pa, and a mixture of the metal and its oxide at intermediate vapor pressures. For the explanation of the obtained results, the kinetic restriction on copper particle growth was proposed. It leads to the main role of a Cu2 molecule participation in the particle formation. The formation of copper (I) oxide particles occurs due to the surface reaction of the decomposition products of which carbon dioxide is the most important. For the explanation of the experimental results,
399 a model is proposed to build a semiempirical phase diagram of the precursor decomposition crystalline products. The phase diagram, showing the boundary of the appearance of copper (I) oxide, is in agreement with the experimental results. Acknowledgements This work has been supported by TEKES, Finland and VTT Chemical Technology via the MATRA research program. Mr. P. R¨aisanen, Dr. M. Ritala and Prof. M. Leskel¨a are gratefully acknowledged for help during the XRD analyses and Mr. Timo Hatanp¨aa¨ for carrying out TGA analysis. The authors thank Dr. J.K. Jokiniemi, Dr. P.V. Pikhitsa, Dr. V. Mikheev and Prof. T. Ward for fruitful discussions and Dr. D.P. Brown for discussions and proofreading the manuscript. References Adams C.R. & T.J. Jennings, 1964. Mechanism studies of the catalytic oxidation of propylene. J. Catal. 3, 549. Altman I.S., P.V. Pikhitsa & M. Choi, 2001. Key effects at nanoparticle formation by combustion techniques. In: Kish L., Granqvist C.G., Marlow W., Siegel R.W. eds. Fundamentals of Gas-Phase Nanotechnology (in press). Ayaappan S., R.S. Gopalan, G.N. Subbanna & C.N.R. Rao, 1997. Nanoparticles of Ag, Au, Pd, and Cu Produced by Alcohol Reduction of the Salts. J. Mater. Res. 12, 398. Bouland D., J.C. Chouard, A. Briand, F. Chartier, J.L. Lacour, P. Mauchien & J.M. Mermet, 1992. Experimental study of aerosol production by laser ablation. J. Aerosol Sci. 23(1), 225. Bowles R.S., J.J. Kolstad, J.M. Calo & R.P. Andres, 1981. Generation of molecular clusters of controlled size. Surf. Sci. 106, 117. Campbell C.T., K.A. Daube & J.M. White, 1987. Cu/ZnO(0001) and Cu(111): Model Catalysts for Methanol Synthesis. Surf. Sci. 182, 458. Champion Y. & J. Bigot, 1996. Characterization of nanocrystalline copper powders preparing by melting in a cryogenic liquid. Mater. Sci. Eng. A217/218, 58. Chou K.-S. & G.-J. Tsai, 1994. Dynamic evaporation behaviour of diketonate compounds of yttrium, copper and barium. Thermochim. Acta 240, 129. Daroczi L., M.T. Beck, D.L. Beke, M. Kis-Varga, L. Harasztosi & N. Takacs, 1998. Production of Fe and Cu nanocrystalline particles by thermal decomposition of ferro- and copper-cyanides. Mat. Sci. Forum 269–272, 319. Dhas N.A., C.P. Raj & A. Gedanken, 1998. Synthesis, characterization, and properties of metallic copper nanoparticles. Chem. Mater. 10, 1446. Ding J., T. Tsuzuki, P.G. McCormick & R. Street, 1996. Ultrafine Cu articles prepared by mechanical process. J. Alloys and Comp. 234, L1.
Du F.-L., Z.-L. Cui, Z.-K. Zhang & S.-Y. Chen, 1997. Behavior of supported nano-copper catalyst in CO oxidation. J. Nat. Gas Chem. 6, 135. ´ & V.A. Uglov, 1991. Nanocrystalline copper powFolmanis G.E. der produced by electrolysis. Poroshkovaya Metallurgiya 2, 5. Gerfin T., M. Becht & K.-H. Dahmen, 1993. Preparation of copper and copper oxide film by metal-organic chemical vapour deposition using β-ketoiminato complexes. Mater. Sci. Eng. B17, 97. Girardin D. & M. Maurer, 1990. Ultrafine metallic powders prepared by high pressure plasma: synthesis and characterization. Mat. Res. Bull. 25, 119. Gredig S.V., R.A. K¨oppel & A. Baiker, 1997. Synthesis of methylamines from carbon dioxide, hydrogen and ammonia over supported copper catalysts. Influence of support. J. Mol. Catal. A: Chemical 127, 133. Haas V. & R. Birringer, 1992. The morphology and size of nano structured Cu, Pd, and W generated by sputtering. Nanostruct. Mater. 1, 491. Hammadi Z., B. Lecohier & H. Dallaporta, 1993. Chemical vapor deposition of metallic copper film in the presence of oxygen. J. Appl. Phys. 73, 5213. Hammerschmidt W., A. Baiker, A. Wokaun & W. Fluhr, 1986. Copper catalyzed synthesis of cyclic amines from aminoalcohols. Appl. Catal. 20, 305. Herley P.J., W. Jones & G.R. Millward, 1989. Electron beam decomposition of copper hydride and the generation of ultrafine particles of copper. J. Mater. Sci. Lett. 8, 1013. Holzschuh H. & H. Suhr, 1990. Deposition of copper oxide (Cu2 O, CuO) thin films at high temperatures by plasmaenhanced CVD. Appl. Phys. A51, 486. Huang H.H., F.Q. Yan, Y.M. Kek, C.H. Chew, G.Q. Xu, W. Ji, P.S. Oh & S.H. Tang, 1997. Synthesis, characterization, and nonlinear optical properties of copper nanoparticles. Langmuir 13, 172. Kaito C., K. Fujita & H. Hashimoto, 1973. Electron-microscopy study of oxidation processes by metal fine particles. Jpn. J. Appl. Phys. 12, 489. Kaito C., T. Watanabe, K. Ochtsuka, F. Chen & Y. Saito, 1993. A new study on the production of clusters from copper ultrafine particles. J. Cryst. Growth 128, 267. Kashu S., M. Nagase, C. Hayashi, R. Uyeda, N. Wada & A. Tasaki, 1974. Preparation and properties of ultrafine metal powders. Proc. 6th Int. Vac. Congr. 1974. Jpn. J. Appl. Phys. 14 (Suppl. 2), 491. Kellerson A., E. Kn¨ozinger, W. Langel & M. Giersig, 1995. Cu2 O quantum-dot particles prepared from nanostructured copper. Adv. Mater. 7, 652. Kijenski J., J. Burger & A. Baiker, 1984. Copper catalyzed disproportionation of benzylamine methyl derivatives. Appl. Catal. 11, 295. Kijenski J., P.J. Niedzielski & A. Baiker, 1989. Synthesis of cyclic amines and their alkyl derivatives from amino-alcohols over supported copper catalysts. Appl. Catal. 53, 107. Kirchheim R., X.Y. Huang, P. Cui, R. Birriner & H. Gleiter, 1991. Free energy of active atoms in grain boundaries of nanocrystalline copper, nickel and palladium. Nanostruct. Mater. 1, 167. Klenov D.O., G.N. Kryukova & L.M. Plyasova, 1998. Localization of copper atoms in the structure of the ZnO catalyst for methanol synthesis. J. Mater. Chemistry 8, 1665.
400 Klier K., 1982. Methanol synthesis. Adv. Catal. 31, 243. Knacke O., O. Kubaschewski & K. Hesselmann, 1991. Thermochemical Properties of Inorganic Substances. Springer-Verlag, Berlin. Knoesel E., A. Hotzel & M. Wolf, 1998. Temperature dependence of surface state lifetimes, dephasing rates and binding energies on Cu(111) studies with time-resolved photoemission. J. Elect. Spectr. Rel. Phen. 88–91, 577. Kodas T.T. & M.J. Hampden-Smith, 1999. Aerosol Processing of Materials. Wiley-VCH, New York. Kofstad P., 1972. Nonstoichiometry, Diffusion, and Electrical Conductivity in Binary Metal Oxides. Wiley-Interscience, New York. Kondrat’ev V.N., 1964. Chemical Kinetics of Gas Reactions. Persimmon Press Oxford, London. Kulikov I.S., 1969. Thermal Dissociation of Compounds. (Termicheskaja dissociatciya soedinenij) Metallurgia, Moscow (in Russian). Lisieski I. & M.P. Pileni, 1993. Synthesis of copper clusters using reverse micelles as microreactors. J. Am. Chem. Soc. 115, 3887. Lisieski I. & M.P. Pileni, 1995. Copper metallic particles synthesized “in situ” in reverse micelles: influence of various parameters on the size of particles. J. Phys. Chem. 99, 5077. Lisieski I., F. Billoudet & M.P. Pileni, 1996. Control of the shape and the size of copper metallic particles. J. Phys. Chem. 100, 4160. Long N.J. & A.K. Petford-Long, 1986. In situ Electron-beaminduced reduction of CuO: a study of phase transformation in cupric oxide. Ultramicroscopy 20, 151. Majumdar D., T.A. Shefelbine, T.T. Kodas & H.D. Glicksman, 1996. Copper (I) oxide powder generation by spray pyrolysis. J. Mater. Res. 11, 2861. Maruyama T. & T. Shirai, 1995. Copper thin film prepared by chemical vapour deposition from copper (II) acetylacetonate. J. Mater. Sci. 30, 5551. Marzouk H.A., J.S. Kim, P.J. Reucroft, R.J. Jacob, J.D. Robertson & C. Eloi, 1994. Evaluation of copper chemical-vapor-deposition films on glass and Si(100) substrates. Appl. Phys. A58, 607. Moini M. & J.R. Eyler, 1988. Formation of small negative and positive cluster ions of gold, silver, and copper by direct laser vaporization. J. Chem. Phys. 88, 5512. Pauleau Y. & A.Y. Fasasi, 1991. Kinetics of sublimation of copper (II) acetylacetonate complex used for chemical vapor deposition of copper films. Chem. Mater. 3, 45. Pelletier J., R. Pantel, C. Oberlin, Y. Pauleau & P. Gouy-Pialler, 1991. Preparation of copper films at ambient temperature by microwave plasma-enhanced chemical vapor deposition from the copper (II) acetylacetonate-argon-hydrogen system. J. Appl. Phys. 70, 3862. Peoples S.M., J.F. McCarthy, L.C. Chen, D. Eppelsheimer & M.O. Amdur 1988. Copper oxide aerosol: generation and characterization. Am. Ind. Hyg. Assoc. J. 49, 271. Pereia R., M. Rufo & U. Schuchardt, 1994. Copper (II) catalyzed oxidation of cyclohexane by tert-butyl hydroperoxide. J. Braz. Chem. Society 5, 83.
Petrov Yu.I., 1986. Clusters and Small Particles. (Klastery i malye chastitcy) Nauka, Moscow (in Russian). Pietrikova A. & E. Kapusanska, 1991. Production of very fine copper powder and control of its properties. Metallic Mater. 29, 626. Runeberg J., A. Baiker & J. Kijenski, 1985. Copper catalyzed amination of ethylene glycole. Appl. Catal. 17, 309. Shannon I.J., F. Rey, G. Sankar, J.M. Thomas, T. Maschmeyer, A.M. Waller, A.E. Palomares, A. Corma, A.J. Dent & G.N. Greaves, 1996. Hydrotalcite-derived mixed oxides containing copper: catalysts for the removal of nitric oxide. J. Chem. Soc. Farad. Trans. 92, 4331. Shiau C.-Y. & J.C. Tsai, 1997. Cu/r-Al2O3 Catalyst prepared by electrolyses method. J. Chin. Ist. Chem. Eng. 28, 55. Smithells C.J. & E.A. Brandes, 1983. Smithells Metals Reference Book. Butterworths, London. Takahashi T., J. Suzuki, M. Saburi & Y. Uchida, 1988. Selective preparation of copper, copper (I) oxide, or copper (II) oxide fine particles from organocopper compounds. J. Mater. Sci. Let. 7, 1251. Teghil R., D. Ferro, L. Bencivenni & M. Pelino, 1981. A thermodynamic study of the sublimation processes of aluminium and copper acetylacetonates. Thermochim. Acta 44, 213. Tijburg I.I.M., Ph.D. Thesis, Utrecht, The Netherlands, 1989. Tonneau D., R. Pierrisnard, H. Dallaporta & W. Marine, 1995. Growth kinetics of copper films from photoassisted CVD of copper acetylacetonate. Journal de Physique IV 5(C5), 629. Tsyganova E.I., G.A. Mazurenko, V.N. Drobotenko, L.M. Dyagileva & Yu.A. Aleksandrov, 1992. Kinetic regularities of thermolysis of yttrium, barium, copper acetylacetinates. J. Gen. Chem. 62, 407 (Translated from Zh. Obshch. Khim. (USSR) 62, 499). Voge H.H. & C.R. Adams, 1967. Catalytic oxidation of olefins. Adv. Catal. 17, 151. Vorobyova S.A., V.V. Mushinsky & A.I. Lesnikovich, 1997. Copper oxide produced by two-phase synthesis in “octanewater” system. Dokl. Acad. Sci. Belarus 41, 62. Vultier R., A. Baiker & A. Wokaun, 1987. Copper catalyzed amination of 1,6-hexanediol. Appl. Catal. 30, 167. Van der Meijden J., Ph.D. Thesis, Utrecht, The Netherlands, 1981. Weins W.N., J.D. Makinson, R.J. De Angelis & S.C. Axtell, 1997. Low-frequency internal friction studies of nanocrystalline copper. Nanostruct. Mater. 9, 509. Winklmayr W., G.P. Reischl, A.O. Linder & A. Berner, 1991. A new electromobility spectrometer for the measurement of aerosol size distributions in the size range from 1 to 100 nm. J. Aerosol Sci. 22, 289. Xu J., X. Sun, W. Chen, X. Fan & W. Wei, 1992. Synthesis of copper ultrafine particles by using gas evaporation. Cailiao Kexue Jinzhan 6, 209. Yurieva T.M., L.M. Plyasova, T.A. Krieger, V.I. Zaikovskii, O.V. Makarova & T.P. Minyukova, 1993. State of copper containing catalyst for methanol synthesis in the reaction medium. React. Kinet. Catal. Lett. 51, 495.