Introduction
Aluminum nitride (AlN) has become an important material in electronic packaging due to its high intrinsic thermal conductivity (320 W.m-1.K-1) [1], its thermal expansion coefficient being close to that of silicon, and its excellent electric and dielectric properties [2, 3]. The thermal conductivity of sintered AlN is greatly reduced by oxygen contamination in AlN grains, because oxygen occupation in AlN lattices generates vacancies that work as phonon scattering sites [1, 4-6]. To obtain low-oxygen, high-thermal conductivity AlN ceramics, the processes from AlN powder production to its sintering has been intensively studied. High-purity AlN powders with low oxygen content has been commercially available. Unfortunately, because AlN is unstable in a water-containing environment, oxygen contamination arises not only from the powder manufacturing process, but also from contact with water molecules in powder processing after manufacturing.
The hydrolysis behaviors of AlN powders and ceramics in water and water-based solutions have been investigated extensively. It has been found that hydrolysis products can be crystalline bayerite, gibbsite, boehmite, or amorphous aluminum hydroxide, depending on the experimental conditions (time, temperature, and pH and solutes in the solutions) [7-12]. The hydrolysis rate is affected by the AlN characteristics and the solutions employed. For example, for AlN ceramics, the hydrolysis rate is higher in alkalis than in acids [13]. For AlN powders, H3PO4 and silicic acid can prevent the hydrolysis by forming a barrier layer of complexes on the surface [10, 12, 14]. AlN powder with higher oxygen content was found to have a better resistance to hydrolysis [15]. Formation of a protective oxide layer on the surface of AlN powder at elevated temperatures can hinder the hydrolysis in water [9, 16, 17], and organic coatings can prevent the hydrolysis to some extent [18, 19].
In contrast, degradation of AlN powders in ambient atmospheres at room temperature has been much less studied. Abid et al. reported that AlN did not readily react with atmospheric moisture at room temperature [20]. However, Kameshima et al. found, by using X-ray photoelectron spectrometry (XPS), that the surface of AlN powders reacted slowly with atmospheric moisture during several years of storage in a capped container [21].
In this paper, AlN powders produced via three major commercial processes have been used to investigate their hydrolysis behaviors in moist air at room temperature. This study provided insight into the degradation phenomena of AlN powders during processing under normal atmospheric environmental conditions.
Experimental
The characteristics of AlN powders studied are listed in Table 1. These powders were produced through three major commercial approaches: chemical vapor deposition (CVD) from triethyl aluminum (powder A), carbothermal reduction and nitridation of alumina (powders B1 and B2), and direct nitridation of aluminum metal (powders C1 and C2). During the hydrolysis, the powder samples were held above distilled water in a glass vessel at room temperature (20°C) under atmospheric pressure. The relative humidity around the samples was 80%. After certain periods of hydrolysis the samples were dried at 100°C for 2 h in air. This drying condition proved to be sufficient since after further heating at 160°C mass change was not observed. X-ray diffraction (XRD) analysis indicated that the drying process did not change phase composition.
Table 1. Characteristics of the AlN Powders.
|
|
|
Nomenclature
|
A
|
B1
|
B2
|
C1
|
C2
|
|
Specific surface area (m2/g)
|
2.0
|
2.63
|
3.31
|
2.8
|
4.0
|
|
Mean particle size (μm)
|
3.0
|
1.55
|
1.3
|
3.25
|
2.41
|
|
O (wt%)
|
0.37
|
0.83
|
0.89
|
1.14
|
1.6
|
|
C (wt%)
|
0.04
|
0.022
|
0.039
|
0.05
|
0.04
|
|
Si (ppm, wt)
|
23
|
38
|
<9
|
|
|
|
Fe (ppm, wt)
|
<10
|
10
|
<10
|
<20
|
20
|
|
Ca (ppm, wt)
|
|
220
|
6
|
|
|
Sample masses were measured before hydrolysis and after drying. The mass change rate, Δm, in the hydrolysis was expressed as
.jpg) (1)
where mt is the mass of the dried sample after hydrolysis for period t, and m0 the initial mass. XRD analysis for phase composition was performed on a Shimadzu XRD 6000 diffractometer (Shimadzu Corp., Japan) with Cu Kα (Kα1 and Kα2) radiation at an interval of 0.02° 2θ. Reflections from Kα2 radiation were removed by using software supplied with the diffractometer. Peak separation for the XRD pattern was performed by using the Jandel software PeakFitTM 4.0 (AISN Software Inc., USA). Diffuse reflectance infrared Fourier transform spectroscopy (DRIFT) was used for surface analysis. The DRIFT spectra of samples without dilution were recorded by using a Shimadzu FTIR 8300 spectrometer (Shimadzu Corp., Japan) equipped with a Triglycine Sulfate detector, at a resolution of 4 cm-1. An aluminum mirror was employed to obtain the background spectra. Scanning electron microscopy (SEM) for microstructural observation was conducted on a Keyence VE 7800 microscope (Keyence Corp. Japan).
Results and Discussion
Mass Change
Mass changes of AlN powders during hydrolysis are shown in Figure 1. Masses of all powders increased with hydrolysis time. The mass increase was due to hydrolysis of AlN to form aluminum hydroxides, as will be analyzed below. At the beginning each powder underwent an induction period when the hydrolysis reaction was slow. With time increase the hydrolysis rate increased at first, and then gradually decreased. Thus the mass change-time curves were S-shaped. The value of Δm of powder A (89.7%) after 417 h of degradation was in good agreement with the calculated value (90.3% for complete conversion of pure AlN to Al(OH)3, the final product of the hydrolysis as will be described below). The small difference could be attributed to impurities, e.g., 0.37wt% oxygen, in the raw powder. Mass changes of the other four powders after the same period of degradation (B1 81.7%, B2 87.3%, C1 79.2%, and C2 85.0%) were less than the calculated values, which was due to incomplete hydrolysis as evidenced by the presence of AlN reflections in their XRD patterns.
|
.jpg)
|
|
Figure 1. Mass change of the AlN powders during hydrolysis in moist air.
|
For each powder, after the mass change per unit surface area reached ~0.006 g/m2, the hydrolysis rate increased quickly. The time required to reach this value of mass change was used in this study as the measurement of the induction period, the values of which are shown in Figure 2. The powders from the carbothermal process showed the longest induction periods.
|
.jpg)
|
|
Figure 2. Induction periods for the hydrolysis of AlN powders in moist air.
|
In previous studies on the hydrolysis of AlN powders in water at room temperature, Bowen et al. found a linear reaction rate without induction period [8], but, other researchers reported the existence of an induction period [10, 11]. The reason could be because, the length of the induction period varies for different powders (as shown in Figure 2) and experimental conditions. If the induction period is short enough, it may not be observed. The existence of an induction period in the hydrolysis could be related to the surface structure of AlN powders. Using the techniques of temperature-programmed desorption mass spectroscopy (TPDMS) and auger electron spectroscopy, Ishizaki et al. found that the surface of AlN powder produced by the carbothermal process was composed of a θ-alumina-like layer containing oxynitride, and those from CVD and direct nitridation were composed of a γ-alumina-like or boehmite-like layer and an oxygen-diffused layer [22, 23]. In the investigation of the surface change during storage over years using XPS analysis, Kameshima et al. also indicated that the surfaces of AlN powders from the carbothermal or direct nitridation processes were composed of an alumina-like layer [21]. The continuous and amorphous oxide/oxyhydroxide layer could protect the AlN against corrosion. When the powders were stored in air this layer could be hydrolyzed slowly and converted to a trihydroxide layer, followed by a fast hydrolysis of AlN [21]. A thicker and denser oxide/oxyhydroxide surface layer was expected to protect AlN particles better. Some authors reported that an oxide surface layer on AlN powders formed by heat treatment at >800°C improved hydrolysis resistance in water [16, 17].
TPDMS analysis indicated that, unlike powders produced by CVD and direct nitridation processes, the powder produced by the carbothermal process showed no desorption of NH3 [22], demonstrating a better resistance of it to moisture attack due to the higher stability of its surface structure. Kameshima et al. reported that the powders from the carbothermal process had better resistance to weathering, and attributed this to thicker surface oxide layer (after measurement of this layer using XPS), than the ones produced by direct nitridation [21]. Our observation that the powders from the carbothermal process (powders B1 and B2) showed the longest induction periods (Figure 2) could thus be attributed to their surface structure that was different from the other powders, as mentioned above. Since every step in the powder handling after manufacturing could influence the surface, powders from the same preparation method differed in the length of induction period.
Phase Composition and Hydrolysis Reactions
XRD analysis indicated that the crystallized phases obtained in the hydrolysis were the same for different powders and for the same powder at different hydrolysis times. A representative XRD pattern obtained from powder A after 417 h of hydrolysis, achieving the highest conversion (Δm = 89.7%) among the five powders employed, is presented in Figure 3. The pattern indicates the presence of polymorphs of aluminum trihydroxide (Al(OH)3), namely bayerite, nordstrandite, and gibbsite. The reflections of bayerite were unambiguously shown, but those of nordstrandite and gibbsite were not apparent due to overlapping. After peak separation it was found that the reflections at 2θ = 18° - 19° were composed of three peaks (see the inset in Figure 3) at 2θ = 18.33°, 18.56°, and 18.79°, corresponding to plane spacings (d) of 4.84 Å, 4.78Å, and 4.72 Å, respectively. The peaks at 2θ = 18.33° and 18.56° were the strongest reflections of gibbsite ((002), d = 4.85 Å) [24] and nordstrandite ((010), d = 4.79 Å) [24], respectively. The peak at 2θ =18.79° was from bayerite ((001), d = 4.71 Å) [24]. DRIFT spectrum of the same sample (Figure 4) showed OH stretching bands characteristic of these three polymorphs of Al(OH)3, as reported in the literature [24, 25]. Accordingly, the AlN powders were hydrolyzed to mixtures of bayerite, nordstrandite, and gibbsite.
|
.jpg)
|
|
Figure 3. XRD pattern for powder A after 417 h of hydrolysis (Cu Kα radiation; after removing reflections from Cu Kα2 radiation).
|
|
.jpg)
|
|
Figure 4. DRIFT spectrum showing crystallized Al(OH)3 (powder A after 417 h of hydrolysis). The number for each band shows its position (cm-1).
|
Many authors have reported bayerite to be a hydrolysis product of AlN powders in water and water-based solutions at room temperature [7, 8, 10, 11, 18]. Svedberg et al. found the presence of a bayerite-gibbsite mixture after hydrolysis of AlN powders in water at 85°C [9]. To the best of our knowledge, the presence of nordstrandite or bayerite-nordstrandite-gibbsite mixture in the hydrolysis products of AlN has not been reported. However, mixtures of these three polymorphs of Al(OH)3 have been obtained from aluminum chloride or sulfate in solutions [26, 27] and from the hydrolysis of alumina powder in water vapor under hydrothermal condition [28].
Because XRD was not sensitive enough to surface change at the beginning of hydrolysis, DRIFT analysis was employed to detect this change. The resultant DRIFT spectra did not indicate formation of crystalline aluminum hydroxides at the beginning of hydrolysis (Δm < ~2%), but showed growth of the broad OH stretching band from 2540 - 3780 cm-1 (Figure 5), suggesting increasing amorphous hydroxide on the surface. The presence of this OH band in the spectra of starting powders was due to hydroxyl containing structures that commonly exist on the powder surface, as evidenced by TPDMS analysis [22, 29]. The increase of this OH stretching band at the beginning of hydrolysis was similar to the observation by Bowen et al. in their hydrolysis of AlN powders in water up to 8 h (>20% conversion), and was attributed by them, after XPS and XRD analysis, to formation of amorphous aluminum oxyhydroxide (AlOOH, which subsequently converted to crystalline bayerite) [8] AlOOH → Al(OH)3 conversion has also been reported in the corrosion of aluminum by water. For instance, immersion of aluminum sheets in water at 20°C has resulted in an amorphous boehmite-like (AlOOH) layer which converted successively to crystalline boehmite and bayerite [30], and aging of aluminum surfaces at room temperature in air at 100% humidity have led to initial formation of an AlOOH type layer which consequently aged to give an Al(OH)3 type surface layer [31]. Accordingly, in our study amorphous AlOOH could be formed initially and transformed subsequently to Al(OH)3, i.e., AlOOH acted as an intermediate product in the hydrolysis.
Accordingly, the overall hydrolysis reaction was
AlN + 3H2O → Al(OH)3 + NH3 (2)
(Ammonia gas was detected during the hydrolysis.) The reaction could proceed in two steps as follows, similar to those proposed by Bowen et al. in their study on the hydrolysis of AlN powders in water at room temperature [8].
|
.jpg)
|
|
Figure 5. DRIFT spectra showing increase of the OH stretching band at the beginning of hydrolysis.
|
|
AlN + 2H2O → AlOOHamorph + NH3 (3)
AlOOHamorph + H2O → Al(OH)3 (4)
|
Microstructural Evolution
The typical microstructural evolution of the AlN powders during hydrolysis is shown in Figures 6 (a) – (d). Evidently the surfaces of the powders were hydrolyzed first. Particles of Al(OH)3 polymorphs, nucleating and growing around the parent particles and their agglomerates, began to appear at a mass change of a few percent and subsequently formed agglomerates. As a result, the agglomerates formed by the reaction products became larger with the proceeding of hydrolysis, enveloping the unreacted AlN inside.
|
|
|
.jpg)
|
|
Figure 6: SEM observation of raw (a) and hydrolyzed powder A at Δm = 8.0% (b), 64.7% (c), and 89.7% (d).
|
Conclusions
AlN powders hydrolyze in moist air at room temperature, resulting in degradation of the powders. The initial hydrolysis product is amorphous AlOOH, which is further converted to a mixture of polymorphs of Al(OH)3 (bayerite, nordstrandite, and gibbsite), forming agglomerates around the unreacted AlN core. In the hydrolysis each powder shows an induction period, which is attributed to slow hydrolysis of the surface oxide/oxyhydroxide layer. The powders produced by the carbothermal process show the longest induction periods.
Acknowledgement
The authors thank the Ministry of Education, Culture, Sports, Science and Technology for financial support of this work through the 21st Century Center of Excellence (COE) Program.
References
1. G. A. Slack, R. A. Tanzill, R. O. Pohl and J. W. Vandersande, “The Intrinsic Thermal Conductivity of AlN”, J. Phys. Chem. Solids, 48 (1987) 641 – 647.
2. L .M. Sheppard, “Aluminum Nitride: A Versatile but Challenging Material”, Am. Ceram. Soc. Bull., 69 (1990) 1801-1812.
3. A. W. Weimer, “Carbide, Nitride, and Boride materials Synthesis and Processing”, Chapman and Hall, London, UK (1997).
4. K. Watari, M. Kawamoto and K Ishizaki, “Sintering Chemical Reactions to Increase Thermal Conductivity of Aluminium Nitride”, J. Mater. Sci., 26 (1991) 4727-4732.
5. K. Watari, K. Ishizaki and T. Fujikawa, “Thermal Conduction Mechanism of Aluminium Nitride Ceramics”, J. Mater. Sci., 27 (1992) 2627-2630.
6. K. Watari, K. Ishizaki and F. Tsuchiya, “Phonon Scattering and Thermal Conduction Mechanisms of Sintered Aluminium Nitride Ceramics”, J. Mater. Sci., 28 (1993) 3709-3714.
7. J. G. Highfield and P. Bowen, “Diffuse-Reflectance Fourier Transform Infrared Spectroscopic Studies of the Stability of Aluminum Nitride Powder in an Aqueous Environment”, Anal. Chem., 61 (1989) 2399-2402.
8. P. Bowen, J.G. Highfield, A. Mocellin and T.A. Ring, “Degradation of Aluminum Nitride Powder in an Aqueous Environment”, J. Am. Ceram. Soc., 73 (1990) 724-728.
9. L. M. Svedberg, K. C. Arndt and M. J. Cima, “Corrosion of Aluminum Nitride (AlN) in Aqueous Cleaning Solutions”, J. Am. Ceram. Soc., 83 (2000) 41-46.
10. S. Fukumoto, T. Hookabe and H. Tsubakino, “Hydrolysis Behavior of Aluminum Nitride in Various Solutions”, J. Mater. Sci., 35 (2000) 2743-2748.
11. H. Tamai, T. Hamauzu, A. Ryumon, K. Aoki, K. Oda and T. Yoshio, “Hydrothermal Corrosion and Strength Degradation of Aluminum Nitride Ceramics”, J. Am. Ceram. Soc., 83 (2000) 3216-3218.
12. K. Krnel and T. Kosmac, “Reactivity of Aluminum Nitride Powder in Dilute Inorganic Acids”, J. Am. Ceram. Soc., 83 (2000) 1375-1378.
13. C. Young and J. Duh, “Corrosion of Aluminum Nitride Substrates in Acid, Alkaline Conditions and Water”, J. Mater. Sci., 30 (1995) 185-195.
14. K. Krnel and T. Kosmac, “Reactivity of Aluminum Nitride Powder in Aqueous Silicon Nitride and Silicon Carbide Slurries”, J. Am. Ceram. Soc., 85 (2002) 484-486.
15. F. Barba, P. Ortega, J. Bermudo, M.I. Osendi and J.S. Moya, “Effect of Oxygen Content on the Corrosion of AlN Powder in Diluted Acid Solution”, J. Eur. Ceram. Soc., 13 (1994) 335-338.
16. Y. Shimizu, S. Takatsuki and M. Egashira, “Effect of Surface-Oxidation Treatment on the Chemical Stability of Aluminum Nitride Powder”, Nagasaki Daigaku Kogakubu Kenkyu Hokoku, 19 (1989) 87-93.
17. Y. Li, T. Qiu and J. Xu, “Effect of Thermal Oxidation Treatment in Air on the Hydrolysis of AlN Powder”, Mater. Res. Bull., 32 (1997) 1173-1179.
18. Y. Zhang and J. Binner, “Hydrolysis Process of a Surface Treated Aluminum Nitride Powder – A FTIR Study”, J. Mater. Sci. Lett., 21 (2002) 803-805.
19. H.B. Shan, Y. Zhu and Z.T. Zhang, “Surface Treatment and Hydrolysis Kinetics of Organic Films Coated on AlN Powder”, British Ceram. Trans., 98 (1999) 146-150.
20. A. Abid, R. Bensalem and B.J. Sealy, “The Thermal Stability of Aluminum Nitride”, J. Mater. Sci., 21 (1986) 1301-1304.
21. Y. Kameshima, S. Kuramochi, A. Yasumori and K. Okada, “Analysis of Surface State and Stability during Storage of AlN Powders by X-ray Photoelectron Spectroscopy”, J. Ceram. Soc. Jpn., 106 (1998) 749-753.
22. N. Saito, Ch. Ishizaki and K. Ishizaki, “Temperature Programmed Desorption (TPD) Evaluation of Aluminum Nitride Powder Surfaces Produced by Different Manufacturing Proceses”, J. Ceram. Soc. Jpn., Int. Ed., 102 (1994) 299-302.
23. N. Saito and K. Ishizaki, “ Auger Electron Spectroscopy of Aluminum Nitride Powder Surfaces”, J. Am. Ceram. Soc., 79 (1996) 1213-1217.
24. K. Wefers and C. Misra, “Oxides and Hydroxides of Aluminum, Alcoa Technical Paper No. 19, Revised”, Alcoa Laboratories, Aluminum Company of Amereica, Pittsburgh, US (1987).
25. H. W. Marel and H. Beutelspacher, “Atlas of Infrared Spectroscopy of Clay Minerals and Their Admixtures”, Elsevier, Amsterdam, the Netherlands (1976).
26. W. H. Gizten, “Alumina as a Ceramic Materials, Special Publication No. 4”, American Ceramic Society, Columbus, OH, US (1970).
27. K. P. Prodromou and A. S. Pavlatou-Ve, “Formation of Aluminum Hydroxides as Influenced by Aluminum Salts and Bases”, Clays and Clay Minerals, 43 (1995), 111-115.
28. Ch. Ishizaki, P. Raharjo, K. Sato and K. Ishizaki, “Mechanism of Hard Agglomerate Formation in a High Purity Sub-micron α-Alumina Powder”, J. Ceram. Soc. Jpn., 109 (2001) 16-22.
29. M. Nakamura, N. Saito, Ch. Ishizaki and K. Ishizaki, “Characterization of AlN Powder Surface by Temperature Programmed Desorption Mass Spectrometry”, Interceram, 50 (2001) 366-372.
30. R. K. Hart, “The Formation of Films on Aluminium Immersed in Water”, Trans. Faraday Soc., 53 (1957) 1020-1027.
31. T. N. Wittberg, J. D. Wolf and P. S. Wang, “Aluminum Hydroxide Growth on Aluminum Surfaces Exposed to an Air/1% NO2 Mixture”, J. Mater. Sci., 23 (1988) 1745-1747.
Contact Details
|