Degradation of Aluminum Nitride Powder in an ... - Wiley Online Library

5 downloads 0 Views 629KB Size Report
Tetard and M. Billy, “The Behaviour of Aluminium Nitride Powder in. Oxygen ... ”K. Wefers and G. M. Bell, “Oxides and Hydroxides of Aluminium,”. Technical ...
journal

J Am Ceram SOC, 73 [3] 724-28 (1990)

Degradation of Aluminum Nitride Powder in an Aqueous Environment Paul Bowen,* James G. Highfield,+Alain Mocellin,***and Terry A, Ring*.* Ecote Polytechnique Federale de Lausanne, Lausanne, CH-1015, Switzerland left open to the atmosphere. Abid et d 2also found AIN to be relatively insensitive to atmospheric moisture, but AlOOH was seen by XRD analysis when AlN powder was reacted . ~ the forwith H 2 0 at 100°C for 10 rnin. Hayashi et ~ 1report mation of AI(OH)3, from 27Al NMR studies, when AIN is aged in H z O for 2 d at room temperature. In the present work we have studied the degradation of AlN in an excess of H 2 0 over 24 h at room temperature, with more emphasis on the kinetics and mechanism of reaction, and characterizing the reaction products using various analytical and spectroscopic techniques (IR; XPS; XRD; C , H , N analysis; SEM; surface area and particle size measurement). The data are fitted to a model which best describes the kinetics and mechanism of the reaction.

The degradation of AIN powder in excess H 2 0 at room temperature for up to 24 h was investigated. Samples were characterized by various techniques ( I R XRD; SEM, XPS; C, H, N analysis; surface area, particle size, and weight change measurements). The reaction rate was found to be significant, with 80% of the AIN being consumed in 24 h. The initial reaction product was found to be a porous, amorphous, hydrated alumina with stoichiometry near AIOOH. After -16 h a crystalline phase, bayerite AI(OH),, was detected which became the predominant phase after 24-h contact. The kinetics of the AIN consumption were found to be first order and the reaction rate linear. The kinetic data fitted an unreacted core model with a porous product layer where the surface chemical reaction controlled the overall kinetics. [Key words: aluminum nitride, hydration, powders, kinetics, bayerite.]

11. Experimental Procedure

(1) Sample PreparationlTreatment I. Introduction The A1N powder used for all the studies was an AIN UCH interest has been shown in AIN as a ceramic maGrade C powder5 with nominal particle size of 1.2 pm, a terial for a variety of applications (e.g., electronic subsurface area of 6.2 mZ/g, and 2.2 wt% oxygen. Boehmite strates' and refractory materials*) because of its advantageous (A100H) and bayerite standards' were also obtained. The physical properties such as high thermal conductivity, high AI2O3standard** used had a median diameter of -0.5 pm. electrical resistivity, and good mechanical properties at high The reaction with HzO was carried out as follows: AlN temperature. In a different context, AlN is also of potential powder (20 g) was mixed with five times its weight of deionuse for the production of AI2O3/TiN composite ceramics via ized water, stirred, and then treated with an ultrasonic horn hot reaction pressing of A1N and Ti02 powder mixture^.^-^ for 2 min (150 W, 20 kHz). The mixture was then stirred for 2 The water sensitivity of AlN has been previously n ~ t e d ; ' . ~ . ~ to 24 h (at 25°C) before being dried. To help remove any exconsequently, nonaqueous solvents were used in the colloidal cess water, 2-propanoltt (400 cm3) was added to the mixture processing of the AlN powder in the production of the tialon and then the slurry was filtered. The resulting paste was dried ceramics. In this paper, studies on the rate and mechanism of in a forced-air oven at 60°C for 1 h. No further weight the reaction between AIN and HzO have been carried out to changes were seen when the dried powder was heated again at assess the possible use of AIN in aqueous suspensions (for 100°C for 1 h. The dried powders were stored in airtight plasboth economic reasons and control of colloidal processing). tic jars for subsequent analysis. Previous studies on the oxidation stability of AIN have con(2) Characterization centrated mainly on the high-temperature region between The various treated powders were characterized using the 900" and 1400°C in air or oxygen,2z"10 where the product was following techniques. A1203and the oxidation rate parabolic. Sat0 et al." found the Elemental analysis**was made with a sample size of -3.0 rate of oxidation of AIN by water vapor between 900" and mg. The aluminum contents of the samples were found by 1250°C to be linear and controlled by the surface chemical remeasuring the weight change on oxidation in air at 1400°C action between AIN and adsorbed water vapor. Above 13S0"C assuming the final oxidation product was A1203 (confirmed they saw a parabolic rate of oxidation controlled by the diffuby XRD). Measurements of oxygen content in the various sion of water vapor through an A1203 layer. Studies at lower powders were then made by difference from the nitrogen, temperatures have been less detailed with respect to the kihydrogen, and aluminum contents. Some oxygen values were netics and reaction mechanism in the reaction between AIN checked by direct analysis@ and found to be consistent (to and water. The reaction with atmospheric moisture andlor +5% relative) with those obtained by material balance. oxygen, discussed by Slack and McNelly: showed the formaX-ray powder diffraction (XRD) analysis" was performed tion of aluminum oxide, protective to further oxidation when on the crystalline phases found in the treated AlN powders. X-ray photoelectron spectroscopy (XPS) data were colJ. Smialek-contributing editor lected?** Gold (Au 4f7,4 and copper (Cu 2P3,z) lines at 84.0

M

Manuscript No. 198418. Received May 16, 1989; approved August 22, 1989. Supported by the Swiss National Science Foundation under Contract No. 4.842.0.85.19. *Member, American Ceramic Society. *Laboratoire de Technologie des Poudres, Department des Materiaux. 'Laboratoire de Chimie Technique. *Laboratoire de Ceramique, Department des Materiaux.

'Hermann C. Starck, Berlin, FRG. 'IAlusuisse, Neuhausen am Rheinfall, Switzerland. **A-16 Powder, Alcoa, Bauxite, AR. "Grade 995 Extra Pure, Merck. "2400 CHN analyzer, Perkin-Elmer, Eden Praire, MN. "LECO TC316 oxygen analyzer, Hermann C. Starck. "D500 X-ray diffractometer, Siemens, Munich, FRG. ***PHI-590-550, Perkin-Elmer.

724

Degradation of Aluminum Nitride Powder in an Aqueous Environment

March 1990

and 932.6 eV, respectively, were used for calibration, and the adventitious carbon 2s peak at 284.6 eV as an internal standard to compensate for any charging effects. The percentage molar fraction of Al, N, and 0 monitored were renormalized to loo%, accounting for the presence of adventitious carbon. The estimated XPS probe depth of the N 1s electrons for these alumina-like samples was estimated to be -70 A (7.0 nm).14 Diffuse-reflectance infrared Fourier transform (DRIFT) spectra from 4000 to 500 cm-' at 4-cm-' resolution were obtained on an FTIR spectrophotometerttt equipped with a diffuse-ref lectance accessory."* Diffuse-ref lectance spectroscopy (DRS) is a versatile and convenient technique to obtain optical absorption spectra from opaque or highly scattering samples, e.g., fine powders, which are difficult to study by conventional means. DRS was originally developed about 20 years ago by Kortum" for application in the UV-visible region. More recently, the advent of commercial Fourier transform infrared spectrometers has enabled its extension to this valuable region of the spectrum, as pioneered chiefly by Griffiths et al., 12~13 under the acronym DRIFT. Spectral signal-to-noise was improved by coaddition of 500 scans, resulting in a data acquisition time of 3 min. Because of the high intensity of most of the IR spectral bands of interest, samples were routinely ground with an excess of KBr (10 to 40 times the weight of the sample, i.e., 0.1 g of AlN). This served to enhance spectral definition. The nitrogen BET surface area of each sample was calculated"$ from the nitrogen adsorption isotherm. The samples were dried at 110°C for 1 h under a vacuum of 4 x lo-' mbar (4 Pa) prior to nitrogen adsorption. The particle size distribution for the parent and reacted AlN powders was analyzed." Slurries for the measurements were prepared by suspending 2.5 g of powder in 20 g of 0.1% polyvinylpyrrolidone (PVP-K30) in 2-propanol solution and then treated for 15 min with an ultrasonic horn (150 W, 20 kHz). Scanning electron microscopy (SEM) studies of the morphology of powders were carried out****before and after HzO treatment. Treated powders were suspended in 2-propano1, deposited on glass slides, and then gold coated for SEM analysis. 111. Results

(1) Characterization

The elemental compositions of the AIN powders before and after reaction with HzO are listed in Table I. This shows a steady decrease in nitrogen content with a corresponding increase in oxygen and hydrogen. Subtraction of the aluminum and nitrogen concentration (assuming they constitute AIN) allows the molar ratio for the aluminum and oxygen to be calculated for the reaction product (see Table I). The XRD data collected on A1N powders after 2- to 24-h contact with water showed no crystalline product other than "'DA3.002, Bomem, Quebec, Canada. '"DRA-2C0, Harrick Scientific Corp., Ossining, NY. 5'*Sorptomatic 1900, Carlo Erba Strumentazione, Rodano, Italy. ?%digraph 5000 D, Micromeritics, Norcross, GA. ****SE250/SE360,Cambridge, Cambridge, U.K.

725

AIN after 8-h contact. The sample after 16-h reaction showed low-intensity reflections for AI(OH)3 (bayerite modificat i ~ n ) . ' ~ .The ' ~ main crystalline component in the sample which had reacted for 24 h was the aluminum hydroxide, bayerite, with AIN still present in a reasonable quantity and weak reflections corresponding to nordstrandite (another Al(OH), m~dification)'~.''as shown in Fig. 1. In the XPS spectra the A1 2p, 0 Is, and N 1s lines were well resolved and easily detected. Because of the similarity in the Al 2p and 0 1s binding energies for AIN, A1203, AlOOH, and A1(OH)3,'8-20no information from the chemical shifts could be extracted. The XPS data do, however, give information on the relative molar percentage of the aluminum, oxygen, and nitrogen in the surface region of the A1N powders. The XPS data collected on these samples (Table 11) showed a rapid decrease in the nitrogen content to -1% after 4 h and none at all being detected after 8- and 16-h reaction. A small nitrogen signal was however seen after 24-h contact (reproduced for several samples), suggesting some inhomogeneity in the reaction of the surface of the AlN or perhaps cleaving of heavily reacted particles. The initial AlN surface shows an O/Al ratio, after subtracting the A1N contribution, near to that of Alz03, supporting previous suggestions21that the oxygen at the surface of AIN powders exists as a noncrystalline AlZO3,albeit probably hydrated to some extent16 (see IR data below). The aluminum-to-oxygen ratio is similar for all reacted samples (at 1.8 to 1.9), approaching that of boehmite. It is possible that there is some dehydration of the samples in the high vacuum of the XPS spectrometer, as indicated in the bayerite XPS data (Table 11). This shows low oxygen content with perhaps loss of the water associated in the bayerite crystals." DRIFT spectra of samples (10% in KBr) reacted with water are presented in Fig. 2 . The main features are the systematic growth of strong bands characteristic of OH stretching and (weaker) bending fundamentals centered at -3300 and 1060 cm-', respectively, a new band at 1510 cm-', and a decrease of the band at 1330 cm-'. From the oxygen analysis, the AIN starting material already has a small oxide content (2.6%), which correlates with the presence of the OH band in Fig. 2(a). In other respects the spectrum, when extended down to 500 cm-' (not shown), closely resembles the reference spectrum of AIN from the literature.22 The very strong AI-N stretching fundamental centered at 700 cm-' can be seen along with a group of features around 1400 cm-', the strongest of which (1330 cm-') is almost certainly an overtone of this fundamental. The band which develops at 1500 cm-' is assigned as the first overtone of an A1-0 stretching fundamental of the growing oxyhydroxide phase, the latter generally occurring in the range 700 to 800 ~ m - l . ' ~No evidence is found for the familiar deformation band of molecular or crystal H 2 0 at 1640 cm-1.24Throughout this series, no correspondence is seen with reference spectra or literature dataZS of hydrated aluminas (AI2O3 nHZO), boehmite (AIOOH), bayerite, gibbsite, or nordstrandite (all different structural forms of Al(OH),). DRIFT spectra of samples after longer exposure times show crystalline features. After 16 h the OH bands and the

Table I. Elemental Composition and O/Al Molar Ratio (of the A1 Oxide/Hydroxide Component) of Powders Before and After Treatment from CHN and Weight Change Analysis Composition (wt%) Reaction time (h)

Al'

N'

Ht

0'

AIN'

O/Al molar ratio for A1 oxide/hvdroxide

0 2 4 8 16 24

64.9 64.3 61.3 59.7 51.1 41.2

32.5 29.8 27.4 23.6 15.3 6.9

0.02 0.05 0.3 0.85 1.7 3.0

2.6 5.9 10.9 15.9 31.9 48.9

95.O 87.1 80.1 69.0 44.7 20.2

1.8 1.9 2.1 1.9 2.5 3.0

*50.25%. '+0.75%. '+5.0%. '*0.4%. "0.8%

Journal of the American Ceramic Society - Bowen et al.

726 OA1N R

A

Bayerite

A

Degrees

(28)

Fig. 1. XRD diffractograms for A1N powders after 0-, 16-, and 24-h reaction with HzO.

Fig. 2. Drift spectra of AIN powders after treatment in H20 for: (a) 0 h, (b) 8 h, (c) 12 h, (d) 24 h, and (e) bayerite reference.

A1-0 overtone at 1500 cm-’ are stronger and signs of the onset of a phase transformation to bayerite are also discernible (Fig. 2(c)). Weak, narrow line-width features at 3656, 3548, and 3440 cm-’ can be seen in the OH stretching region and a new peak at 980 cm-’, all bands characteristic of the bayerite Al(OH)3 standard shown in Fig. 2(e). After 24 h (Fig. 2(d)), the spectrum indicates that the crystallization process is essentially complete. The change in nitrogen BET surface area values versus time of reaction shows a marked increase from 8.3 to 29.7 mz/g between 2- and 4-h reaction and a maximum value after 16 h of 137.7 mz/g. The surface area then shows a decrease from the 16-h value to 68.1 m’/g. The particle size distributions for the parent AlN and powders after reaction show an increase in the median diameter (dso,diameter at 50% cumulative mass) as time of reaction increases. The parent A1N powder median is 1.2 pm, and after 24 h this has increased to 3.8 pm. This could be due to agglomeration, loss of fines during processing, or real particle growth. The SEM studies show significant agglomeration to have taken place on reaction (Fig. 3), but the individual particle shape and size are still very similar to the parent AlN even after 24-h reaction. (2) Kinetic Dnta It is usual to study the oxidation rate of a material by following the oxygen weight gain as a function of timeP6

Table 11. Surface Mole Fractions of Oxygen, Nitrogen, and Aluminum from 0 Is, N Is, and A1 2p XPS Spectra for AIN Powders After Reaction with HZOwith Similar Data for Aluminum Oxide/Hydroxide Standards Reaction time (h)

0

0 2 4 8 16 24 AlOOH boehmite AI(OH)3bayerite A1203alumina

37.5 50.9 64.1 64.6 65.5 63.7 67.2 70.7 57.1

‘210%.

Vol. 73, No. 3

Fraction (mol%)* N A1

18.3 10.5 1.1 1.2

44.3 38.6 34.7 35.4 35.2 35.2 32.8 29.8 42.9

O/Al molar ratio at surface

1.5 1.8 1.9 1.9 1.9 1.9 2.1 2.4 1.4

Figure 4 shows the percentage oxygen weight gain (from data in Table I) for the AIN powder on reaction with HzO over 24 h which shows a linear rate. The kinetic data are discussed in terms of a mechanistic model in the following section. IV. Discussion (1) Reaction Products The parent AIN contains -3 wt% oxygen which, from the XPS and elemental analysis results, has a stoichiometry close to that of A1203but with a slightly higher oxygen content, suggesting a partially hydrated surface. The IR data support this idea, showing evidence for OH in the AlN spectrum. No crystalline Al2O3was detected in its XRD pattern or IR spectrum. This suggests a partially hydrated noncrystalline A1203 surface layer as previously reported.” On reaction with HzO, no crystalline phases are detected after 8 h and only small quantities after 16 h. The XPS analysis shows the gradual disappearance of the N 1s signal and an increase in the 0 1s signal indicative of a growing oxide layer. Assuming the nitrogen in the sample remains as AlN, this surface layer approaches the stoichiometry of boehmite (AIOOH), though no evidence is seen in either IR or XRD data for the presence of a crystalline b ~ e h m i t e . ~ ’In* ~the ~ early stage of reaction, i.e., up to -8 h, DRIFT shows a systematic growth of the amorphous, hydrated oxide layer. This is an encouraging result in view of the inhomogeneous nature of the samples. A more detailed discussion of quantitative measurements using DRIFT and its susceptibility to possible artifacts due to the above inhomogeneities is presented el~ewhere.’~ Surface area measurements show this amorphous hydrated layer to be highly porous with a much higher surface area than the parent AIN (130 cf. 6 mz/g). Between 16 and 24 h of reaction, both XRD and IR analysis show the formation of the crystalline phase bayerite, AI(OH)3, and the specific surface area correspondingly decreased. From previously reported studies on the production of alumina gels;’ the formation of crystalline products was seen to be slow (i.e., over months) in the preparation of alumina gels from sulfate or chloride solutions when the OH- concentration was low. In contrast, mixtures of bayerite, nordstrandite, and gibbsite were formed within hours when the OH- level was 3 times that of the A13+level. Such a situation could be envisaged in

March 1990

Degradation of Aluminum Nitride Powder in an Aqueous Environment

727

,WTOXYGEN GAIN -%

50-

MEASURED 30-

I

I

I

I

2

8

16

24

TIME (hrs) Fig. 4. Measured percentage oxygen weight gain of

reacted powder vs time.

representation of the degradation of AlN powder in excess H 2 0 . In the unreacted core model the reaction starts at the outer surface of the particle and moves inward, leaving a layer of reacted material. The unreacted core of parent material will shrink in size as the reaction continues. For an irreversible reaction there are three regimes of control envisaged: (i) mass transfer of reactant (liquid or gas) through a film surrounding the particle to the surface of the solid, (ii) diffusion of the reactant through the product layer to the surface of the unreacted core, and (iii) chemical reaction of reactant and solid at the core surface. The slowest step will be rate controlling. Kinetic equations for the cases where each of these steps is rate controlling have been developed by L e ~ e n s p i e l ~ ~ for spherical particles:

(i) Mass transfer through fluid film t = TMXB

(4)

(ii) Diffusion through product layer

t = T,[1 - 3(1 - X B ) ” + ~ 2(1 - XB)]

Fig. 3. SEM micrographs showing particle size and morphology of parent A1N (A) and after 8-h reaction (B) in H 2 0 .

our reaction mechanism whereby only amorphous products are formed up to a point when the OH- concentration exceeds (by the dissolution of liberated ammonia from the AIN/H*O reaction in progress) the threshold level for the formation of crystalline products. The pH of the reaction mixture was seen to slowly increase from 7 at the outset to 10 after 5.5 h and remained constant at this value of 10 thereafter. The transformation from the amorphous monohydroxide to the crystalline bayerite is generally believed to be a dissolution-recrystallization p r o ~ e s s . From ~ ~ , ~ the ~ data collected here it is not possible to discern whether this transformation to a crystalline product takes place via solid state or by a dissolution-recrystallization mechanism. Recent studies on the aging of aluminum surfaces at room temperature in air at 100% humidity have also shown the initial formation of an AlOOH type layer which consequently ages to give an Al(OH)3type surface layer.33 A possible stoichiometric representation of the reactions taking place could be written as follows: AlN

+ 2H20 -+

NH3

+ HzO

A100Harnorph

A100H,,,,,~

+ NH3

+ OH-

(2)

+ H2O “;AI(OH)3,xsral

(3)

NH:

(2) Kinetic Model Two simple idealized models, the progressive-conversion model and the unreacted-core model, for fluid particle reactions have been treated by L e ~ e n s p i e lSeveral .~~ experimental facts indicate the unreacted core model as being the preferred

(5)

(iii) Surface chemical reaction (first-order reaction) = T R [~ (1 - x ~ ) ’ ’ ~ ] where t is reaction time, T‘s are the times for complete reaction of a particle for each step, and XB is the volume fraction of the unreacted core with respect to the total particle volume. If the mathematical functions expressed in Eqs. (4),(5), and (6) are plotted versus time, then the model that best describes the reaction under study should give a straight line. Figure 5 shows the fit to two of the three regimes considered for the unreacted (or shrinking) core model. The fit is very good for both the liquid film resistance and chemical reaction control, with the square of the coefficients of regression, R2, of 0.998and 0.988,respectively. (The product layer diffusion controlled regime gave a poor fit, with an Rz of 0.897.)In order to ascertain if one of these models more closely describes the reaction in question, calculations of kl (mass transfer coefficient from TM)and ks (first-order chemical reaction rate constant from TR)were made using the following relationships:

pB is the A1N density, b is the stoichiometric coefficient of AIN (equal to unity from Eq. (l)),R is the radius of the particle (taken as 1.0 pm, the AlN dso median), Ca is the concentration of reactant water, and TMand TR are given by the gradient of plots in Figs. 5(a) and (b), respectively. The comparison of a theoretical value of kl with that calculated from above gives a discrepancy of lo8,whereas the first-order reaction rate constant calculated from TR is within a factor of 50 for the consumption of 80% of a 1.2-pm AIN particle as seen in the reaction data (Table I). This suggests that an unreacted core model where chemical reaction is rate controlling best describes the reaction under consideration.

Journal of the American Ceramic Society - Bowen et al.

728

There are a number of results from the reaction product characterization which support this choice of kinetic model. (i) The fundamental particle size and morphology do not change significantly (although significant agglomeration is shown from the particle size measurements). (ii) The amorphous hydroxide-like phase shows high porosity (i.e., product layer diffusion resistance should be small). (iii) XPS analysis shows a rapid decrease in nitrogen content and increase in oxygen in the uppermost surface layers of the powder. The linear reaction rate seen is also indicative of a porous product layer, which is expected when the molar volumes of the parent and product material are significantly differentz6 as seen for the oxidation of calcium3’ and alumin~rn’~ at low temperatures. Although there are limitations in the assumptions made in the unreacted core model (such as uniform particle morphology and size, a sharply defined reaction front, and the absence of temperature gradients within the particle) the process seems simple enough in this case for the unreacted core model to describe the kinetic data well.

V. Conclusions

A1N powder in excess HzO at room temperature reacts initially to give an amorphous layer of stoichiometry close to AIOOH. After -16 h this is transformed into a crystalline hydroxide, bayerite AI(OH)3. The kinetics of the reaction are well described by an unreacted core model where the chemical reaction at the product layer/unreacted core interface controls the reaction rate. The AlN consumption is found to be first order and the corresponding reaction rate linear, which would be expected for a porous product layer. After contact with HzO for 24 h, -80% of the parent AlN powder is consumed. The use of H 2 0 in the colloidal ceramic processing of AIN is not viable when contact times are of the order of hours (unless oxygen contents of 10% or so can be tolerated). The use of organic solvents such as 2-propano13(found to be the best of 20 or so organic solvents by Grobety and Mocellin)’ is preferable.

Acknowled ments:

The authors would like to thank Dr. H-J. Mathieu and Mr. N. &anthopoulos for XPS analysis, Mr. B. Senior for SEM Analysis, and Dr. P. Moeckli for XRD analysis.

fn (XB)

0 0

_____.._

2

8

16

24

TIME ( h r s )

Fig. 5. Plots of two of the controlling regimes for unreacted core model fitted to AlN degradation data.

Vol. 73, No. 3

Ryferences H. Kolker and W. Grellner, “Aluminum Nitride-A New Substrate Material,” Ceram. f i r u m Int., 65 [3/4] 75-79 (1988). ’A. Abid, R. Bensalem, and J. Sealy, “The Thermal Stability of AIN,” J. Muter. Sci., 21, 1301-304 (1986). ’B. Grobety, T. Sperisen, and A. Mocellin, “Influence of Powder Processing on Mechanical Properties of Tin/A1203 Particulate Composites Prepared by Reaction Hot Pressing of AIN/Ti02”; pp. 315-20 in Science of Ceramics, Vol. 14. Edited by D. Taylor. Institute of Ceramics, Shelton, U.K., 1988. 4A. Mocellin and G. Bayer, “Chemical and Microstructural Investigations of High-Temperature Interactions Between AIN and TiOz,”J. Muter. Sci., 20, 3697-704 (1985). ’B. Grobety and A. Mocellin, “Preparation of AIN/Ti02 Powder Compacts Using Colloidal Methods”; to be published in Cerum. Int. 6G. A. Slack and T. F. McNelly, “Growth of High Purity AIN Crystals,” J. Cryst. Growth, 34, 263-79 (1976). 7S. Hayashi, K. Hayamizu, and 0. Yamamoto, “”AI High-Resolution Solid-state NMR Study of Hydration of Ultrafine Powder of Aluminium Nitride,”Bull. Chem. Soc. Jpn., 60 [2] 761-62 (1987). ‘V. A. Lavrenko and A. F. Alexeev. “Oxidation of Sintered Aluminium Nitride,” Cerum. Znt., 9 [3] 80-82 (1983). 9D. Tetard and M. Billy, “The Behaviour of Aluminium Nitride Powder in Oxygen Between 900” and 1300°C”; pp. 512-19 in Reaction Kinetics in Heterogeneous Chemical Systems. Edited by P. Barret. Elsevier Scientific, New York, 1975. ”T.Sato, K. Haryu, T. Endo, and M. Shimada, “High Temperature Oxidation of Hot-Pressed Aluminium Nitride by Water Vapour,” J. Muter. Sci., 22,2277-80 (1987). ”G. Kortum, Reflectance Spectroscopy. Springer-Verlag, Berlin, FRG, 19F2. P. R. Griffiths and M. P. Fuller, Advances in Infrared and Raman Spectroscopy, Vol. 9; p. 63. Heyden, Philadelphia, PA, 1982. I3P. R. Griffiths and J. A. Haseth, Chemical Analysis, Vol. 83, FourierTransform Infrared Spectrometry; pp. 536-63. Wiley, New York, 1986. I4M.P. Seah and W. A. Dench, “Quantitative Electron Spectroscopy of Surfaces-A Standard Data Base for Electron Inelastic Mean Free Paths in Solids,” SurJ Interface Anal., 2-11 (1979). “Powder Diffraction File, Card No. 20-11. Joint Committee on Powder Diffraction Standards, Swarthmore, PA. 16B.C. Lippens and J. 5. Steggerda, Physical and Chemical Aspects of Adsorbents and Catalysts; pp. 171-211. Edited by B. G. Linsen. Academic Press, London, U.K., 1970. ”Powder Diffraction File, Card No. 24-6. Joint Committee on Powder Diffraction Standards, Swarthmore, PA. “R. L. Tapping, I. Aitchison, and D. G.W. Goad, ‘Applications of AES and XPS to Characterization of Aluminium and its Compounds,” Cun. J. Spectrosc., 28 [3] 87-94 (1983). I9C. D. Wagner, D. E. Passoja, H. F. Hillery, T. G . Kinisky, H. A. Six, W.T. Jansen, and J. A. Taylor, ‘Auger and Photoelectron Line Energy Relationships in Aluminium-Oxygen and Silicon-Oxygen Compounds,” J. Vuc. Sci. Technol., 21 [4] 933-44 (1982). ”3. A. Kovacich, J. Kasperkiewicz, D. Lichtman, and C. R. Aita, “Auger Electron and X-ray Photoelectron Spectroscopy of Sputter Deposited Aluminium Nitride,”J. Appl. Phys., 55 [8] 2935-39 (1984). 21M. Zulfequar and A. Kumar, “Electrical Conductivity and Dielectric Behaviour of Hot-Pressed AIN,”Adu Cerum. Muter., 3 [4] 332-36 (1988). ”R.A. Nyquist and R.O. Kagel, Infrared Spectra of Inorganic Compounds. Academic Press, New York, 1971. =M. C. Steggman, D. Vivien, and C. Mazieres, “Studies of the Infrared Vibration Modes in Aluminium Oxyhydroxides Boehmite and Diaspore,” Spectrochim. Actu, Part A, 29, 1653-63 (1973). “K. Nakamoto, Infrared Spectra of Inorganic and Coordination Compounds; pp. 159-61. Wiley, New York, 1963. ZSD.L. Frederickson, “Characterization of Hydrated Aluminas by Infrared Spectroscopy,” Anal. Chem., 26, 1883-85 (1954). 26K. Hauffe, Treatise on Solid State Chemistry, Vol. 4, Reactivity of Solids. Edited by N. B. Hannay. Plenum Press, New York, 1976. ”Powder Diffraction File, Card No. 21-1307. Joint Committee on Powder Diffraction Standards, Swarthmore, PA. %R. Tettenhorst and D. A. Hofmann, “Crystal Chemistry of Boehmite,” Clays Clay Miner., 28 [5] 373-80 (1980). 29J. Highfield and P. Bowen, “Diffuse-Reflectance FTIR Spectroscopic Studies of into the Stability of Aluminium Nitride Powder in Aqueous Environment,” Anal. Chem., 61 [21] 2399-402 (1989). ’OW. H. Gitzen, ‘Alumina as a Ceramic Material,” Special Publication No. 4. American Ceramic Society, Columbus, OH, 1970. ”A. C. Pierre and D. R. Uhlman, “Gelation of Aluminum Hydroxide Sols,”J. A m . Cerum. SOC.,70 [l] 28-32 (1987). ”K. Wefers and G. M. Bell, “Oxides and Hydroxides of Aluminium,” Technical Paper No. 19, Alcoa Research Laboratories, 1972. ”T.N. Wittberg, J. D. Wolf, and P. S. Wang, ‘Muminiurn Hydroxide Growth on Aluminium Surfaces Exposed to an Air/l% NOz Mixture,” J. Muter. Sci., 23, 1745-47 (1988). 340. Levenspiel, Chemical Reaction Engineering; pp. 357-408. Wiley, New York, 1972. ”U. Evans, The Corrosion and Oxidation of Metals; pp. 18-48. Edward Arnold, London, U.K., 1967. 36T.B. Grimley and M.W. Trapnell, “The GaslOxide Interface and the Oxidation of Metals,” Proc. R . Soc. London, A, 234, 405-18 (1956).