Enhancement of Lithium Amide to Lithium Imide Transition via

Sep 16, 2006 - Tippawan Markmaitree,† Ruiming Ren,†,‡ and Leon L. Shaw*,†. Department of Materials Science and Engineering, UniVersity of Conn...
0 downloads 0 Views 375KB Size
20710

J. Phys. Chem. B 2006, 110, 20710-20718

Enhancement of Lithium Amide to Lithium Imide Transition via Mechanical Activation Tippawan Markmaitree,† Ruiming Ren,†,‡ and Leon L. Shaw*,† Department of Materials Science and Engineering, UniVersity of Connecticut, Storrs, Connecticut, and School of Materials Science and Engineering, Dalian Jiaotong UniVersity, Dalian, China ReceiVed: January 10, 2006; In Final Form: May 28, 2006

The decomposition of lithium amide (LiNH2) to lithium imide (Li2NH) and ammonia (NH3) with and without high-energy ball milling is investigated to lay a foundation for identifying methods to enhance the hydrogen uptake/release of the lithium amide and lithium hydride mixture. A wide range of analytical instruments are utilized to provide unambiguous evidence of the effect of mechanical activation. It is shown that ball milling reduces the onset temperature for the decomposition of LiNH2 from 120 °C to room temperature. The enhanced decomposition via ball milling is attributed to mechanical activation related to the formation of nanocrystallites, the reduced particle size, the increased surface area, and the decreased activation energy. The more mechanical activation there is, then the more improvement there is in enhancing the decomposition of LiNH2. It also is found that the activation energy for the decomposition of LiNH2 without ball milling is 243.98 kJ/mol, which is reduced to 222.20 kJ/mol after ball milling at room temperature for 45 min and is further reduced to 138.05 kJ/mol after ball milling for 180 min. The rate of the isothermal decomposition at the later phase of the LiNH2 decomposition is controlled by diffusion of NH3 through the Li2NH layer.

I. Introduction The reaction path for hydrogen storage in Li3N was established to consist of two steps1-6 as shown in reactions 1 and 2.

Li3N + H2 T Li2NH + LiH

(1)

Li2NH + H2 T LiNH2 + LiH

(2)

Reaction 1 can store and release 5.0 wt % H2, whereas the corresponding value for reaction 2 is 6.5 wt % H2.1 Reaction 2 has an enthalpy of absorption of -45 kJ/mol.1 When this enthalpy of absorption is used and the entropy change associated with absorption is assumed to be -130 J/K mol, the thermodynamic equilibrium temperature for absorbing and desorbing hydrogen under a pressure of 1 bar would be 73 °C for reaction 2. However, previous studies1,7 have shown that reversible hydrogen uptake and release at 1 bar corresponding to reaction 2 are achieved at ∼280 °C. The temperature discrepancy between the thermodynamic prediction and the experimental measurements can presumably be attributed to the slow kinetics of the absorption and desorption processes. It was shown that the reverse process of reaction 2, (i.e., desorption of the lithium amide (LiNH2) and lithium hydride (LiH) mixture) proceeds with two elementary reactions as defined by reactions 3 and 4.8,9

LiNH2 ) 1/2Li2NH + 1/2NH3

(3)

/2NH3 + 1/2LiH ) 1/2LiNH2 + 1/2H2

(4)

1

As shown in 3 and 4, one complete cycle of these two elementary successive reactions consumes 50% LiNH2 and 50% * Corresponding author. E-mail: [email protected]. Address: 97 North Eagleville Road, U-3136, Storrs, Connecticut 06269-3136. Phone: (860) 486-2592. Fax: (860) 486-4745. † University of Connecticut. ‡ Dalian Jiaotong University.

LiH in the powder mixture. Such successive reactions would continue until all LiNH2 and LiH completely transform to lithium imide (Li2NH) and H2. It has been established that reaction 4 takes place extremely fast in the order of microseconds.8 As such, the rate-limiting step for reaction 2 would be reaction 3 if reaction 3 has a slow reaction rate. If this is indeed the case, the key to reduce the hydrogen absorption and desorption temperatures corresponding to reaction 2 would be to enhance the reaction rate of reaction 3. In the early 1950s, studies on reaction 3 were conducted.10 Recently, this reaction was reexamined by several investigators.11,12 It was shown that ball milling can enhance the decomposition of LiNH2.11 For example, the temperature for 10% weight loss because of the evolution of NH3 was decreased from about 380 to 350 °C for LiNH2 without and with ball milling, respectively. The activation energy for the decomposition of the ball-milled LiNH2 was determined and was found to be about 128 kJ/mol.11 In addition, it was shown that LiNH2 decomposes at higher temperatures than Mg(NH2)2, which is a result consistent with the thermodynamic analysis.12 In this study, the reaction kinetics of reaction 3 was investigated further. In particular, the effect of high-energy ball milling on the decomposition rate of LiNH2 was studied systematically. A wide range of analytical instruments was utilized to provide unambiguous evidence of the effect of mechanical activation. The variation of the activation energy for the decomposition of LiNH2 with the degree of mechanical activation was related to the structural and energy state changes induced by ball milling, and the rate-controlling mechanism for the decomposition of LiNH2 was proposed. II. Experimental Section Lithium amide with 95% purity was purchased from Alfa Aesar. High-energy ball milling was performed using a modified Szegvari attritor. The modified Szegvari attritor was shown to be effective in preventing the formation of the dead zone and

10.1021/jp060181c CCC: $33.50 © 2006 American Chemical Society Published on Web 09/16/2006

LiNH2 to LiNH via Mechanical Activation producing uniform milling products within the powder charge.13 Furthermore, a previous study demonstrated that the seal of the canister of the attritor is airtight and there is no oxidation during ball milling of LiH.14 The canister of the attritor and the balls (6.4 mm in diameter) were made of stainless steel. The loading of the balls and the powder to the canister was performed in a glovebox filled with ultrahigh-purity argon that contained Ar 99.999%, H2O < 1 ppm, O2 < 1 ppm, H2 < 3 ppm, N2 < 5 ppm, and THC < 0.5 ppm (to be referred to as Ar of 99.999% purity hereafter). The ball-to-powder weight ratio was 60:1, the milling speed was 600 rpm, and the milling temperature was maintained at 20 °C, which was achieved by water cooling at a flow rate of 770 mL/min. Ball milling time was a variable in this study and was adjusted from 45 to 180 min depending on the specific experiment. To assess the effect of the Fe impurity from the wear of stainless steel balls, some LiNH2 powder samples also were ball milled using WC balls, and their decomposition behaviors were compared with those samples ball milled using stainless steel balls. All the samples before and after high-energy ball milling were subjected to various characterizations and handled in a glovebox filled with Ar of 99.999% purity. The thermogravimetric analysis (TGA) was conducted using a TA instrument (TGA Q500). LiNH2 samples of 25-55 mg were loaded into a Ptmicrobalance pan with a short exposure to air (less than 30 s). The system was then flushed immediately with Ar of 99.999% purity for 90 min before heating from 20 °C to 550 °C with a heating rate of 10 °C/min. The argon flow rate was maintained at 60 mL/min for the entire holding and heating process. The outlet gas from TGA was constantly analyzed using a quadrupole residual gas analyzer (RGA) equipped with a mass spectrometer (model ppt-c300-F2Y). The gases monitored included H2, NH3, N2, O2, CO2, and H2O. The release of ammonia from lithium amide also was analyzed using a gas chromatograph-mass spectrometer (H/P 6890 GC/MS). This GC/MS instrument provided dynamic thermal desorption conditions and direct gas composition analysis at any pre-set temperature. Specifically, the sample to be analyzed was loaded into a syringe injector purged with helium for 1 min before being injected to the analysis chamber, which was filled with flowing helium and set at a desired temperature. The outlet gas resulting from the 2 min holding at the analysis chamber was analyzed using the mass spectrometer with the examination range spanning from 10 to 200 Da. The specific surface area (SSA) of the powders before and after high-energy ball milling was determined by the BET method using a SSA analyzer from QUANTACHROME Corporation (NOVA 1000). The particle morphology of the powders was examined using an environmental scanning electron microscope (ESEM 2020) operating at 20 kV and a chamber pressure of 3 Torr. The SEM samples were prepared by putting them on a sample holder covered with conductive adhesive carbon tape and then coating them with gold in 2.2 kV voltage and 20 mA current for 2 min using an E5100 SEM coating unit from Polaron Instrument Inc. LiNH2 powders with and without high-energy ball milling were analyzed using X-ray diffraction (XRD - Rigaku RU200B series). The operation conditions for the XRD data collection were Cu KR radiation, 40 kV, 40 mA, 5 °/min, and 0.02 °/step using a D5005 ADAVANCE diffractometer. In addition to monitoring phase transformations, XRD also was used to estimate the crystallite size and internal strains of LiNH2. The XRD peak broadening was attributed to the refinement of

J. Phys. Chem. B, Vol. 110, No. 41, 2006 20711

Figure 1. SEM images of LiNH2 powder ball milled at room temperature for different times with (a) 0, (b) 45, and (c) 180 min.

crystals, introduction of internal strains, and instrumental effects. The correction for instrumental broadening was done by assuming that the as-received LiNH2 contained no internal strains and had large crystallite sizes. In other words, the asreceived LiNH2 powder was treated as the standard. This assumption was quite reasonable because most of the as-received LiNH2 powder particles had a particle size larger than 10 µm and many reached 100 µm as will be shown later. The (112) reflection of LiNH2 was used to estimate the average crystallite size of LiNH2 through the Scherrer formula,15 whereas the effective internal strain within LiNH2 was estimated using the (204) reflection of LiNH2 with the aid of the Stokes and Wilson formula.16 The estimations using such an approach were bound to result in finer grain sizes and higher internal strains than what they were in LiNH2. Nevertheless, as a first order of estimation, such an approach should provide a reasonable evaluation of crystallite sizes and internal strains because the XRD broadening at low angles was dominated by small grain sizes, and at high angles, the XRD broadening was dominated by internal strains.16,17 Recently, a detailed XRD analysis,18 which used the Rietveld method in conjunction with Levenberg-Marquardt nonlinear least-squares fit (LM-fit) and line-broadening analysis, demonstrated that this approach was indeed the case for nanocrystalline Al alloys that were subjected to severe plastic deformation. III. Results and Discussion 3.1. Powder Characteristics of LiNH2 with and without Mechanical Activation. The SEM examination (Figure 1) reveals that the as-received LiNH2 powder particles are angular and have a bimodal particle-size distribution with 90 vol % of the powder being ∼120 µm and 10 vol % of the powder lower than 6 µm. After high-energy ball milling for 45 min, the average particle size has been reduced substantially (Figure 1b). Most of the powder particles (∼90 vol % of the powder) are reduced to the range of 1-10 µm, although some large particles of 30-60 µm are still present. After 180 min of ball milling (Figure 1c), the average particle size is further reduced with 95 vol % of the powder ranging from 0.5 to 8 µm. The remaining

20712 J. Phys. Chem. B, Vol. 110, No. 41, 2006

Markmaitree et al.

Figure 2. XRD patterns of LiNH2 powder ball milled at room temperature for different times as indicated.

TABLE 1: Powder Characteristics of LiNH2 with and without Ball Milling ball milling time (min)

SSA (m2/g)

crystallite size (nm)

0 45 180

3.72 40.71 46.65

>100 5.9 5.5

effective internal strain

equivalent particle size (µm)a

0 0.0239 0.0314

1.37 0.13 0.11

a The equivalent particle size was calculated from the SSA by assuming that all the particles were spherical.

5 vol % of the powder has particle sizes less than 50 µm. In addition, all of the particles become rounded. Shown in Figure 2 are XRD patterns of LiNH2 powder with and without ball milling. It is noted that the as-received LiNH2 contains some Li2O and LiOH. Some of the Li2O and LiOH are present in the as-purchased condition and some are formed in the course of the XRD data collection. The ball-milled LiNH2 exhibits more LiOH and Li2O because of their finer particles and hence more oxidation in the course of the XRD data collection. The presence of a small amount of Li2O and LiOH, however, does not alter a general trend that shows the broadening of LiNH2 peaks as the ball-milling time increases. The peak broadening indicates the refinement of the crystallite size of LiNH2 and the presence of the lattice microstrain within the LiNH2 crystals.16-19 On the basis of the peak broadening, the crystallite size and effective internal strain of LiNH2 after ball milling have been estimated using the Scherrer formula and the Stokes and Wilson formula, respectively. The results are summarized in Table 1. It can be seen that the crystallite size of LiNH2 has been reduced to lower than 10 nm after 45 min of ball milling. Additionally, the crystallite size of LiNH2 decreases further, although slightly, as the milling time increases to 180 min. When combined with SEM analysis (Figure 1), this result suggests that the main function of ball milling from 45 to 180 min is breaking down of large particles. In contrast, refinement of small particles is very limited for ball milling beyond 45 min. Such a phenomenon is believed to stem from the fact that ball milling is a mini-forging of many powder particles trapped between two colliding balls.20 The collision force is mainly carried by large particles, whereas smaller particles can slide past each other. As a result, large particles fracture and become smaller, while small particles stay about the same or only decrease slightly in size. Table 1 also reveals that the effective internal strain within the ball-milled LiNH2 increases with the ball-milling time, which

Figure 3. TGA curves of lithium amide without and with ball milling for different times. The heating rate in the analysis is 10 °C/min, and a flowing argon atmosphere is employed.

suggests an increase in the defect density within LiNH2 as the ball-milling time increases. Because the estimations of the crystallite size and effective internal strains are performed using the as-received LiNH2 powder as the standard, it can be concluded that high-energy ball milling conducted in this study reduces the crystallite size of LiNH2 from larger than 100 nm to near 10 nm and introduces a large amount of defects to the LiNH2 crystals. The (SSA) of the LiNH2 powder before and after high-energy ball milling is measured. The SSA results are included in Table 1 along with the equivalent particle size calculated from the SSA by assuming that all the particles are spherical. Note that the SSA of LiNH2 increases with the ball-milling time. When compared with the as-received LiNH2, the high-energy ball milling has increased the SSA by 1 order of magnitude, and the equivalent particle size is decreased also by 1 order of magnitude. It is interesting to note that the equivalent particle size calculated from the SSA is lower than the particle size observed using SEM. The discrepancy suggests the presence of porous aggregates and/or rough surface of particles in LiNH2 powders, especially those with ball milling. It also is interesting to note that the equivalent particle size calculated from the SSA, although very small, is still much larger than the crystallite size estimated from the XRD peak broadening. This comparison suggests that a single LiNH2 particle after ball milling contains many crystallites. In summary, high-energy ball milling has reduced the particle and crystallite sizes of LiNH2 and has increased its specific surface areas as well as its defect concentrations. All of these will make LiNH2 more reactive and easy to decompose to Li2NH and NH3. The investigation of LiNH2 decomposition behavior has confirmed this expectation as discussed in the next section. 3.2 Decomposition Behavior of LiNH2 with and without Mechanical Activation. The decomposition behavior of LiNH2 was investigated using TGA. Shown in Figure 3 are TGA traces of LiNH2 with and without ball milling. It is clear that highenergy ball milling has greatly enhanced the weight loss process of LiNH2. The weight loss of LiNH2 without ball milling starts at 120 °C, which is reduced to 50 °C for LiNH2 ball milled for 45 min. After ball milling for 180 min, the onset temperature for weight loss is further reduced to room temperature. Moreover, the temperature for any given weight loss also has been reduced substantially. These data unequivocally indicate that the high-energy ball milling performed in this study has mechanically activated the LiNH2 crystals and accelerated its decomposition rates. Furthermore, when the ball-milling time

LiNH2 to LiNH via Mechanical Activation

J. Phys. Chem. B, Vol. 110, No. 41, 2006 20713

Figure 5. GC/MS analysis of LiNH2 powder with and without mechanical activation (i.e., 0 and 180 min of ball milling).

Figure 4. Composition analysis of the effluent gas from TGA with (a) no ball milling and (b) ball milling at room temperature for 180 min.

is longer, the mechanical activation increases and the decomposition temperature decreases. Such mechanical activation can be attributed to the reduced particle and crystallite sizes of LiNH2 as well as the increased specific surface areas and defect concentrations. More discussion on this topic will be provided in Section 3.3. According to reaction 3, a complete decomposition of 1 mol of LiNH2 should result in the formation of 0.5 mol of Li2NH and 0.5 mol of NH3. The latter would flow away with the flowing argon under the TGA experimental condition employed in this study. Thus, a complete decomposition of LiNH2 should lead to 37% weight loss. The less than 37% weight loss shown in Figure 3 is believed to be because of the presence of some Li2O and LiOH in the starting powder (see Figure 2). To confirm that the weight loss observed during the TGA is indeed because of reaction 3, the effluent gas from the TGA has been constantly analyzed using an RGA. The composition of the effluent gas analyzed is shown in Figure 4 as a function of the temperature measured during the TGA. The onset temperature for the detectable NH3 release is 220 °C for LiNH2 without ball milling and is reduced to 110 °C after ball milling for 180 min. The presence of H2 and N2 in the effluent gas is detected almost at the same temperature as NH3. This phenomenon is believed to be related to the decomposition of the released NH3 according to the following reaction

2NH3 ) N2 + 3H2

(5)

The thermodynamic data indicate that the standard Gibbs free energy changes for reaction 5 are 32 875, 11 954, and -9559 J/mol N2 at 298, 400, and 500 K, respectively.21 Thus, at room temperature, NH3 is very stable. However, at 400 K (127 °C),

the decomposition of NH3 is expected to occur under the TGA condition employed in this study in which flowing argon will continuously move N2 and H2 away from the TGA chamber, while NH3 is continuously produced from reaction 3. Therefore, the gas composition analysis confirms that the weight loss observed during the thermogravimetric study is indeed related to the decomposition of LiNH2 as defined by reaction 3. A comparison between the onset temperature of the detectable NH3 release from the gas composition analysis and the onset temperature of the detectable weight loss from the TGA reveals that these two onset temperatures do not coincide with each other (see Figures 3 and 4). Repeated and careful experiments confirm that this is always the case. As such, the discrepancy has been attributed to the time it takes for the effluent gas to travel from the TGA chamber to the RGA. Because of this travel time, there is a delay in the composition analysis when compared with the weight loss analysis. Since the heating rate is 10 °C/ min during the TGA, and the temperature shift between the TGA and the composition analysis is about 100 °C for both milled and unmilled powders, the travel time is about 10 min. In an effort to remove any doubt about the release of NH3 at the low-temperature range (