Abstract

THE EFFECTS OF AGING TEMPERATURE AND AGING TIME ON THE OXYGEN STORAGE CAPACITY OF Pt-Rh/CeZrO₂ CATALYSTS

C.E.Hori^1* * To whom correspondence should be addressed , K.Y.Simon Ng¹, A.Brenner², K.M.Rahmoeller³ and D.Belton³

¹Department of Chemical Engineering, Wayne State University, Detroit - Michigan, USA

²Chemistry Department, Wayne State University, Detroit - Michigan, USA

³General Motors Research and Development Center, Warren - Michigan, USA

¹Faculadade de Engenharia Química, Universidade Federal de Uberlandia,

Phone +55 34 239-4189, Fax: +55 34 239-4188, Av. João Naves de Ávila 2160,

Campus Santa Mônica - Bloco 1K, Uberlandia - MG, Brazil

E-mail: cehori@ufu.br

(Received: January 5, 2000 ; Accepted: September 19, 2000)

INTRODUCTION

In the early 1980’s catalyst manufacturers started to add ceria to automotive catalysts in order to improve their performance. The addition of ceria was proven to have several benefits, including promotion of the water-gas shift reaction and stabilization of the alumina support, but probably its most important role was as an oxygen storage agent (Trovarelli, 1996 and Taylor, 1993). Oxygen storage refers to the ability of ceria to rapidly undergo oxidation and reduction, depending on the conditions in the exhaust. It has been demonstrated that this oxygen storage property of ceria can widen the air/fuel ratio window where three-way catalysts effectively remove CO, hydrocarbons and nitrogen oxides from the exhaust.

In recent years, the emission standards for automotive applications have become significantly more stringent (Yamada et al., 1997; MuBmann et al., 1997). New legislation in Europe, California and the remainder of the United States dictates that the catalysts must very quickly (in less than one minute) reach temperatures where near 100% conversion is achieved. The most obvious solution to this requirement is to mount the converter closer to the engine so the catalyst can reach operational temperatures faster. However, this positioning means that the catalyst will experience higher temperatures during normal warmed-up conditions making the durability of the catalyst a more important issue (Yamada et al., 1997; Bartley et al., 1993).

It is already known that when automotive catalysts are exposed to high temperatures for long periods of time, they lose their ability to efficiently remove pollutants from the exhaust. The sintering of the noble metal particles, a well-studied phenomena (Flynn and Wanke, 1974; Ruckenstein and Pulvermacher, 1973; Pulvermacher and Ruckenstein, 1974), is partially responsible for this loss in activity. Another cause of catalyst degradation is a loss of the oxygen storage function needed to damp air-to-fuel ratio excursions in the exhaust. Degradation of the oxygen storage function occurs in part because the ceria-based particles in the catalyst grow larger over time at high temperatures. This growth in particle size is, to some extent, reflected as a loss in BET surface area. Degradation of oxygen storage capacity (OSC) can also occur when the noble metal promotors dispersed on the oxygen storage materials (OSMs) sinter. It is well established that noble metals promote ceria reducibility (Trovarelli, 1996; Nunan et al., 1992; Yao and Yao, 1984). Thus, as the noble metals sinter this promotional effect is lost to some degree (Nunan et al., 1992; and Mikki et al.1990).

With regards to the sintering of ceria, it has been proposed that when ceria particles grow very large, oxygen release becomes limited by diffusion of the oxygen from the bulk of the ceria particles to the surface, rendering the bulk of these particles unavailable for oxygen storage purposes (Trovarelli, 1996; de Leitenburg et al., 1996). In the last five years, several studies have shown that durability of ceria based OSM can be improved by doping ceria with other rare-earth elements, such as La, Nd, Y and Zr (Kubsh et al., 1991; Permana et al., 1997; Fornasiero et al., 1995; Ozawa et al., 1993; Hori et al., 1998, 1999). It has been suggested that in the Ce-Zr system the addition of zirconia to the ceria lattice improves O^-2 mobility in the lattice; therefore, more of the bulk oxygen is available for reduction. In our previous work (Hori et al., 1998, 1999), we showed that diffusion of O^2- from the bulk to the surface is not the rate-limiting process in the reduction of these materials. We proposed instead that the energy of incorporation of a vacancy into the oxide lattice limits the reducibility of ceria particles. By extension, the incorporation of a dopant into the lattice favorably alters the energy of vacancy incorporation and thus improves its oxygen storage capabilities.

In this study, we extend our previous work to obtain a better understanding of aging processes that occur in ceria-zirconia OSMs. The time scales for aging were 16 hours or less, which is representative of the number of hours a converter might experience peak operating temperatures, such as 900 to 1000^oC. The primary goal was to determine the relationship between noble metal surface area and oxide BET surface area in defining the oxygen storage capacities (OSC) of aged ceria-zirconia catalysts. The secondary goal of this study was to establish a correlation between time and temperature in the aging of OSC in the Pt-Rh-CeZrO₂ system.

EXPERIMENTAL

The support used in this study was a ceria-zirconia solid solution containing 75 mol% of Ce and 25 mol% of zirconia. This Ce_0.75Zr_0.25O₂ mixed oxide was prepared by dissolving cerium (IV) ammonium nitrate and zirconium nitrate in water at the desired Ce:Zr ratio. Then the Ce and Zr hydroxides were precipitated by adding excess (~100%) ammonium hydroxide. Finally, the precipitate was washed with deionized water and then calcined in air at 500^oC for 1 hour in a muffle furnace.

The noble metals were added by incipient wetness impregnation and the precursors used were tetraammineplatinum chloride and hexaamminerhodium chloride. The pore volume of the support was 0.25 cm³/g. Three Pt loadings were used, 1 wt%, 0.5 wt% and 0.2 wt% and the Pt/Rh ratio was kept at 5. After impregnation, the samples were dried in an oven at approximately 100^oC for 1 hour and then calcined in a muffle furnace at 500^oC for 1 hour. The aging procedure was carried out in a tube furnace and the gas stream fed into the furnace was cycled (0.1 Hz) between 5% O₂/N₂ and 5% H₂/N₂. This procedure was designed to mimic both the conditions of excess and deficiency of oxygen that a catalyst experiences in a car. The gas stream was saturated with water at room temperature before entering the furnace. All the samples were first aged at 825^oC for 4 hours to stabilize the oxygen storage capacities (OSC), and we consider these to be the fresh catalysts with zero aging hours. Subsequently, three higher aging temperatures (900^oC, 950^oC and 1000^oC) were employed for exposure times of 1 to 16 hours. At a given temperature only one catalyst sample was used. Exposure time was recorded in a cumulative manner and the catalyst was returned to the aging furnace after each set of measurements.

In this study, we also present some results from a previous study (Hori et al., 1999). In this case, the samples were prepared using the same Ce_0.75Zr_0.25O₂ support that had its BET surface area stabilized by heating for 24 hours at 950^oC in a muffle furnace. The support was loaded with 0.5 wt% of Pt by ion adsorption. The precursor used was H₂PtCl₆,and five consecutive adsorptions, each delivering about 0.1% of Pt, were used to obtain total loading of 0.5 wt% of Pt. In order to assure the decomposition of the Pt precursor after each adsorption, the sample was calcined at 500^oC in a muffle furnace for 2 hours and then placed in a tube furnace at 400^oC in flowing H₂ (5% in N₂) for 5 hours.

The noble metal dispersions of the samples were measured by CO chemisorption at room temperature. These experiments were carried out using a multi-function reactor system (RXM-100 from Advanced Scientific Design, Inc.). Typically, 0.5-0.7 g samples were used for these measurements. In order to ensure that the noble metals were in their metallic state, the samples were gently reduced in pure H₂ (20 cm³/min.) at 300^oC for 1 hour. These conditions were chosen to minimize CO adsorption on the support (Golunski et al., 1995). After the reduction procedure the sample was evacuated at 300^oC until no more degassing could be detected, which would normally take 2 hours. The sample was then cooled to room temperature before CO chemisorption was performed. We also performed chemisorption experiments on bare Ce_0.75Zr_0.25O₂ after aging at 825^oC for 4 hours and at 1000^oC for 16 hours and no chemisorbed CO could be detected. The dispersions were calculated using the assumption that each exposed metal particle adsorbs one CO molecule. Since there was no way that Pt and Rh could be distinguished by CO, the calculations were done based on a weighted average of the molecular weights and densities of Pt and Rh.

When the CO chemisorption experiment was finished, the same sample was evacuated again at room temperature, usually for one hour, before starting the BET surface area experiment. After this evacuation, liquid N₂ was used to cool the sample to 77 K, at which point N₂ adsorption was measured. The initial BET surface areas ranged from 28 to 37 m²/g after NM impregnation. These differences are probably not related to the noble metal content, since the amount of noble metal added was very small. The variation in the initial BET surface areas could be due to the fact that fresh samples (4 hours at 825^oC) were prepared in two different tube furnaces. Although the aging times at a given temperature were always the same, the time taken to reach the desired temperature varied from furnace to furnace.

The oxygen storage capacities of the samples were obtained in a flow-through reactor. For these experiments, 0.05 g of catalyst and a flow of 1000 cm³/min (STP) were used. The inlet of the reactor was connected to a series of four microvalves that allowed different gas streams to be fed into the sample. The experiments were performed at 250^oC and 300^oC. In the beginning of each run the sample was pre-treated at 250^oC or 300^oC in a pure N₂ flow in order to clean the surface of the catalyst. A gas stream containing 0.5% O₂ in N₂ was used to oxidize the samples. In these experiments, the concentrations of 0.25% CO₂ (for calibration purposes) and 1% CO as well as a stream of pure N₂ for purging were used. The sample was completely oxidized at the beginning of each experiment. After that, a 50 s pulse of CO₂ was sent through the reactor to calibrate the mass spectrometer. Finally, a 50 s pulse of CO was sent through the reactor. Each measurement was repeated at least three times and the results were averaged. The total amount of CO₂ produced was detected using a quadrupole mass spectrometer and was called the oxygen storage capacity of the sample.

RESULTS AND DISCUSSION

The noble metal dispersions as well as the average crystallite sizes calculated by CO chemisorption are shown in Table 1. Samples with lower noble metal contents were able to maintain higher dispersion over the entire range of aging temperatures; however, since there was less noble metal in them, these samples had a lower specific metal surface area. The samples with 0.5%Pt-0.1%Rh and with 0.2%Pt-0.04%Rh had somewhat similar average particle sizes after being aged at any given temperature. On the other hand, the samples with 1.0%Pt-0.2%Rh had larger particles from the start. When considering samples in the initial hours of the aging process, for any given aging temperature the 1.0%Pt-0.2%Rh samples had average particle sizes, typically twice as large as the ones found in the samples with lower noble metal contents. For the most severely aged samples (1000^oC and long periods of exposure), all the samples had a similar particle size, independent of the noble metal loading used. When comparing samples with the same noble metal loading but aged at different temperatures, one can find some unexpected results (Table 1). For instance, 1%Pt-0.2%Rh aged for 4 hours at 900^oC had a slightly lower dispersion than the same sample aged at 950^oC, and the same happened for 0.2%Pt-0.04%Rh aged for 16 hours at 950^oC and 1000^oC. We took these small variations as experimental errors, since the dispersions measured were low.

The dispersion data were plotted according to the following equation developed by Ruckenstein and Pulvermacher, (1973, 1974):

where n is the sintering order, D_o is initial metal dispersion, D_t is metal dispersion at time t, and C is a constant. The sintering order can be obtained by plotting log(D_o/D_t) vs. log (t). For the data in this study, the values found for n ranged between 0.8 and 2. Previous studies foun in the literature reported values of n ranging between 2 and 20 for short periods of exposure (less than one hour), and for long periods of aging, the values were between 1 and 4 (Harris, 1986; Beck and Carr, 1988, 1993). Although the idea of using values of n to determine the prevailing sintering mechanism is still the subject of debate (Lee, 1980), there is general agreement that high values of n correspond to the crystallite migration mechanism, while lower values of n are an indication that the atomic migration mechanism is predominant. The general trends observed in this study suggest that atomic migration is the dominant sintering mechanism (Harris, 1986; Beck and Carr, 1988, 1993). Considering that the fresh samples were exposed to 825^oC for 4 hours, it is not surprising that, in the subsequent aging, particle sizes were sufficiently large so that the atomic migration mechanism would prevail. We take the similarity of our results to those in previous papers to mean that we are measuring dispersions properly and that the noble metals sinter on a ceria-zirconia support in a fashion similar to that of Pt on alumina.

Correlation of OSC with BET and NM Surface Areas

The OSC measured at 300^oC for all the samples is shown in Figure 1. The symbols in Figure 1 represent the actual experimental data and the lines are a fit of the data with the functional form and parameters given in Table 2. Several functional forms were tried in order to describe the relationship between OSC, time of exposure and aging temperature. We found that power laws in both time and temperature, exponential and logarithmic functions and combinations of power laws in time and exponential functions in aging temperature gave poor fits to our data. A reasonably good fit was obtained with the following equation, which resembles the forms used to describe nonelementary reaction kinetics, such as enzymatic reactions:

where OSC is the oxygen storage capacity; IOSC is the initial oxygen storage capacity at zero aging hours; T is the temperature in Kelvin, t is the aging time in hours and a, b, c and d are parameters of the curve fit (Table 2). The different symbols in Figure 1 represent OSC measured from samples with different noble metal loadings. Recall that zero aging hours represents samples that were stabilized by aging at 825^oC for 4 hours. This stabilization treatment was required in order to achieve reproducible OSC at zero aging hours. Because the samples were stabilized prior to aging, we did not observe a precipitous drop in OSC upon further aging at 900^oC for several hours. At an aging temperature of 900^oC, we observed little or no difference in the OSC of the three differently loaded samples up to 8 hours of aging time. However, after aging for 16 hours, the lowest NM loaded catalysts showed about one half the OSC measured for the higher loaded catalysts.

The center panel of Figure 1 presents OSC data for samples aged at 950^oC. As explained in the experimental section, in this study zero hour aging data were equivalent, but as the samples were aged for successively longer periods, OSC decreased slightly faster than that observed at 900^oC (upper panel). Similar to samples aged at 900^oC after 8 hours very little difference in metal loadings was discernible, but after aging for 16 hours, the catalysts with the two highest metal loadings had roughly twice the OSC of the lowest loaded sample. Again, OSC in all the catalysts was only two to five times lower than for zero hours, and these samples which were stabilized at 825^oC showed an exceptional ability to resist aging at a relatively high temperature. The lower panel in Figure 1 presents the OSC data for catalysts aged at 1000^oC where the effects of time and temperature had a more marked effect on OSC. After 16 hours of aging, the OSC of all the catalysts was approximately five to ten times lower than that of the zero-hour samples. At this aging temperature, the benefit of having high NM loadings for the OSC can be clearly observed. After 16 hours of aging, the highest NM loaded sample had an OSC roughly five times greater than the lowest loaded catalyst.

In order to try to establish a correlation between OSC and the other properties of the samples, the noble metal area (Figure 2) and the BET surface area (Figure 3) were also measured at the same points in the aging process at which OSC was measured. Figures 2 and 3 show and compile all the data by plotting the noble metal area vs. OSC (Figure 2) and BET surface area vs. OSC (Figure 3). In Figures 2 and 3, all the samples from this study as well as those from a previous study, which contained 0.5 wt% Pt supported on the same CeZr support (Hori et al., 1999). Differences in preparation of these catalysts were presented in the experimental section, but the important point is that the supports used in the previous study (light gray symbols) had their surface area stabilized at 7.7 m²/g before the addition of noble metal. Therefore, the surface area was kept constant as the sample was aged. Figure 2 shows the correlation between oxygen storage capacity and noble metal area. The OSC was measured at both 250^oC and 300^oC. The most important observation was that all catalysts fall on the same curve regardless of NM loading. This is what one would expect since the fundamental quantity in the system is number of NM sites in the catalyst. Interestingly, the samples from the previous study containing 0.5 wt% Pt also fall on the same curve indicating that in terms of increasing the OSC, Rh and Pt are similar on a surface area basis. This study did not address whether Rh and Pt have different tendencies to maintain NM surface area upon aging. The data showed a strong correlation of OSC and noble metal area at low NM surface areas and a much weaker correlation at higher noble areas. These effects were observed independent of the OSC measurement temperature.

Based on Figure 2 we would, not surprisingly, conclude that OSC is strongly correlated with NM surface area. The point where the curves inflect is dependent on the temperature at which the OSC measurement was made. For experiments performed at 300^oC, the inflection was around 0.1 m²/g, but at 250^oC, this inflection was closer to 0.2 m²/g. It is easy to envision that, as the temperature of the OSC measurement is increased, the noble metal area required to "fully" promote reaction decreases well below 0.1 m²/g. In other words, Figure 2 would show a step function in OSC vs. NM surface area very close to 0 if the measurements were done at higher temperatures. This is important because practical catalysts typically operate above 500^oC. However, in our apparatus the CO₂ formation reaction, which is the basis of the OSC measurement, becomes mass transport limited above 300^oC. Since mass transport properties in our powder reactor are dramatically different from those of a real monolith catalyst in automotive exhaust, we chose not to make measurements in the mass transport limited regime.

Practical Considerations of Time and Temperature in the Aging of Pt/Rh/Ce_0.75Zr_0.25O₂ Catalysts

It is well known that automotive catalysts tend to deactivate under conditions routinely encountered during the lifetime of a vehicle (100,000 miles/10 years). Conservatively, over 10 years or 100,000 miles the catalyst will be functioning for about 2000 hours (50 miles per hour average speed). Most of this time is spent under relatively mild operating conditions that give rise to catalyst temperatures below 800^oC with a small fraction spent at higher temperatures (we assumed above 900^oC). The secondary goal of this study was to address the following question: how aging time and temperature sensitive are ceria-zirconia OSMs? In older catalyst technologies that contained ceria as the OSM, it was commonplace to view severity of aging as a step function in aging temperature. Within this model, time spent below some critical temperature was irrelevant if a short period of time above the critical temperature was encountered. Above the critical temperature the catalyst deactivated very quickly. In this part of the paper, we attempt to quantify the time and temperature variables of the simple model put forth above for ceria-zirconia OSM. Our initial assumptions were that a) 825^oC was below the critical temperature for catalyst deactivation and b) a catalyst would spend about 1% of its useful life (20 hours) at peak operating temperatures between 900 and 1000^oC. Within these assumptions, we wanted to compare time vs. temperature trade-offs for three important physical properties of a catalyst: OSC, NM surface area, BET surface area.

The decay of noble metal surface area as a function of aging temperature and time of exposure is shown in Figure 4. The different symbols represent NM surface areas measured for samples with different noble metal loadings. As expected, the zero-hour samples showed the effect of increased noble metal loading with the highest noble metal surface areas measured for 1%Pt-0.2%Rh samples, followed by 0.5%Pt-0.1%Rh and finally, the lowest noble metal content samples. The noble metal area decreased most rapidly during the first two to four hours of exposure at a given temperature with a more gradual decay occurring over longer periods. Given the strong correlation between OSC and NM surface area it is not surprising that the same functional form (Table 3) used to fit the OSC data (Figure 1) could also be used to describe the decay of NM surface area. The lines in Figure 4 are the actual fit obtained with the parameters in Table 3.

Turning now to the fit of the data, the time dependence for the noble metal area is in a power law form with the exponent varying from 1 to 1.4 between 900 and 1000^oC, depending on the noble metal loading. Basically, the noble metal area decreases with t^-1 in this temperature range, without a strong influence of aging temperature on the time. The majority of temperature dependence is in the exponential term (Table 3), which increases when the aging temperature is raised from 900^oC to 1000^oC. For the least temperature-sensitive catalyst, 1%Pt-0.2%Rh, the exponential term, exp(a+b/T), varied from 0.5 to 2 between 900 and 1000^oC, an increase by a factor of 4. In contrast, for the more temperature-sensitive 0.2%Pt-0.04%Rh catalyst the exponential term, exp(a+b/T), ranged from 0.85 to 6.5 between 900 and 1000^oC, a 7.6x increase.

To give a sense of the effect of aging temperature on NM surface area, we give examples for high- and low-loaded catalysts. For instance, when aged at 900^oC for 8 hours, the low-loaded catalyst (0.2%Pt-0.04%Rh) experienced a drop in NM surface area from its initial value of 0.2 m²/g to 0.055 m²/g. When aged at 950^oC, the equivalent noble metal area (0.055 m²/g) was obtained in half the time (4 hours), and when aged at 1000^oC, a 0.055 m²/g NM surface area was achieved in 2 hours. Based on these data, we conclude that this catalyst sees a fourfold increase in aging severity for every 100^oC above 900^oC. This severity factor is strongly dependent on NM loading, as is evidenced by the temperature sensitivity of the most highly loaded catalyst (1.0%Pt-0.2%Rh). Aging the 1.0%Pt-0.2%Rh sample at 900^oC for 8 hours decreased the noble metal area fivefold from an initial value of 0.5 m²/g to 0.1 m²/g. When aged at 950^oC, it still took 8 hours to achieve 0.1 m²/g. However, when aged at 1000^oC, a 0.1 m²/g NM surface area was achieved in approximately 4 hours. Qualitatively speaking, the highly loaded part sees a twofold increase in aging severity for every 100^oC increase in aging temperature.

Turning now to the BET surface area of the catalysts, Figure 5 shows the BET surface area of all the samples as a function of aging temperature and time of exposure. A single curve fit (solid lines in Figure 5) was done using all the data points independent of noble metal content. The parameters for the BET surface area fit are also shown in Table 3. The decay of the BET surface area could be fit with the same functional form used to describe the decay of NM surface area in agreement with the idea that the two phenomena are coupled. The time dependence for the decay of the BET surface area is ~t^-0.65. Our curve fit of the data shows that the BET surface area has roughly the same temperature sensitivity as the noble metal area in the lowest loaded catalyst. If we examine the exponential term of the fit, exp(a+b/T), we see that this term varies by a factor of 6.9 between 900 and 1000^oC for the BET surface area which is comparable to the factor of 7.5 seen for the NM surface area of the lowest loaded catalyst. In contrast, the exponential term varies only by a factor of 2 for the NM surface area of the highest loaded catalyst.

To illustrate the similarities of and differences between aging of BET and NM surface areas, we make the same 8 hours aging comparison for BET surface area that was done for the NM surface area. Comparing degradation in the BET surface area at different temperatures, our curve fit predicts that 8 hours of aging at 900^oC is equivalent to 4 hours at 950^oC or ~2 hours at 1000^oC. To restate this, this is the same behavior as seen for the low-noble-metal-loaded catalysts, which have the smallest NM particles. This correlated behavior suggests that sintering in the smaller, more mobile NM particles is driven strictly by the surface area of the support, but the larger (see Table 1), less mobile particles of the more highly loaded catalysts are not as sensitive to the support area. Thus, the correlation between NM surface area and BET surface area becomes poorer as the particles become sufficiently larger.

Finally, for OSC the fit shows that the time dependence for OSC is, like that for the NM surface area, highly dependent on noble metal loading and only weakly dependent on aging temperature. For the highest loaded catalysts we observe a dependence of ~t^-0.7 which increases to t^-1.5 for the 0.2%Pt-0.04%Rh sample. This increased time sensitivity with decreased loading is what was observed for the NM surface area. As for temperature sensitivity, OSC shows a sensitivity to temperature similar to those of the NM and the BET surface areas. For OSC the ratio of the exponential term, exp(a+b/T), at 900^oC to that at 1000^oC is 3 for the highly loaded catalysts (compared to 4 for the NM surface area and 3 for the BET surface area) and 8.5 for the lowest loaded catalysts (compared to 7.5 for the NM surface area and 3 for the BET surface area). All three properties show similar sensitivity to aging time and temperature. Roughly speaking, the properties show a near t^-1 dependence on time and this time dependence is weakly influenced by aging temperature. Lumping together all three properties we can generalize temperature dependence in the following manner. An increase in temperature from 900 to 1000^oC gives a two to four fold increase in aging severity. Overall, we find that these ceria-zirconia OSMs are not deactivating at an extremely rapid rate (10x decrease in OSC or NM surface area in 1 hour) below 1000^oC.

SUMMARY

The effects of aging time and temperature on the OSC of Pt-Rh-promoted ceria-zirconia solid solutions were measured and correlated with the BET surface area and NM surface area in the catalysts. The NM surface area is better correlated with OSC than is the BET surface area. However, materials with a low BET surface area are likely to have low NM dispersion and thus poor OSC. Comparing our results here to our previous work suggests that Pt and Rh have similar abilities to promote OSC on a NM surface area basis; however, we did not investigate which of these metals is most resistant to high temperature sintering. On the practical side, our results demonstrated that, even when operating at 900^oC with alternating rich and lean conditions, these materials deactivate slowly; however, the observed deactivation rate is NM-loading dependent. As the aging temperature is increased to 1000^oC, the deactivation rate for all three properties increased two to four times with the lower NM loaded catalysts being more sensitive to aging temperature.

NOMENCLATURE

Publication Dates

History