Environ. Sci. Technol. 2005, 39, 800-806
Glass Transitions in Peat: Their Relevance and the Impact of Water GABRIELE E. SCHAUMANN* Department of Environmental Chemistry, Institute of Environmental Technology, University of Technology Berlin, KF3, Strasse des 17 Juni 135, D-10623 Berlin, Germany EUGENE J. LEBOEUF Department of Civil and Environmental Engineering, Vanderbilt University, 278 Jacobs Hall, Nashville, Tennessee 37235
This contribution aims to expand the macromolecular view of fractionated natural organic matter (NOM) to organic matter in whole soils. It focuses on glass transition behavior of whole soil organic matter (SOM) and its interrelation with water through use of differential scanning calorimetry (DSC) and thermomechanical analysis (TMA). Three processes of structural relaxation related to macromolecular mobility were distinguished. Process I occurs in thermally pretreated and very low water-content samples and corresponds to classic glass transition behavior. Process II occurs in water-containing samples, where water is believed to act as an antiplasticizing agent in the peat at water contents below 12%, causing decreased macromolecular mobility and increased glass transition temperature. We suggest the formation of hydrogen bond-based cross-links being responsible for this antiplasticizing effect. Process III represents a slow swelling process induced by water uptake with a time constant of swelling in the order of days, with water acting as a plasticizing agent. Results from this work are of particular importance for environmental systems as changes in environmental conditions (e.g., water content, temperature) may induce slow structural relaxation processes in NOM over periods of time ranging from days to weeks. These influences on NOM macromolecular mobility lead to continuous changes in physicochemical properties that may greatly influence sorbate-sorbent interactions in surface and subsurface environments.
Introduction Macromolecular models of natural organic matter (NOM) have been proposed (1, 2) as a means to help explain nonideal sorption of organic contaminants in soils. Similar to synthetic organic macromolecules, matrixes of NOM are suggested to consist of glassy and rubbery domains. Glassy domains, described as rigid, condensed organic matter, are considered responsible for slow desorption, non-Fickian diffusion, nonlinear sorption, and sorption/desorption hysteresis, whereas rubbery domains may be responsible for increased diffusion rates, linear sorption, and partitioning-like processes (3). Complementary nonlinear sorption and slow desorption rates are observed characteristics of black carbon * Corresponding author phone: +49(30)314-73173; fax: +49(30)314-29319; e-mail:
[email protected]. 800
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 39, NO. 3, 2005
(BC) and other highly condensed carbonaceous subfractions of organic matter (OM) (4-6). It is suggested that these OM fractions, representing glassy components of macromolecular NOM (5, 6), may largely control observed sorption behaviors of hydrophobic organic compounds. Explanations for origins of slow desorption kinetics and hysteresis present in glassy NOM include micropore deformation (7), where sorbate chemical potential induced swelling causes a change in sorbent characteristics. Explanations for swelling-induced changes in NOM were also used by Altfelder et al. (8) to describe the influence of moisture content on sorption kinetics in soil samples. When a rubbery, amorphous polymer is cooled fast through the glass transition region, it transforms to a glassy matrix with nonequilibrium structure (9). Although called rigid (3), glassy matrixes undergo a process called structural relaxation or physical aging during which the macromolecular structure changes gradually with time, tending toward an equilibrium state (9). Physical aging is believed partially responsible for the aging of contaminants in NOM and SOM (1). The relaxation rate of individual macromolecular chains decreases with increasing glassy character, and it increases with increasing temperature when approaching the glass transition temperature (Tg). Glass transitions are typically reflected by a significant change in the heat capacity in a rather narrow temperature range at Tg (10). The kinetic nature of the transition in combination with the nonequilibrium character of the glassy state and the temperature dependence of structural relaxation implies that the glass transition may be strongly influenced by thermal or sample history. Cross-linking reduces the side chain mobility and therefore increases glassy character and glass transition temperature. The glass transition temperature of a polymer/water gel follows the Fox-Flory equation (11):
CW 1 - C W 1 ) + T g TW TP g
(1)
g
P where CW is the water content, and TW g and Tg are the glass transition temperatures of water (136-170 K; 12) and the dry polymer, respectively. Tg decreases with increasing water content if TPg > TW g , which is commonly known as a plasticizig effect (13, 14). While water commonly acts as a plasticizer, a limited number of studies report antiplasticizing properties of water (15-17). The discovery of glass transitions in humic substances by differential scanning calorimetry (DSC; 1) and of different domain mobilities in humic acids by NMR (18) support a macromolecular view for SOM. Their relevance for sorption behavior has been shown by several authors (2, 3, 19). Thus, humic and fulvic acids appear to reveal macromolecular characteristics, although they generally possess relatively low molecular weights (20). However, the transfer of phenomenological observations from humic substances to SOM is not trivial. Although humus fractionation does not significantly change the functional group composition (21), the macromolecular structure of SOM may be influenced by extraction and precipitation procedures. Thus, evidence for glass transitions in humic substances does not automatically imply glass transition behavior in nonfractionated SOM. The initial difficulty in identifying glass transitions in whole soils was explained by the heterogeneity of SOM, which could result in a continuum of glass transitions over a wide temperature range (22). Subsequent studies, however, reported glass transitions for several whole soils (23, 24). These
10.1021/es0490931 CCC: $30.25
2005 American Chemical Society Published on Web 12/22/2004
transitions may result from a continuum of glass transition temperatures (11) with highest intensity at Tg. It may alternatively be assumed that the chemically heterogeneous mixture reveals a quasi-homogeneous macromolecular suprastructure. Furthermore, a uniform, structure-independent process may be responsible for the observed glass transitions. A variety of soil-relevant compounds may affect the macromolecular characteristics of SOM. While polyvalent cations such as Ca2+or Pb2+ can act as cross-linking agents stabilizing the macromolecular structure (18), the omnipresent water may act as a plasticizer (11). Large moisture dynamics existing in field conditions may subject the macromolecular matrix of SOM to continuous transformations at varying rates. 1H NMR relaxometry studies point at swelling in SOM with time constants (i.e., reciprocal rate constants) of up to 6 d (25, 26). Swelling may represent a matrix relaxation process accelerated by the incorporation of water. This is also supported by the finding that water appears to act as a swelling/plasticizing agent in humic substances (1) and in a soil sample (23). Despite the relevance, a void of knowledge still exists on possible factors that influence the macromolecular character of SOM. This study seeks to further examine the glass transition behavior of an air-dried peat sample by DSC and thermomechanical analysis (TMA). Water content and thermal history are used as experimental variables as a means to probe the hypothesis that water is primarily responsible for the differences in observed transition behaviors as well as the hypothesis that water may act as a plasticizing and an antiplasticizing agent in the same peat sample.
Experimental Section Sample Preparation. The sample used in this study originates from the Rhinluch fen, 60 km northwest of Berlin, which has been in agricultural use over the last 200 yr. While the upper layers are strongly decomposed and pedogenetically altered, the deeper layers are dominated by sedge (carex) and reedpeats (phragmites) with visible fiber structure (27). The peat sample of this study was taken from the deeper layers (1 m) and contained 420% water and 55% organic matter (dry mass basis). The sample was air-dried and then stored in a water atmosphere of 76% relative humidity (RH) at 20 °C to obtain a reproducible moisture content (12% on dry mass basis). Subsamples were oven-dried at 105°C for 12 h or vacuumdried at 20 °C to separate effects of water evaporation and temperature. These samples were stored in a desiccator under a dry atmosphere until used, with final water contents of 2.0 ( 0.6%. Thermal Pretreatment. The differently dried samples were investigated by DSC without thermal pretreatment, aiming at the current amorphous state (23). We additionally examined the effects of thermal pretreatment, which was applied in ref 24 to remove physisorbed water and to quantify material properties, in two separate procedures. In procedure I, water evaporation was allowed by punching three holes into the lid of a filled DSC pan and heating isothermally at 120 °C for 30 min (28, 29). In procedure II, water evaporation was prevented by tempering the sample at 120 °C for 30 min in a hermetically sealed pan. Hydration. Hydration effects were examined by equilibrating air-dried and oven-dried samples at 20 °C over water in a glass desiccator (2.0 L), which resulted in 85 ( 5% RH with fluctuations due to sampling events. At designated points of time within a 4-week period, subsamples were taken and investigated by DSC. The time span between the first contact with water atmosphere, and each sampling is referred to as hydration time. The water content was determined gravimetrically in a separate container, which was weighed each time when subsamples from the other container were taken. The water content (Θ) and Tg were fitted as a function of
hydration time using a first-order model by minimizing χ2 with the Simplex minimization method (Microcal Origin Version 5.0, Microcal Software Inc., Northhampton, MA). The resulting functions are as follows:
Θ(t) ) (Θ0 - Θ∞)e-t/τW + Θ∞
(2)
Tg(t) ) (T0g - T∞g )e-t/τT + T∞g
(3)
Θ(t) represents the water content at hydration time t, Θ∞ is the water content at infinite hydration time, and τW is the time constant of the first-order process. In similar fashion, we may define Tg(t) and T∞g to represent glass transition temperatures at time t and at infinite time, respectively, where τT is the time constant. DSC Experiments. TA Instruments model 2920 DSC and model Q1000 DSC (TA Instruments, New Castle, DE) using a heating rate of 10 K min-1 from -50 to 110 °C with nitrogen as purge gas were employed for DSC analysis. As the initial phase up to -20 °C lacks expressiveness due to technical restrictions, it is not shown in the figures. Hermetically sealed pans were used in most experiments, while selected DSC experiments were conducted with standard aluminum pans. A total of 2-5 mg of the sample was placed into the sample pans, and the pans were sealed before the DSC experiment. All samples were stored in their respective sample pans following DSC experiments for subsequent investigation. DSC data were analyzed using Thermal Advantage V3.9 software (TA Instruments). The change in the heat flow (sample heat capacity) at Tg is indicated by an inflection point in the thermogram. Operationally, three tangent lines are applied for evaluation. Tg is defined as the temperature at the halfheight of the central tangent line. The change of heat capacity (∆Cp) is calculated from the height of the central tangent line. Modulated DSC (MDSC). MDSC employs a modulated heating protocol that allows distinguishment of fast processes following temperature modulation (reversing heat flow), from slow or nonreversing processes (nonreversing heat flow). Reversing and nonreversing heat flow are calculated by Fourier transformation. Classical glass transitions reveal reversing characteristics, while structural relaxation and evaporation are of a nonreversing nature (28, 29). Distinguishment of structural relaxation and evaporation may be possible through differences in thermogram shape, but it is not trivial to achieve a separation of these processes. MDSC was conducted from -40 to 110 °C at 2.5 K min-1 with a 90 s modulation period and (2 °C temperature amplitude. Thermomechanical Analysis (TMA). TMA involved use of a TA Instruments model 2940 thermomechanical analyzer (TA Instruments, New Castle, DE) using the instrument’s dilatometer mode at 2.5 K min-1 with a force of 10 mN and nitrogen purge gas. Samples were placed in a glass receiver containing a layer of silica. TMA experiments were conducted in open systems. The thermograms reflect the effect of heating on the sample volume, measured as dimension change. Glass transitions typically reflect a significant change in the thermal expansion coefficient and thus are often identified by a change in the thermogram slope (14). Data Analysis. The errors given with estimates of eqs 2 and 3 are the calculated standard errors of fitting. The estimated error of Tg ((1 °C) and of ∆Cp ((40%) are represented by the standard deviation of sample repetitions. Values have been compared on the basis of a t-test (P < 0.05).
Results DSC of Dried Samples. Figure 1 illustrates representative DSC thermograms of the dried samples with and without VOL. 39, NO. 3, 2005 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
801
FIGURE 1. Representative DSC thermograms of samples without thermal pretreatment (a: air-dried, c: hydrated, d: vacuum-dried), and thermally pretreated with water evaporation in a standard pan (e1) and in a punched hermetic pan (e2). Plot b represents a thermally pretreated sample in a hermetic pan (without water evaporation) and is representative of thermal behavior of samples without thermal pretreatment in standard and open pans.
TABLE 1. Summary of Transition Temperatures (Tg) and Changes of Heat Capacity (∆Cp) for the Transitions Determined by Standard DSC Measurements for Different Pretreatment Methodsa sample state
pan type
without thermal pretreatment
air-dried
open/ punched
nd
air-dried
standard
nd
air-dried
hermetic
oven-dried
hermetic
vacuum-dried
hermetic
air-dried f hydrated
hermetic
oven-dried f hydrated
hermetic
65 ( 1 °Cc 0.15 ( 0.06 J g-1 K-1 47 ( 1 °C 0.09 ( 0.04 J g-1 K-1 46 ( 1 °C 0.03( 0.01 J g-1 K-1 51 ( 3 °Cd 0.09 ( 0.06 J g-1 K-1 52 ( 2 °Cd 0.09 ( 0.06 J g-1 K-1
with thermal pretreatment 37 ( 10 °Cb 0.03 ( 0.01 J g-1 K-1 17 ( 10 °Cb 0.02 ( 0.01 J g-1 K-1 nd nd nd nd nd
a The given errors characterize the standard deviation. Transitions that could not be detected are indicated as nd. b Estimated uncertainty due to strong influence of thermal history. c Maximum deviation is (2 °C. d Values correspond to the average of the transitions for hydration times between 13 and 28 d; errors characterize the corresponding standard deviation.
thermal pretreatment. The air-dried sample (plot a) reveals a step transition at 65 °C. The average transition temperature is 65 °C, with a maximum deviation of ∆Tg ) (2 °C and standard deviation of ∆Tg ) (1 °C (see Figure SI-2). The transitions reveal an average ∆Cp ) 0.15 J g-1 K-1 with a standard deviation of ∆(∆Cp) ) (0.06 J g-1 K-1 (Table 1). The value of ∆Cp and the detectability of the transitions is strongly influenced by the quality of pan sealing. If not sealed hermetically, water evaporation overlaps the transition and makes evaluation of the transition difficult (plot d in Figure SI-2). The transition temperature of the vacuum-dried sample (plot d) equals 46 ( 1 °C, where the thermogram is also representative of the behavior of the oven-dried sample (47 ( 1 °C). Oven-drying and vacuum-drying reduced ∆Cp to 0.086 and 0.032 J g-1 K-1, respectively (Table 1). In all 802
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 39, NO. 3, 2005
FIGURE 2. MDSC thermogram of the air-dried sample (water content 12%) representing total, reversing, and nonreversing heat flow. experiments without thermal pretreatment, the transition vanished in subsequent runs and re-appeared after several days of storage in the respective pan at room temperature (data not shown). In ashed samples, the transition disappears (data not shown), suggesting that the organic matter is likely responsible for the observed transitions. The strong transition observed in the hermetic pan (plot a) at 65 °C can neither be detected in the standard pan nor in open pans or after thermal pretreatment without water evaporation (plot b). Allowing water evaporation together with the thermal pretreatment, however, produces transitions located at Tg ) 37 °C with ∆Cp ) 0.026 J g-1 K-1 (punched hermetic pan, plot e1) and at Tg ) 17 °C with ∆Cp ) 0.018 J g-1 K-1 (standard pan, plot e2). The strong difference in Tg (∼20 °C) underscores the sensitivity of the transition to thermal history. As indicated by the lower ∆Cp, the extent of these transitions is significantly smaller than the initial transition in the air-dried sample (plot a). In contrast to the nonthermally pretreated samples, the transitions remain detectable in subsequent runs following thermal pretreatment, and their Tg appears to change with each heating/ cooling cycle (data not shown). Sample storage of the thermally pretreated samples in their respective pans modifies the thermal behavior, which is then similar to that of the vacuum-dried sample (plot d). Modulated DSC. Figure 2 shows a MDSC thermogram of the air-dried sample (12% water) in the hermetic pan. Distinct transitions can be seen in the total and in the nonreversing heat flow at Tg ) 57 °C, and a slight transition can be detected in the reversing heat flow. The change of heat capacity is 0.14 J g-1 K-1 in the nonreversing component and 0.03 J g-1 K-1 in the reversing component. Thus, only a small portion (
