Environ. Sci. Technol. 1999, 33, 4250-4255
Evolution of the Surface Area of a Snow Layer LAURENCE HANOT AND F L O R E N T D O M I N EÄ * CNRS, Laboratoire de Glaciologie et Ge´ophysique de l'Environnement, BP 96, 38402 St. Martin d'He`res Cedex, France
Atmospheric trace gases can partition between the atmosphere and the snow surface. Quantifying this partitioning requires the knowledge of the surface area (SA) of snow. Eleven samples were taken from a 50 cm thick snow fall at Col de Porte, near Grenoble (French Alps) between January 20 and February 4, 1998. Fresh snow and 3, 8, and 15 day-old snow were sampled at three different depths. Surface hoar, formed after the fall, was also sampled. Air and surface snow temperature, snow density, and snow fall rate were measured. Snow temperature always remained below freezing. Snow SA was measured using methane adsorption at 77.15 K. Values ranged from 2.25 m2/g for fresh snow to 0.25 m2/g for surface hoar and surface snow after 15 days. These values are much too high to be explained by the macroscopic aspect of snow crystals, and microstructures such as small rime droplets must have been present. Large decrease in SA with time were observed. The first meter of snowpack had a total surface area of about 50 000 m2 per m2 of ground. Reduction in SA will lead to the emission of adsorbed species by the snowpack, with possible considerable increase in atmospheric concentrations.
Introduction Snow can cover more than 50% of land masses in the Northern hemisphere in winter (1, 2). Because snow has a large surface-to-volume ratio, an important interaction potential between ice and atmospheric trace gases exists. Snow can adsorb atmospheric trace gases such as HNO3 and nonpolar organics (3-6), thus removing them from the gas phase and modifying atmospheric composition. Adsorbed gases can lead to heterogeneous photochemistry, that can release reactive trace gases such as formaldehyde and NOx to the atmosphere (7, 8). Quantifying the amount of trace gases adsorbed on snow and the rate of surface reactions requires the knowledge of the surface area (SA) of snow. For adsorbed species such as nonpolar organics, the evolution of their concentration in the snow cover will be determined to a large extent by the evolution of snow SA (9), as a decrease in SA will result in emission of the adsorbed gas to the atmosphere. Thus, a complete understanding of the exchange of trace gases between snow and the atmosphere requires the knowledge of snow SA and of its evolution after deposition. Understanding the evolution of snow SA will also find applications in the reconstruction of past atmospheric composition from ice cores analyses (10, 11). * Corresponding author phone: (33) 476 82 42 69; fax: (33) 476 82 42 01; e-mail:
[email protected]. 4250
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 33, NO. 23, 1999
Surprisingly, snow SA measurements have stirred little interest. Isolated measurements were performed by Adamson et al. (12), who measured a surface area of 0.2 m2/g for fresh snow sampled in a blizzard on Mount Gorgonio (about 2400 m elevation) and 1.3 m2/g for snow sampled near Denver 1 day after its fall. The method used was nitrogen adsorption at 77 K. Using a similar method, Jellinek and Ibrahim (13) obtained a value of 7.77 m2/g for a snow sample of unspecified origin. More recently, using CH4 adsorption at 77 K Chaix et al. (14) obtained a value of 0.057 m2/g for fresh snow sampled outside our laboratory, near Grenoble, on February 9, 1995, when the temperature was -2 °C. Finally, using N2 adsorption at 77 K, Hoff et al. (6) obtained SA values in the range 0.060.37 m2/g for six snow samples, taken , usually within a few hours of falling, at temperatures below freezing .. Considering the small number of available SA measurements, and the experimental difficulties encountered by Hoff et al. (6), which decrease their reliability, we have measured the SA of 11 fresh and aged snow samples. The purpose of this initial study is to obtain reliable SA values and to propose preliminary interpretations of what determines initial SA and what causes its evolution. A more complete future investigation strategy is also proposed to improve our understanding of snow microphysics.
Experimental System SA was determined by measuring the adsorption isotherm of CH4 on snow at liquid nitrogen temperature (77.15 K) by a volumetric method. CH4 was used rather than N2, because its saturating vapor pressure is P0 ) 12.77 mbar (15), making real gas corrections unnecessary. The setup is essentially that of Chaix et al.14 Methane is introduced in volume V1. Its pressure P1 is measured by a capacitance manometer, and the number of molecules n1 is deduced from the ideal gas equation. Methane is then expanded into volume V2, which contains the snow at 77 K. The pressure P2 is measured, from which the number of molecules still in the gas-phase n2 is deduced. The number of adsorbed molecules is then n1 n2, in equilibrium with P2. Increments of methane are then added, and an adsorption isotherm is thus recorded. Volumes V1 and V2 were measured by expanding helium from a volume calibrated by weighting the volume filled with water. Two different calibrated volumes were used, and both measurements gave the same values. For each sample the dead volume in V2 (i.e. V2 - snow volume) was measured by expanding helium from V1 (1029 cm3) into V2 (1890 cm3). Helium does not adsorb on ice. It diffuses into ice (16), and we tested whether this diffusion could affect our measurements. We exposed 47.3 g of snow placed in V2 to 3.20 mbar of helium (similar to what we use for our dead volume measurements) and measured continuously for 150 min the helium pressure drop. After 5 min, the pressure was 3.19 mbar, meaning that 4.7 × 1017 molecules of He had dissolved in the ice. After 150 min, pressure had dropped to 3.03 mbar. Because our dead volume measurements lasted less than 5 min, we concluded that diffusion of He into our snow samples was negligible. Snow Sampling. All snow samples were taken in January and February 1998 at Col de Porte, at an elevation of 1330 m, about 10 km north of Grenoble (210 m elevation), in the Chartreuse range, western French Alps. The sampling site was on the north slope of the pass, 30 m from the meteorological station of the Centre d’Etude de la Neige (CEN), operated by Me´te´o-France. Air temperature and humidity, surface snow temperature, snowpack height, liquid and solid precipitation rates, and wind speed are recorded 10.1021/es9811288 CCC: $18.00
1999 American Chemical Society Published on Web 10/19/1999
FIGURE 1. Snowpack height and snow fall rate at Col de Porte (1330 m asl, 45°12′N, 5°44′E). Time is GMT.
FIGURE 2. Air temperature (1.5 m above surface) and surface snow temperature (measured by IR emission) at Col de Porte.
continuously by CEN. Air temperature is measured in a ventilated shelter maintained at a height of 1.5 m above the snow surface. Surface snow temperature is measured within 0.4 °C by its infrared emission in the 8-14 µm wavelength range, so that the ice thickness probed is less than 20 µm, and the temperature measured will indeed be that of the surface crystals. The sun can hit the sampling site but only about 2 h a day and at near grazing incidence in the early afternoon in winter, and minimal amounts of sun energy are deposited in the snowpack. The sampling procedure is as follows. A trench is dug with a snow shovel. Temperature is measured at several heights in the snowpack and in the air with an alcohol thermometer. The top layer is sampled first, and layers are progressively removed to sample lower layers. The aspect of the crystals is examined with a 8× magnifying glass. A sampling spoon is used to fill glass vials with snow, taking great care to minimize snow packing. The spoon and the vials are thermally equilibrated at snow temperature. The vials are then immediately dropped in a l-N2 dewar, to prevent snow metamorphism between sampling and measurement. Snow density is measured by sampling and weighting a core using a PVC tubing 5.7 cm i.d. Because the snow is sampled over a finite depth, the distance of a snow sample from the surface is estimated with an accuracy of (3 cm. The snow is transferred to volume V2 in a cold room at -15 °C. To avoid condensation of air moisture on snow, the vials are thermally equilibrated to -15 °C before transfer. V2 is then connected to the vacuum line, placed in a l-N2 dewar, and evacuated with a turbo-molecular pump. Frost may form if the transfer procedure is inadequate. Frost formed at low temperature will be made up of amorphous, microporous ice with high SA, resulting in a significant artifact. This can be detected by performing a desorption isotherm, and the presence of high SA, porous frost will result in an hysteresis loop (17). The adequate timing between cooling and pumping was found to ensure that no frost was formed during transfer.
FIGURE 3. Photograph of a surface hoar crystal sampled at Col de Porte on January 28, 1998. Dimension of crystal shown is 18 mm.
Results Eleven surface area measurements were performed. The first sampling took place on January 20, 1998, at 13 h GMT, during a snow fall. CEN measurements (Figure 1) show that this snow fall had started the previous day at 18 h, lasted 1.5 days, and deposited 48 cm of snow. Another distinct snowfall had been observed from January 18 to 19, which deposited 24 cm of snow, but that fall was not sampled. The fresh snow crystals were dendritic and heavily rimed. This same layer was then sampled 3, 8, and 15 days later. The air and surface snow temperatures during this period are shown in Figure 2. On the last two samplings, a layer of very large (up to 6 cm) planar surface hoar crystals aggregates had formed on the snow surface (Figure 3), and this superficial layer was also sampled. The parameters measured during sampling are summed up in Table 1. The snow fall was a bit thicker
at our sampling site, 53-55 cm, about 30 m from the CEN measurement spot, than suggested by Figure 1. Adsorption isotherms were measured for all samples. Type 2 isotherms were obtained. A typical isotherm (sample 3) is shown in Figure 4. The classical BET treatment (18) was then applied to each isotherm. This treatment is based on the fact that according to eq 1 a straight line should be obtained if the left part of the equation is plotted as a function of P/P0
P/P0 n(1 - P/P0)
)
( )
C-1 1 P/P0 + nm‚C nm‚C
(1)
where P and P0 are CH4 pressure and saturating vapor pressure, respectively, n is the number of molecules adsorbed in equilibrium with P, nm is the number of adsorbed molecules when a monolayer is completed, and C is the BET constant, related to the BET net heat of adsorption, ∆Q, by RT ln C ) ∆Q. The slope and intercept of this BET plot then readily yield C and hence ∆Q and nm, from which the surface area is deduced if the area of a molecule is known. We used a methane molecular area of 19.18 Å2 (14). For each isotherm, at least nine data points were taken, at least five (on two occasions, but usually 7 or 8) of which in the region of P/P0 between 0.03 and 0.25, where the BET plot is found to be linear in most studies. The linearized BET plot of sample 3 is shown in Figure 5, and the surface areas, net heats of adsorption, and mass of each sample are summed up in Table 2. Because the equations yielding the surface area and ∆Q are complex, we did not evaluate our errors analytically. Rather, we performed sensitivity analyses by modifying all the parameters used in our calculation spreadsheet. The only parameter that had a significant effect was the dead volume VOL. 33, NO. 23, 1999 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
4251
TABLE 1. Snow Sampling Conditions date time 20/01 13 h 23/01 13 h 20 28/01 14 h
4/02 9 h 20
a
weather conditions
T air (°C)
no.
depth (cm)
T snow (°C)
snowing light wind sunny no wind
-5.7
sunny no wind
-2.5
1 2 3 4 5 6
3 50 3 30 50 0
-4.2 -1.0 -9.5 -4.5 -1.0 -2.5
0.04 0.1 0.14 0.18 0.17 a
7 8 9 10 11
3 50 0 5 40
-10.5 -3.0 -6.5 -12.2 -3.5
0.19 0.22 a 0.21 0.24
sunny no wind
-2.5
-6.5
Layer not thick enough to measure density.
b
density
comments heavily rimed dendritic crystals rounded grains rounded grains,
