Acidic Vapors Above Saturated Salt Solutions Commonly Used for Control of Humidity
ROBERT L. OPILA . JR ..MEMBER. CHARLES J. WESCHLER. AND RUDOLF SCHUBERT
Abstract – The gaseous transfer from a saturated NaCl solution, of chlorine but not sodium, to copper coupons has been demonstrated in a bell jar that was sealed for nine years. Thermodynamic calculations indicate that the active agent. HCI(g), is in equilibrium with H* (aq) and CI (aq) in the saturated salt solution. Auger electron spectroscopy was used to analyze the surface of the copper coupons, and several equivalent monolayers of chlorine were found on the surface. However, the amounts of chlorine found were significantly less than might be expected to be deposited from air in equilibrium with the saturated salt solution. These results are compared with other reported work.
Because laboratory experiments that are designed to accelerate corrosion rates frequently utilize a saturated salt solution to maintain constant with, various saturated salt solutions have been calculated.
INTRODUCTION
LABORATORY EXPERIMENTS are frequently designed to accelerate corrosion rates by increasing pollutant concentrations, temperature, or relative humidity (RH). The RH can be controlled by either dynamic or static methods. The former require more labour and equipment (gas cylinders, flow meters, special glassware, etc.). Static control is much simpler; RH is controlled through the use of saturated salt solutions in a sealer chamber [1]-[5]. However, a study by Schubert and Tompkins [6] indicated that components of the salt solutions can be transferred to the surface of metal coupons exposed within such a chamber. This process can further accelerate corrosion and alter surface chemistry in a manner that may not have been anticipated in the original design of the laboratory experiment.
In this paper we will describe an experiment in which copper coupons were placed on wooden blocks in a sealed bell jar containing a saturated NaCI solution. The bell jar remained sealed and undisturbed for slightly more than nine years. The saturated NaCl solution maintained a constant humidity and has also been found to be a source of vapor phase HCI. The vapour phase HCI in turn, interacted with the surface of the copper coupons. The expected concentration of vapour-phase HCI within the bell jar was calculated from simple thermodynamic arguments. This calculated concentration was subsequently found to be low, due to the unanticipated release of formic and acetic acids from wood support blocks and subsequent dissolution of these species in the saturated salt solution. The magnitude of this effect has been estimated.
Auger electron spectroscopy was used to analyze the surface of the copper coupons, and several equivalent monolayers of chlorine were found on the surface. However, the amounts of chlorine found were significantly less than might be expected on the basis of HCI released from the saturated salt solution. These results are compared with reported work of others.
Because laboratory experiments that are designed to accelerate corrosion rates frequently utilize a saturated salt solution to maintain constant RH, the partial pressures of gases phase species in equilibrium with various saturated salt solutions have been calculated.
EXPERIMENTAL
Samples
Oxygen-free high conductivity (OFHC) Cu coupons (1mm thick) were cut to 14 x 14 mm. samples were cleaned ultrasonically in twice distilled water, followed by 95-percent ethanol. Four coupons were used for the exposure experiment, and eight coupons were set aside for future reference.
Chemicals
ACS reagent grade NaCI was used as the solute in the saturated salt solution. The solvent was freshly distilled 18-MΩ water. The solution was prepared in a 250-ml glass beaker, had a final volume of approximately 200 ml, and sufficient water was present to keep the excess NaCI covered with water. Buffers for pH measurements were Fisher Certified ph4.00 and pH 7.00.
Equipment and Procedures
The exposure chamber is schematized in Fig. 1. Positioned within a 13-1 bell jar were a 250ml beaker containing the saturated NaCI solution and two kiln-dried wooden blocks. Two copper coupons were placed directly on one of the wooden blocks, and two additional copper coupons rested upon aluminum studs stuck into the other wooden block. Several NaCI crystals were placed on one of the coupons resting on the wooden block and one of the s coupons on aluminum stud. The remaining two coupons were left bare. On June 7,1976 the bell jar was sealed with silicone high-vacuum grease and remained undisturbed at ambient temperature (22-25⁰C) until August 29, 1985 (over nine years). The initial Auger analyses were performed within hours of opening the chamber.
The Auger investigation utilized a Perkin Elmer-Physical Electronic 595 Scanning Auger Microprobe and spectra were recorded using a 5-kV,0.5uA electron beam rastered over 4x10⁴cm² area. For depth profiling, a 2-kV,0.2mA Ar beam was rastered over 0.10cm². under those conditions, the sputtering rate through thermally grown SiO₂ is approximately 50 A/min.
The pH measurements were made with a Fisher Accumet Model 825 MP pH Meter using two- point standardization with pH 7.00 and pH 4.00 buffers. The electrodes consisted of a universal glass pH electrode and a Calomel reference electrode.
RESULTS
After sitting undisturbed for slightly more than nine years, the bell jar was opened. An odor characteristic of acetic acid emanated from under the chamber. More will be said about the source and consequences of the acetic acid below. The copper coupons were visually inspected. The two coupons on which several crystals of NaCI had been placed were severely corroded. Indeed, the surface of these two coupons, including the corrosion products, appeared moist before the bell jar was opened; this was consistent with the presence of salts that deliquesce at 75-persent RH or less. The two bare coupons were not visibly corroded. However, they were slightly darker in appearance than the control coupons that had remained in ambient air during the same time interval. An Auger spectrum of a bare coupon that was exposed in the bell jar is shown in Fig.2. The major elements detected on this coupon are carbon(c), chlorine (CI), oxygen (O), and copper (Cu). The large carbon signal is not unusual for surfaces exposed to the ambient environment. A depth profile is shown in Fig.3. where the Auger intensities of C, Cu, CI, O, and sulfur (S) peaks are plotted as a function of sputtering time. The depth profile was interrupted occasionally, and the entire Auger spectrum was recorded. Throughout this exercise, no sodium (Na)was detected. The intensity of the Auger CI signal initially increases as a function of sputtering time, reaches a peak, and then decreases. The C Auger intensity was greatest at the surface. The S and O Auger intensities peaked together at longer sputtering times than the CI peak.
As described in the Experimental Section, eight coupons from the same sheet of OFHC copper were not enclosed in the bell jar, but instead were stored, protected from the dust, in ambient air. Several of these coupons were cleaned following the original procedure and inspected by Auger spectroscopy. The Auger spectrum of the surface of one of these coupons is shown in Fig.4. The Cand CI peaks are less intense, and the Cu, O, and S peaks are more intense than the corresponding peaks for the coupons from the bell jar. Na and N are present, but only in small amounts. During depth profiling the C, CI, O, and S Auger signals decrease monotonically with time. There is significantly les CI present on the sample stored in room air than on that exposed in the bell jar.
Auger analysis was performed on the reverse side of the bare copper that rested on the wood block (i.e., the side in direct contact with the wood). An Auger spectrum from this surface is displayed in Fig.5. The C Auger peak dominates the spectrum, although O and Cu are also evident. The depth profile showed only a monotonic decrease in the C Auger intensity. Some S was detected as sputtering proceeded, but no CI was seen. The strong C signal may be partially due to acetic and formic acid outgassing from the wood (see below). The lack of a CI signal on this side (the “wood side”), as contrasted with the significant CI signal on the reverse side, is evidence that the source of the CI is not the wood block.
Approximate elemental compositions of the surface films can be determined by dividing the Auger peak-to-peak intensity by the appropriate sensitivity factors, using handbook values [7], and normalizing. Such compositions are summarized in table 1for several areas on each sample after various sputtering times. The approximate atomic concentrations for each coupons confirm that the NaCI solution and not the wood block is the source of the CI.
By assuming a sputtering rate equivalent to SiO₂, (50 A/min) the depth profile data can be converted to approximate elemental compositions. The amount of CI in the near-surface region may be roughly estimated by integrating the depth profile. Integration of the CI depth profile shown in Fig.3, after converting to atom percent and assuming one monolayer is 5 A thick, gives 4.5 monolayers of CI. Taking 2x10¹⁵ atom-cm² monolayer, approximately 1x10¹⁶ atoms of CI were transferred to the surface from the salt solution. This estimate is likely to be good to within an order of magnitude.
The presence of CI without Na on the surface of the bare copper coupons exposed in the bell jar indicates that the CI was not transferred to the copper surface by the surface “creep” of a NaCI solution. Vapor-phase transfer via gas phase HCI is the most probable mechanism. HCI (g) is in equilibrium with the H₃O+ (aq) and CI¯ (aq) present in the saturated NaCI solution. The equilibrium vapor pressure of HCI over the saturated NaCI solution will depend directly on the concentration of H₃O+ (i.e., the pH) of the solution. The pH of this solution was measured on two occasions-shortly after the bell jar was first opened and eleven days later (see Experimental). Between the two sets of measurements, the bell jar was sealed. The pH was found to be 3.145 ± 0.003 in the first set of five measurements and 3.152 ± 0.002 in the second set of four measurements. A pH value of 3.15 will be used in subsequent analysis of this system. The reported [8] concentration of chloride ion [CI¯], in a saturated NaCI solution at 25̊ C is 5.42 molar (M) or 6.15 molar (m). The solution density is 1.1979 g/cm³.
DISCUSSION
The vapor pressure of HCI(g) in equilibrium with a given concentration of H+ (aq) and CI¯ (aq) can be calculated from thermodynamic considerations. At 298 K and 1 bar, the free energy of formation ΔG⁰ of HCI (g) is -95.3 kJ/mol and ΔG⁰ of H(aq) + CI (aq) is -131.2 kJ/mol [9]. Consequently, ΔG ⁰for the reaction
H+ (aq)+CI¯ (aq)= HCI (g)
Equation (4) can be used to calculate the vapor pressure of HCI (g)within the bell jar for the conditions described in the results section. The mean activity coefficient of a 6.0-molal solution of NaCI IS 0.986 [10]. Hence, the activity of the chloride ion in the saturated salt solution can be approximated by its molality of 6.15m. The activity of the hydrogen ion, 7.1 x 10¯⁴, is derived from the saturated salt solution’s measured Ph of 3.15. (The pH of a solution is defined as the negative base 10 logarithm of the hydrogen ion activity.) However, the equation pH = -log is employed with the understanding that it is, strictly speaking, a formal relationship, as is itself undefined – see Harned and Owen [10, p.443]. Consequently, the vapor pressure of HCI(g), in equilibrium with the pH 3.15 saturated NaCI solution contained within the exposure chamber, was 1.7 x 10¯ torr. This relatively high vapor pressure is directly linked to the unexpectedly lo Ph of the saturated NaCI solution. Water in equilibrium with air typically has a pH of about 5.6 due to dissolved CO₂ [11]. The high ionic strength of the saturated NaCI solution will increase the pH, but the magnitude of this effect is relatively small, in addition to being in the wrong direction (for a 6.15-m NaCI solution equals 2.54 [10], where yi is the molal activity coefficient of species i; this value would correspond to a pH of 7.20 if there were no CO₂ (aq). Similarly, alkali dissolved from the borosilicate glass of the beaker might also be expected to increase the Ph of the solution, but this effect is also in the wrong direction. The measured hydrogen ion activity, two orders of magnitude greater than initially expected, is believed linked to the odor of acetic acid that was noted when the bell jar was opened. The only sources of acetic acid were the three small wood blocks (2x3.2x8.5 cm) placed within the bell jar to hold the specimens. Various woods are known to outgas acetic acid [12]-[14]. The type of wood used in this experiment (maple) outgasses both acetic and formic acids, with the concentration of acetic being approximately ten times greater than that of the formic acid [12] [13]. An acetic acid concentration of 2.9x10-₂ M or a formic acid concentration of 2.9 x 10- M is consistent with a pH of 3.15 (k (acetic) = 1.76 x 10¯⁵ ; K (formic) = 1.77 x 10¯⁴ [15]. Unfortunately, we were not in a position to measure the actual concentration of acetic and formic acid in the saturated salt solution at the time the bell jar was opened. However, given the presence of the wood blocks, it is not unreasonable that the above concentration would be present during the exposure experiment. For acetic acid, a vapor phase concentration of approximately 8.7 x 10¯² torr would lead to such a solution concentration ( ΔG for CH₃ COOH (aq)→ CH₃ COOH (g) at 298 K is 22.5kJ [9]). It is interesting to note that the calculated vapor pressure of HCI(g) within the bell jar, 1.7 x 10⁶ torr, is actually higher than reported indoor HCI concentrations and also higher than or comparable to average outdoor concentrations. Rice et al. [16] measured mean HCI (g) concentrations between 4.1 x 10¯⁸ and 1.5 x 10¯⁷ torr inside air-conditioned buildings at six sites (“urban” and “industrial”) across the United States. Matusca et al. [17] report outdoor HCI (g) concentrations between 5.1 x 10¯⁸ and 7.1 x 10¯⁷ torr in rural areas, and between 1.0 x 10¯⁷ and 1.5 x 10¯⁶ torr in urban areas. In other words, the equilibrium concentration of HCI (g) above a saturated NaCI solution (pH 3.15), intended to maintain fixed humidity, is greater than that encountered in many real-world applications.
In a hypothetical environmental chamber that contained a saturated sodium chloride solution, in the absence of other factors that influence the solution pH other than atmospheric CO₂, the pH would be approximately 5.6. In such a case, the vapor pressure of HCI in equilibrium with the solution at 25 C would be approximately 6 x 10¯⁹ torr.
A similar calculation can be performed for the vapor pressure CI₂ over the salt solution. In this case, the reaction is 2H (aq) + 2CI¯ (aq) + ½ O₂(g)→CI (g) + H₂o (aq) and has ΔG, = 25.3 kJ/mol. This yields CI₂ vapor pressures of 2.4 x 10¯⁷ torr at pH 3.15, 7.6 x 10¯¹²torr at pH 5.4 and 4.8 x 10¯¹⁵ torr at pH = 7, significantly less than the vapor pressure of HCI at these pH’s. In fact, the equilibrium vapor pressure of CI₂ is unlikely to be reached because oxidation by oxygen in acid solutions are kinetically slow unless appropriate catalysts are present [11]. Given the purity of the H₂O and NaCI that were used in this experiment, the presence of a suitable catalyst is unlikely. Since the vapor pressure of CI₂ is 0.1 times the vapor pressure of HCI at pH 3.15, we will neglect CI₂ for the purposes of the following discussion.
Returning to the system that has been the focus of this study, the vapor pressure of HCI is expected to have remained relatively constant over the nine years period after equilibration was achieved (several days). Consequently, the kinetic theory of gases can be used to describe the transport of HCI(g) to the surface of the Cu coupon. An HCI(g) partial pressure of 1.7 x 10 torr results in an HCI wall collision rate approximately 3 x 10¹⁴ cm¯². In nine years, there would be approximately 1 x 10²³ collisions/cm² with the Cu coupon, yet the Auger analysis indicated the presence of only about 1 x 10¹⁶ CI atoms/cm². In other words, only one in 10⁷colliding atoms remained on the copper surface. Actually, this description oversimplifies the interaction between HCI(g) and the copper surface. The copper surface was covered with a thin film of adsorbed water. During the nine-year exposure, the RH within the bell jar was fixed by the saturated NaCI solution at 75 percent. At 75 percent RH and 30⁰C, the aqueous film on a gold surface is 30-35 A thick [18]. Similar studies indicate that approximately ten monolayers of water are present on a copper surface at 75 percent RH]19]. Given that the copper surface is more polar than the gold surface, 30-35 A can be used as a minimum value for the thickness of the aqueous film on the copper coupons within the bell jar. This aqueous layer is in equilibrium with HCI (g), HCOOH (g), and CH COOH (g). Consequently, it contains hydrogen ions, chloride ions, formic and acetic acids, and formate and acetate ions. If the aqueous surface film behave as bulk water, the solution concentration of these species could be calculated from thermodynamic considerations and the vapor-phase concentrations of the related compounds. However, in such a thin film, solid/liquid and liquid/air interface effects are expected to be strong and to propagate throughout the water layers. According to corrosion literature [20], unless an oxidizing agent is present, copper is expected to display little reactivity under such conditions. Indeed, the amount of chlorine detected on the copper surface may reflect a quantity determined by stability constants for various copper and chlorine containing complexes (i.e., values obtained from equilibrium considerations), rather than a quantity determined by irreversible reactions of chloride ions with copper over the entire nine-year period.
However, it should be noted that under similar conditions (vapor pressure of HCI between 1.3 x 10¯⁶ and 3.8 x 10¯⁶ torr) Rice et al. [21] report a much greater corrosion rate than we observed in the present study. There are several possible explanations for this: 1) If the rate-determining step in the HCI corrosion of copper is strongly pH dependent, different pH’s in the aqueous films on the copper surfaces could lead to different corrosion rates. However, available information suggests that the surface pH’s in the two sets of experiments are comparable. 2) Initial corrosion rates on copper are expected to be much greater than final corrosion rates (Graedel et al ) [22]. However, the small amount of product we observe does not allow for a large rate even in the initial day of exposure. 3) The work of Rice et al, was conducted in a flowing system (50-500 1/min) while the present study considers test coupons in a static system. However, given the small dimensions within the bell jar, it seems unlikely that mass transport limitations would occur. 4) The presence of acetic acid in the “bell jar” experiment may generate a copper acetate film of sufficient thickness to significantly deter HCI/Cu corrosion processes (see, e.g., Tompkins and Allara [23]). The Auger studies are inconclusive with regards to the existence of a copper acetate film. The carbon Auger intensity is typical of surfaces exposed to the ambient environment, and the oxygen Auger intensity is not correlated with the carbon Auger intensity as would be expected for an acetate film. However , other work has shown that copper oxalate and formate salts will decompose to CO₂ under electron and ion bombardment, resulting in no detectable carbon Auger signal from these species [24]. Further experiments are planned to elucidate the reasons for the discrepancy in corrosion rate between the present study and that reported by Rice et al. [21].
The mechanism described by scenario 4) above is probably operative on the bottom, “wood” side of the coupon. Since the mean free path for air is approximately 6 x 10¯⁶cm at 760 torr and the surface roughness of the wood blocks was approximately 3 x 10¯³ cm, significant amounts of HCI could have been present in the regions between the Cu coupon and the wooden block. Yet negligible amounts of CI were detected on the bottom side of the Cu coupon. Because of the proximity of the wooden block, a thicker adsorbed film, undoubtedly containing significant amounts of formate and acetate, likely formed on the bottom side of the coupon, and this adsorbed film could inhibit the corrosion of Cu by HCI, as described above.
Nevertheless, the work of Rice et al. [21] does demonstrate that low concentrations of HCI, comparable to those that can be generated above saturated salt solutions, have the potential to cause significant amounts of copper corrosion. The concentrations of HCI(g) generated in a environmental chamber containing a saturated NaCI solution can react to an even greater extant with other materials (e.g., aluminum). Furthermore, other saturated salt solutions, commonly used in environmental chambers for the static control of RH, can also generate significant concentrations of potentially corrosive vapors.
Table II lists calculated vapor pressures of gas phase species, mostly acids, in equilibrium with various saturated salt solutions in a closed system at 25⁰ C. The list is by no means described in ASTM Standard E 104, “Standard practice for Maintaining Constant Relative Humidity by Means of Aqueous Solutions” [1]. The entries in Table II have been calculated using the same approach outlined above to calculate the vapor pressure of HCI(g) in equilibrium with a saturated solution of NaCI. Values of ΔG were derived from the appropriate quantities as tabulated in “The NBS Tables of Chemical Thermodynamic Properties” [9]. And the concentrations of the ionic species were taken from “Solubilities of Inorganic and Organic Compounds” [8].
Vapors pressures of the gas phase species have been calculated for three different hydrogen ion activities: pH 7, Ph4.6, and pH 3.15. At the hydrogen ion activity most likely to exist in simple exposure studies, pH 6.6, numerous saturated salt solutions generate HCI (g) and HNO₃ (g) concentrations on the order of 10¯⁸ to 10¯⁹ torr; HBr (g) concentrations are typically two orders of magnitude smaller. Certain saturated solutions, prepared from the salt of a weak acid or a weak base, fix the hydrogen ion activity. The vapor pressure of the corresponding gas phase species is also fixed. For acetate salts, the vapor-phase concentrations of acetic acid are greater than 10¯⁶ torr. In the case of ammonium salts, the vapor pressure of NH₃ (g) above the saturated solution can be very high (e.g., 3 x 10¯² torr for (NH₄) SO₄ and 8 x 10¯⁴ torr for NH₄CI). In many studies such concentrations are sufficiently high for these species to become important components of the exposure experiment.
CONCLUSIONS
We have demonstrated the gaseous transfer of CI from a saturated NaCI solution to nearby Cu surfaces. Thermodynamic calculations indicate that the active agent is HCI (g), in equilibrium with H+ (aq) and CI¯ (aq) in the saturated salt solution. Investigators must be aware of the vapor pressures of gas phase species in equilibrium with various saturated salt solutions. However, the authors are unaware of any discussion of tis point in the literature that describes such solutions for maintaining constant humidity. We have presented partial pressures of corrosive gases in equilibrium with a variety of saturated salt solutions (table II). This table is representative, not comprehensive. However, ΔG⁰ values and salt solubilities are published ]8], [9] that permit calculations, similar to those used to prepare Table II, for any of the salt solutions commonly used for the control of humidity. Adequate knowledge of the vapor-phase chemicals in an exposure experiment requires that such calculations be made, or referred to, any time salt solutions are used for the static control of humidity in an enclosed air space.