Introduction

The growing use of chemicals containing chloride salts used in the snow and ice control operations, which are also called deicers, raised concerns about their impact on transportation infrastructure, vehicles and the environment. Deicer products are used to remove the ice from streets and roads once it has formed, while anti-icing products are used to prevent the formation of ice.

According to field studies in Finland, some agencies have begun to voice preference for potassium-based deicers [1–3]. Lindroos and Nysten [3] recommend in their report that groundwater quality problems associated with deicing can also be mitigated by using alternative deicing chemicals. Long-term monitoring studies showed that if potassium formate is applied as winter maintenance, this results in low load of organic carbon into surface water bodies [3, 4]. One of the key conclusions by Nysten et al. [4] is that potassium formate is more efficient against black ice compared to sodium salt. On the other hand, it was shown that for the removal of packed snow from roads, sodium salt is more effective.

The performance of a deicer is mainly determined from its eutectic and effective (cf. description further below) temperature. At ambient temperatures below the eutectic or effective temperature, a deicer will not perform at its best. The eutectic temperature is defined in the phase diagram of a eutectic system by the eutectic point and the eutectic concentration. Liquid and solid curves merge at the eutectic point, i.e., the eutectic of a deicer defines the lowest temperature at which it can melt ice. Effective deicers have eutectic temperatures well below the expected ambient temperature range. In Table 1, the eutectic temperatures and concentrations of three deicers are summarized [5].

Table 1 Eutectic concentrations and temperatures of deicers [5]
Full size table

Know-how on the practices in deicing and cost-efficient use of potassium formate has to be developed. Potassium-based deicers may increase costs, but last longer and can be more powerful at melting ice, especially at cold temperatures. There are many parameters playing a role in the effectiveness of deicer products used in winter maintenance. Understanding the performance characteristics and negative impacts of deicers is crucial for effective and sustainable winter maintenance operations. A more precise selection of deicers should then be possible.

We present a study to evaluate various deicers in terms of their performance, like thermal properties or ice-melting capacity. Differential scanning calorimetry (DSC) has been identified as a promising laboratory test for evaluating deicers, and it was demonstrated how such results can be utilized in decision-making framework [6–12].

Experimental

Materials

Analytical grade sodium chloride (NaCl >99.8 mass%) was purchased from Carl Roth, Germany, and used without further purification as a reference. Different aqueous deicers in the range of 20–50 mass% were provided by Evonik Industries AG, Germany, used without further purification, and are summarized in Table 2.

Table 2 Selected parameters of the aqueous deicers studied
Full size table

DSC experimental setup and protocol

A Setaram DSC (microDSC7evo) and a Netzsch DSC (204 Phoenix) were used in our study. The Setaram DSC operates between −45 and 120 °C and the Netzsch DSC between −150 and 600 °C. The DSC devices were selected according to temperature requirements. All experiments with deicing solutions were conducted at atmospheric pressure.

In the Setaram DSC device, two Hastelloy cells were enclosed in the same temperature-controlled furnace. Both cells are sealed during the measurement with the sample in the sample cell, while the reference cell is left empty. The Setaram DSC was equipped with a thermostat for cooling, and dry nitrogen was used as purge gas.

The furnace of the Netzsch DSC is a disk-shaped sensor on which the crucibles are placed. The reference crucible is left empty during the experiments. The Netzsch DSC uses liquid and gaseous nitrogen for cooling purposes. The influence of different kinds of sample cells was studied, too. We used cold-weldable aluminum crucibles being sealed with a handy sealing press, and stainless steel crucibles (high-pressure cells) that could be joint together gastight with a mounting tool. The calibrations of both DSC were made using high-purity gallium and pure distilled water at different heating rates.

For the dynamic experiments in the Setaram DSC, the furnace temperature was ramped down from room temperature down to −45 °C at rate of 1 °C min−1. The temperature was kept for 20 min to freeze the liquid completely and subsequently raised from −45 to 20 °C with heating rates of 0.2, 0.5 and 1 °C min−1. As an example, a thermal analysis curve of the Setaram DSC for sodium chloride (20 mass%) with a heating rate of 1.0 °C min−1 is shown in Fig. 1. Two endothermic peaks are observed during heating in Fig. 1. The first peak, at around T = −21 °C, is due to the melting of the eutectic mixture of ice and NaCl. The subsequent peak is due to the non-isothermal melting of ice, as a result of progressive dilution of NaCl in the aqueous phase [13, 14]. The Netzsch DSC was used in a lower temperature range down to −80 °C, and three different heating rates for the dissociation were used as well. A representative dissociation curve of the experiments conducted with Netzsch DSC including furnace temperature and heat flux over time of potassium formate (50 mass%) is given in Fig. 2. All samples were run in triplicate, at least. If there were deviations between repeated measurements in the used DSC device, further experiments were conducted in the other DSC device for control purposes.

Fig. 1
figure 1

Exemplary thermal analysis curve of sodium chloride (20 mass%), heating rate 1.0 °C min−1 (Setaram DSC)

Full size image
Fig. 2
figure 2

Representative dissociation curve of potassium formate (50 mass%), heating rate 0.5 °C min−1 (Netzsch DSC)

Full size image

For pure, eutectic or stoichiometric composition, DSC traces show a sharp melting peak, and the onset temperature T onset corresponds to the phase transition temperature, i.e., the start of melting. For non-stoichiometric and non-eutectic compositions, the DSC curve shows a broad melting peak, and the equilibrium temperature is defined by the end of melting: It is the temperature when the last crystals vanish. A representative DSC dissociation curve is shown in Fig. 3, in which the phase change temperatures were characterized following the definitions given in [13, 14]: The onset temperature was determined in case of eutectic melting. T onset can be determined at the intersection of the auxiliary line through the descending peak slope with the baseline. For the progressive melting peak, we determined two different temperatures (T peak, T end). Tpeak denotes the extremum peak temperature, whereas T end is the temperature corresponding to the inflection point of heat flow curve. T end (β = 0) is the inflection point temperature at zero heating rate with β as heating rate. This temperature was extrapolated by the inflection point temperature T end at 0.2, 0.5 and 1.0 °C min−1. The inflection point is the peak point of the first derivative of the heat flow curve, which was calculated based on the heat flow and temperature data [13, 14].

Fig. 3
figure 3

Representative DSC dissociation curve with characteristic temperatures, reproduced from [13]

Full size image

Results and discussion

The effective temperature of a deicer is the lowest temperature limit at which the deicer remains effective and achieves ice melting. It corresponds to the temperature threshold below which ice crystals start to form in diluted deicer solution. Above the effective temperature, there is no ice crystal present, so that there is no risk of slippery pavement.

If the apparent transition temperature is plotted as a function of heating rate (Fig. 4), the extrapolation to zero heating rates (β = 0) should give the best estimation for the phase transition temperature. In the literature, it has been pointed out that T peak does not represent the end of the phase transition, but the recorded temperature indicates the maximum of energy exchange [13–16]. The last solid particle dissolves at the time corresponding to the inflection point T end. With higher heating rates, peaks are shifted to higher temperature. Then, the peak area gets bigger (broader and higher) and the converted mass per time unit is larger. The influence of heating rate on temperature shifts can be eliminated by the extrapolation to zero heating rate [17]. The extrapolation to zero heating rate followed the GEFTA procedure described in [18], where the thermal inertia of the measuring system (“thermal lag”) is considered. The extrapolation to zero heating rates (β = 0) should give the best estimation for the phase transition temperature and was compared to the literature data.

Fig. 4
figure 4

Extrapolation to zero heating rate for melting of potassium formate (20 mass%) compared to the literature value [5]

Full size image

In general, we observed in our investigations that the DSC curves were very reproducible for each heating rate. The experiments with different kinds of sample cells with the Netzsch DSC resulted in better reproducible curves when using the high-pressure cells assuming better contact face between bottom of cell and disk-shaped oven. Seven aqueous deicers gave similar curve shapes, so that we present here only curves with potassium formate (20 mass%). The melting temperatures of aqueous deicers were underestimated using T peak (β = 0). Extrapolating T end (β = 0) to the zero heating rate gives better results than using T peak (β = 0). In Fig. 4, it can be seen that the extrapolated T end (β = 0) for potassium formate (20 mass%) is very close to the literature value given in [5]. The freezing points and heat of fusion of aqueous deicers studied are summarized in Table 3, where freezing points given in the literature are listed as well [5].

Table 3 Freezing points and heat of fusion of aqueous deicers studied
Full size table

The freezing point of sodium chloride used as a reference material in the concentration of 20 mass% was determined to be approximately −18 °C. Compared to two other aqueous deicers with the same concentration of 20 mass% potassium formate and potassium carbonate, the DSC measurements showed up with clearly higher freezing points (KF approx. −11 °C, KC approx. −8 °C). In general, the freezing points decrease with increasing salt concentration, which was stated in our studies. The potassium formate-based deicer with the highest concentration (50 mass%) results in the lowest freezing point (KF approx. −55 °C) as the lowest effective temperature of all studied aqueous deicers.

Figure 5 shows a deviation plot with [(T Lit − T (β = 0))/T Lit] × 100 % on the ordinate and the solution concentration on the abscissa. It was found that besides one value (T peak for K2CO3, 30 mass%), all other values of [(T Lit − T (β = 0))/T Lit] are smaller than 1 %, whatever the solute concentrations were. This temperature difference is highly acceptable.

Fig. 5
figure 5

Temperature deviation of T end or T peak at zero heating rate (β = 0) from reference temperature over different solute concentrations

Full size image

During measurements of K2CO3 (30 mass%), we observed with both DSC devices that white crystals were left as residue in the sample cells. The heat flow curves of K2CO3 (30 mass%) showed some disturbances, thereby preventing determination of the inflection point and T end, respectively. Only T peak was determined in this case.

By using calorimetric methods, time-consuming performance tests following SHRP can be overcome. An empirical correlation of DSC measurements and ice-melting capacity tests according to SHRP at 0 °C is described in the literature [6, 11], where the extremum peak temperature T peak is used as the characteristic temperature and a heating rate of 3.6 °C min−1 in the DSC measurements. In our DSC experiments, we used three different heating rates (0.2, 0.5 and 1.0 °C min−1). We characterized two temperatures, T peak and T end, and found that those temperatures underestimate and overestimate the melting temperatures of aqueous deicers. Compared to the literature values, the extrapolated T end (β = 0) at zero heating rate generated clearly better results than T peak and T end. The effective temperature of a deicer should therefore use extrapolated T end (β = 0) to assess the deicer performance under certain road weather conditions.

The change in heat flow was calculated in [6, 11] by subtracting the total heat of fusion for pure water from the measured heat flow of the characteristic peak. It is noted in [6, 11] that the more powerful a deicer is, the less external heat is needed to melt the ice. This was confirmed by our experiments. The potassium formate-based deicer with the highest concentration (KF 50 mass%) results in the lowest melting enthalpy of all studied aqueous deicers.

Conclusions

Calorimetric methods were used to study different aqueous deicers in the range of 20–50 mass% to determine the thermal properties (freezing point, melting enthalpy) of aqueous brines. We developed procedures for two different DSC apparatuses and extrapolated the characteristic temperatures T end and/or T peak to zero heating rate. T end (β = 0) at zero heating rate, which was extrapolated with three different heating rates, generated better results than T peak and T end. The extrapolated temperature T end (β = 0) should be used as effective temperature for the selection of a deicer.

Freezing points of three different aqueous solutions with varying concentrations were determined. The results matched the literature data very well. The DSC curves were highly reproducible for each heating rate.

A more powerful deicer needs less external heat for melting, i.e., the deicer with the highest concentration (KF 50 mass%) results in the lowest melting enthalpy.

The deicer performance was characterized by calorimetric methods, with the advantages of DSC measurements: being quick and easy to perform.