RAPID INDUCTION OF HETEROGENEOUS ICE NUCLEATION IN A BIOLOGICALLY COMPATIBLE COOLANT

Abstract

Hypothermia is gaining recognition as an important medical treatment. To treat local cases of injury such as stroke, or certain surgical procedures, there is a need to induce local hypothermia. To treat shock or cardiac arrest survivors, there is a need to rapidly induce global hypothermia. Rapid induction of hypothermia has been achieved in animal research, but it has yet to be achieved clinically using a simple, widely practicable method. The clinical need for therapeutic hypothermia represents an engineering opportunity to develop an easy to use coolant that is sterile, biologically compatible, and maximizes coolant heat capacity.

Here we present an initial characterization of a prototype platform technology designed to create a sterile, biologically compatible, high heat capacity coolant that has the potential to be used in all of these clinical applications. The coolant is a specially processed micro-particulate ice saline slurry, that can be easily pumped into a patient through surgical tubing, syringes, or minimally invasive surgical instruments. The device induces heterogeneous ice nucleation in a saline stream that has been super-cooled from room temperature to a temperature below the saline freezing point. Currently, the device begins continuous production of ice slurry that contains ~30 % ice by mass within 10 minutes. The nominal ice particle diameter is smaller than 100 μm. This work represents a significant first step toward addressing clinical needs for rapid human cooling.

1. INTRODUCTION

Cardiac arrest and stroke are significant public health concerns, affecting thousands of people daily in the United States. Established medical treatments have poor success, with cardiac arrest resuscitation efforts returning ~2% of out of hospital victims to their daily routine, and stroke resuscitation efforts returning only ~70% to their daily routine.^1-2 One of the reasons that the outcomes from these diseases are so poor is irreversible injury to the brain and heart begins within minutes following onset of ischemia. Ironically, mechanistic research into hypoxia induced cell death has demonstrated that cells can survive prolonged hypoxia; instead cell death is initiated by the return of oxygen, called reperfusion injury.^3-4 Ischemia followed by reperfusion leads to a burst of mitochondrial free radical production which usually leads to cell death.^5-7-4 Reperfusion injury induced cell death is attenuated by therapeutic hypothermia targeted at the burst of free radical production immediately following reperfusion.³ Reperfusion injury is the most likely explanation for the difference in the number of patients who achieve return of spontaneous circulation after cardiac arrest, ~30%, and the final survival numbers, ~2%.

Recently, two randomized control trials demonstrated that the induction of mild hypothermia, core temperature in the 33-34 °C range, in comatose patients who achieved return of spontaneous circulations (ROSC) had significantly improved outcomes when compared to patients who did not receive the treatment.^8-9 This led the AHA to make a class IIa recommendation that therapeutic hypothermia become the standard of cardiac arrest care in 2005.¹⁰ The clinical efficacy of therapeutic hypothermia in the treatment of stroke has yet to be proven,^11-12 but the efficacy is well demonstrated in animal models.^13-14

The average time to target temperature in the HACA and Bernard studies was several hours, a timescale which is consistent with cooling rates recently reported in the treatment of stroke and cardiac arrest,^15-18-12 and hypothermia induction device promotional literature.¹⁹ However, resuscitation experiments in rodents have demonstrated that intra-ischemia cooling, which targets early reperfusion injury, provides the most protection, either survival in cardiac arrest or reduced infarct size in stroke, and that the conferred protection from hypothermia decays rapidly with time to target temperature.^20-13-14-21 For these reasons, it is preferable to induce hypothermia before reperfusion, either ROSC in cardiac arrest or clot removal in stroke. This history highlights a gap in our clinical ability to induce therapeutic hypothermia to treat multiple ischemic injuries, to rapidly cool the ischemic area before reperfusion, and to induce hypothermia in the pre-hospital setting. Technologies that address this gap will further fulfill the promise of better outcomes, survival for cardiac arrest and reduced infarct size for stroke, for the 100,000s of patients annually who suffer from these emergent ischemic injuries.

2. ICE PARTICULATE SLURRY RATIONALE

Clinical adoption of therapeutic hypothermia has been relatively slow. There are several practical reasons for this. The human body is well adapted to preserve core body temperature, making it challenging to cool the patient. Most hypothermia induction technologies use external heat exchangers to cool the body, forcing the cooling technologies to overcome corporeal heat conservation and generation mechanisms. The process generally takes hours to achieve, and proper patient temperature maintenance is both critical and challenging.²²

One simple method which has gained widespread adoption is to rapidly deliver 2L of ice cold saline into the patient. Vascular delivery improves heat transfer efficiency, and the 2L can be delivered in less than 10 minutes. This results in 1.5 °C of cooling. 2L of additional fluid is generally well tolerated by cardiac arrest patients, but the addition of higher volumes to achieve additional cooling can stress the damaged heart, and lead to subsequent heart failure. This technique has two significant advantages: it is relatively simple and fast and it requires little extra training for nurses and physicians. The drawback is that too little cooling is achieved.

We are investigating phase-change coolants as a way to increase the cooling achieved with a rapid 2L fluid infusion. Ice has a latent heat of 334 KJ/Kg. Therefore, an engineered two-phase, ice and saline, coolant that is 50% ice by mass can result in an ice-particulate slurry that has more than twice the cooling capacity of ice cold saline alone.¹⁹ The major obstacle to creating a successful ice-slurry is that it must be easy to pump and flowable. If chunk ice is mechanically sheared into sub millimeter particles, the resulting particles are dendritic (have many jagged edges), and the slurry cannot be pumped because the ice particles will clump together. The ice particles will build up at any irregularity in the flow pathway, and only the liquid phase will be able to pass through. A solution to this is to alter the shape of the ice particles using chemical or thermal smoothing. Once the sharp, dendritic features of the particles have been removed, the ice-slurry can be easily pumped as a two-phase coolant that is biologically compatible, melts to form normal saline, can be rapidly introduced into the patient, providing rapid cooling.

Originally, ice-particulate slurry was made in a blender. Desired proportions of ice, water, and salt were added to the blender, and the ice was broken down by an initial blending stage. At the end of this stage, the slurry would have ice particles with a diameter of ~100 μm, but the slurry could not be easily pumped because the edges of the particles were dendritic. A second smoothing stage was performed, where a second addition of salt to the mixture, followed by slower mixing would smooth the particles and make the slurry pumpable. This was a very successful proof of concept method,²³ but was limited by many issues, most notably: sterility, repeatability, and air bubble entrainment.

3. SLURRY ON DEMAND PROTOTYPE VISION

The slurry on demand prototype, Fig. 1, was designed to reside inline in an IV tubeset between the patient and standard medical grade saline bags. The device houses a heat exchanger that would cool room temperature saline to the freezing point and grow ice particles, as well as a mixing chamber where high concentration saline is added to smooth the ice particles, after which the slurry is pumped into the patient. The prototype is a closed system which ensures sterility and eliminates air entrainment. Designing the device to reside in-line shortens the delivery tube length, reducing ice loss in the tubing, and ensures homogenous delivery because the slurry is never stored or stationary. The device houses a counter-current scraped-surface (CCSS) heat exchanger. A secondary coolant flows between the slurry production heat exchanger and a second heat exchanger which resides in a dry-ice alcohol bath. Utilizing the phase change of the dry-ice enables the formation of ice in room temperature saline, without the use of large refrigeration units which would consume significant electrical power. The envisioned process is very similar to the blender process; a low concentration saline (0.45%) is cooled from room temperature to the freezing point of the saline. Ice particles will begin to form, and will be scraped off of the heat exchanger. Ice will continue to form in the low concentration saline stream, until the ice fraction is ~50%. Concurrently, there is a high concentration saline stream (3%) that is also being cooled from room temperature down to the freezing point of the low concentration stream (0.45%). The high concentration saline stream and the slurry stream are mixed, which leads to chemical smoothing. The slurry then exits the device, and is delivered to the patient through standard IV tubing.

Figure 1.

Image of device prototype, with labels. Coolant flow direction is depicted by the blue arrow and saline flow direction is depicted by the red arrow.

4. HEAT TRANSFER

The current prototype is comprised of three heat exchangers. The principle heat exchanger, where the saline is subcooled, is a scraped surface counter current (SSCC) heat exchanger. A copper coil heat exchanger is used to reduce the coolant, 70% potassium formate, temperature, and a third heat exchanger is used to set the inlet saline temperature. Two peristaltic pumps set the saline/slurry and the coolant flowrates. An electric motor is used to turn the driveshaft and scraper blades. During experimental operation, saline of concentrations, 1%, 4.7%, or 10% is pumped into the heat exchanger. Before entering the scraped surface heat exchanger, the saline temperature is set to a desired inlet temperature, either 10 °C or 3 °C using a heat exchanger and water bath. Simultaneously, the potassium formate is pumped through a heat exchanger submerged in a dry-ice propanol bath. The driveshaft, which has eight scraper blades, turns at 100 rpm, effectively scraping the surface every 0.08 s. A schematic of the temperature profile and operating conditions is shown in Fig. 2 below.

Figure 2.

Schematic of heat exchanger wall, with scraper blade, coolant, and saline. Representative temperature profile is shown. T_C is the bulk coolant temp, T_W is the saline side wall temp, and T_S is the bulk saline temp.

The heat flux for the SSCC heat exchanger can be calculated using

$$q=\mathit{UA}\mathrm{\Delta }{T}_{\mathit{lm}}$$ (1)

where U is the total heat transfer coefficient, A is the heat exchange surface area, and ΔT_lm is the log mean temperature, defined in this case as

$$\mathrm{\Delta }{T}_{\mathit{lm}}=\frac{\mathrm{\Delta }{T}_2-\mathrm{\Delta }{T}_1}{\ln \left(\mathrm{\Delta }{T}_2/\mathrm{\Delta }{T}_1\right)}$$ (2)

where

$$\begin{array}{lll}\mathrm{\Delta }{T}_1& =& T_{{\mathit{Saline}}_{\mathit{in}}}-T_{{\mathit{coolant}}_{\mathit{out}}}\\ \mathrm{\Delta }{T}_2& =& T_{{\mathit{Saline}}_{\mathit{out}}}-T_{{\mathit{coolant}}_{\mathit{in}}}\end{array}$$ (3)

where T_{fluidlocation} represents the temperature of the specified fluid at the inlet or outlet of the SSCC heat exchanger. The heat exchanger is made from stainless steel; the table below contains pertinent geometric information.

5. EXPERIMENTAL METHOD

Dry ice was placed in a dewar filled with propanol. When the propanol reached -70 °C, the coolant heat exchanger was submerged in the ice bath. The secondary coolant is a 70% solution of potassium formate, which has a freezing temperature lower than -70 °C and is non-toxic. The motor which turns the scraper blades was set at 100 rpm, effectively scraping the surface every .08 seconds. Temperatures were measured at the coolant inlet, coolant outlet, saline inlet, and saline outlet. Coolant and saline flowrates were varied with each experiment. The actual flowrates are reported in the results section

6. THREE ICE PRODUCTION MODES

STABLE ICE FORMATION: WALL SCRAPING

Saline and coolant flow rates were matched at 150 ml/min. Dry ice was placed in a dewar filled with propanol. The saline was 1% by mass. A special chamber was fabricated for the device, which provided a window to observe the scraper blade, and to visualize ice particles as the exited the heat exchanger.

The temperature increase for the coolant was 25 °C, from -15 °C to 10 °C. The temperature drop for the saline was 18 °C, from 25 °C to 7 °C. Ice formation on the wall was consistent, and significant ice would be mixed into the saline stream, as shown in Fig. 3. However, because the saline outlet temperature was 7 °C, the ice would melt quickly. Attempts to use pre-cooled saline or to slow the saline flowrate were not successful. Pre-cooled saline would result in a solid ice plug inside the heat exchanger. Slowing the saline flowrate resulted in warmer saline outlet temperatures and less ice scraping. While it seems paradoxical that a slower saline flowrate would result in a higher outlet temperature, the slower flowrate results in a higher percentage of the heat exchanger surface area being coated with ice. Ice fouling lowers the heat exchanger efficiency, resulting in warmer saline. Ice fouling of the heat exchanger was the major obstacle encountered during these experiments, and became the first obstacle to overcome.

Figure 3.

Ice scraped off of the heat exchanger wall. Picture taken through experimental observation window.

UNSTABLE ICE FORMATION: HETEROGENEOUS ICE NUCLEATION IN SUPERCOOLED SALINE IN THE HEAT EXCHANGER

Saline flowrate was varied between 215 ml/min and 130 ml/min. Saline concentration was raised to 4.7% by mass. Coolant flowrate was increased to 400 ml/min in an attempt to maintain a more uniform heat exchanger wall temperature. Ice formation was observed through a special window to observe the scraper blade, and to visualize ice particles as the exited the heat exchanger.

Adjusting the saline flow rate through the slurry production heat exchanger provided control over the slurry outlet temperature. Using 4.7% saline depressed the freezing point of the saline that the ice fouling behavior changed. Temperature profiles and the heat transfer analysis for an exemplary run are shown in Fig. 4. When the saline flow rate was reduced from 180 ml/min to 130 ml/min, the saline exiting the heat exchanger was super cooled. Heterogeneous ice nucleation occurred spontaneously in the supercooled saline inside the heat exchanger, resulting in a small volume of highly ice loaded slurry. During heterogeneous nucleation, the surface of the heat exchanger gets coated in ice, and the heat transfer efficiency drops. At this point, ice is shaved off of the heat exchanger surface, and the device behaves as it did in the previous result. A time series of pictures of heterogeneous ice nucleation in the heat exchanger are shown in Fig. 5. Clearly heterogeneous ice nucleation in supercooled saline can result in a highly ice loaded slurry. Ice fouling of the heat exchanger surface was still a problem. Heat transfer data are shown above in Table 2.

Figure 4.

Coolant and saline temperature measurements during nucleation and scraping event. Light shading represents nucleation, dark shading represents ice scraping. (+) represents coolant inlet temperature, (×) represents coolant outlet temperature, (▲) represents saline inlet temperature, (◆) represents saline outlet temperature, (■) represents slurry temperature after ice nucleation occurs.

Figure 5.

Time series photographs of heterogeneous ice nucleation within the heat exchanger. The resulting slurry is highly ice loaded, but cannot be produced continuously.

Table 2.

Heat transfer data, averaged over measurement area shown in Figure 4. Averages are shown plus or minus standard deviation.

	q	UA	U
	[W]	[W/K]	[W/m2 K]
Coolant	225.2±1.3	17.2±0.11	1651.1±10.4
Saline	184.4±0.59	14.1±0.04	1304.6±3.93

STABLE ICE FORMATION: HETEROGENEOUS ICE NUCLEATION IN SUPERCOOLED SALINE OUTSIDE THE HEAT EXCHANGER

The saline flowrate was 323 ml/min. For these runs, the saline that entered the device was cooled to 6 °C before entering the heat exchanger. This allowed us to maintain higher saline flow rates to minimize the chances of ice nucleation occurring within the heat exchanger. Saline concentration was 4.7% by mass. Coolant flowrate was increased to 520 ml/min in an attempt to maintain a more uniform heat exchanger wall temperature.

Fig. 6 shows the coolant and saline temperature dataset from this experiment. The saline outlet temperature passes below the freezing point of the saline, -2.8°C shown as a dashed line in Fig, 6, at roughly 5 minutes. The saline stream is maintained at a supercooled temperature for the remainder of the run, about 13 minutes. To induce ice nucleation outside of the heat exchanger, a seed ice particle was inserted into the saline flow on the end of a thin metal strip. Once heterogeneous ice nucleation is induced, the original seed particle can be removed, and ice nucleation will continue. In addition to flow stagnation points we have successfully induced heterogeneous ice nucleation using a seed ice particle and droplet impact on a flat fluid surface. As can be seen in Fig. 6, the temperature of the slurry is higher than the temperature of the supercooled saline stream. This is a result of the release of the heat of fusion during the phase change from liquid to solid. Slurry was produced for a total of 6 minutes during this experiment, thus a total of 1.9L of slurry was produced. This data demonstrates our ability to produce saline-ice particulate slurry creating de novo ice particle in a supercooled saline stream. Heat transfer data is provided in Table 3 on the left. As discussed in the next section this method creates ice slurries that are 15-30% ice by mass. A picture of the supercooled saline undergoing heterogeneous nucleation in a 60 ml syringe is shown in Fig. 7. Heat transfer data are shown above in Table 3.

Figure 6.

Temperature profiles for coolant, saline, and slurry as a function of time. Light shading represents duration of heterogeneous nucleation. Dark shading represents sampling for ice concentration measurement. (+) represents coolant inlet temperature, (×) represents coolant outlet temperature, (▲) represents saline inlet temperature, (◆) represents saline outlet temperature, (■) represents slurry temperature after ice nucleation occurs.

Table 3.

Heat transfer data, averaged over measurement area shown in Figure 6. Averages are shown plus or minus standard deviation.

	q	UA	U
	[W]	[W/K]	[W/m2 K]
Coolant	244.4±3.6	18.2±0.15	1745.7±14.5
Saline	223.6±2.53	16.7±0.2	1541.6±18.6

Figure 7.

Heterogeneous nucleation of supercooled saline after impact with a fluid interface. Volumetric flowrates are in the range of 200-350 ml/min.

7. Quantifying Slurry Quality

Ice saline slurry is a two phase coolant. One of the properties of coolants of this type is that the ice particles are in a constant state of change. By this we mean that they are constantly changing size and shape as heat passes between the ice and the fluid on the molecular level, even in a perfectly insulated container. This makes ice saline slurry hard to store, and it makes the properties of the ice saline slurry hard to determine. Because perfect insulation is nonexistent, the ice fraction of the slurry also decreases rapidly as time passes. Therefore, it is preferable to measure slurry characteristics as the slurry is being produced, as opposed to try to characterize the slurry after the fact. We have developed several methods to characterize the slurry.

METHOD TO MEASURE ICE MASS FRACTION

Ice crystals formed in a saline solution will contain only water molecules, provided the temperature of the saline is above the eutectic point of the solution (-21.1C for NaCl). Therefore ice formation will cause an increase in the salt concentration of the fluid. An increase in salt concentrations will cause an increase in the liquid density, which can be measured. The formula to measure the ice concentration is

$${\varphi }_{{\mathit{mass}}_{\mathit{ice}}}=1-\frac{{\varphi }_{{\mathit{mass}}_{{\mathit{Saline}}_{\mathit{old}}}}}{{\varphi }_{{\mathit{mass}}_{{\mathit{Saline}}_{\mathit{new}}}}}$$ (4)

where φ_{mass_ice} is the mass concentration of the ice, φ_{mass_Saline_old} is the mass concentration of the original saline, and φ_{mass_Saline_new} is the mass concentration of the saline after ice nucleation. Saline is drawn in aliquots from the delivery tubing of the device during heterogeneous nucleation. Ice is prevented from entering the sample with a sub-micron sized metal filter. The advantage of this technique is that the sample is taken during production, so there is minimal error in the measurement due to ice melt.

This method is robust, and has indicated that heterogeneous ice nucleation in a supercooled saline stream results in a slurry that contains 15-30% ice by mass. The 30% ice measurements have been further confirmed by the initiation of slurry pumping problems during those experiments. When a fluid becomes ~30% solid by volume, non-Newtonian viscous behavior is expected, and was observed. A further advantage of this technique is that it can easily be scaled down and built into the device for continuous ice fraction measurement and quality control.

METHOD TO MEASURE HEAT FLUX DEPENDENCE ON ICE PARTICLE SIZE

Heat transfer is strongly related to the amount of surface area available for heat exchange. For ice particulate slurry, the surface area to ice volume ratio increases as the ice particle size decreases. Therefore, it is expected that heat transfer, or heat flux, will be improved in the slurry if the ice particle size decreases. We created different slurries using the original blender technique, but blended the slurries for two different times, 2 min and 1 min. As shown in Fig. 8, increases in blending time results in smaller nominal ice particle radii. To measure the heat flux, the slurry was pumped through a copper tube that was submerged in a well stirred warm water bath. A thin film heat flux sensor was mounted on the copper tube – water bath interface, and measured.

Figure 8.

Plot of nominal ice particle size as a function of blending time. Particle size was measured using ImageJ and microscopic images of the slurry.

The results for these experiments are shown in Fig. 9. The smaller ice particles, radius of 63 μm, resulted in more heat flux than the larger ice particles. The heat flux measurement had to be normalized to ice volume to account for differences in ice volume that arose from the differences in slurry blending time. Heat flux can also be measured in-line during slurry production. If the heat flux measurement is taken continuously and coupled to the ice volume measurement, this will allow characterization of the ice content and particle size, in real time, for quality assurance.

Figure 9.

Heat flux measurement, normalized by ice volume, as a function of ice particle size. It is clear that smaller particles increase heat transfer

QUANTIFYING ICE PARTICLE SIZE AND SHAPE

Images of the ice particles are challenging to obtain. The lamp that illuminates the image emits a significant amount of heat. The slurry causes a significant amount of condensation on the microscope slide. To overcome these issues, we developed an insulated double pane dish to image the ice particles. The images were taken with a Nikon film SLR camera, model number 6006, attached to a Nikon brightfield microscope, model Eclipse TE300. We made two blender slurries, one without chemical smoothing, one with chemical smoothing, and one with prototype 1. Pictures were taken at 10X magnification, at two different times.

The images from the ice slurries, Fig. 10, show differences between the ice particles formed by the three methods. The two blender slurries have a variety of ice particle sizes. This is especially clear in the T2 column, where some melting has occurred. The remaining ice particles in the blender slurries are quite large. The ice particles remaining in the heterogeneous ice nucleation slurry are more monodisperse. A more monodisperse slurry will have better fluid characteristics and a more predictable melting behavior in the blood stream.

Figure 10.

Images of ice particulate slurry made by three different methods. Pictures were taken at two different times; T2 was taken about 1 minute after T1.

8. CONCLUSIONS

The first prototype to produce slurry on demand has been a success. The device is capable of generating ice in the saline stream through two different mechanisms. Ice can form on the heat exchanger surface and be scraped into the stream. The saline can also be supercooled, at which point heterogeneous ice nucleation can be induced. If ice nucleation occurs within the heat exchanger, the heat exchanger interface is fouled with ice, and heat transfer efficiency drops significantly. If ice nucleation does not occur within the heat exchanger, the prototype is capable of producing liters of supercooled saline at flowrates in the 200-400 ml/min range. In this mode, the prototype is capable of producing high volumes of ice-particulate slurry that contain 15-30% ice by mass, when heterogeneous ice nucleation is induced outside of the heat exchanger. One distinct advantage of heterogeneous ice nucleation outside of the heat exchanger is that any tubing diameter can be selected. This makes it easier to scale down the inner diameter of the slurry delivery tubing. This will make it easier to adapt the final prototype to different medical indications such as stroke or laparoscopic surgery, and increases the likelihood that the slurry could be used in existing hypothermia induction technologies.

The methods we have developed to quantify slurry quality have provided insight into the nature of the ice formed by the original blender method and the first prototype. Microscopic images of the microparticulate ice slurries reveal that the prototype produces ice particles of a smaller diameter than the blender method, and the ice particle size distribution is more monodisperse. A more monodisperse ice slurry will have more reliable viscous properties.

The current prototype would be drastically improved through a modification of the heat exchanger. Counter current heat exchangers are relatively efficient, but are not ideal for this application. Ice fouling of the heat exchanger interface changes the temperature profile on the coolant side of the heat exchanger. Instead it would be preferable to develop a heat exchanger that provided a uniform wall temperature that was slightly below the freezing temperature of the low concentration saline. This would reduce the effect of ice fouling, and simplify the system making slurry production more reliable.

Table 1.

Prototype heat exchanger geometry

Device Dimensions [mm]	66 × 57 × 152
Internal Surface Area [m2]	0.0108
Saline Volume [ml]	24
Annular Gap [mm]	4.94

NOMENCLATURE

A

Heat exchanger surface area

q

heat transfer rate

T

temperature

U

total heat transfer coefficient

ΔT_LM

Log Mean temperature difference

φ

mass fraction