Share this article on:

Intravenous Air: The Partially Invisible Phenomenon

Varga, Christopher PhD; Luria, Isaac MD; Gravenstein, Nikolaus MD

doi: 10.1213/ANE.0000000000001604
Patient Safety: Original Laboratory Research Report
Continuing Medical Education

BACKGROUND: Air injection is carefully avoided during IV solution administration; however, ambient air is dissolved in all liquids used for intravenous (IV) therapy. A portion of this gas will come out of solution in the form of bubbles as the solution is warmed to body temperature in a fluid warming system and/or within the body. We sought to quantify the proportion of the gas theoretically dissolved in room temperature crystalloid and 4°C blood products that comes out of solution in the IV tubing on warming to 37°C.

METHODS: Equilibrium-dissolved air calculations were performed for sodium chloride (0.9%), packed red blood cells, and fresh frozen plasma at various temperatures according to Henry’s Law. Outgassed gas volumes were experimentally measured for room temperature sodium chloride (0.9%) and 4°C blood products (packed red blood cells and fresh frozen plasma) warmed to 37°C during infusion into a body temperature water bath. The measured gas volumes were quantified as a fraction of the theoretical outgassing volumes required to maintain equilibrium saturation.

RESULTS: Measured outgassed volumes in the IV tubing in milliliters of gas per liter of fluid were 1.4 ± 0.3 mL/L (n = 6) for sodium chloride (0.9%), 3.4 ± 0.2 mL/L (n = 6) for packed red blood cells, and 4.8 ± 0.8 mL/L (n = 6) for fresh frozen plasma when these fluids were warmed to body temperature from their respective starting temperatures. Theoretical outgassed gas volumes required to maintain equilibrium saturation for the same fluids and temperatures are 4.7 mL/L for sodium chloride (0.9%), 8.3 mL/L for packed red blood cells, and 10.9 mL/L for fresh frozen plasma. As a fraction of the theoretical outgassing volumes, the measured air volumes represented 30%, 41%, and 44%, respectively, for sodium chloride (0.9%), packed red blood cells, and fresh frozen plasma. Prewarming crystalloid solutions to 37°C before administration significantly reduced the outgassing.

CONCLUSIONS: A significant and potentially clinically relevant amount of the resident dissolved gas in room temperature crystalloid, and 4°C packed red blood cells and plasma solutions comes out of solution on warming to body temperature. A nontrivial fraction of this outgassing is also expected to occur within the body circulation based on the results of this study. This can be substantially prevented by prewarming.

From the *CareFusion, Yorba Linda, CA; University of Florida College of Medicine, Gainesville, FL; and Department of Anesthesiology, University of Florida College of Medicine, Gainesville, FL.

Accepted for publication August 3, 2016.

Funding: None.

The authors declare no conflicts of interest.

Reprints will not be available from the authors.

Address correspondence to Nikolaus Gravenstein, MD, Department of Anesthesiology, University of Florida College of Medicine, 1600 SW Archer Rd, PO Box 100254, Gainesville, FL 32610. Address e-mail to ngravenstein@anest.ufl.edu.

Delivery of air into the circulation through intravenous (IV) fluid administration is an important clinical concern. The severity of venous air embolism generally is understood to depend on both the volume and the rate of venous air introduction.1,2 Even small gas embolizations that bypass or overwhelm the pulmonary capillary system may cause significant clinical consequences. There is ultimately a continuum of venous air-related events ranging from no discernible effect to death, with an underlying assumption that this generally scales with delivered air volume and introduction rate.3–5 Evidence of the biological and clinical effects of gas bubbles within the circulation exists including, tissue ischemia, inflammation, and activation of coagulation leading to thrombus formation.5 As a consequence, day-to-day clinical concern over potential negative outcomes due to delivery of air bubbles through IVs remains high, and a zero-tolerance approach to air delivery is common. Patients with a direct or potentially direct path from the venous circulation to the left side of the heart, such as those with a probe-patent foramen ovale, represent an estimated 25% to 40% of the general adult population.6,7 In these individuals, even small quantities of air introduced into the venous circulation can move to the left side of the heart and from there cause vascular obstruction in the brain or elsewhere.

The practical implementation of a zero-tolerance approach to the delivery of air/bubbles during IV fluid delivery is far from trivial. Abundant attention routinely is directed at elimination of air in IV lines before use. Because IV tubing sets are packaged dry, IV lines initially contain only air, and complete purging of all air from the IV set (ie, priming) before connection to the patient is complicated by recesses at access points where air easily remains trapped despite fluid flow through the tubing. In addition to the visible air residing in the IV tubing set, there is a less commonly considered, but significant, volume of air delivered into the venous circulation during IV fluid administration. This is the volume of air that is dissolved initially in the infusate at ambient or chilled conditions, a fraction of which must ultimately come out of solution as the infusate is actively or passively warmed, eventually equilibrating to body temperature. The equilibrium concentration of dissolved gas in a liquid at a given temperature and pressure is described by Henry’s Law8; the equilibrium-dissolved gas concentration is a decreasing function of increasing temperature. Thus, as infusate temperature is increased, there is excess gas dissolved in the liquid with respect to the new equilibrium condition, resulting in a state of supersaturation that drives bubble formation. This temperature-driven outgassing will necessarily generate bubbles in the IV set and/or subsequently liberate gas within the body, depending on whether body temperature equilibration of the infusate occurs extra- or intracorporeally. The use of simple air traps and/or more complex bubble removal devices/air filters within an IV tubing set can prevent ingress of externally generated air volumes but does nothing to address the invisible volume of air that may routinely enter the circulation dissolved within the infusate.

In the hospital setting, IV fluids commonly are delivered at room temperature. A predictable portion of the dissolved air in these fluids will inevitably come out of solution in the circulation as the body internally warms the liquid from room temperature (∼20°C) to body temperature (∼37°C), reducing the equilibrium solubility of air in the liquid. Some IV fluids are heated before administration. In this external-heating scenario, bubbles commonly and visibly form inside the IV tubing or within the IV fluid warmer itself as outgassing occurs due to the temperature rise. By comparing the equilibrium amount of dissolved air in the liquid at an initial temperature (eg, fluid temperature in the IV bag) to that at the heated-state temperature (ie, body temperature), the amount of air that must ultimately exit the liquid to re-establish equilibrium can be quantified. Because there is no free surface inside the IV tubing for direct diffusion of the air out of solution, any air leaving the liquid must do so through the formation of a multitude of bubbles at distinct nucleation sites on the tubing surfaces (heterogeneous bubble nucleation).9 In smooth-walled IV tubing, this limitation is expected to result in a fraction of the initially dissolved air remaining dissolved in the fluid as it enters the patient circulation. In a study examining outgassing during equilibration of Champagne, only 20% of the dissolved gas actually exited the liquid in the form of the familiar tiny bubbles, while the remaining dissolved fraction (ie, 80%) exited directly through the free surface.10 This result was opposite of expectations, and although supersaturation ratios are far less in warmed IV fluids, this suggests that the availability of bubble nucleation sites in the IV tubing set and IV fluid warmer surfaces will ultimately govern the amount of air that is liberated externally, while the remaining dissolved fraction may be liberated as tiny bubbles inside the body.

In summary of the aforementioned, we note that (1) more air is dissolved in IV fluids and blood products when they are cold than when they are warmed; (2) gas comes out of solution when IV fluids or blood products are warmed from room or refrigerated temperatures to a higher temperature, forming bubbles; (3) these volumes of gas that must come out of solution can be calculated using Henry’s Law; (4) when administering unwarmed IV fluids or blood products, this outgassing must occur within the body, potentially causing intravascular bubbles; (5) when IV fluids or blood products are warmed during or before infusion, some of the air will come out of solution in the tubing and these bubbles may be captured before infusion; (6) some of the outgassed bubbles will adhere to the tubing, and some will flow out the distal end; (7) the remaining dissolved air will outgas within the body; and (8) it may be useful to find better methods to more completely degas solutions before IV administration.

In this study, we present calculations of equilibrium volumes of outgassed air for common IV fluids and administration temperatures using available Henry’s Law constants.11–13 We subsequently compare these values to results from laboratory experiments wherein outgassed bubbles were collected from solutions of varying initial temperatures heated to body temperature while passing through a standard IV set. We provide these calculations and experimental data to provoke a conversation about whether the phenomenon of outgassing should be of clinical concern, and if so, should intentional mechanisms be put in place to remove air from IV fluids before delivery into the circulation?

Back to Top | Article Outline

METHODS

The equilibrium concentration of a given gas species dissolved in a liquid is governed by the Henry’s Law equation

CV

CV

where Ci (mol/L) is the equilibrium concentration of species i, k(Hi) (T) (mol/bar-L) is the Henry’s Law constant for species i and the corresponding liquid at temperature T (°K), and Pi (bar) is the partial pressure of gas species i above the liquid. Henry’s Law constants were calculated for oxygen and nitrogen according to the equation

CV

CV

where k(Hi(298.15)) is the Henry’s Law constant at the reference temperature of 25°C (298.15°K) for species i, and Bi is the temperature dependence constant for species i. (B(O2) = 1500°K, B(N2) = 1300°K).11 For simplicity in calculations, we have ignored trace gases and assumed that ambient air is represented by 21% oxygen and 79% nitrogen by volume. It is of interest to note that because of the greater solubility of oxygen in water compared with nitrogen that the composition of “air” dissolved in water is closer to 36% oxygen and 64% nitrogen.

Equilibrium-dissolved air calculations were performed for three common IV fluids, including sodium chloride (0.9%), packed red blood cells (PRBCs), and fresh frozen plasma (FFP) using available Henry’s Law constants for oxygen and nitrogen in water.11 Sodium chloride (0.9%) contains slightly less dissolved air (∼5% less) than pure water due to a “salting-out” effect; Sechenov “salt-effect” parameters12,13 were used in our calculations to appropriately scale the pure-water solubility values for oxygen and nitrogen. For blood products, our calculations account for the dissolved air in the estimated fraction of water contained in each constituent/formulation only (ie, red blood cells ∼70% water, plasma ∼92% water), and do not account for the possibility of any binding of oxygen to desaturated hemoglobin or solubility of oxygen and nitrogen in the lipid component of FFP.

Experiments (Table 1) were performed at The University of Florida in Gainesville, FL, under approval of the UF Institutional Review Board. Room temperature 1000-mL bags of 0.9% sodium chloride (Baxter Healthcare, Irvine, CA), expired units of human PRBC (LifeSouth Community Blood Centers, Gainesville, FL), and expired human FFP (LifeSouth Community Blood Centers) were hung on standard IV poles. The blood products were removed from a 4°C refrigerator just before the experiment. Infusion of the fluids was controlled using an Alaris PC infusion pump (CareFusion Inc, San Diego, CA). Smartsite infusion sets (Ref 2426-0007; CareFusion Inc) were used in all cases in conjunction with a 42-inch extension set (Ref BC 291-W; Codan US Corporation, Santa Ana, CA) as per the clinical routine in our operating rooms. For PRBC and FFP infusion, Y-type blood sets with standard blood filters (Ref 2C7627; Baxter Healthcare, Deerfield, IL) were used. All tubing was primed to clear any visible air from the tubing and associated injection ports and stopccocks. The IV tubing downstream of the infusion pump modules (∼50″) was submerged in a digital temperature-controlled, 28-L water bath (Model WB28A11B; PolyScience, Niles, IL) to heat the liquids to 37°C, as well as to maintain this temperature at the air collection point. This generic method of fluid heating was chosen over a commercially available fluid warmer to avoid a device-specific bubble formation contribution to the outgassed volumes. We note that commercial IV fluid warming devices often heat to temperatures greater than body temperature (eg, 38°C–41°C), however, because the fluids ultimately equilibrate to body temperature in the circulation, we focused on this final equilibration condition. Infusion of room temperature sodium chloride (0.9%) with no heating was treated as a control. A gas-over-water collection technique was employed in which 10-mL inverted graduated cylinders filled with 37°C water from the water bath were used to collect and measure the quantity of outgassed air during administration of each liquid. The IV tubing terminated in the inverted submerged graduated cylinder and any air exiting the IV tubing displaced fluid and could thereby be collected and measured in the inverted graduated cylinder. The air volume measurement precision was 0.1 mL. In addition to quantifying the outgassed air volume that exited the IV tubing and accumulated in the graduated cylinder, we also measured the volume of visible outgassed air that remained in the IV tubing; the latter was accomplished by applying a 40-mL room temperature 0.9% saline flush (drawn from an IV bag and without visible bubbles) just downstream of the infusion pump to move this remaining air to the submerged, inverted collection cylinder. Ambient temperature and IV fluid temperatures (in the fluid bags and at the submerged air collection location) were measured using a handheld thermocouple meter (Fluke Corporation, Everett, WA) with type-K thermocouples. The infusion rate was fixed at 300 mL/h for all experiments to simulate an intermediate clinical fluid administration rate. The volume of outgassed air collected in each case was normalized to milliliters of air per liter of liquid to account for differing initial volumes of sodium chloride (0.9%), PRBC, and FFP. Delivered fluid volumes were monitored using the infusion pump and were confirmed by weighing the IV bags and tubing sets before and after each experiment using a digital balance (EK Series; A&D Inc, San Jose, CA). Corrections were applied to the cylinder-collected air volumes to account for the presence of water vapor in equilibrium with the outgassed air at 37°C.

Table 1

Table 1

The effect of prewarming IV fluid on outgassed air collection was examined by heating 1000-mL bags of 0.9% sodium chloride to 37°C for 12 hours and subsequently infusing them after 2 different delay periods. A first delay period of 30 minutes served as a representative delay time between removal of an IV bag from a prewarmer and the start of its infusion into a patient. A second delay period of 12 hours was tested as a worst-case scenario of removing an IV bag from a prewarmer at the start of the day and infusion commencing at the end of the day. Each experimental condition was run 6 times.

We note that IV fluid heating devices often heat to temperatures slightly higher than body temperature (eg, 38°C–41°C), however, because the fluids ultimately equilibrate to body temperature in the circulation, we focused on this final equilibration condition.

Back to Top | Article Outline

RESULTS

Equilibrium-Dissolved Gas Calculations

Table 2 contains Henry’s Law calculations of the equilibrium-dissolved mass (mg-gas/L-fluid) of oxygen and nitrogen for each fluid type at representative initial temperatures for IV bags hung in the operating room, as well as the equilibrium-dissolved mass of each gas in each fluid at body temperature (37°C). Body temperature represents the eventual state of each infusate in the circulation. Unwarmed sodium chloride (0.9%) is typically introduced at room temperature (20°C–25°C), whereas unwarmed/undiluted blood products such as PRBC and FFP typically begin at 4°C. Table 2 additionally reports each dissolved gas quantity as a volume (milliliters of gas per liter of fluid) normalized to 37°C. This enables direct calculation of the volume of gas that must exit the fluid (ie, via outgassing) to re-establish equilibrium saturation when the fluid temperature is increased from an initial value to a final condition of body temperature. This is accomplished by taking the difference between the respective dissolved air volume at the initial temperature and the dissolved volume at 37°C. For example, consider delivery of 1 L of sodium chloride (0.9%) solution introduced at a room temperature of 20°C. We observe that when this liter of fluid is heated to body temperature (ie, 37°C), there must be 20.9 mL − 16.2 mL = 4.7 mL of air eventually liberated from solution, irrespective of whether this occurs externally within the IV set, internally in the circulation, or in some combination thereof.

Table 2

Table 2

Figure 1

Figure 1

Figure 1 provides a useful visual representation of the impact of initial IV fluid temperature for each of the 3 fluids considered in our calculations. The horizontal axis represents the starting temperature of the fluid in the IV bag as it is introduced. The vertical axis represents the volume (mL) of air per liter of fluid that must come out of the liquid to re-establish equilibrium at body temperature. Because of the drastically different temperatures at which blood products and crystalloids are typically introduced, we observe a large difference in the amount of air that must be liberated. For example, for 1000 mL of FFP introduced at 4°C, we note that nearly 11 mL of air must be liberated to re-establish equilibrium saturation at 37°C, whereas for a 1000-mL bag of sodium chloride (0.9%) introduced in a warmer operating room at 25°C, a much smaller ∼3-mL air volume must be liberated through outgassing to re-establish equilibrium.

Back to Top | Article Outline

Experimental Air Collection Results

Table 3 and Figure 2 contain experimental results from the present study. Experimental condition 1 represents a control condition; that is, outgassed air was collected during infusion with no appreciable temperature change of the liquid. Under this circumstance, we expect to collect near-zero outgassed air volume. As anticipated, when infusing room temperature sodium chloride (0.9%) to a room temperature water bath, the collected air volume was miniscule, averaging only 0.1 mL of air per liter of fluid. Experimental condition 2 is representative of typical surgical infusion of crystalloids, wherein bags are typically hung at room temperature. With the sodium chloride (0.9%) used for this case, we observed that, on average, 1.4 mL of air came out of each liter of fluid when heated from room temperature (ie, ∼20°C) to body temperature (∼37°C). This volume represents only 30% of the predicted volume of outgassing (ie, 4.7 mL) from Henry’s Law calculations for this temperature change. For PRBC (experimental condition 3) heated from 4°C to body temperature, the average volume of air collected per liter of fluid was 3.4 mL, whereas for FFP (experimental condition 4) over the same temperature change, we collected, on average, 4.8 mL of air per liter of fluid. Similar to the sodium chloride infusion, these measured values account for less than the predicted volumes from Henry’s Law calculations, at 41% and 44%, respectively. Experimental conditions 5 and 6 involved preheating 1000-mL bags of sodium chloride (0.9%) to 37°C for 12 hours to assess if this might be useful to degas the solutions before infusion, with 2 different delay times between the end of the prewarming period and initiation of infusion into the 37°C water bath. In experimental condition 5, the delay time was 30 minutes, which resulted in an initial IV fluid temperature of 34.0°C and an average collected air volume of just 0.3 mL/L of liquid. The degassing due to prewarming was effectively preserved in this case due to the relatively short prewarming to infusion delay time. In contrast, for experimental condition 6, the delay time was extended to 12 hours, resulting in the IV bag temperature dropping back to room temperature before infusion. In Table 3, we observe that the average outgassed air volume was 1.0 mL/L of liquid for this case, indicating that the degassing benefit of the prewarming period was effectively lost within this longer delay period of 12 hours.

Table 3

Table 3

Figure 2

Figure 2

Figure 2 illustrates the split of the total outgassed air in each test case between bubbles that freely exited the IV line in the flowing stream and were collected in the inverted graduated cylinder (the proxy for a patient) and bubbles that remained in the IV lines attached to the tubing walls. The discrepancy in this split across the 3 fluids tested is quite remarkable. For sodium chloride (0.9%), the bulk of the outgassed air (∼85%) exited the IV tubing, with very little residual air trapped on the tubing walls, whereas for the PRBC infusion, nearly 75% of the volume of air bubbles formed during outgassing remained trapped in the IV tubing (before flushing). For FFP, the split between these 2 volume categories was 50/50.

Back to Top | Article Outline

DISCUSSION

We used theoretical calculations to predict outgassed air volumes for common IV fluids and temperatures encountered in clinical practice, and compared these with air collection experiments using the same fluids. The results confirm that there is a potentially clinically significant volume of air that predictably (ie, theoretical > measured) remains dissolved in the infusate as it enters the patient. Because this volume of air is invisible to the caregiver, we expect that it is not commonly considered as a potential source of negative impacts, unlike visible air bubbles. This invisible air volume will be the largest (as a fraction of total outgassed gas) when IV fluids are administered with no external fluid warming in place. In this scenario, outgassing will occur almost exclusively within the circulation, and air volumes reaching the circulation could be as high as 11 mL/L of fluid infused at 4°C and around 5 mL/L of fluid initiated at room temperature. Inclusion of an air trap/vent or bubble removal device within the IV set will be ineffective in preventing this dissolved air volume from entering the circulation. In a surgical or critical care environment where IV fluid warming is more routinely implemented, some fraction of the dissolved air will be motivated to come out of solution within the IV tubing or warmer itself. The present study suggests that this fraction is in the range of 30% to 45% of the total dissolved gas. The use of this fraction within the context of our theoretical calculations suggests that the remaining dissolved air volume (ie, the invisible volume) that reaches the patient circulation is still substantial. For example, our experimental results with blood products (ie, PRBC and FFP) infused at 4°C and warmed to body temperature suggest that 5 to 6 mL of air per liter of fluid could potentially remain dissolved and enter the circulation where it would later be liberated internally. For common crystalloids such as sodium chloride or lactated Ringer’s solution infused at room temperature and warmed to body temperature, our results suggest that the remaining dissolved air volume reaching the circulation could be as large as 3 mL of air per liter of fluid. We note that lactated Ringer’s solution, although not tested experimentally in the present study, has a similar salt concentration to sodium chloride (0.9%) and thus has nearly identical theoretical outgassing volumes. In situations where large volumes of fluid are infused to a patient, the dissolved air volumes reported here on a per liter of fluid basis would accumulate to produce even higher circulating volumes. In contrast to the case of infusion of fluids with no external warming, inclusion of a bubble or air removal device within the IV set for cases where an IV fluid warmer is used allows the possibility of at least “catching” the fraction of air that is liberated within the IV set. This raises the question of whether additional mechanisms could or should routinely be put in place before or during IV administration to encourage additional external outgassing, or in the ideal case, to completely degas the IV fluid.

To this end, we investigated the effect of preheating IV fluids before use and found that if IV administration was initiated a short time after removal of the fluid from the prewarmer, the collected air from outgassing within the IV set was reduced to near zero. On the other hand, we found that if initiation was delayed by 12 hours (chosen to mimic an end of the day scenario) with the IV bag back in a room temperature environment, the effect of the prewarming degassing was virtually lost. Bulk prewarming of IV bags may therefore provide a reasonable approach to reducing dissolved air volumes in IV fluids before administration, but practical limitations and work flow must be considered.

For the outgassed air that was collected in our experiments in the present study, we attempted to quantify the split between bubbles that freely exited the IV tubing and bubbles that remained adhered to the IV tubing walls after administration. This attachment phenomenon is obviously flow rate dependent, and in the present experiments, we evaluated only a single fluid flow rate. For the representative flow rate tested, we found a large discrepancy in this bubble behavior and the air volume split across the fluid types. At the extremes, the bulk of the outgassed air for sodium chloride infusion freely left the tubing with very little remaining on the walls, whereas for PRBC, the opposite occurred, with the bulk of the outgassed air remaining in the IV tubing after infusion. For FFP, the split was much more even.

Back to Top | Article Outline

CONCLUSIONS

The results of the present study illuminate a potentially important aspect of IV fluid delivery, which is the partially visible and invisible (dissolved) volume of air that is routinely and unintentionally delivered to patients. The quantity and relevance of any outgassed air will be directly proportional to the total volume of fluids delivered, to the fluid temperature history, and to the anatomic vulnerability of the individual patient. If IV fluid is prewarmed and administration is initiated a short time after removal of the fluid from the prewarmer, the collected air from outgassing within the IV set is reduced to near zero.

Back to Top | Article Outline

DISCLOSURES

Name: Christopher Varga, PhD.

Contribution: This author helped design the study, conduct the study, analyze the data, and write the manuscript and attests to the integrity of the original data and the data analysis reported in this manuscript.

Name: Isaac Luria, MD.

Contribution: This author helped conduct the study, analyze the data, and write the manuscript.

Name: Nikolaus Gravenstein, MD.

Contribution: This author helped design the study, conduct the study, analyze the data, and write the manuscript and attests to the integrity of the original data and the data analysis reported in this manuscript.

This manuscript was handled by: Sorin J. Brull, MD.

Back to Top | Article Outline

REFERENCES

1. Mirski MA, Lele AV, Fitzsimmons L, Toung TJ. Diagnosis and treatment of vascular air embolism. Anesthesiology. 2007;106:164–177.
2. Toung TJ, Rossberg MI, Hutchins GM. Volume of air in a lethal venous air embolism. Anesthesiology. 2001;94:360–361.
3. Munson ES, Merrick HC. Effect of nitrous oxide on venous air embolism. Anesthesiology. 1966;27:783–787.
4. Oppenheimer MJ, Durant TM, Lynch P. Body position in relation to venous air embolism and the associated cardiovascular-respiratory changes. Am J Med Sci. 1953;225:362–373.
5. Barak M, Katz Y. Microbubbles. Chest. 2005;128:2918–2932.
6. Hagen PT, Scholz DG, Edwards WD. Incidence and size of patent foramen ovale during the first 10 decades of life: an autopsy study of 965 normal hearts. Mayo Clin Proc. 1984;59:17–20.
7. Foster PP, Boriek AM, Butler BD, Gernhardt ML, Bové AA. Patent foramen ovale and paradoxical systemic embolism: a bibliographic review. Aviat Space Environ Med. 2003;74:B1–64.
8. Henry W. Experiments on the quantity of gases absorbed by water, at different temperatures, and under different pressures. Phil Trans Royal Soc (London). 1803;93:274–276.
9. Jones SF, Evans GM, Galvin KP. Bubble nucleation from gas cavities: a review. Adv Colloid Interface Sci. 1999;80:27–50.
10. Liger-Belair G. The physics and chemistry behind the bubbling properties of champagne and sparkling wines: a state-of-the-art review. J Agric Food Chem. 2005;53:2788–2802.
11. Wilhelm E, Battino R, Wilcock RJ. Low-pressure solubility of gases in liquid water. Chem Rev. 1977;77:219–262.
12. Battino R, Rettich T, Tominaga T. The solubility of nitrogen and air in liquids. J Phys Chem Ref Data. 1984;13:563–600.
13. Benson BB, Krause D. The concentration and isotopic fractionation of oxygen dissolved in freshwater and seawater in equilibrium with the atmosphere. Limnol Oceanogr. 1984;29:620–632.
© 2016 International Anesthesia Research Society