Additional Carbon Dioxide Removal by Oxygenated Dialysis Fluid: Insights for the Development of a Novel Lung and Kidney Assist Device

Abstract

RenOx, a novel artificial lung and kidney assist device, combines gas exchange and dialysis fibers for integrated respiratory and renal support, with dialysis fibers intended for toxin clearance and filtration. However, when kidney support is not needed, dialysis fibers could be repurposed for additional respiratory support for patients in exacerbated cases, and to compensate losses in CO₂ transfer caused by the partial replacement of gas exchange fibers by dialysis fibers. We analyzed the feasibility of extracorporeal gas transfer via dialysis membranes with fully oxygenated and decarboxylated dialysis fluid in a closed circuit, quantifying O₂ and CO₂ exchange during standardized in-vitro tests with blood. Oxygenated dialysate was pumped through a dialyzer with a similar dialysis fiber area (0.6 m²) to the RenOx (adult size). Gas transfer efficiency was evaluated at blood-to-dialysate flow ratios of 1, 3, and 6. Average CO₂ removal from 12 to 35 ml/L_blood was achieved by adjusting blood-to-dialysate flow ratio, approaching the full metabolic requirement of adult patients (40 ml/L_blood). Maximum oxygen supply was 15 ml/L_blood. Blood pH and hematocrit were within physiological range. This study proposes a simple method to enhance lung support in the RenOx, advancing research on CO₂ removal by dialysis.

Extracorporeal membrane oxygenation (ECMO) offers a lifesaving alternative to >20,000 patients/year¹ as bridge-to-transplant or bridge-to-recovery (venovenous ECMO).^2,3 Lung support is provided by blood flowing around hollow-fiber membranes in the oxygenator, in a way that oxygen diffuses from gas to blood, and carbon dioxide is eliminated from blood.⁴ However, outcome is often limited because 70% of ECMO patients develop acute kidney injury, thus requiring continuous renal replacement therapy (CRRT).^5,6 Current treatment for concurrent lung and kidney support involves two distinct extracorporeal circuits, one for ECMO and another for CRRT, as presented in Figure 1A. This arrangement requires a higher number of pumps, cannulas, tubing, and devices, and thus, increases the probability of inflammation, hemorrhage, blood clotting, and blood cell damage.^7–10

Figure 1.

Current treatment of ECMO patients who suffer from severe renal failure is suboptimal involving separate circuits for ECMO and CRRT, increasing complications (A). Novel integrated lung and kidney support (RenOx) in a single device and circuit is developed to decrease current drawbacks (B). When kidney support is not required, the RenOx system can be easily modified to supply extra lung support by relocating the dialysis compartment for gas exchange (C). CRRT, continuous renal replacement therapy; ECMO, extracorporeal membrane oxygenation.

One attempt to overcome disadvantages related to ECMO and CRRT in separate circuits is the development of a novel artificial lung that integrates kidney support (RenOx).^9,11,12 Combined support is delivered by a single membrane bundle integrating gas exchange fiber layers and dialysis fiber layers.^9,11,12 This allows a simplified circuit with less tubing, pumps, cannulas, and extracorporeal surfaces to be applied, as presented in Figure 1B, targeting a safer and less costly treatment.

RenOx provides gas exchange, toxin removal, and filtration (Figure 1B).^9,11,12 However, patients with severe pulmonary and renal impairment, often present different needs in terms of organ support along their treatment trajectory.^6,13 In this sense, optimal timing for kidney support depends on their degree of fluid overload and accumulation of metabolites.^6,14 Typically, kidney support is performed for a period of time (up to a few days) until renal function parameters are satisfactory, and resumed later if deemed necessary. In contrast, lung support needs to be continuously and typically longer than kidney support, sometimes exceeding >20 days.¹⁵ Thus, it is expected that RenOx patients require kidney support to be switched on and off during treatment to attend these needs. The RenOx device is adapted to allow for this need.^9,11,12 However, when kidney support is stopped, dialysis fibers are temporarily unused, and solely contribute to the mixing of blood flow and thus to convective oxygen transfer.¹¹

Therefore, we hypothesize that RenOx’ dialysis fibers can be used to enhance gas transfer, especially CO₂ exchange, in patients who temporarily do not require kidney support. In this modified arrangement (Figure 1C), RenOx provides the single function of enhanced lung support for a determined period, i.e., gas exchange by both gas exchange and dialysis fibers. Supplementary CO₂ removal can assist in the treatment of patients with sudden and severe CO₂ retention, such as those with chronic obstructive pulmonary disease (COPD).^16,17 In addition, a higher gas exchange surface area could aid patient weaning. Therefore, using dialysis fibers to boost gas exchange in RenOx can be advantageous for two key reasons. First, additional gas exchange can compensate for the loss in CO₂ transfer caused by the replacement of 25% of gas exchange fibers by dialysis fibers in the RenOx.^11,12 Second, this re-purposes dialysis fibers when kidney support is unnecessary, with minimal modification to the RenOx circuit, Figure 1C, maximizing the usage of available surface area.^11,12 In a previous study, we identified that the replacement of gas exchange fibers resulted in a minor loss of CO₂ transfer capacity of up to 15%.¹¹ Although 40 ml/L_blood to comply with the metabolic requirement of an adult patient was still provided,¹¹ increasing CO₂ transfer capacity could be beneficial for patients with an acute exacerbation.

This study proposes a new method to transfer CO₂ (and oxygen) using dialysis fibers. For that, fully oxygenated (and decarboxylated) dialysis fluid is recirculated in a closed loop through dialysis fibers maintaining a continuous driving force for gas exchange while preserving essential blood components such as minerals and hormones, and reducing dialysate waste.¹⁸ Although previous literature explored O₂ and CO₂ transfer with dialysis fibers (see Table S1, Supplemental Digital Content, https://links.lww.com/ASAIO/B577) these mainly focused on respiratory dialysis and extracorporeal CO₂ removal (ECCO₂R). In opposition to respiratory dialysis and ECCO₂R¹⁹ that focus on eliminating CO₂ at low blood flows (<500 ml/min), ECMO patients also require sufficient oxygenation, and thus treatment at higher blood flows (0.5–7 L/min, adults). Furthermore, respiratory dialysis often applies a lower blood (Q_b) to dialysate (Q_d) flow ratio, i.e., Q_b/Q_d <1. To our knowledge, no standardized method exists to provide gas transfer through dialysis fibers suitable for the RenOx operation (see Table S1, Supplemental Digital Content, https://links.lww.com/ASAIO/B577).

This study aims to quantify the amount of carbon dioxide and oxygen that can be transferred by fully decarboxylated dialysis fluid recirculated through dialysis fibers in a setup similar to the RenOx. For that, commercial dialyzers with comparable surface area to our RenOx device (0.6 m²) were in-vitro tested with blood-to-dialysate flow ratios (1, 3, 6) in standardized conditions with full blood to realistically represent gas transfer during final RenOx application.

Materials and Methods

Blood Preparation

Gas transfer tests with fully oxygenated dialysis fluid were performed according to ISO 7199:2024²⁰ (blood-gas exchangers) and ISO 8637:2016²¹ (hemodialyzers). ISO 8637:2016 does not specify the use of blood for clearance tests, but ISO 7199:2024²⁰ requires the use of full blood for gas exchange efficiency tests of oxygenators. Therefore, porcine blood was selected due to regional availability and its similarity to human blood.

Porcine blood was collected at a local abattoir (Haaksbergen, the Netherlands), filtered with a nylon filter to remove slaughtering residues, and stored in a canister prepared with 15.000 IU/L heparin sodium solution (LEO Pharma, Lier, Belgium) to obtain full anticoagulation (activated clotting time ³1,000 s), 6 ml/L_blood of 0.9% saline solution, and 1.8 ml/L_blood of 50% glucose solution (Braun, Kronberg, Germany).

The canister was continuously stirred to avoid blood cells’ sedimentation. Saline solution at 0.9% was added to achieve the reference hemoglobin level of 12 ± 1 g/dl_blood as required by ISO 7199:2024.²⁰ Furthermore, blood was adjusted to obtain 0 ± 5 mmol/L base excess.²⁰

Dialysate Preparation

Dialysis fluid was 0.9% saline solution. Before inserting the dialysis fluid in the circuit, sodium bicarbonate solution at 8.4% (Fresenius Kabi, Enschede, the Netherlands) was added to achieve the same HCO₃⁻ concentration as previously reached in the blood. This was necessary to standardize the bicarbonate concentration before sampling to avoid an overestimation of the CO₂ clearance in the results.

Gas Transfer Tests With Decarboxylated Dialysis Fluid

The gas transfer test circuit was composed of a blood conditioning loop and a test loop arranged in parallel, as illustrated in Figure 2. Before blood, circuit was primed with saline solution and de-aired. The conditioning loop (Figure 2) consisted of a roller pump (HL20; Getinge Group, Gothenburg, Sweden), a reservoir, and a commercial oxygenator (both 8F INSPIRE; Sorin Group, Mirandola, Italy). This oxygenator with an integrated heat exchanger–maintained blood at 37 ± 1°C, eliminated O₂, and added CO₂ to maintain the standardized venous conditions outlined in ISO 7199:2024.²⁰ For this, mass flow controllers (Analyt-MTC, Müllheim, Germany) adjusted the flow rates of O₂, CO₂, and N₂ through the deoxygenator. A blood-to-gas flow ratio of 3 was maintained during tests. The conditioning circuit kept blood at standardized venous conditions throughout the experiment.

Figure 2.

Schematic (left) and photo (right) of the experimental circuit used for evaluating gas transfer efficiency between oxygenated dialysis fluid and blood. Standardized venous blood flowed in the conditioning loop (left) through a test D, to which the dialysis fluid fully oxygenated in the test loop (right) was supplied. P and F transducers are shown. Arrows indicate flow directions for blood (red) and dialysate (blue). D, device; F, flow; P, pressure.

The dialyzers (SmartFlux-HFP.06; Medica, Medolla, Italy, same batch) had a surface area of 0.6 m² comparable to the area of dialysis fibers in the RenOx (0.5 m²). The dialyzer membranes consisted of PUREMA H hollow fibers (Solventum, Wuppertal, Germany), see Table S2, Supplemental Digital Content, https://links.lww.com/ASAIO/B577. Blood passed through the fibers, whereas dialysate circulated around them in counter-current flow.

Blood-gas transfer rates were assessed in the test loop. In the test circuit (Figure 2), a second roller pump (HL20; Getinge) sent dialysis fluid from a small dialysate reservoir through an oxygenator (Quadrox-i Adult; Getinge), where pure oxygen was used as sweep gas to obtain fully oxygenated and decarboxylated dialysate, which was then pumped through the dialysate compartment of the dialyzer. The sweep oxygen flow was adjusted to be three times the dialysate flow rate.

Six different combinations of blood flow (Q_b) and dialysate flow (Q_d) rates were selected (Table 1). Q_b/Q_d ratios of 1, 3, 6 were selected to achieve expected CO₂ transfer below, within, and above the maximum required CO₂ exchange of 40 ml/L_blood.^22,23 In addition, blood flow rates within the maximum blood flow rate of 300 ml/min specified for the dialyzer (SmartFlux-HFP.06; Medica) were chosen. Thus, although higher blood flows are aimed for the RenOx system, selected blood-to-dialysate flow ratios of 1, 3, and 6 are comparable to the ones envisioned for clinical application.

Table 1.

Combinations of Blood (Q_b) and Dialysate Flow Rates (Q_d) (ml/min) Used for the Gas Transfer Tests With Their Respective Ratios

Qb (ml/min)	Qd (ml/min)	Qb/Qd
200	33	6
200	66	3
200	200	1
300	50	6
300	100	3
300	300	1

Blood flows were chosen according to the instructions for use of the dialyzer used for these tests.

Pure diffusion, thus zero net filtration, across the dialyzer was targeted.²⁴ For that, pressures on the dialysis side were controlled by an external, adjustable clamp. Flows and pressures were monitored using flow sensors (Transonic ME 6PXL; Transonic Systems, Ithaca, NY) and pressure sensors (Meritans-DXTPlus; Merit Medical Systems, South Jordan, UT).

The procedure was repeated twice for each flow rate ratio and for each of the five experiments, resulting in a total of 10 observations per flow rate permutation. The concentrations of O₂ and CO₂ were assessed using an i-STAT Alinity point-of-care blood analyzer with CG8+ cartridges (Abbott, Chicago, IL).

The concentration of O₂ and CO₂ transferred was computed as described previously¹¹ (see Tables S3 and S4, Supplemental Digital Content, https://links.lww.com/ASAIO/B577). CO₂ transfer accounted for the major contributions of bicarbonate ions (in mmol/L) and partial pressure of carbon dioxide (pCO₂) (in mm Hg) in blood.²²

Urea and creatinine content were monitored in whole blood and dialysate, although equilibrium was expected (see S5, Table S5, Supplemental Digital Content, https://links.lww.com/ASAIO/B577).

Statistical Analysis

SPSS Statistics 29 (IBM, Armonk, NY) was used for the statistical analysis. Experimental data illustrated a non-normal distribution. The Wilcoxon signed-rank test verified if blood flow rate (200 and 300 ml/min) significantly affected gas transfer. The Friedman test was used to determine significant differences between the three Q_b/Q_d ratios for both CO₂ and O₂ transfer. The Wilcoxon signed-rank test was applied to determine which Q_b/Q_d ratios resulted in significant differences. The test outcomes were adjusted with a Holm-Bonferroni correction. p values <0.05 were regarded as significant.

Results and Discussion

Results: Carbon Dioxide Removal

Figure 3 shows the amount of CO₂ removed by fully oxygenated and decarboxylated dialysis fluid at tested blood flow rates and blood/dialysis fluid flow ratios (Q_b/Q_d = 1, 3, 6). Results in Figure 3 are normalized in CO₂ removal per liter of blood to facilitate comparison between flow rates, and the metabolic requirement of 40 ml/L_blood of CO₂ elimination for full adult support (assuming a 0.8 respiratory quotient).²³

Figure 3.

Removed CO₂ in ml/L_blood as a function blood-to-dialysate flow ratios measured at blood flows of 200 and 300 ml/min. Boxplots show median (n = 16), and hinges show first and third quartile. p values after Holm-Bonferroni correction are shown. The dashed red line shows the metabolic requirement of CO₂ removal for an adult (40 ml/L_blood).²³ Statistically significant p-values are marked with *.

Blood flow rates of 200 and 300 ml/min did not result in statistically significant differences in CO₂ exchange (p > 0.1). Therefore, experimental data were grouped to evaluate the effect of the three tested Q_b/Q_d ratios on CO₂ elimination. Q_b/Q_d statistically significantly influenced CO₂ removal rate (Friedman test p = 0.0001, n = 16). Specifically, a Q_b/Q_d ratio of 1 led to statistically significantly higher CO₂ elimination than a Q_b/Q_d ratio of 6 (p = 0.001 after Holm-Bonferroni correction, n = 16) and a Q_b/Q_d ratio of 3 (p = 0.003). There was no statistical significance in CO₂ elimination between Q_b/Q_d of 3 and 6 (p = 0.07).

Discussion: Carbon Dioxide Removal

CO₂ elimination increased as dialysate flow increased, in other words, when Q_b/Q_d ratio decreased. This trend can be attributed to a reduced boundary layer thickness on the dialysate side and a higher convective transport of CO₂ at higher dialysate flow rate, improving overall mass transfer efficiency. A comparable trend was shown by May et al.¹⁸ and Cove et al.²⁵ Therefore, the recirculation of fully oxygenated dialysate across dialysis fibers can provide supplementary CO₂ clearance between 12 and 35 mL/L_blood, on average, almost matching the full 40 ml/L_blood metabolic requirement of CO₂ removal for an adult patient.²³ When required, higher CO₂ removal may be achieved with a lower Q_b/Q_d equal to 3 or 1.

Although this study did not directly test the RenOx device, results show that supplementary CO₂ clearance can be relevant to compensate for possible losses of CO₂ exchange caused by the replacement of gas exchange fibers by dialysis fiber in the RenOx. Supposing an adolescent patient of 50 kg requires solely lung support, the minimal amount of 40 $\mathrm{m}{\mathrm{l}}_{\mathrm{C}{\mathrm{O}}_2}$/L_blood needs to be eliminated,²³ thus, approximately 140 ml of CO₂ need to be removed from the total 3.5 L of patient’s blood volume.²⁶ Considering a maximum 10 $\mathrm{m}{\mathrm{l}}_{\mathrm{C}{\mathrm{O}}_2}$/L_blood is lost due to fiber replacement, the RenOx operating at a typical blood flow of 3.5 L/min combined with fully oxygenated dialysate at Q_b/Q_d equal to 6, could provide these necessary 10 $\mathrm{m}{\mathrm{l}}_{\mathrm{C}{\mathrm{O}}_2}$/L_blood to compensate for losses, so that combined, at least 40 $\mathrm{m}{\mathrm{l}}_{\mathrm{C}{\mathrm{O}}_2}$/L_blood (or 140 $\mathrm{m}{\mathrm{l}}_{\mathrm{C}{\mathrm{O}}_2}$) is eliminated.

With the current setup, our method provides limited respiratory dialysis potential, since our dialyzer (0.6 m²) could remove, in average, a maximum of 35 $\mathrm{m}{\mathrm{l}}_{\mathrm{C}{\mathrm{O}}_2}$/L_blood. Indeed, Figure 4 shows that our results are in the lower range compared with reported respiratory dialysis systems. This can be explained, since most literature used different dialysate compositions, a high blood flow to dialysis fiber area, and/or a single-pass dialysis system. Traditional ECCO₂R and respiratory dialysis operate below 1 L/min blood flow,¹⁹ therefore, target higher CO₂ transfer rates around 150 ml/L_blood (e.g., 45 $\mathrm{m}{\mathrm{l}}_{\mathrm{C}{\mathrm{O}}_2}$/min at 300 ml_blood/min in the PrismaLung device).²⁸ However, considering that our results ultimately target RenOx application, RenOx operates at higher blood flows than ECCO₂R and respiratory dialysis, targeting different patient groups,¹⁹ thus a smaller CO_2exchange/L_blood is sufficient to achieve the required metabolic support.

Figure 4.

Comparison between CO₂ removal rate (ml/L_blood) obtained in this study and previously reported respiratory dialysis systems.^18,25,27

Our method could be adapted to increase CO₂ removal with a larger dialyzer and a smaller Q_b/Q_d ratio. This is relevant considering that the recirculation of fully oxygenated dialysate presents advantages compared with previous methods. First, compared with single-pass studies, a closed loop reduces dialysate usage, costs, and prevents essential components such as minerals, hormones, and drugs to be removed by the dialysate.¹⁸ Second, compared with previous closed-loop systems, our method appears simpler, solely requiring a membrane lung in the dialysate circuit to drive CO₂ elimination. Moreover, the membrane lung is added to the dialysate circuit, avoiding additional artificial surface area in contact with blood and volume in the blood circuit, and thus hemodilution.

Gas transfer fluctuations might be attributed to operational constraints. Peristaltic pumps, used in the conditioning and test circuits, generate a pulsatile intermittent discharge. Therefore, variations in flow could affect our results, since samples are taken at a specific moment during the experiments. This is more pronounced at lower dialysate flow rate combinations, for which the measurement to error ratio is more significant. Similar pumps were used in our previous study¹¹ leading to comparable fluctuations.

Results: Oxygen Transfer

Different blood flows of 200 and 300 ml/min did not statistically significantly influence O₂ exchange, however, changes in Q_b/Q_d ratio lead to significant differences in oxygenation (Friedman test with p < 0.0001, n = 16). Pair-wise analysis demonstrated that a Q_b/Q_d ratio of 1 resulted in higher O₂ transfer rates than a Q_b/Q_d ratio of 6 (p = 0.001), also higher than a Q_b/Q_d ratio of 3 (p = 0.001), as presented in Figure 5. However, no statistically significant differences were found between the results for Q_b/Q_d ratio of 3 and 6 (p = 0.09).

Figure 5.

O₂ transfer (ml/L_blood) as a function of blood to dialysate flow rate ratios measured at blood flow of 200 and 300 mL/min. Boxplots show median (n = 16), and hinges show first and third quartile. p values <0.05 indicate statistically significant differences. Dashed red line represents the typical metabolic requirement of O₂ uptake for an adult (50 ml/L_blood).²³

For oxygen, the same general trend was observed as for CO₂, implying that gas exchange increased overall with lower Q_b/Q_d (Figure 5). Similarly, an increase in the dialysate flow reduces oxygen transfer limitations on the dialysate side and lead to a higher oxygen exchange. Oxygen is less soluble in the dialysis fluid than carbon dioxide,²² therefore, less mass of oxygen can be transported toward the blood, as a result, overall oxygen transfer was about 50% lower than CO₂ transfer in our system. Experimental data are available in Table S4, Supplemental Digital Content, https://links.lww.com/ASAIO/B577.

Discussion: Oxygen Transfer

Oxygenated dialysis fluid provided extra oxygen up to 15 ml/L_blood on average (Q_b/Q_d = 1). If similar oxygen exchange rates can be maintained at higher blood flows, dialysis fibers could provide up to 50 $\mathrm{m}{\mathrm{l}}_{{\mathrm{O}}_2}$/min in the possible RenOx application at a blood flow of 3.5 L/min, meeting ~30% of the metabolic requirement of a 50 kg patient. Therefore, combining the contribution of gas exchange fibers (50 $\mathrm{m}{\mathrm{l}}_{{\mathrm{O}}_2}$/L_blood)¹¹ and dialysis fibers repurposed for gas exchange (~15 $\mathrm{m}{\mathrm{l}}_{{\mathrm{O}}_2}$/L_blood), RenOx could theoretically supply nearly 130% of the oxygen respiratory requirement of a 50 kg patient. From a clinical perspective, supplementary oxygen delivered by dialysis fibers could be relevant to treat mild hypoxemia in individuals undergoing hemodialysis, or acute higher demands for oxygen, e.g., in septic ECMO patients.^29,30

In the view of respiratory dialysis, our study validates previous research by Tange et al.,^31,32 with comparable oxygen transfers to their baseline results.³² Nevertheless, compared with previous methods,³² the combination of a dialyzer with a membrane lung in the dialysate circuit appears to be a simpler alternative offering continuous gas exchange.

Monitoring of Other Relevant Blood Parameters

Blood pH and hemoglobin levels

Blood pH was maintained within the physiological range of 7.32–7.45 for venous and arterial blood,³³ indicating no pH imbalances. In addition, hemoglobin levels were maintained between 11.8 and 12.4 g/dl with no overall blood hemodilution or concentration (Table S6, Supplemental Digital Content, https://links.lww.com/ASAIO/B577). The equilibrium of blood pH suggests that the proposed methodology could overcome one of the main limitations of previous respiratory dialysis studies, reported by Cove et al.²⁵ However, future studies will show if a similar equilibrium of blood pH is achieved at higher blood and dialysate flow rates.

Study Limitations

Although this study shows as a proof of concept that recirculating fully oxygenated (and decarboxylated) dialysis fluid through dialysis fibers can be used for extrarespiratory support with the RenOx, our study has limitations. Blood volumes of a maximum of 300 ml/min could be tested in the current setup; thus, future studies should assess if similar gas exchange rates can be maintained at higher flow rates of blood and dialysate as planned for the RenOx. In addition, a membrane bundle combining gas exchange fibers and dialysis fibers needs to be used with blood flowing outside of dialysis fibers as intended in the RenOx, which could affect mass transfer. In a previous study, we could show that the transfer of small molecules was similar between inside-out and outside-in blood flow during dialysis,¹² therefore, we expect that gas transfer will also be similar between these two modes. However, tests with a RenOx prototype at desirable blood and dialysate conditions are still necessary to assess performance and risks of extra gas exchange in realistic clinical conditions and will be performed in a future study.

Conclusions

We assessed the feasibility to exchange CO₂ and O₂ by transporting fully oxygenated and decarboxylated dialysis fluid through a dialyzer in a closed circuit. This was assessed in standardized blood-gas transfer tests according to ISO 7199:2024. The ratio between blood flow and dialysate flow influenced both CO₂ and O₂ transfer with higher gas exchange capacity being linked to the lowest tested Q_b/Q_d ratio equal to 1. By controlling the Q_b/Q_d ratio, it would be feasible to adjust CO₂ removal rate from 12 to 35 ml/L_blood on average, nearly matching the full metabolic requirement of an adult patient (40 ml/L_blood). In addition, a maximum of 15 ml/L_blood oxygen was supplied. Supplementary gas exchange by dialysis fibers is valuable in the development of a combined lung and kidney assist system, first, to compensate for the combination of gas exchange and dialysis fibers in a single bundle, and second, to provide additional respiratory support, when necessary, especially for patients with acute CO₂ accumulation such as COPD patients.

Acknowledgments

The authors gratefully acknowledge the German Research Foundation (DFG) Priority Program SPP2014 “Towards an Implantable Lung” and their incentive “Women in Science Grant” for funding support.

Supplementary Material

mat-72-71-s001.pdf ^{(464KB, pdf)}