1. Introduction
MBO replace temporarily shunted lungs in cardiopulmonary bypass, being a mature technology intensively used worldwide during heart surgery. Yet, relatively low oxygen transfer rates to the erythrocytes (RBC) are still a shortcoming (Matsuda & Sakai, Reference Matsuda and Sakai2007), making of mass transfer enhancement a path for MBO technical/medical progresses (Lim et al., Reference Lim2006; Yeager & Roy, Reference Yeager and Roy2017). The CFL built up at the membrane/blood interface, typical of in vitro laminar blood flows (Popel & Johnson, Reference Popel and Johnson2005), is a major resistance to the oxygen transport to the erythrocytes. Although recognized that the blood-side resistance to oxygen transport is directly dependent on the blood flow characteristics in very narrow channels (characteristic lengths typical of microfluidics), little attention has been given to this subject, particularly on how interposition of disruptors at the membrane surface acts as potential reducer of CFL thickness (Lim et al., Reference Lim2006; Yeager & Roy, Reference Yeager and Roy2017), which will constitute the main focus of the present work.
2. Objective
Effects on CFL thickness of interposing hemicylindrical disruptors within dog or horse blood flows inside rectangular microchannels, surrogate of MBO mimicking their hemodynamics (Yeager & Roy, Reference Yeager and Roy2017), are quantified. Different disruptor geometries, as displayed in Figures 1 (c,d), are placed at one wall for cavity-flow, or alternately at opposing walls for zigzag-flow – see Figures 1 (a,b), with arrangements shown in Figure 1 (e).
Figure 1. Illustrative SEM images of some microchannels and sketch of the cavity and zigzag disruptors arrangements: (a) SEM image of cavity microchannel (
$ \times 30 $
); (b) SEM image of zigzag microchannel (
$ \times 35 $
); (c) SEM image of CFL disruptor with the largest penetration length (
$ \times 250 $
); (d) SEM image of CFL disruptor with the smallest penetration length (
$ \times 400 $
); (e) top view sketches of the used zigzag and cavity microchannel arrangements and corresponding dimensions (
$ {D}_0 $
– disruptor diameter;
$ W $
– microchannel width;
$ {L}_e $
– distance between disruptors;
$ {x}_{\mathrm{disr}} $
– disruptor penetration length. Note – Visual inspection of magnified images acquired both with an optical microscope (objective ×40) and SEM (up to ×500) allowed for the assessment of the good quality of the microchannels (walls smoothness and good definition and perpendicular level of the walls).
3. Methods
Microchannels were manufactured with biocompatible PDMS, frequently adopted for in vitro blood flow studies (Haeberle & Zengerle, Reference Haeberle and Zengerle2007; Ng et al., Reference Ng, Gitlin, Stroock and Whitesides2002), by photolithography (mould) and soft-lithography (microdevices).
Blood-flow images, acquired with a high-speed/high-resolution CMOS camera (Optronics-CR600x2) connected to a microscope (Sampaio et al., Reference Sampaio, Lopes and Semiao2015), were digitally post-processed with MATLAB techniques to accurately determine the CFL thickness: walls were located with histogram threshold (Kim et al., Reference Kim, Ong, Yalcin, Intaglietta and Johnson2009; Otsu, Reference Otsu1979) combined with Sobel/Prewitt filters; RBC/plasma interface identified with multi-threshold clustering (Sampaio et al., Reference Sampaio, Lopes and Semiao2015).
Calibrating the used syringe pump (blood-flow generator), Nexus-5000-Chemyx, yielded as maximum flowrate uncertainty 6.0%. Blood, kept at 4 °C, was heated in a thermostatic bath (25 °C) before experiments, and waved gently for 1 minute (distributing the RBC and preventing sedimentation). Differently aged dog/horse blood viscosity was measured in a rotary viscometer DV-II + Pro.
Microchannels geometry was characterized by SEM images (Figure 1), testifying their surface smoothness (Silva et al., Reference Silva, Leal and Semiao2009), perpendicularity and disruptors good definition. Several similar images allowed comparing the actual height and width of all used microchannels with the designed ones (maximum error: 4%). Key parameters are displayed in Table 1.
Table 1. Actual height and width (assessed from SEM images) of the manufactured microchannels. Dimensions are defined in Figure 1 (e). R: reference microchannel (no disruptors); Ci: cavity-flow type; Zi: zigzag-flow type.
Note – design size values were: height = 40 μm; width = 400 μm.
4. Results and discussion
Different aged blood viscosity results, horse (hematocrit,
$ {H}_t\approx $
40%) and dog (
$ {H}_t\approx $
46%), displayed in Figure 2, evidence its negligible variation with blood age. Therefore, experiments performed up to 1 week after blood collection appear not to affect CFL results.
Figure 2. Effect of aging on the viscosity of blood of different animals [⋄ 3 hours after collecting blood; □ 10 hours after collecting blood; △ 24 hours after collecting blood;
$ \times $
48 hours after collecting blood;
$ \ast $
3 days after collecting blood; ∘ 7 days after collecting blood;
$ + $
15 days after collecting blood]: (a) horse blood (
$ {H}_t\approx $
40%, at 25 °C); (b) dog blood (
$ {H}_t\approx $
46%, at 25 °C).
Figures 3 (a-d) show CFL thicknesses for horse and dog blood flows for different flowrates (microchannels R, C1 and Z1). Figures 3 (a,b) evidence the systematic and more pronounced CFL thickness increase with flowrate in microchannel R. Disruptors interposition reduces always the CFL thickness (ca. 20%), which is desirable for MBO. However, comparing dog and horse blood flows, CFL thickness is ca. 6% smaller for the former: its larger hematocrit renders smaller the space available at the central region for RBC to migrate. Figures 3 (c,d) evidence the spatial periodic character of blood flows after some disruptors for the zigzag arrangement: CFL thickness keeps practically invariable between the 13th and 14th disruptors. This might suggest the zigzag arrangement as preferable.
Figure 3. Thickness of CFL as function of the volumetric flowrate and blockage ratio (non-dimensional degree of the disruptor penetration into the flow). (a,b) Horse and dog blood flows [□ microchannel R (free of CFL disruptors); △ zigzag-flow type microchannel Z1; ▲ cavity-flow type microchannel C1]; CFL measured at the wall region opposing the 1st disruptor in microchannels Z1 and C1; (c,d) CFL thickness after some disruptors for horse and dog blood flows in microchannel Z1; CFL measured [□ right above 13th disruptor; ■ right above 14th disruptor; △ wall opposite to the 13th disruptor; ▲ wall opposite to the 14th disruptor]; (e) Thickness of CFL at the wall opposing the disruptor as function of the blockage ratio for horse blood flows in zigzag-flow type microchannels Z1, Z4 and Z5 [⋄ 10
$ \mu \mathrm{l}/\min $
, 1st disruptor;
10
$ \mu \mathrm{l}/\min $
, 2nd disruptor;
10
$ \mu \mathrm{l}/\min $
, 14th disruptor; △ 25
$ \mu \mathrm{l}/\min $
, 1st disruptor; ▲ 25
$ \mu \mathrm{l}/\min $
, 2nd disruptor; ▲ 25
$ \mu \mathrm{l}/\min $
, 14th disruptor]; (f) Thickness of CFL at the wall opposing the disruptor as function of the blockage ratio for dog blood flows in cavity-flow type microchannels C1, C4 and C5 [∘ 10
$ \mu \mathrm{l}/\min $
, 1st disruptor; • 10
$ \mu \mathrm{l}/\min $
, 5th disruptor; □ 25
$ \mu \mathrm{l}/\min $
, 1st disruptor; ■ 25
$ \mu \mathrm{l}/\min $
, 5th disruptor].
Note – For the cavity arrangements, CFL exhibited a non-linear increase of its thickness with the flowrate at the downstream disruptors region.
CFL variation with the diameter was inconclusive (microchannels C1, C2, and C3) or incipient (microchannels Z1, Z2 and Z3). Moreover, CFL thickness kept unchanged with disruptors positioning (microchannels C1, C6, C7, Z1, Z6 and Z7).
Figures 3 (e,f) reveal a systematic CFL thickness reduction with the blockage ratio increase (microchannels C1, C4, C5, Z1, Z4 and Z5): between 10–20% for dog blood in cavity-type flows; and 10–15% for horse blood in zigzag-type flows.
Images in Figure 4 suggest a limit for the blockage ratio: when
$ {x}_{\mathrm{disr}} $
is greater than
$ {D}_0/2 $
, a CFL localized at the disruptor/wall edge, with low velocities and large residence times, may potentiate clot formation (Completo et al., Reference Completo, Geraldes and Semiao2014), which is not desirable for MBO: erythrocytes inside such CFL require much more time (ca.
$ \times $
6.5) to flow away from it than those outside it.
Figure 4. Illustration, for a blood flow around a disruptor with its penetration length into the flow
$ {x}_{\mathrm{disr}} $
larger than its radius
$ {D}_0/2 $
, of the large residence time of a RBC inside the CFL upstream and close to the disruptor edge (highlighted with a circle marker) in comparison with that of another RBC at the plasma/erythrocytes interface (highlighted with a square marker). Flow images at that region for different instants: (a) t = 0 s; (b) t = 0.01 s; (c) t = 0.02 s; (d) t = 0.03 s; (e) t = 0.04 s; (f) t = 0.26 s.
Note – These images, extracted from hundreds of the sequential images acquired with a frame rate of 2000 fps (0.5 ms between consecutive frames), exhibit a partial top view of a disruptor with the flow coming from the top to the bottom of each image. There was no experimental evidence of a similar phenomenon for blood flow around disruptors with penetration lengths
$ {x}_{\mathrm{disr}} $
smaller than or equal to their radius
$ {D}_0/2 $
.
6. Conclusion
Interposition of disruptors in animal blood microchannel flows revealed effective in reducing CFL thickness (ca. 20%), suggesting zigzag arrangement as slightly better, a feature desirable for MBO as it promotes larger oxygen transport rates to the erythrocytes. Variation of the disruptors diameter and positioning yielded no relevant CFL thickness improvements. Blockage ratio (non-dimensional measure of the disruptor penetration into the flow) proved to be quite effective in reducing CFL thickness (ca. 10–20%). However, this comes at the cost of risking clot formation (undesirable for MBO) for disruptors with a penetration length greater than its radius.
Acknowledgments
The authors are deeply grateful to Dr. Mario Velhinho from ProBiologica, Portugal, who gently provided the horse blood, and to Dr. Belmira Carrapiço from Faculdade de Veterinária of the University of Lisbon, who gently provided the dog blood. Also, Eng. Isabel Nogueira from MicroLab is acknowledged for her support in acquiring SEM images.
Author contributions
Prof. Viriato Semiao and Prof. Goncalo Silva conceived and designed the study. The MSc students Duarte Dias and Duarte Sampaio manufactured the microchannels, conducted experiments and data gathering and gave support to data analysis. Prof. Viriato Semiao, with support of Prof. Goncalo Silva, wrote the article.
Funding information
This work was supported by FCT – Fundação para a Ciência e Tecnologia, Portugal (through IDMEC, under LAETA, project UIDB/50022/2020, IDMEC, Instituto Superior Técnico, Universidade de Lisboa, Lisboa, Portugal), (G.S. grant No. SFRH/BPD/111228/2015), (V.S. Grant No. PTDC/CTM-BIO/6178/2014).
Data availability statement
Readers can contact the authors if they want access to such materials.
Conflict of interest
The authors have no conflicts of interest to declare.
Supplementary Materials
To view supplementary material for this article, please visit http://dx.doi.org/10.1017/exp.2020.60.
Open access
Comments
Comments to the Author: The manuscript addresses a very important and timely topic. The presented experimental results are very interesting and will be valuable for further work on the improvement of mass transfer in blood oxygenators.
Specific comments:
Introduction:
• First sentence: use ‘temporarily’ instead of ‘interim’.
• Line 7: not sure ‘fluid mechanics’ is the right expression here, as this basically refers to the science. Maybe replace by ‘flow characteristics of the blood flow...’, or similar
Methods:
• Please explain how exactly the thickness of the CFL layer was determined.
• Last sentence of methods: was the actual height taken at one specific location or at more points? How big is the variation within one channel? The error is meant between set and real height/width?
• Was wall smoothness measured? Or how do you testify from the SEM images?
Results and Discussion:
• Last sentence on page 2: the sentence is not very clear, please re-formulate.
• Figure 4: How were the locations of individual RBCs and their residence times determined in the flow images? Please explain.
• Page 5, last sentence of first paragraph: have the same visualizations as in Fig. 4 been made for other penetration lengths or how do the authors predict that the risk of clot formation only occurs for penetration length greater than radius? Please explain.