Monday, 8 December 2014

10 billion frames per second videos of single pulses of light

There has been quite some buzz about doing ultrafast photography. This involves using ultra short pulses of light to illuminate a scene and take images at billions of frames per second. However there is always a trick and in this case the pulse of light is not the same pulse of light throughout the movie. What you do is repeat the measurement and delay the opening of the shutter on the camera. This is the same way we take ultra-fast spectrum of molecules we rely on the repeatability of the measurement and change when we observe the molecules after firstly exciting them. Here is a video made up of hundred of different pulses that when you put them altogether you can see how a pulse of light hits an apple.
An exciting new paper by Liang Gao et al. in nature shows a way to take a movie of a single pulse of light. This opens up the ability to see non-repeatable events such as nuclear explosions, optical rogue waves or gravitational waves. These events happen on the femtosecond (10-15s) timescale that are one off events.

Before we get into the nuts and bolts of how it works here are the amazing movies they captured of a single pulse of light.

Light hitting a mirror will be reflected at the same angle it approaches the mirror. Something else you can see if the evanescent field (light tunnels through the material) which decays away exponentially over time. We can use this to perform spectroscopy on a surface.
Light will travel slower in a lower refractive index material here is a comparison of air and resin where the light in the resin is travelling slower.
Refraction occurs at the surface of a lower refractive index material and a higher refractive index material.
This is my favourite, Seeing the glow of molecules after they have been excited. They capture a phenomenon called fluorescence where light of a higher energy (green in this case) is absorbed by the molecule (Rhodamine) the excited molecules loses some of that energy to vibrations and then drops down to the ground state emitting lower energy light (red).

Here are all of the different experiments.

So how does it work. Here is my basic explanation feel free to correct me in the comments. Firstly an ultrafast laser pulse is needed these are generated using special laser cavities that bounce a laser pulse inside a cavity hundreds of times each time they destructively interfere inbetween pulses and constructively interfere only for a few femtoseconds - if you have ever made a diffraction pattern in on a wall (spatially) think of this as a diffraction pattern in time (temporally). The camera used is quite similar to the old CRT TVs but in reverse instead of electrons being scanned over a phosphorous screen that converts the electrons to light. Light hits a phosphorous screen where the photons of light are converted to electrons these are accelerated towards a camera sensor like in a cellphone and an image is created. However you apply a voltage ramp across the path of the electrons so they are deflected depending on when they arrive (you reduce the voltage as a function of time). This means electrons from light that arrived earlier will be deflected more by the electric field than electrons from light that arrives near the end. You can think of this as sophisticated light painting where you get a blur of colour by keeping the shutter open on a camera.
Here is a schematic of the optics used.

Now how do you extract out a video from a blured image on the image sensor. You need some sort of pattern that will be the same at every time step. The way they do this is using a coded mask that randomly patterns the image of the object. You can see this in the diagram below.

Using matrix methods on the computer you can reconstruct a whole video from a single blurred image. What do you want to observe at 10 billion frames per second?

Gao, L.; Liang, J.; Li, C.; Wang, L. V. Nature 2014, 516, 74–77.

Sunday, 30 November 2014

Low-cost Laser Cut Colorimeter

There are many low cost spectrometers on the internet that make use of compact discs and low cost CCD detectors from cell phones or webcams to determine the different colours that are emitted or absorbed by objects. These spectrometers are great for many applications where you want to look at the relative intensity of the different colours such as what colours are absorbed by a molecule or looking at whether the light is spread out or whether only specific colours are emitted as is the case with LEDs and fluorescent lights. These spectrometers have one large flaw - auto-exposure. Autoexposure means that the time that the CCD takes to collect light from the sensor is dynamically adjusted. Some expensive cameras allow this to be adjusted however most cell phones and web cams do not. This means that comparisons between different spectra intensities are difficult and getting reliable measurements of intensity out of these instruments is near impossible.

This is where the colorimeter comes in. By using only a single coloured LED and a light dependent resistor a low cost reliable measurement can be made of how a substance absorbs light. There are a large number of colour based test kits for nitrates, phosphates, pH and nitrites that could use this colorimeter for environmental monitoring. Other applications include measuring yeast growth rates through a turbidity measurement and if an LED is used as the detector a fluorescence measurement could be done which may be of interest in the environmental monitoring of oil in the environment.

You can buy educational colorimeters for 147 USD ( 144USD (, a nice open source model is made by IORodeo for $85USD and commerical systems are even more expensive.

Instead of a complete computer control system the whole idea of this project was to strip the colorimeter down to the least number of components at the lowest cost possible to get it out to schools, citizen scientists and researchers. There have been many colorimeters build recently that make use of 3D printed parts but due to the time needed to print we opted for laser cutting which is cheap and quick to fabricate. To avoid the need for a box which would involve more materials and assembly, a layering of laser cut sheets was used. Borrowing from the royal society colorimeter we make use of a multimeter which can read the resistance of a light dependent resistor the CdSe or CdS LDRs are cheap and can detect similar colours to our eyes (however they are not very good in the IR). This is the cheapest option low cost multimeters are ~$5 and many schools already have multimeters. Using a button battery (CR2032) allows the device to be portable and also current limits the LED so there is no need for a resistor in series with the LED. However as the battery's energy is used the system will need to be recalibrated as the voltage will drop. Most CR2032 button battery have around 225mAh, an LED drawing ~16-18mA should allow at worst 12 hours of operation.

The three layers made of 3.5mm acrylic was laser cut using a 30W CO2 laser. The top layer is the shielding layer which holds the battery in place, the middle layer contains the light dependent resistor on the left and the LED on the right. Any 5mm LED will work, the LED is connected to a button battery and the final layer clamps the other side of the LED legs onto the battery.

The pins for the multimeter plug into the holes where the LDR pins have been pulled through this forms a good tight connection meaning no soldering needed. A clip is used to stabilise it and keep the legs of the LED connected to the battery.

The cost of the different components not including the multimeter came to under $2NZD. I will do one more check on the laser cut designs and upload them to thingiverse for anyone to use in the next few days.

Nitrate colorimetric test

Excess nitrates in waterways leads to algal blooms and can kill wildlife. Monitoring streams and waterways is therefore an important target. To begin with a 3D printed spectrometer based on the design was used to look at what colours are absorbed by the API nitrate test kit dye (a ~$10USD test kit that can do a 100 tests).

3D printed spectrometer with the iphone LED flashlight as the light source. Another cell phone was used to collect the spectra.

From left to right; the LED from the iphone contains blue, green and red light, the dye without any nitrate added shows an absorbance in the blue, the solution appears yellow so this makes sense, the dye with the dye and nitrate show the absorption of the green light. The line plots were done in ImageJ using the profile plot.

Standards were then prepared of zinc nitrate in a range which is common in fresh water 50ppm, 100ppm 150ppm and 200ppm. 

The API test kit contains two solutions that need to be added and then a colour develops over time. The colour becomes more red as the green is absorbed by the dye-nitrate complex.

Colour card that came with the API nitrate test kit showing the range of colours.

Human error in reading the card makes the test not very accurate and the difference between what different people deem to be a particular colour varies. By measuring using a colorimeter you can measure much smaller changes in nitrates and remove the variability introduced through the colour card.

Colorimeter with the blue LED we later used a green LED as well.

A test tube covered with tape was used to block the ambient light from the room. You could also use a toilet roll or a film canister.

Taking the logarithm of the resistance (which is a measure of the absorbance) should give a value proportional to concentration. Using the blue LED (wavelength of 495nm) we get a linear relationship between the nitrate concentration and log(R).

Next we tried a green LED as we saw that green was being removed when the nitrate was being added. The sensor showed a non-linear response with absorbance but a linear response in intensity.

The bottom plot shows the nitrate concentration vs. the resistance which is linear this indicates the non-linearity is logarithmic and most likely due to the sensor. To work out if the non-linearity was the sensor not responding linearly or the test we used a spectrometer that we could control the exposure time and get a good reading of the absorbance.

Absorbance vs. wavelength for the different solutions you can see that as nitrate is added the dye starts absorbing in the green part of the spectrum.

Plotting the peak absorbance at 550nm as it changes with the nitrate concentration a linear relationship is found and therefore test is linear and the sensor is non-linear for the intensities it was dealing with.It could also be the multimeter which may not have the best electronics (highest impedance opamp).

This is OK as you can fit a polynomial or even better a logarithmic fucntion (as the error is probably logarithmic) and use that to determine the nitrate concentration from an unknown by plugging the absorbance into the equation.

The next step is to see if quantitative fluorescent measurements could be done by using another LED as a detector ( This could allow for measurement of the oil in water for example. If you choose an LED with a green or red emission they have build in filters so a combination of a blue (405nm) LED and a green LED as a detector could be used to determine quantitatively the amount of oil in water. It looks like devices already exist that use LEDs to do this measurement. This may interest the publiclab community as the characterisation of the type of oil could be done with a low cost spectrometer and the concentration measured using the colorimeter. (

This work was done as part of preparation for the year of light 2014 with Sandra Jackson the teacher fellow currently in the Photon Factory.

Sandy preparing the standard solutions of Zinc nitrate.

Here are some links to interesting pages on colorimeters that the Photon Factory colorimeter borrowed from.

Breadboard with LED powered by 9V battery and LDR connected to a multimeter.

First design that caught my attention with a LED and LDR on a pcb.

Nitrate and phosphate detection using LED and LDR that included tuning of the LDR circuit.

IORodeo low cost colorimeter with great posts on doing nitrate measurements and integration with an Arduino and software.

Good open source review of using LEDs as sensors for fluorescence and for absorbance measurements

Article using LEDs as detectors (for fluorescence)

Laser cut LDR LED box with a multimeter as the measuring device

Michigan tech colorimeter using an Arduino and a 3D printed case.

This was all integrated into an opensource colorimeter for water quality measurements.
(Article on water quality (unfortunately not open source)
OS water platform.JPG

Wednesday, 5 November 2014

Decapentaene pi orbitals - Particle in a box

While tutoring in physical chemistry this year I wanted to show the students how simple models like the particle in a box can come quite close to describing the electron motion in real molecules. So I used molcalc a great tool that lets you build molecules and then run calculations in Gamess and display the orbitals in the browser. I plotted the solutions in python for comparison.

Tuesday, 4 November 2014

Shrinking down chemical analysis - microTAS 2014

Just got back from microTAS conference 2014 (micro total analysis systems). The conference was all about miniaturised chemical detectors. A lot of what was shown was lab-on-a-chip devices which makes use of semiconductor chip technology to fabricate channels in plastics to automate lab work.

One of the questions I often get when I show off microfluidic chips is "What's so amazing about that?". Most people don't have a macroscale reference to see how amazing fluid moves in small channels. So the day before we had to leave for San Antonio we decided to do an experiment for the video competition (very quickly!!) where we went from a macrofluidic mixer to a microfluidic mixer. We ended up placing in the top 3 videos at microTAS the only educational video.

Just to give a taster as to what was at microTAS here are some of the amazing devices that were being shown off.

Digital microfluidics
A.H.C. Ng gave a presentation on a digital microfluidic system from the wheeler lab from the University of Toronto that automated the synthesis of gold nanoparticles with DNA bound only on one side of the nanoparticle which can then self assemble into structures.

I was surprised to find out that the digital microfluidics platform they developed is open source and is available online. All is needed is a high voltage power supply and some pcbs and an arduino. They also got a video into the top three videos at microTAS which shows off their open source digital microfluidic device Dropbot.

Droplet to digital microfluidics
Steve C. C. Shih presented a talk on his new droplet to digital system from the Joint Bioenergy Institute USA. A droplet maker was used to put single yeast cells into each drop. These were then transferred over to a digital microfluidic device that mixed the droplets with an ionic liquid. The idea was to see how ionic liquids damage the yeast. As ionic liquids can be used to help convert cellulose into sugars for biofuel production. However the ionic liquids can damage the yeast cells which are meant to eat the sugars.

Graphical abstract: A droplet-to-digital (D2D) microfluidic device for single cell assays

They designed and made the whole system using arduino's. The device really leveraged the best out of both technologies and has been published in lab on a chip last week

 image file: c4lc00794h-f3.tif

Inertial microfluidics
I went to a workshop run by Assoc. Prof. Di Carlo on inertial microfluidics. Two main forces dictate if a microfluidic flow will be laminar (Steady and predictable) or turbulent (unsteady and difficult to model); inertial forces, which relates to the momentum of the fluid and viscous forces, where the fluid resists being sheared. Often when modelling microfluidics we ignore inertia and arrive at Stokes flow. However a recent review by Assoc. Prof. Di Carlo looked at including these effects and the different phenomenon that can be observed.

You are also able to arrange particles in a flow due to the way the particles deform the flow. This deformation is felt by other particles and leads to nicely ordered particles in channels.

Di Carlo Inertial Microfluidics
You can use this to put individual particles or cells into single droplets which is super useful for doing lots of cell based analysis.

Dino Di Carlo
It can also be used to align particles along a channel.

As fluid moves around a corner or around a barrier you get vortices set up which allow you to more easily mix solutions. Which as I showed with the mixer at the beginning of this post is quite difficult.

 uFlow was demonstrated a program written in python that allows for calculation of the flow pattern at the outlet. If you were to do this with a finite element modelling program it would take hours to run each time. To reduce this time they calculated advection maps (transport map) for pillars at different places in the channel and operated on the concentration profile by these maps to make the simulation run in real time. This was a really interesting way to speed up simulations in microfluidics and I am thinking about what else could be simulated in this way.

Wednesday, 3 September 2014

Nanotech wedding ring

New Zealand is known for the Lord of the Rings trilogy and so I took the opportunity recently when I got married to make the precious. But I couldn't resist including some nanotechnology in the design.

Shane Hartdegen a great teacher and designer of jewellery made both my wife and my wedding rings and I wanted to explain the design as we had a lot of fun with the materials. There are four concentric rings of different metals.

From the outside in there is electrochemically recycled gold, followed by white gold which was made from forged gold and palladium, then a copper layer giving a nice orange colour and inside is titanium which would normally be grey in appearance but by heating it in the flame a very thin layer of titanium oxide is formed that allows light to interfere and produce this blue-purple colour.

I was intrigued about how a grey metal could produce so many amazing colours. It all comes down to a small layer of oxide that is formed when you heat treat the metal. This nanocoating is on the order of the wavelength of light.

Figure of a 50 nm in (a) and (b), 550 nm in (c) and (d), 750 nm in (e) and (f), layer of Titanium dioxide grown on a layer of indium tin oxide a conductive film. The layer was grown with atomic layer deposition and is as close as I could find to oxide growth on Titanium metal however you would probably not get such a porous structure. 

What optical effect causes some colours to be reflected (such as blue and purple in this case) but not other colours such as yellow. The phenomenon responsible is called interference and occurs as the light bouncing off the top of the oxide layer and off the oxide metal interface interfere. Only certain wavelengths will constructively interfere and be reflected.

To work out what wavelengths will be reflected you can use similar formulas to Bragg's law which is used for measuring the thickness of layer of crystals, however that require much shorter wavelength x-rays.

This effects allows us to determine the thickness of the layer by determining the colour of the film (as has been done by harry sherwood on his blog)

A more common way to deposit the oxide layer is by using electricity through a process called electrolysis where titanium ions in the solution are neutralised by the electrons given from the anode allowing for the depositing of the metal. This method allows for a wide range of colours to be created.

Electrolysis was the method used to recycle the gold used in the ring. Most gold is made through dissolving rocks in cyanide solutions. These are harmful for the environment. We decided to recycle gold from old jewelery. First the jewelery was dissolved in an acid solution followed by electrolysis onto the anode and finally smelting it to form the metals needed.

Sunday, 15 June 2014

Onehunga high school visit to the Photon Factory

Here are some notes for the students coming from Onehunga High School to take part in practical science projects. Below is a list of the different projects.

During the course of the day we are going to be drawing up some slides about the project you have chosen including:
  • Background research
  • Design of the device
  • Building the device and testing
  • The questions will you ask with what you have made
Clean drinking water using natural microorganisms
3.4 million people die every year from water-related diseases. This is largely due to pathogens living and breeding in the water. There are many different ways to clean water of microorganisms; filters and solar precipitators to name just a couple. However, there is a way to filter out microorganisms using other microorganisms. Using a biosand filter (a bucket full of sand) that slowly empties, water can be sanitised simply and naturally. A thick biofilm forms on the surface of the sand and breaks down bacteria. Then the sand allows for filtering of water and provides a dark place where the bacteria die from lack of food. How the biofilm and each different element work is still not fully understood. We will make a biosand filter and look at what experiments can be done to improve our understanding of the biosand filter.

Timelapse images for long term changes in nature
The climate is changing due to an increase in greenhouse gases in the atmosphere. This rapid change is going to have a huge impact on the environment around us, particularly on plants which are dependent on the weather cycles to survive. Due to the slow movement of plants it is hard to look at trends in their behaviour. Using special time lapse photography we can watch the long term movements of plants using a simple webcam and computer software. By stitching the images together  using software a video can be made and analysed. We will be building an enclosure for a webcam to be set up outside to collect images of plants growth. We will also look into using algorithms to exaggerate the movement of plants to observe subtle changes in the plants as they grow.
Revealing Invisible Changes In The World - YouTube

3d printed spectrometer using a CD grating and a cell phone
Have you ever wondered why something appears blue or red? What determines a material's colour? A spectrometer is a device that can measure the light that comes off an object. We are going to make one of these using a normal CD which has lots of small pits in its surface which allow it to reflect light of different colours in only one direction. This means that if white light enters in at a particular angle, the colours spread out at different angles and can be measured individually. These have been used to look at the different chemicals are in stars and on planets. They are also used to detect contaminants in water. In the Photon Factory we use this to look at how energy moves around molecules and we are working on making artificial leaves that turn sunlight into energy. You will build your own spectrometer and attach it to a cellphone to find out how you can help the world by conducting experiments with your own spectrometer.

Invisible light camera reveals when plants are stressed
We can see from blue light (~400 nanometers) to red light (~600 nanometers). Outside this range we are blind, however continuing past the red end of our vision will show signatures in plants that would otherwise be invisible. Plants absorb red and blue light but reflect green light. But what about the invisible light? Plants reflect near infrared light (NIR) (~700nm - ~1200nm) very well also. However, when they are under stress the part of the plant that reflects the NIR becomes damaged and the plant reduces its reflectance. This can happen when there is a pest invasion or when the plant is not getting the right nutrients or water. Using a simple webcam we will remove the infrared filter from the camera to see the invisible light, then add a red filter. This replaces the red channel with the infrared to show stress in the plants.

Biochar and efficient stoves for safer cooking conditions and carbon sequestration
The majority of the world still uses open fires for cooking food. This presents a health risk due to smoke inhalation and also leads to pollutants in the environment. Using a different way of burning wood, a highly efficient stove (gasifier) can be built that firstly turns the wood into a gas and then burns that gas in another part of the burner. This gasifier can then burn the fuel more efficiently with fewer harmful emissions. One of the byproducts from this process is charcoal, a solid form of carbon, that is very difficult to break down. When wood grows it uses and stores carbon dioxide, a greenhouse gas, and when it dies microbes break down the wood and the carbon dioxide is released. By heat-treating the wood carbon can be trapped in the ground for thousands of years. The application of charcoal to the soil (this type of charcoal is known as biochar) was first done with soils along the Amazon. Ancient Amazonians would put food waste and charcoal together to produce a very potent fertiliser. We will build a gasifier out of a can and a computer fan, then design some experiments to test out the effects of the charcoal on different plants and soils.

This is just the beginning. If you want to continue doing the project you were assigned or another group's, then you can enter the high school science fair.

Friday, 10 January 2014

Floating raisins and sad shellfish

The floating raisin is a classic Christmas kitchen experiment. A raisin is dropped into a carbonated beverage (often champagne) and after a minute or so the raisin rises to the top of the drink and then sinks back to the bottom to repeat the process. I posted a video on Facebook of the experiment and had a few people ask me how it works.

Why is champagne bubbly? The bubbles in champagne and other carbonated drinks are bubbles of carbon dioxide gas (CO2).
These bubbles of CO2 are formed by tiny yeast microbes during a second fermentation step in producing champagne. During this step, the yeast eats the sugar breaking it down into carbon dioxide.

Why don't we see bubbles of CO2 when champagne is corked? This is due to a dynamic equilibrium inside a closed system - dynamic in that chemicals are always moving from one state to another (in this case carbon dioxide in solution and carbon dioxide gas) and an equilibrium in that the rate of change going from the liquid to the gas phase is the same as going from the gas into the liquid. No bubbles form, as that would indicate more carbon dioxide moving from the liquid to the gas and the system would be out of equilibrium.
Another time you may have come across this is when you leave your drink bottle in the car on a hot day with a little bit of water in the bottom. When you come back to open the bottle you hear a release of gas when the lid is opened. This is due to the water in the gas phase which is in equilibrium with liquid water until you open the bottle. We call this vapour pressure in chemistry. Below is a video of water at the gas-liquid interface.

So we know the bubbles come from CO2 escaping due to a non-equilibrium open system.  

Considering that CO2 is usually thought of as a gas (at room temperature) how is it stable in water? Hydrogen bonding holds the answer to CO2 stability in water. In this case, oxygen is electronegative (has higher negativity) than hydrogen, which is comparatively more positive. When oxygen comes near the small positive hydrogen, the positive and negative attract to form a very strong intermolecular (between molecules) bond. This stabilises the CO2 in water, stopping it from grouping together and forming a bubble.

The bonding H2O around CO2 [1], [2]
However, it is important to keep in mind that due to thermal energy (temperature) there is always change at a molecular level. This perfectly bonded sphere is the representation of an average of these bonds, which are themselves constantly changing.

When the cork is popped the pressure at the top is released and so there is no gas pushing on the liquid. Without this pressure, the gas at room temperature will push the water molecules apart and form a bubble.
Above you can see a bubble of CO2 with water on either side. However, water is quite heavy and will push on this bubble until the CO2 goes back into the water and no bubbles form in the liquid. [4]

If no bubbles can form in a liquid how do they form in champagne? This is due to microscopic dirt and scratches on the inside of the glass on which the bubbles can form. Only on a solid, where CO2 forms weak bonds with the surface, can enough CO2 molecules get together to form a bubble. When this bubble reaches about 1 micron (a hair being about 100 microns in diameter) it is large enough to continue to grow by itself in the liquid. [5-7]
Above is a bubble formed on a cotton fibre. It can be seen to form, grow and then detach and continue to grow as it travels through the liquid. Champagne bottles and glasses can be scratched using lasers or mechanically to tune just how quickly the champagne will stay bubbly. This video explains it with some simple experiments.

Coming back to the raisin in the champagne, it is easy to see how bubbles can form on its wrinkled surface.

Why do the bubbles grow larger than a micron on a raisin and actually stick to it? In short, the bubble of CO2 likes the sugary surface of the raisin more than it wants to float. As a result, the raisin is pulled up to the top of the glass due to the buoyancy of the bubbles. Once at the surface there is no liquid to hold the gas in bubbles so it escapes into the air and the raisin sinks again.

Super-heated water
A dangerous example of the need for impurities on which bubbles can form is super-heated water. (Don't try this at home - after this I have an experiment you can try at home.) Pure water (with a complete lack of impurities) can be heated above boiling in the microwave for a few minutes. When it comes in contact with a spoon (as my father found out while flatting to disastrous consequences) or sugar (as the video below shows) it explodes as gas is released.

Super-cooled water
Instead of this rather dangerous experiment, what you can try at home is super-cooling water. If you put purified water into the freezer it will cool below freezing until either the introduction of an impurity or a shockwave (for example, hitting the bottle) starts the ice crystal formation.

The really important research that is going on in this area is to do with CO2 storage. Due to the blanket of CO2 around the earth, our atmosphere is heating up. Something slowing down this increase in CO2 are the oceans, which are absorbing excess CO2. However, this presents two long-term problems - firstly, as the earth heats up further less CO2 can be stored in the oceans, so this is a finite solution, and secondly, it acidifies the ocean making it very difficult for shellfish to make their shells.

Sebastian is sad because his shell is slowly dissolving in the ocean's acidity.

The picture below shows pteropod shells that have already started dissolving in the Southern Ocean.

A shell placed in seawater with increased acidity slowly dissolves over 45 days.<div class='credit'><strong>Credit:</strong> A shell placed in seawater with increased acidity slowly dissolves over 45 days.</div>
A more scientific representation of a sad shellfish.

Shellfish use calcium carbonate to form their shells. The shell is made out of calcium with the carbonate ions helping to make sure the calcium isn't dissolved in water. But by flooding the oceans with CO2 we are removing carbonate and forming bicarbonate, therefore removing the molecules needed for shells to not dissolve.

This is why the study of CO2 in water and also trapping CO2 by adsorbing them onto solids is of great importance to the earth and why the chemistry of floating raisins is much more interesting and important than you could have imagined.

[1] H. Sato, N. Matubayasi, M. Nakahara and F. Hirata, Which carbon oxide is more soluble? Ab initio study on carbon monoxide and dioxide in aqueous solution, Chem. Phys. Lett. 323 (2000) 257 - 262.

[2] G. K. Anderson, Enthalpy of dissociation and hydration number of carbon dioxide hydrate from the Clapeyron equation, J. Chem. Thermodynamics 35 (2003) 1171-1183.

[3] G.A. Gallet, F. Pietrucci, W. Andreoni, J. Chem. Theory Comput. 8, 4029-4039 (2012)

[4] W. L. Ryan et al., J. Coll. Interf. Sci. 1993, 157, 312. DOI: 10.1006/jcis.1993.1191

[5] G. Liger-Belair et al., Langmuir 2004, 20, 4132. DOI: 10.1021/la049960f

[6] G. Liger-Belair et al., J. Phys. Chem B, 2005, 109, 14573. DOI: 10.1021/jp051650y

[7] G. Liger-Belair et al., J. Phys. Chem B, 2006, 110, 21145. DOI: 10.1021/jp0640427

For more of the physics behind champagne check out this great review: