When we wrote a recent blog on the composition of water from Silica Rapids on Mt Ruapehu, following a GeoNet News story on how volcanology field technician Karen Britten collects the water samples, we omitted the critical middle part of the story – how we get from the water sample to the composition results. That part is mostly carried out by the New Zealand Geothermal Analytical Laboratory (NZGAL) at GNS Science’s Wairakei Research Centre near Taupō.

Today’s blog is a little different from usual, we chat with Karl Laurence, Senior Laboratory Technician, and ask him a lot of questions on how our Silica Rapids water sample was analysed at NZGAL. The details on how the analyses are made are a bit geeky at times, but it’s a really interesting insight into a how a chemical laboratory like NZGAL works. So, lets get into it.

What are the steps the Silica Rapids water sample went through when it arrived at the lab – from reception to final results?

At Silica Rapids and similar locations, Karen collects a few different water samples and prepares them based on the analyses we will conduct. This might include filtering to remove undissolved particles, acidification with nitric acid (to preserve dissolved cations, positively charged components) or exclusion of atmospheric oxygen using a rubber seal. Karen delivers these samples to our lab; we register them in our Laboratory Information Management System (LIMS) and the required analyses show up on the worklists of the technicians working at NZGAL. Each technician is expert in their own area, such as spectroscopy, gas and/or liquid chromatography and wet chemistry, and performs their part of the analysis using specialist equipment. The equipment can be as simple as a pH meter, or as complex as a plasma-source mass spectrometer.

After all analyses are complete, the results are collated and reviewed by a ‘Key Technical Person’ – a senior NZGAL member who is responsible for certifying the results. They conduct a few ‘sense checks’ on the results, such as comparing the latest results with the site history and investigating any unexpected differences, or confirming the results are internally consistent across all the analyses, something called an ion balance (more on that later).

Finally, a Certificate of Analysis report is generated from our LIMS system containing all results for Silica Rapids with detail on methodologies used and limits of detection for each analyte, and this is emailed to Karen. Job done, simple!

Do other samples that NZGAL analyses take a different path through the lab? Can you give an example?

Most samples follow roughly the same path as described above. However, we receive all sorts of weird and wonderful sample types such as geothermal scale, rocks, clays and geothermal gases. We’re excited to receive this type of sample because it often requires creative chemistry to produce accurate, reliable, and meaningful analytical results. These all have their own preparation and analysis requirements, for example, a sample of solid scale from a geothermal pipeline may need digestion with hydrofluoric acid before analysis, or to be crushed up before it can be fed into an instrument for analysis. Some sample preparations are sent away to specialist laboratories for specific analyses, such as the Stable Isotope Laboratory at our National Isotope Centre where measurements of 18O (oxygen isotope) and 2H (hydrogen isotope) are made. Notwithstanding these differences, the story is always roughly the same: receive sample, prepare sample, analyse sample, collate and quality check results, send report.

For Silica Rapids the Volcano Monitoring Group (VMG) gets concentrations of 19 analytes from NZGAL. How much water does Karen deliver? Do you have to split up the water for different kinds of analytes?

For the Ruapehu Silica Rapids analysis, Karen takes several samples based on the analyses that will be performed in the lab:

1. Rubber Seal: (Glass Bottle with flexible sealed top to exclude atmospheric oxygen) – 300 mL. Measurements made: F⁻, HCO₃2⁻, pH, dissolved H₂S

2. ‘Raw’: (water sampled directly into a plastic bottle) – 250 mL. Measurements made: ¹⁸O, ²H (subcontracted to GNS Stable Isotope Laboratory)

3. Filtered Sample: (water sampled through a 0.45µm cellulose acetate filter into a plastic bottle) – 100 mL. Measurements made: dissolved NH₃, NO₃-N, SO₄²⁻, Cl⁻, Br⁻

4. Filtered and Acidified Sample: (water sampled through a 0.45µm cellulose acetate filter into a plastic bottle, then acidified to pH<2 with nitric acid – 100 mL. Measurements made: Al³⁺, Ca²⁺, K⁺, Li+, Mg²⁺, Na⁺, SiO₂, As3₊, B³⁺, Fe²⁺

All up, 750 mL of water is sampled for our Silica Rapids analyses.

The analysis procedure itself – most of us remember analysing things in a chemistry lab at school, we had some liquid in a test tube and added a chemical, got a reaction and used that to evaluate what was in the test tube. NZGAL is a bit more sophisticated than that isn’t it? You have machines do the analyses?

The essence of our analyses is exactly how you’ve described it – you have a sample, you expose it to some new condition (heat/cold, pressure, electric/magnetic field, irradiation by light, presence of reactive compound, etc), this effects some kind of change in the sample, we measure the magnitude of this change, and interpolate that change along a calibration curve to calculate the concentration of whatever analyte we’re interested in.

The classic high-school titration reactions you mention are an excellent example of this principle and demonstrate how observations we can make at a macroscopic level relate to a molecular level. For example, how much acid we added to a sample before the colour changes reveal detail at the molecular level, such as how many molecules are present in every litre of sample. Our lab equipment functions much the same, but the reactions we’re performing are more controlled and usually more extreme! One example is an instrument called an ICP-OES (inductively coupled plasma - optical emission spectrometer) that we use for measuring the presence of positively charged cations like sodium and magnesium. We take our water sample, turn it into a fine aerosol (think fly spray) by mixing it with argon gas, and then introduce it to a superheated region of plasma (at a temperature of around 10,000°C). At this temperature the atoms and ions are excited to higher energy states. As they cool, they emit characteristic wavelengths of light which we can measure with a spectrometer and use that infer the precise concentration of the atom or ion that emitted it.

That’s pretty technical so here’s the TL;DR (too long, didn’t read) summary of the above – Karen gives me a sample, I heat it up to the temperature of the surface of the sun, it emits light, and I measure the colour and intensity of this light to figure out what elements are in it and how much there is. Easy peasy!

How many different analysis machines are needed for Silica Rapids samples? How many machines in total?

Lots – each instrument performs a specialist task. For Silica Rapids, we use six different machines, most with long, complicated names. For chemistry geeks, here’s the list. For the rest of you, jump to the next question: 1. Agilent 5900 SVDV ICP-OES (cations)

1. Lachat QuikChem 500 FIA (dissolved reactive phosphorous, dissolved NH₃, dissolved H₂S)

2. Ion Selective Electrode (F⁻)

3. ManTech Auto-titrator (HCO₃²⁻, pH)

4. Thermo Fisher Dionex ICS-5000 Ion Chromatograph (other anions)

5. Isoprime isotope ratio mass spectrometer (GNS Stable Isotopes Lab – ¹⁸O and ²H)

We have many more machines, also with complicated names (including ICP-MS, TCD-Gas Chromatographs, Discrete Analyser, Liquid Scintillation Analyser, and HPLC) that are used for analyses beyond what the Silica Rapids samples require.

How much of the water sample do the machines need? Teaspoon, tablespoon, cup?

Each instrument varies in its consumption. Some use as much as 50 mL (about 10 teaspoon’s worth), while others use only 20 µL (a very small fraction of a teaspoon). Typically, instruments that are doing wet chemistry (that is, reaction chemistry like a titration) use more, while those performing non-destructive analysis like spectroscopy use less.

If it’s all done by machines, what’s the role of someone working in a lab at NZGAL?

Even a Formula 1 race car needs a driver and a team behind it. The instruments we use allow us to measure small effects to identifying and quantifying the chemical species that are present in a sample. They require a knowledgeable person who understands the chemistries of the samples, who can develop and improve methods, perform instrument maintenance and troubleshooting, and be a technical expert for our external auditors. Final reports need an experienced eye to certify the quality of the results and track down the cause of anything unexpected. Anyone can learn the basic procedure to run a sample, but the method development process requires deep knowledge of the scientific principles of the instrument operation and a tenacity to push past the many roadblocks and dead ends they will encounter.

Accuracy and uncertainty are important questions in any discussion about data. How accurate are the analyte concentrations from Silica Rapids water?

Our methods are validated for selectivity, freedom from interference, linear dynamic range, method detection limit (MDL) and uncertainty of measurement (UoM). All real-world measurements are subject to some degree of uncertainty due to random or systemic error of the measurements, and variation between the sample chemistry and calibration standard chemistry. The UoM is what most would consider when thinking of “accuracy and uncertainty”. UoM is usually high close to the detection limit (when there is little of what you are looking for in the sample) and low in the mid-high calibration range (where there is a lot in the sample).

One tool, mentioned above, that we use to check aggregate error across all the analyses is an ion balance calculation. Natural waters are electrically neutral, so the sum of charges of cationic (positive) species must equal the sum of charges of anionic (negative) species. After all analyses have been completed, the results of the positive and negatively charged species are summed and compared to determine whether there is a positive or negative bias. NZGAL accepts a ±5% bias either way and investigates the cause of any ion balance that exceeds this.

Does accuracy vary by analyte and concentration?

Yes – each analyte on each instrument will have its own UoM. In fact, some analytes can be measured multiple ways – for example, chloride (Cl⁻) can be measure by ion chromatogram or by automated titration. These techniques rely on fundamentally different chemical techniques, and therefore have different UoMs.

What do you do with all the analysis values once you have the concentrations of everything in the Silica Rapids water? Do they go into a computer somewhere?

We store all sample and analytical data digitally in our LIMS (Laboratory Information Management System) database for easy retrieval and review of historic trends. While hardcopy results are destroyed after 5 years, our digital results exist into perpetuity.

Do you keep the Silica Rapids water in a bottle somewhere so in 5 years’ time we could ask you to re-run the samples through the analysis machines and re-calculate the concentrations?

Our lab serves both the GeoNet programme and the domestic and international geothermal power industry, so our sample volume is high at many thousands of bottles per year. As such, it is impractical to store samples for such a long period. We keep sample bottles for at least 6 months from issuance of results and are happy to re-run tests if checks are needed.

How long does it take to do all the Silica Rapids analyses?

NZGAL takes five working days on average from sample reception to issuance of results.

That’s it for now

Analysis of GeoNet data is an important step that often doesn’t have the visibility it deserves. How data are analysed might often be skipped over, but it is an important part of what we do and without it there would be no data to blog about. We’d like to thank Karl Laurence at NZGAL for his time answering our questions, and the wider NZGAL team for analysing our water samples.

You can find our earlier blog posts through the News section on our web page just select the Data Blog filter before hitting the Search button. We welcome your feedback on our data blogs and if there are any GeoNet data topics you’d like us to talk about please let us know!

Ngā mihi nui.

Contact: info@geonet.org.nz