HowTo – Cations analysis [90%]

Priority: low (cations data are only supplementary)
Updating: process is mature, HowTo is in rapid progress

This is about cations analysis via ICP in external labs. It covers from the making of sub-aliquots from acidified aliquots through transcription of lab results into the database.

Change log:

When	Who	Comments
2023 06 22	Sp17	Began from memory of handling of samples and interaction with Shree K. Giri. This reflects practice for the past several years primarily with Shree when we had our own ICP machine.
2023 06 29	Sp17	Fleshing out while preparing first 2023 batch for Shree’s lab.
2023 06 30	Sp17	Mention that detection limits also scaled upward to reflect dilution by preservative acid.

Related HowTos:

1. Objectives and overview

Determine concentrations of cations in water samples.
Merge returned, cleaned results into our tabular database.

This is the overall process:

   graph TD;
   A[acidified 40 mL<br>aliquots of<br>field samples]
   B((prepare 12 mL<br>sub-aliquots))
   L[new 15 ml<br>containers,<br>numbered]
   Rack[rack for<br>15 mL tubes]
   A-->B-->C
   L-->B
   C[sub-<br>aliquots]
   D[/manifest/]
   E((rack in<br>sequence<br>number order))
   Rack-->E
   F((carry<br>to lab))
   C-->E-->F
   D-->F

   F-->H[/results<br>from lab/]
   I((apply<br>detect limits,<br>insert sample IDs,<br> adjust dilution))
   J((transform<br>row/column to<br>tabular database))
   N((supply EQUIS<br>metadata))
   K[(extended<br>database)]
   H-->J-->K
   H-->I-->K
   H-->N-->K

2. Data quality considerations

The analyses are done by external labs, thus we must trust the lab’s handling of samples and data processing as well as their ability to operate machines and keep them in good working order.
Our sample identification is too complex for an external lab, thus (as with the DEC lab) we provide simple sequence numbers for all samples sent to a lab and we translate back and forth between these and our more complex sample IDs. This way the lab is less challenged and the burden is entirely with us to translate back and forth.
If we ever must depend on these data, we sometimes should use split-sample tests between multiple labs, and repeat sample tests with the same lab. We can also fortify environmental samples and make artificial samples as blind or disclosed tests of lab performance. We did two cross-lab checks for cations in 2022. Results were OK.
The labs we use have different reliabilities. Shree K. Giri’s work in USDA lab is exemplary. The CNAL lab shut down for reorganization shortly after our last use of them. Later data interpretation should take into account lab quality.
The labs tend not to report lower and upper detection limits. We can ask them afterwards.

3. Sub-aliquot preparation and storage for cations analysis

For the 2021-2025 DEC project, we preserve 40 mL acidified with 1 mL of 1M HNO₃. We store these in fridge indefinitely, hopefully not for more than 4 months. Labs want 12 mL per sample, thus the 40 mL provides room to prepare two repeats.

Thus this section is about sub-aliquots made for analysis starting from aliquots made for preservation, i.e. sub-aliquots which are sub-sub-samples.

graph LR;
A[original sample]-->B[acidified 40 mL]-->C[12 mL for analysis]
D[1mL 1N nitric acid]-->B

We are sending batches of 30-70 samples at a time to labs.

We use 15 mL new polypropylene centrifuge containers to hold the 12 mL for ICP analysis. Containers are not reused. We discard most of these containers as plastic waste after a single use.

The aliquot containers are not rinsed with sample water, to preserve volume of the acidified aliquot. We also refrain from washing containers with deionized water. We rely on low contamination 15 mL containers. Recent stock is sterilized and sealed, preferred over bulk which provide uncapped containers and open tubes in separate bags.

The acidified aliquots are stored in fridges rather than freezer, so the sub-aliquots can be made without thawing.

Labels on the sub-aliquot containers should include both a simple sequence number, for the cations lab, and our complete sample identifiers, and indicate that they were preserved to pH 2 by adding HNO₃. Add the sequence number on the container cap, as we do with the acidified aliquot container.

We email a manifest of samples in spreadsheet form to the lab before providing the aliquots. We provide a printed version of this with the sub-aliquots.

We provide the containers in racks matching the conical 15 mL tubes, sorted in sequence number order. If we submit any fortified or artificial samples, those may get some kind of special numbering, since to use the plain numbers would overlap with our own .

4. Handing over samples to lab

Since the external labs are on the Cornell campus, we hand carry. Make arrangements in advance via email with the lab for when to arrive. The lab will store the samples in fridge until they get to testing them.

We usually get the racks back, but don’t count on it. Labs dispose of the 12 mL tubes.

5. Data transcription and interpretation

5.1 Format received

The data come back via an emailed spreadsheet. We begin by saving this under another name to preserve what we received as a backup, then we edit the spreadsheet to bring it closer to copy/paste potential for our tabular database. The lab spreadsheet and our expanded version should be stored in Box: /DEC/samples/cations/ and our file name should include the date of analysis.

The lab spreadsheet uses our sequence numbers as the sample identifiers. The lab may as a courtesy copy in our own sample ID’s; if they do not, we begin by inserting four columns and copying in our sample IDs: site ID, sampling location ID within site, date, time.

The labs use various ways to indicate normal results and results that are out of reliable analytical range. Some labs may provide numbers for all regardless of the reliability. We have to impose the detection limit upper and lower bounds via transforming their data.

Our acidified aliquots, thus the sub-aliquots, are all diluted by adding 1 mL of acid to 40 mL of sample, thus the results are all underestimated by a factor of 41/40 = 1.025. Insert a new formula column and multiply all lab numbers by this value, and round to the same number of decimal places in the lab original. (Labs usually provide extra decimal places, our computed values can be fewer.)

Remember that the detection limit also scales by any dilution factor. EQUIS provides for entering a dilution factor; the 1.025 should not be used there since it is so small. The EQUIS dilution field should be used when a sample is diluted to get into detection range.

The cation symbols in their results will usually be chemical symbols. We need to transform the lab’s analyte symbols into our tabular database conventions. For cations, these are spelled out with an initial capital letter.

As with the DEC lab results, the cations results will have one row per sample and one column per cation. There may also be rows with results for their Quality Control samples, which have known concentrations. We can ask the lab what the known concentrations are, or they may be obvious by having results like ‘39.997’, clearly the known was 40. We accept their judgment that the QC numbers are adequate by their standards, or else they would have redone the work. We merge their QC result into our database and create artificial “samples” in the database to represent them.

Ultimately the QC results are supposed to go into EQUIS. As of mid 2023 we have not done this yet, thus we will need to consult more EQUIS documentation and perhaps DEC EQUIS specialists for how to encode them. We have the option, as before, of treating these as equivalent to field samples.

5.2 Transforming into our tabular database (EQUIS-driven)

The main effort is to transform separate columns per cation into more rows. If there are 20 cations, that generates 20 database rows per lab row. A new column is needed in the database, containing the name of the cation analyte. Sample ID columns for the 20 copies are identical. This is a long and tedious process if the numbers of samples and cations are large. For example, if we submitted 60 samples and get 20 cation results for each, we must create 1200 new rows in our destination data table. Since nearly all cations results are “detects”, there is no sparseness to simplify proofreading like there is in the mostly non-detect pesticide results.

The process involves a long copy and paste session, with intense focus. Take breaks during the conversion session to reduce the frequency of mistakes. This is scriptable, but each time we get data back there is a different format, so we end up not investing the time to write conversion scripts.

ToDo: Invest time to test scripting, to see how long script writing will take versus manual copy and paste. There should be some carryover between batches because the fundamental unrolling of the row/column format into one test result per row is always present. The same script may be usable for the DEC lab pesticide results.

The detailed conversion from lab-spreadsheet-to-database is described in the HowTo for DEC lab data conversion. The first difference is the need to apply an upper detection limit rather than a lower. We can use “gt” as the qualifier, for “greater than.”

5.3 Too-high values, known interferences

If there are ever critical needs to understand quantities more than “>300” for example, the samples can be resubmitted after dilution. The diluent should be deionized water, and the ending pH.

Years ago Shree noted precipitates in very many field samples that we had thawed. The settled crystals must be removed from the containers before ICP analysis; in one case we poured off the liquid. We began in 2023 to acidify samples to head off this precipitation.

ToDo: do these precipitates also affect anions and pesticide results?

There were elevated zinc concentrations in some 2022 samples. These were limited to our steel monitor wells, absent from owner wells and our PVC wells. Our zinc-galvanized steel wells exhibit some white powder at the surface, possibly zinc oxide crystals. We decided to do without zinc analytical results; we carry them into the database and flag them with useit=0 to omit from most reporting.

ToDo: would the elevated zinc have any effect on other analytes?
ToDo: insert here the paper citation about zinc and galvanized wells.

6. Administrative

We pay for analysis via Cornell interdepartmental funds transfer. Funds for the USDA lab go to the Cornell Food Science department. SWL staff prepare the invoice from lab to ourselves as a courtesy to the USDA lab staff. BEE department admin staff then arrange with the Food Science admin staff for the transfer.

If we ever use another Cornell lab, such as Cornell Nutrient Analysis Lab (CNAL) used in the pass, they prepare the invoice that we pass along to the BEE department admin staff along with the DEC project account number and a business purpose phrase, such as “Analysis of project samples for cations (DEC pesticides)”.