README.md
In blongworth/HybridGIS: Data, Functions, and Analysis for the NOSAMS Hybrid Gas Ion Source
HybridGIS
The HybridGIS project contains functions, analysis scripts and data for
the NOSAMS hybrid gas ion source.
Installation
You can install the development version from
GitHub with:
# install.packages("devtools")
devtools::install_github("blongworth/hgis")
Running make will download the NOSAMS results files if a connection to
the NOSAMS file server is available. This currently uses an absolute
path to the server of /mnt/shared/USAMS/Results. This should be
converted to a URL, but until then, modify RES_DIR in Makefile to
match it’s location on your system. make will eventually run the
analysis scripts as well.
Functions
Loading the package loads functions for flow calculation, efficiency,
data import, data reduction and plotting.
library(HybridGIS)
Project Summary
The information below summarizes the operating parameters, development,
and questions for the hybrid gas ion source on USAMS.
Source parameters
Best currents for the source are around 10uA 12C- or roughly 1/10 of
graphite currents. This is in line with what other hybrid sources
produce. Typical “good” currents are between 5-10uA. The source
typically produces best current at around 5uLs-1 CO2. Literature values
give 1.5uLs-1 as optimal, but we’ve never seen that, and pressures,
capillaries and flows in at least one of these papers doesn’t make
sense.
Source currents are variable by day and by target. “Good” currents for a
given day vary between 5-10uA, and it’s not yet clear what causes this
variability. Alignment, source condition, and settings of the ionizer
and cesium are all factors. Current also varies by target for a given
set of gas and source conditions. Some targets are clearly “bad”,
producing currents much lower than other targets for a given set of
conditions. Capillary clogging can be an issue, especially when sampling
from vials with acid.
Precision taken as the standard error of the mean of repeated runs of a
sample is around 1%. Precision is typically a little worse than
predicted by counting statistics. More work is needed to look at whether
this reflects variability between samples, but initial test of paired
replicates show similar standard deviation for samples that show good
stable currents.
Measured value of tank modern gas normalized to solid graphite standards
ranges from agreeing with solid OX-I to 7% lower than expected for low
current samples. Normalizing to tank CO2 is probably the best method to
account for variability in hybrid source performance.
The system has a blank contribution from the target and source of about
100nA 12C- or 1% of a sample producing 10uA 12C-. This current varies by
day and by target and can be as high as 1uA. Helium does not seem to add
significantly to this blank. If the blank current is constant, the blank
contribution is inversely related to the sample current, ie. if the
sample current is 5uA, 2% of that will be from the blank and so on. This
can be modeled with a “current balance model” following the same methods
used for constant mass contamination.
The above measurements are at optimum sample current of 6-10uA 12C-.
Precision and measured ratio both decrease as sample current drops below
the optimum range. The ratio effects may be related to the “constant
current” blank discussed above. More work is needed to see whether this
is just a blank effect or whether there are other factors at low
current.
Open split inlet
The open split interface supplies a helium-CO2 mixture to a split where
a capillary delivers gas to the source and the rest is vented. The
current split is a VICI Luer adapter on the outflow side of the dual
needle (Figure 1). The split has fairly low volume. Testing is needed to
determine minimum sample flow to avoid atmosphere entering the split.
Currently, 250cm of 50um capillary defines flow rate from the open split
into the source, with a bulkhead connector at the source cage. Precision
of measurements of modern CO2 has been better than 1% (std error of mean
of measurements of a target) for high current samples, with precision
dropping off as current decreases below 6uA or so. Accuracy is
reasonable within the large errors on gas measurements. Uncorrected
measurements of dead CO2 range from 1% to 5% of modern depending on
sample current.
Measurement of a small number of replicates shows that between sample
reproducibility is around 1%. Samples displaced by helium have slightly
higher variability, but more tests are needed to better assess
reproducibility.
Helium displacement from vials
The test setup for extracting CO2 from septa-sealed vials is as follows:
a double needle is inserted into the vial through the septa. One side of
the needle is connected to helium which is used to force CO2 from the
vial through the other side of the needle. The outflow from the vial
enters an open split where a capillary is used to admit sample to the
ion source. The system is currently set up to run with samples starting
at atmospheric pressure.
First tests of displacing pure CO2 from a vial into the open split using
helium are promising. Helium flowing at 100uLm-1 produces current from
7mL of CO2 for over 2 hours. Currents do decrease with time due to
dilution, but by supplying higher than optimal CO2 flow to the source
initially, stable currents can be maintained for some time. Fraction of
CO2 in sample gas should follow this relationship:
$$C(t) = e^{-\frac{r}{V}t}$$
Where C(t) is the fraction CO2, r is the rate of helium
displacement flow, and V is the initial volume of the CO2-filled vial.
Setting flow to the source to 10uL/min provides excess CO2 to source
until the helium dilution ratio increases to > 1:1. Another approach
may be to backfill vials with helium and run at a higher dilution ratio
and flow to the source. This should produce a smaller change in He:CO2
over the course of a run. MICADAS runs at 95% He:CO2, which would mean
producing < 0.5mL CO2 in a 7mL vial and backfilling with helium
before displacing to the source. Some intermediate dilution may be
appropriate. Alternatively, vials could be flushed with helium prior to
acidification, rather than evacuating, leading to some concentration of
CO2 in He at slightly > 1 ATM.
Testing long runs of dilution of CO2 with displacing He and diluting He
with CO2 show that the setup is still optimized for low flow to the
source, even with larger diameter delivery capillaries. There appears to
be a constriction in the split-source tubing that is keeping flows low.
Current focus is on the large-sample 7mL method, so this will be
diagnosed as time allows.
Targets have shown “pulses” of higher current. Not sure why this is, but
I suspect it’s either buildup of graphite or other available carbon on
the target, or target switching to the “high state” as sometimes happens
with solid targets.
First tests with carbonate in vials were hit-or-miss. Problems with
capillary clogging and low currents led to poor results for several
samples. A pair of NOSAMS2 samples produced good currents, results
agreed well, but were about 3% higher than expected Fm. A second set of
carbonates produced better currents (\~5uA) and better results.
Agreement of replicates varied from 3 to 16 permil. Near-modern NOSAMS2
was very close to the consensus value, while C-2 and C-1 were elevated
by 2 and 3.6% respectively, indicating that a blank correction is
needed.
Two more recent carbonate tests produced stable currents and
backgrounds. Since sample, target blank, and sample blank currents were
relatively constant across samples, a large blank correction can be used
to correct results. Samples from both runs showed a linear relationship
between difference from expected Fm and Fm of the sample, but in one
sample set the relationship was positive, and in the other, negative.
Since the intercept with 0-offset was close to 0 Fm, I think this is a
problem with normalization, rather than blank correction. Standards run
at the end of the day are higher than those run at the beginning, which
may give clues as to what’s going on.
Questions
Q: Background from the source/target is relatively stable at roughly
100nA or 1% of maximum CO2 current. Is the relative blank contribution
proportional to CO2 current?
A: Based on limited data, it’s looking like yes, this follows a constant
contamination model.
Q: If the relative blank contribution is proportional to sample current,
can we correct for a varying sample current over a run (due to variable
dilution, pressure, etc.)?
A: If the sample current varies widely, this will likely require taking
data in shorter segments and blank correcting each segment using the
mean current for that segment. We may be able to mitigate by setting
capillary length such that CO2 flow starts above optimal and drops to
optimal by the end of the acquisition.
Q: Does standard error of the mean as a sample error agree with between
target variability?
A: Initial data from identical samples run under similar conditions show
standard deviation of around 1-1.5% for modern samples.
To do
- Set up for acidification of samples under helium to produce diluted
samples
- Test closed system with vial displacement.
- More repeated measurements under identical conditions to determine
reproducibility
- Test breakthrough with new open split
- Establish best helium flow for displacement
- Fine tune capillary length for best flow to source over pCO2 range
of full size sample
- How to handle < 1 Atm in vials. Backfill?
- Measure more test samples prepped on Gilson (7mL CO2, Atm pressure)
- Test higher dilution (5% CO2 in He)
- Define blank, determine whether current-balance correction is
needed.
- Refine normalization and error propagation for data reduction.
Development
2021-04-16 - Carbonate test #4. Similar test to last Friday, with C-1,
C-2, and NOSAMS-2. Stable currents and good data.
2021-04-09 - Carbonate test #3. Good data and stable currents from
TIRI-F, TIRI-I, and NOSAMS-2.
2021-04-02 - High dilution test #2.
2021-03-26 - Startup after tank opening and high dilution test.
2021-03-05 - Carbonate test #2. Much better currents and data than
previous attempt.
2021-03-01 - Test currents and tuning after source cleaning.
2021-02-12 - Clean source in preparation for next tests. Lots of black
residue on CO2 delivery arm. Flushed steel delivery capillary.
2021-02-05 - Tested a series of carbonate samples prepared on Gilson.
Tested C-1, TIRI-F, TIRI-I, C-2, and NOSAMS2. Used \~32mg carbonate to
produce \~7mL CO2. Problems with clogging and low currents limited
quality of tests.
2021-01-29 - Ran a series of vial tests with modern and dead CO2 to
assess reproducibility.
2021-01-22 - Ran pure CO2 at different pressures to look at effect of
current on precision and ratio. Repeated test for live and dead tank
CO2.
2021-01-15 - Pure CO2 and displacement tests using vial and double
needle. Tested precision with vial setup by flowing CO2 at 200kPa (high
flow, roughly 400ul/min) into open split. Followed this by switching to
helium flow to vial at 100kPa (200ul/min). Repeated test with a second
vial. Built second dual needle setup for use in vial filling (can also
use Gilson).
2021-01-08 - Precision tests with direct CO2-source connection. Tested
source pressure and current vs CO2 regulator pressure. Tested flow vs
pressure with leak checker. Several capillary swaps. Acquired data at
several flow rates for precision testing.
2021-01-07 - Set up for precision test using 7m x 50um capillary from
CO2/he via Valco valve to cage wall.
2020-12-31 - Add CO2/He valve to vial test setup. VICI 4-way valve
connected with 1/16" ss tubing from CO2 and He regulators allows
switching between gasses delivered to a glass capillary leading to outer
needle of dual needle. Leak checking and capillary adjustments for good
flow range at reasonable pressures.
2020-12-24 - Build dual needle setup. Supply displacement helium through
outer 1/16" tubing, sample mixture through inner needle. Use VICI Luer
adapter on top of needle as open split with inserted capillary to
source.
2020-12-18 - Continue testing lower open split flows with leak checker,
run dead and live gas at lower flows. Spoke with Josh B. about double
needle setups.
2020-12-11 - Open split flow reduction. Test flow controller for helium,
adjust needle valves to allow lower CO2 flows. Measure blanks of target,
helium, CO2 and diluted CO2.
2020-12-04 - Measure blank contributions of target/source, helium, CO2,
and diluted CO2. Install flow controller for helium.
2020-12-03 - Prepare and load new HGIS test wheel.
2020-11-17 - Diagnose issue with low currents, measure diluted and pure
live and dead gas.
2020-11-03 - Troubleshoot low currents, measure diluted and pure live
and dead gas.
2020-10-15 - Troubleshoot low currents. Lots of tune and target position
testing.
2020-10-13 - Prepare and load new HGIS test wheel.
2020-10-06 - Dilution tests of live and dead gas. Test stability of
currents over time.
2020-09-29 - Starting up HGIS. Load sample wheel. Reterminate
capillaries at cage wall and open split.
Figures
Presentations
Slides for a presentation given at SNEAP 2021:
SNEAP 2021 Presentation
blongworth/HybridGIS documentation built on Dec. 19, 2021, 10:41 a.m.
R Package Documentation
Browse R Packages
We want your feedback!
Note that we can't provide technical support on individual packages. You should contact the package authors for that.
HybridGIS
The HybridGIS project contains functions, analysis scripts and data for the NOSAMS hybrid gas ion source.
Installation
You can install the development version from GitHub with:
# install.packages("devtools")
devtools::install_github("blongworth/hgis")
Running make will download the NOSAMS results files if a connection to
the NOSAMS file server is available. This currently uses an absolute
path to the server of /mnt/shared/USAMS/Results. This should be
converted to a URL, but until then, modify RES_DIR in Makefile to
match it’s location on your system. make will eventually run the
analysis scripts as well.
Functions
Loading the package loads functions for flow calculation, efficiency, data import, data reduction and plotting.
library(HybridGIS)
Project Summary
The information below summarizes the operating parameters, development, and questions for the hybrid gas ion source on USAMS.
Source parameters
Best currents for the source are around 10uA 12C- or roughly 1/10 of graphite currents. This is in line with what other hybrid sources produce. Typical “good” currents are between 5-10uA. The source typically produces best current at around 5uLs-1 CO2. Literature values give 1.5uLs-1 as optimal, but we’ve never seen that, and pressures, capillaries and flows in at least one of these papers doesn’t make sense.
Source currents are variable by day and by target. “Good” currents for a given day vary between 5-10uA, and it’s not yet clear what causes this variability. Alignment, source condition, and settings of the ionizer and cesium are all factors. Current also varies by target for a given set of gas and source conditions. Some targets are clearly “bad”, producing currents much lower than other targets for a given set of conditions. Capillary clogging can be an issue, especially when sampling from vials with acid.
Precision taken as the standard error of the mean of repeated runs of a sample is around 1%. Precision is typically a little worse than predicted by counting statistics. More work is needed to look at whether this reflects variability between samples, but initial test of paired replicates show similar standard deviation for samples that show good stable currents.
Measured value of tank modern gas normalized to solid graphite standards ranges from agreeing with solid OX-I to 7% lower than expected for low current samples. Normalizing to tank CO2 is probably the best method to account for variability in hybrid source performance.
The system has a blank contribution from the target and source of about 100nA 12C- or 1% of a sample producing 10uA 12C-. This current varies by day and by target and can be as high as 1uA. Helium does not seem to add significantly to this blank. If the blank current is constant, the blank contribution is inversely related to the sample current, ie. if the sample current is 5uA, 2% of that will be from the blank and so on. This can be modeled with a “current balance model” following the same methods used for constant mass contamination.
The above measurements are at optimum sample current of 6-10uA 12C-. Precision and measured ratio both decrease as sample current drops below the optimum range. The ratio effects may be related to the “constant current” blank discussed above. More work is needed to see whether this is just a blank effect or whether there are other factors at low current.
Open split inlet
The open split interface supplies a helium-CO2 mixture to a split where a capillary delivers gas to the source and the rest is vented. The current split is a VICI Luer adapter on the outflow side of the dual needle (Figure 1). The split has fairly low volume. Testing is needed to determine minimum sample flow to avoid atmosphere entering the split.
Currently, 250cm of 50um capillary defines flow rate from the open split into the source, with a bulkhead connector at the source cage. Precision of measurements of modern CO2 has been better than 1% (std error of mean of measurements of a target) for high current samples, with precision dropping off as current decreases below 6uA or so. Accuracy is reasonable within the large errors on gas measurements. Uncorrected measurements of dead CO2 range from 1% to 5% of modern depending on sample current.
Measurement of a small number of replicates shows that between sample reproducibility is around 1%. Samples displaced by helium have slightly higher variability, but more tests are needed to better assess reproducibility.
Helium displacement from vials
The test setup for extracting CO2 from septa-sealed vials is as follows: a double needle is inserted into the vial through the septa. One side of the needle is connected to helium which is used to force CO2 from the vial through the other side of the needle. The outflow from the vial enters an open split where a capillary is used to admit sample to the ion source. The system is currently set up to run with samples starting at atmospheric pressure.
First tests of displacing pure CO2 from a vial into the open split using helium are promising. Helium flowing at 100uLm-1 produces current from 7mL of CO2 for over 2 hours. Currents do decrease with time due to dilution, but by supplying higher than optimal CO2 flow to the source initially, stable currents can be maintained for some time. Fraction of CO2 in sample gas should follow this relationship:
$$C(t) = e^{-\frac{r}{V}t}$$
Where C(t) is the fraction CO2, r is the rate of helium displacement flow, and V is the initial volume of the CO2-filled vial.
Setting flow to the source to 10uL/min provides excess CO2 to source until the helium dilution ratio increases to > 1:1. Another approach may be to backfill vials with helium and run at a higher dilution ratio and flow to the source. This should produce a smaller change in He:CO2 over the course of a run. MICADAS runs at 95% He:CO2, which would mean producing < 0.5mL CO2 in a 7mL vial and backfilling with helium before displacing to the source. Some intermediate dilution may be appropriate. Alternatively, vials could be flushed with helium prior to acidification, rather than evacuating, leading to some concentration of CO2 in He at slightly > 1 ATM.
Testing long runs of dilution of CO2 with displacing He and diluting He with CO2 show that the setup is still optimized for low flow to the source, even with larger diameter delivery capillaries. There appears to be a constriction in the split-source tubing that is keeping flows low. Current focus is on the large-sample 7mL method, so this will be diagnosed as time allows.
Targets have shown “pulses” of higher current. Not sure why this is, but I suspect it’s either buildup of graphite or other available carbon on the target, or target switching to the “high state” as sometimes happens with solid targets.
First tests with carbonate in vials were hit-or-miss. Problems with capillary clogging and low currents led to poor results for several samples. A pair of NOSAMS2 samples produced good currents, results agreed well, but were about 3% higher than expected Fm. A second set of carbonates produced better currents (\~5uA) and better results. Agreement of replicates varied from 3 to 16 permil. Near-modern NOSAMS2 was very close to the consensus value, while C-2 and C-1 were elevated by 2 and 3.6% respectively, indicating that a blank correction is needed.
Two more recent carbonate tests produced stable currents and backgrounds. Since sample, target blank, and sample blank currents were relatively constant across samples, a large blank correction can be used to correct results. Samples from both runs showed a linear relationship between difference from expected Fm and Fm of the sample, but in one sample set the relationship was positive, and in the other, negative. Since the intercept with 0-offset was close to 0 Fm, I think this is a problem with normalization, rather than blank correction. Standards run at the end of the day are higher than those run at the beginning, which may give clues as to what’s going on.
Questions
Q: Background from the source/target is relatively stable at roughly 100nA or 1% of maximum CO2 current. Is the relative blank contribution proportional to CO2 current?
A: Based on limited data, it’s looking like yes, this follows a constant contamination model.
Q: If the relative blank contribution is proportional to sample current, can we correct for a varying sample current over a run (due to variable dilution, pressure, etc.)?
A: If the sample current varies widely, this will likely require taking data in shorter segments and blank correcting each segment using the mean current for that segment. We may be able to mitigate by setting capillary length such that CO2 flow starts above optimal and drops to optimal by the end of the acquisition.
Q: Does standard error of the mean as a sample error agree with between target variability?
A: Initial data from identical samples run under similar conditions show standard deviation of around 1-1.5% for modern samples.
To do
- Set up for acidification of samples under helium to produce diluted samples
- Test closed system with vial displacement.
- More repeated measurements under identical conditions to determine reproducibility
- Test breakthrough with new open split
- Establish best helium flow for displacement
- Fine tune capillary length for best flow to source over pCO2 range of full size sample
- How to handle < 1 Atm in vials. Backfill?
- Measure more test samples prepped on Gilson (7mL CO2, Atm pressure)
- Test higher dilution (5% CO2 in He)
- Define blank, determine whether current-balance correction is needed.
- Refine normalization and error propagation for data reduction.
Development
2021-04-16 - Carbonate test #4. Similar test to last Friday, with C-1, C-2, and NOSAMS-2. Stable currents and good data.
2021-04-09 - Carbonate test #3. Good data and stable currents from TIRI-F, TIRI-I, and NOSAMS-2.
2021-04-02 - High dilution test #2.
2021-03-26 - Startup after tank opening and high dilution test.
2021-03-05 - Carbonate test #2. Much better currents and data than previous attempt.
2021-03-01 - Test currents and tuning after source cleaning.
2021-02-12 - Clean source in preparation for next tests. Lots of black residue on CO2 delivery arm. Flushed steel delivery capillary.
2021-02-05 - Tested a series of carbonate samples prepared on Gilson. Tested C-1, TIRI-F, TIRI-I, C-2, and NOSAMS2. Used \~32mg carbonate to produce \~7mL CO2. Problems with clogging and low currents limited quality of tests.
2021-01-29 - Ran a series of vial tests with modern and dead CO2 to assess reproducibility.
2021-01-22 - Ran pure CO2 at different pressures to look at effect of current on precision and ratio. Repeated test for live and dead tank CO2.
2021-01-15 - Pure CO2 and displacement tests using vial and double needle. Tested precision with vial setup by flowing CO2 at 200kPa (high flow, roughly 400ul/min) into open split. Followed this by switching to helium flow to vial at 100kPa (200ul/min). Repeated test with a second vial. Built second dual needle setup for use in vial filling (can also use Gilson).
2021-01-08 - Precision tests with direct CO2-source connection. Tested source pressure and current vs CO2 regulator pressure. Tested flow vs pressure with leak checker. Several capillary swaps. Acquired data at several flow rates for precision testing.
2021-01-07 - Set up for precision test using 7m x 50um capillary from CO2/he via Valco valve to cage wall.
2020-12-31 - Add CO2/He valve to vial test setup. VICI 4-way valve connected with 1/16" ss tubing from CO2 and He regulators allows switching between gasses delivered to a glass capillary leading to outer needle of dual needle. Leak checking and capillary adjustments for good flow range at reasonable pressures.
2020-12-24 - Build dual needle setup. Supply displacement helium through outer 1/16" tubing, sample mixture through inner needle. Use VICI Luer adapter on top of needle as open split with inserted capillary to source.
2020-12-18 - Continue testing lower open split flows with leak checker, run dead and live gas at lower flows. Spoke with Josh B. about double needle setups.
2020-12-11 - Open split flow reduction. Test flow controller for helium, adjust needle valves to allow lower CO2 flows. Measure blanks of target, helium, CO2 and diluted CO2.
2020-12-04 - Measure blank contributions of target/source, helium, CO2, and diluted CO2. Install flow controller for helium.
2020-12-03 - Prepare and load new HGIS test wheel.
2020-11-17 - Diagnose issue with low currents, measure diluted and pure live and dead gas.
2020-11-03 - Troubleshoot low currents, measure diluted and pure live and dead gas.
2020-10-15 - Troubleshoot low currents. Lots of tune and target position testing.
2020-10-13 - Prepare and load new HGIS test wheel.
2020-10-06 - Dilution tests of live and dead gas. Test stability of currents over time.
2020-09-29 - Starting up HGIS. Load sample wheel. Reterminate capillaries at cage wall and open split.
Figures
Presentations
Slides for a presentation given at SNEAP 2021:
SNEAP 2021 Presentation
R Package Documentation
Browse R Packages
We want your feedback!
Note that we can't provide technical support on individual packages. You should contact the package authors for that.
Embedding an R snippet on your website
Add the following code to your website.
For more information on customizing the embed code, read Embedding Snippets.