In chepec/periodicdata: Properties of the chemical elements
Introduction
In this vignette we will demonstrate how the dataset was collected, by --- drumroll --- collecting it!
We will also show how you can generate the familiar IUPAC periodic table from this dataset
together with some ggplot2 code.
Almost all of the properties in this dataset were scraped from the
periodictable.com website which provides "up to date, curated data provided by Mathematica's
ElementData function from Wolfram Research, Inc." as HTML/CSS tables.
The Wolfram's ElementData() function,
in turn, lists multiple sources,
both scholarly and other websites.
Despite most of the data originating in Mathematica (by Wolfram Research),
the quality of the data on periodictable.com
website is actually rather poor, appearing to be suffering from whatever
conversion was applied from the original Mathematica format.
Obvious typos and other easy-to-spot errors in the scraped data have been corrected
by yours truly in this dataset.
library(dplyr)
library(magrittr) # extract2()
library(httr)
library(rvest)
library(stringr)
library(readr)
library(tibble)
library(knitr)
library(usethis)
library(ggplot2)
library(ggrepel)
library(scales)
library(here)
options(
digits = 7,
width = 84,
continue = " ",
prompt = "> ",
warn = 0,
stringsAsFactors = FALSE)
opts_chunk$set(
dev = 'svg',
#out.width = "100%",
fig.width = 7.10,
fig.height = 4.39,
fig.align = 'center',
echo = FALSE,
eval = TRUE,
cache = TRUE,
collapse = TRUE,
results = 'hide',
message = FALSE,
warning = FALSE,
tidy = FALSE)
# dont reset scrape_afresh here, it is reset automatically if
# inst/extdata/periodicdata-raw.rdata is not found
# this just sets it to the default FALSE
scrape_afresh <- FALSE
# periodic table dimensions (sensitive setting, layout will likely break if changed)
ptable.width <- 9
ptable.height <- 5.25
# Starting from this URL, we crawl the page and collect the URLs to all property pages, along with the name of each property.
# We put them together in a dataframe, so it is clear which URL belongs to which elemental property.
element_data <- read_html("http://periodictable.com/Elements/001/data.html")
elemental_properties <- data.frame(
property = element_data %>%
html_nodes(xpath = "//a") %>%
html_text())
elemental_properties$links_raw <-
element_data %>%
html_nodes(xpath = "//a") %>%
html_attr("href")
# drop any rows that do not begin "../../"
elemental_properties <-
elemental_properties[grep(
pattern = "^\\.\\./\\.\\./",
x = elemental_properties$links_raw), ]
# remove the leading ../.. (relative link signifier)
elemental_properties$links_trail <- gsub(
pattern = "^\\.\\./\\.\\./",
replacement = "",
x = elemental_properties$links_raw)
# keep only the rows that begin with "Properties"
elemental_properties <-
elemental_properties[grep(
pattern = "^Properties",
x = elemental_properties$links_trail), ]
# make full urls
elemental_properties$url <- paste0(
"http://periodictable.com/",
elemental_properties$links_trail)
# modify the URL so it sorts the table by atomic number
# (this is a feature of periodictable.com, they offer lists sortered in different ways
# for each property)
elemental_properties$url <- sub(
pattern = "\\.html$",
replacement = ".an.html",
x = elemental_properties$url)
# we are done, go ahead and drop the no longer required columns
# links_raw, links_trail
elemental_properties <- elemental_properties[, -c(2,3)]
# oh, and add a sanitised version of the property names
elemental_properties$sanitized <- gsub(
# replace spaces with underscores
"\\s+", "_",
# replace "%" with "percent"
gsub("%", "Percent",
# remove parentheses and dashes
gsub("[()-]", "",
# remove apostrophes
gsub("'", "", elemental_properties$property))))
# NOTE: Melting point and Boiling point are duplicated because they are displayed twice
# on the data page for Hydrogen, both under "Overview" and "Thermal properties".
# Other properties could be too, so we should deduplicate the dataframe.
elemental_properties <- unique(elemental_properties)
# reset the row numbering
row.names(elemental_properties) <- seq(1, dim(elemental_properties)[1])
Collect periodic table data from periodictable.com
Each property, e.g., density, will have a value for each element of the periodic table.
This value is just a string, and depending on the type of the property, it may be just
a number, or a quantity with a unit, or some text with various attributes.
At this stage, we don't mind the internal structure of the value, we just want to
prepare for collecting them.
# get the name of all the elements (two-step process)
element_names <-
read_html(elemental_properties$url[1]) %>%
html_nodes("table") %>% extract2(1) %>%
html_nodes("table") %>% extract2(8) %>%
html_nodes("td") %>%
html_text()
# here we initialise the properties dataframe, which we will use later
# to read in all the properties (building the frame using ..._join() calls)
properties_raw <- data.frame(
Name = element_names[
read_html(elemental_properties$url[1]) %>%
html_nodes("table") %>%
extract2(1) %>%
html_nodes("table") %>%
extract2(8) %>%
html_nodes("td") %>%
html_attr("align") == "left"])
Some property pages (listed below) are difficult to parse, usually because of bad HTML.
For now, we just skip those pages (no big loss).
skip_properties <- c(
# has multiple values per element, plus formatting errors in HTML (cell 1 and 120)
"Ionization_Energies",
# replacement has 120 rows, data has 118
"NFPA_Label",
# has formatting errors in HTML (cell 1 and 120)
"Names_of_Allotropes",
# formatting errors in HTML, specifically in cell 1 and 120
"Discovery",
# formatting errors in HTML, specifically in cell 1 and 120
"Lattice_Angles",
# formatting errors in HTML, specifically in cell 1 and 120
"Lattice_Constants",
# replacement has 120 rows, data has 118
"Crystal_Structure",
# formatting errors in HTML, specifically in cell 1 and 120
"Known_Isotopes",
# formatting errors in HTML, specifically in cell 1 and 120
"Stable_Isotopes",
# formatting errors in HTML, specifically in cell 1 and 120
"Isotopic_Abundances")
# drop those rows from elemental_properties
elemental_properties <-
elemental_properties[-which(elemental_properties$sanitized %in% skip_properties), ]
Ok, let's populate our properties dataframe with some scraped data.
# This chunk does all the scraping of periodictable.com
# But it tries its best not to do any scraping
# If you really want to scrape the website and rebuild the entire dataset from
# scratch (this takes some time!), go ahead and delete the file "inst/extdata/periodicdata-raw.rdata"
# before running this chunk!
if (!file.exists(here::here("inst", "extdata", "periodicdata-raw.rdata"))) {
scrape_afresh <- TRUE
for (k in 2:length(elemental_properties$property)) {
message(paste0(
"Reading property page (", k, " of ", length(elemental_properties$property), "): ",
elemental_properties$property[k]))
# On these property pages, the source HTML appears to be malformed (but recoverable)
# The HTML source looks weird: in essence, a number of cells <td></td> are repeated (but not all)
# so that the table which is expected to contain 118 *2 = 236 cells contains something like 299 instead.
# To be able to ingest such a table, we need to detect and get rid of the duplicated td cells.
# My approach is based on the fact that duplicated td cells always lack any CSS attributes, whereas the cells
# we want to keep always specify align (either left or right).
td_nodes <-
read_html(elemental_properties$url[k]) %>%
html_nodes("table") %>% extract2(1) %>%
html_nodes("table") %>% extract2(8) %>%
html_nodes("td")
# The formatting of periodictable.com's HTML is unfortunately not consistent across the property pages
# On most, the label and value are inside td blocks neatly aligned left or right, but on other property
# pages the value part is repeated inside free-standing td blocks lacking any CSS attributes.
# To distinguish between these two cases (i.e., remove the duplicate values), I check for the align attr:
check_htmltable_align <-
td_nodes %>%
# here we are just checking for the existence of "align" attribute
# will return NA for the td cells that lack align attribute, and those are the duplicates we want to remove
html_attr("align")
if (any(is.na(check_htmltable_align))) {
# if it contains any NAs, print a message (for easier debugging)
# example: https://periodictable.com/Properties/A/RTECSNumber.an.html
message(paste0("Detected funky formatting on property page ", elemental_properties$property[k]))
}
# record the full table, element name/value pairs as a single vector
property_table <-
td_nodes[which(!is.na(check_htmltable_align))] %>% html_text()
property <- structure(data.frame(
# just the element names
Name =
property_table[td_nodes[which(!is.na(check_htmltable_align))] %>% html_attr("align") == "right"],
# the value of the current property (this column is renamed below)
value =
property_table[td_nodes[which(!is.na(check_htmltable_align))] %>% html_attr("align") == "left"]),
.Names = c("Name", elemental_properties$sanitized[k]))
## NOTE: You CANNOT rely on the order of the element names staying consistent across property pages,
## in particular not on "funky" property pages. Learned this the hard way.
# So here we use what should be a reliable way of growing our properties df with each new property:
properties_raw <- left_join(properties_raw, property, by = "Name")
}
properties_raw <- properties_raw %>% arrange(as.numeric(Atomic_Number))
# save this dataframe to file
# (it is not large, but re-scraping the contents takes time)
save(properties_raw, file = here::here("inst", "extdata", "periodicdata-raw.rdata"))
write_csv(properties_raw, here::here("inst", "extdata", "periodicdata-raw.csv"))
} else {
load(file = here::here("inst", "extdata", "periodicdata-raw.rdata"))
}
At the present time, the following property pages had formatting that deviated from
the others, as detected by our logic in the previous chunk:
Reading property page (33 of 81): Ionization Energies
Detected funky formatting on property page Ionization Energies
Reading property page (34 of 81): DOT Hazard Class
Detected funky formatting on property page DOT Hazard Class
Reading property page (36 of 81): RTECS Number
Detected funky formatting on property page RTECS Number
Reading property page (39 of 81): Names of Allotropes
Detected funky formatting on property page Names of Allotropes
Reading property page (46 of 81): Discovery
Detected funky formatting on property page Discovery
Reading property page (49 of 81): CID Number
Detected funky formatting on property page CID Number
Reading property page (69 of 81): Lattice Angles
Detected funky formatting on property page Lattice Angles
Reading property page (70 of 81): Lattice Constants
Detected funky formatting on property page Lattice Constants
Reading property page (79 of 81): Known Isotopes
Detected funky formatting on property page Known Isotopes
Reading property page (80 of 81): Stable Isotopes
Detected funky formatting on property page Stable Isotopes
Reading property page (81 of 81): Isotopic Abundances
Detected funky formatting on property page Isotopic Abundances
Extend the dataset with more properties
If you want to add more elemental properties to the data, that should be done at this point,
i.e, before we split quantities into magnitudes and units (see below).
So far, we have extended the original dataset with the following properties:
Autoignition_Point. From periodictable.com. Has since been removed from the website.Flashpoint. From periodictable.com. Has since been removed from the website.Gmelin_Number. From periodictable.com. Has since been removed from the website.Heat_of_Combustion. From periodictable.com. Has since been removed from the website.Production and Price. Global annual production and price of the elements, from @Vesborg2012.
property <- structure(data.frame(matrix(c(
# Name # Autoignition_Point
"Hydrogen", "535.5°C",
"Lithium", "179.°C",
"Carbon", "485.°C",
"Sodium", "115.°C",
"Magnesium", "472.°C",
"Aluminum", "400.°C",
"Silicon", "150.°C",
"Phosphorus", "300.°C",
"Potassium", "440.°C",
"Titanium", "250.°C",
"Chromium", "400.°C",
"Iron", "100.°C",
"Zinc", "460.°C",
"Cadmium", "250.°C",
"Tin", "630.°C",
"Tellurium", "340.°C",
"Samarium", "150.°C",
"Thorium", "130.°C"),
ncol = 2, byrow = T)),
.Names = c("Name", "Autoignition_Point"))
properties_raw <- left_join(properties_raw, property, by = "Name")
property <- structure(data.frame(matrix(c(
# Name # Flashpoint
"Hydrogen", "-18.°C",
"Sodium", "4.°C",
"Magnesium", "500.°C",
"Aluminum", "645.°C",
"Phosphorus", "30.°C",
"Bromine", "-18.°C"),
ncol = 2, byrow = T)),
.Names = c("Name", "Flashpoint"))
properties_raw <- left_join(properties_raw, property, by = "Name")
property <- structure(data.frame(matrix(c(
# Name # Heat_of_Combustion
"Lithium", "-298.J/(Kg K)",
"Carbon", "-393.J/(Kg K)",
"Magnesium","-668.J/(Kg K)",
"Silicon", "-9055.J/(Kg K)",
"Potassium","-182.J/(Kg K)",
"Calcium", "-990.J/(Kg K)",
"Germanium","-536.J/(Kg K)"),
ncol = 2, byrow = T)),
.Names = c("Name", "Heat_of_Combustion"))
properties_raw <- left_join(properties_raw, property, by = "Name")
property <- structure(data.frame(matrix(c(
# Name # Gmelin_Number
"Hydrogen", "Gmelin3",
"Neon", "Gmelin16211",
"Sodium", "Gmelin563",
"Magnesium", "Gmelin16207",
"Aluminum", "Gmelin16248",
"Silicon", "Gmelin61188",
"Phosphorus", "Gmelin8863",
"Argon", "Gmelin821",
"Potassium", "Gmelin15203",
"Helium", "Gmelin16294",
"Calcium", "Gmelin16277",
"Scandium", "Gmelin16304",
"Titanium", "Gmelin16313",
"Vanadium", "Gmelin49047",
"Chromium", "Gmelin16274",
"Manganese", "Gmelin699",
"Iron", "Gmelin6845",
"Cobalt", "Gmelin49017",
"Nickel", "Gmelin16229",
"Copper", "Gmelin16269",
"Lithium", "Gmelin30",
"Zinc", "Gmelin16321",
"Gallium", "Gmelin16287",
"Germanium", "Gmelin93634",
"Arsenic", "Gmelin16247",
"Selenium", "Gmelin8872",
"Bromine", "Gmelin1182",
"Krypton", "Gmelin16201",
"Rubidium", "Gmelin16244",
"Strontium", "Gmelin16302",
"Yttrium", "Gmelin16319",
"Beryllium", "Gmelin16256",
"Zirconium", "Gmelin16322",
"Niobium", "Gmelin16213",
"Molybdenum", "Gmelin16205",
"Ruthenium", "Gmelin16241",
"Rhodium", "Gmelin16245",
"Palladium", "Gmelin16239",
"Silver", "Gmelin1253",
"Cadmium", "Gmelin16276",
"Indium", "Gmelin1081",
"Boron", "Gmelin16262",
"Tin", "Gmelin16300",
"Tellurium", "Gmelin1153",
"Iodine", "Gmelin1160",
"Xenon", "Gmelin1125",
"Cesium", "Gmelin15188",
"Barium", "Gmelin16266",
"Cerium", "Gmelin16275",
"Praseodymium","Gmelin16238",
"Carbon", "Gmelin8868",
"Neodymium", "Gmelin16212",
"Samarium", "Gmelin16301",
"Europium", "Gmelin15985",
"Gadolinium", "Gmelin16286",
"Terbium", "Gmelin16311",
"Dysprosium", "Gmelin16278",
"Holmium", "Gmelin16291",
"Erbium", "Gmelin16280",
"Thulium", "Gmelin16307",
"Nitrogen", "Gmelin150",
"Ytterbium", "Gmelin16014",
"Lutetium", "Gmelin16202",
"Hafnium", "Gmelin16293",
"Tantalum", "Gmelin16312",
"Tungsten", "Gmelin16317",
"Rhenium", "Gmelin16243",
"Osmium", "Gmelin16234",
"Iridium", "Gmelin16298",
"Platinum", "Gmelin79607",
"Gold", "Gmelin16246",
"Oxygen", "Gmelin485",
"Mercury", "Gmelin1623",
"Thallium", "Gmelin16308",
"Lead", "Gmelin16240",
"Bismuth", "Gmelin16267",
"Radon", "Gmelin16242",
"Radium", "Gmelin40437",
"Fluorine", "Gmelin16281",
"Thorium", "Gmelin16314",
"Uranium", "Gmelin16315",
"Plutonium", "Gmelin40432"),
ncol = 2, byrow = T)),
.Names = c("Name", "Gmelin_Number"))
properties_raw <- left_join(properties_raw, property, by = "Name")
# Note, I have rather lazily entered year 2010 for *all* values from Vesborg,
# despite that paper noting that some values were taken from previous years.
# If time permits, go back and check @Vesborg2012 and update year field in
# those cases.
vesborg2012 <- tibble::tribble(
~Symbol, ~Production, ~Price, ~value, ~year, ~mb, ~comment,
"Fe", 12.4 , -0.24, 12.1, 2010, "M", "",
"Cl", 10.7 , NA, NA, 2010, "", "NaCl",
"N", 11.1 , 0.8 , 11.9, 2010, "M", "NH3 only",
"Na", 11.0 , NA, NA, 2010, "M", "NaCl",
"O", 10.9 , -1.4 , NA, 2010, "M", "O2 only",
"H", 10.8 , 0.9 , 10.7, 2010, "M", "H2 only",
"S", 10.8 , -0.7 , 10.1, 2010, "B", "Oil by-product",
"Al", 10.6 , 0.3 , 10.9, 2010, "M", "",
"Ca", 10.5 , NA, NA, 2010, "M", "Ca compounds",
"P", 10.5 , -0.28, 10.8, 2010, "M", "Phosphate rock",
"K", 10.4 , NA, NA, 2010, "M", "K compounds",
"Cr", 10.3 , 0.4 , 10.7, 2010, "M", "",
"Cu", 10.2 , 0.9 , 11.1, 2010, "M", "",
"Mn", 10.1 , 0.6 , 10.7, 2010, "M", "",
"Zn", 10.1 , 0.26, 10.3, 2010, "M", "",
"Ba", 9.9 , NA, NA, 2010, "M", "Mostly BaSO4",
# "Mg", 9.8 , 0.5 , 10.3, 2010, "M", "Mg compounds",
# "Mg", 8.9 , NA, NA, 2010, "", "Mg metal",
# I have joined Mg compounds and Mg metal from Vesborg2012 into a single row
"Mg", 9.85, 0.5 , NA, 2010, "M", "Metallic and compounds",
"Si", 9.8 , 0.4 , 10.2, 2010, "M", "Si only, no SiO2",
"Pb", 9.6 , 0.28, 9.9, 2010, "MB", "Also a by-product of Zn",
"Ar", 9.5 , -0.21, 9.3, 2010, "M", "",
"F", 9.4 , NA, NA, 2010, "M", "Mostly CaF2",
# "Ti", 9.4 , 1.4 , 10.4, 2010, "M", "TiO2 + FeTiO3",
# "Ti", 8.1 , NA, NA, 2010, "", "Ti metal",
# I have joined Ti compounds and Ti metal from Vesborg2012 into a single row
"Ti", 9.42, 1.4 , NA, 2010, "", "Metallic and oxides",
"B", 9.2 , 1.4 , 10.6, 2010, "M", "B compounds",
"C", 9.2 , 0.18, 9.4, 2010, "M", "Graphite only",
"Ni", 9.1 , 1.3 , 10.4, 2010, "M", "",
"Zr", 8.9 , 0.5 , 9.4, 2010, "M", "Mostly ZrO2",
"Br", 8.8 , 0.3 , NA, 2010, "M", "",
"Sr", 8.6 , 0.8 , 9.4, 2010, "M", "Sr compounds",
"Mo", 8.4 , 1.5 , 9.9, 2010, "BM", "By-product Cu",
"Sn", 8.4 , 1.3 , 9.4, 2010, "M", "",
"Sb", 8.2 , 1.2 , 9.4, 2010, "M", "",
"Co", 7.9 , 1.6 , 9.5, 2010, "B", "",
"Nb", 7.8 , 2.0 , 9.8, 2010, "M", "",
"W", 7.8 , 2.8 , 10.6, 2010, "M", "",
"V", 7.8 , 1.4 , 9.2, 2010, "M", "",
"U", 7.7 , 2.0 , 9.7, 2010, "M", "Radioactive",
"Ce", 7.6 , 2.0 , 9.6, 2010, "M", "With La and Nd",
"As", 7.6 , 0.21, 7.8, 2010, "B", "",
"La", 7.5 , 2.0 , 9.5, 2010, "M", "With Ce and Nd",
"I", 7.5 , 1.4 , NA, 2010, "M", "",
"Th", NA, 1.9 , NA, 2010, "M", "Radioactive",
"Ag", 7.4 , 3.0 , 10.4, 2010, "B", "Mostly by-product, approx. 30% primary product",
"Li", 7.4 , 1.5, 8.9, 2010, "M", "",
"Nd", 7.3 , 2.5, 9.8, 2010, "M", "With La and Ce",
"He", 7.3 , 0.5, 7.8, 2010, "B", "By-product of natural gas",
"Cd", 7.3 , 0.4, 7.7, 2010, "B", "By-product Zn",
"Y", 6.8 , 2.3, 9.1, 2010, "B", "By-product HREEs",
"Bi", 6.8 , 1.5, 8.3, 2010, "B", "By-product Pb",
"Pr", 6.7 , 2.5, 9.2, 2010, "B", "By-product LREEs",
"Ne", 6.7 , 2.5, NA, 2010, "B", "By-product Ar",
"Au", 6.4 , 4.7, 11.1, 2010, "M", "",
"Sm", 6.4 , 2.2, 8.6, 2010, "B", "By-product LREEs",
"Se", 6.4 , 2.1, 8.5, 2010, "B", "By-product Cu",
"Gd", 6.3 , 2.3, 8.6, 2010, "B", "By-product HREEs",
"Hg", 6.3 , 1.7, 8.0, 2010, "M", "",
"Dy", 6.0 , 3.4, 9.4, 2010, "M", "Co-product Eu",
"Ta", 6.0 , 2.7, 8.7, 2010, "M", "",
"Be", 5.8 , NA, NA, 2010, "M", "",
"Er", 5.8 , 2.5, NA, 2010, "B", "By-product HREE",
"In", 5.7 , 2.9, 8.6, 2010, "B", "By-product Zn",
"Te", 5.7 , 2.5, 8.2, 2010, "B", "By-product Cu",
"Yb", 5.7 , 2.7, NA, 2010, "B", "By-product HREE",
"Eu", 5.6 , 3.8, 9.4, 2010, "M", "Co-product Dy",
"Kr", 5.6 , 2.6, NA, 2010, "B", "By-product Ar",
"Ho", 5.4 , 3.5, NA, 2010, "B", "By-product HREE",
"Tb", 5.3 , 3.7, 9.0, 2010, "B", "By-product HREE",
"Pd", 5.3 , 4.3, 9.6, 2010, "B", "By-product Ni, Pt",
"Pt", log10(2e5), log10(30e3), NA, 2020, "M", "@Chatenet2022",
"Ge", 5.1 , 3.2, 8.3, 2010, "B", "By-product Zn",
"Ga", 5.1 , 2.9, 7.9, 2010, "B", "By-product Al",
"Tm", 5.1 , 4.0, NA, 2010, "B", "By-product HREE",
"Lu", 5.0 , 4.0, NA, 2010, "B", "By-product HREE",
"Hf", 4.9 , 2.8, 7.7, 2010, "B", "By-product Zr",
"Re", 4.7 , 3.5, 8.2, 2010, "B", "By-product Mo, (Cu)",
"Xe", 4.6 , 3.4, 8.0, 2010, "B", "By-product Ar",
"Rh", 4.4 , 4.7, 9.1, 2010, "B", "By-product Pt",
"Ru", 4.4 , 3.7, 8.1, 2010, "B", "By-product Pt",
"Ir", log10(7.7e3), log10(203e3), NA, 2018, "B", "By-product of Pt. Production 2018 from @Minke2021, price 2021 from @Chatenet2022",
"Tl", 4 , NA, NA, 2010, "B", "By-product Cu, Zn",
"Cs", 4 , NA, NA, 2010, "M", "Co-product Li, Ta, Pb",
"Os", 3.7 , 4.5, NA, 2010, "B", "By-product Pt",
"Rb", 3.6 , 4.6, 8.2, 2010, "M", "Co-product Cs, Ta, Li",
"Sc", 3.3 , 3.1, NA, 2010, "B", "By-product")
# for purposes of quality control (easier check for duplicates, etc.) let's add
# atomic number
vesborg2012 %<>%
left_join(properties_raw %>% select(Symbol, Atomic_Number), by = "Symbol")
As a crude check that we have not introduced any (grave) transcription errors
in this data, we will now recreate some of the plots from the source paper.
p.data <-
vesborg2012 %>%
select(Symbol, Atomic_Number, Production)
ggplot(p.data, aes(y = 10^Production, x = as.numeric(Atomic_Number))) +
geom_line(colour = "gray", size = 0.75) +
geom_point() +
# formatting this axis title nicely required scales::
# https://stackoverflow.com/questions/10762287/how-can-i-format-axis-labels-with-exponents-with-ggplot2-and-scales
scale_y_log10(
name = "Global annual production/(kg/yr)",
breaks = scales::trans_breaks("log10", function(x) 10^x),
# note, for this trans_format() to work as intended that data needs to in absolute numbers
labels = scales::trans_format("log10", math_format(10^.x))) +
scale_x_continuous(name = "Atomic number, Z") +
geom_text_repel(
aes(label = Symbol))
Ok, that looks quite like figure 1 in @Vesborg2012. I find no obvious discrepances.
Let's also recreate the only figure in the SI (log(price) vs log(production)).
p.data <-
vesborg2012 %>%
select(Symbol, Atomic_Number, Production, Price)
ggplot(p.data, aes(y = 10^Price, x = 10^Production)) +
# geom_line(colour = "gray", size = 0.75) +
geom_point() +
scale_x_log10(
name = "Global annual production/(kg/yr)",
breaks = scales::trans_breaks("log10", function(x) 10^x),
labels = scales::trans_format("log10", math_format(10^.x))) +
scale_y_log10(
name = "Price/(USD/kg)",
breaks = scales::trans_breaks("log10", function(x) 10^x),
labels = scales::trans_format("log10", math_format(10^.x))) +
geom_text_repel(
aes(label = Symbol))
Ok, that looks about right.
# note that the production and price values above are log10(),
# which is how they are presented in @Vesborg2012
# unfortunately I have not found the source values anywhere (haven't asked the authors yet, though)
# this means a quite significant loss of accuracy in the following conversion, but
# we do it anyway because I want to be explicit regarding the units
vesborg2012 %<>%
mutate(Production = ifelse(
is.na(Production),
"",
# formatC ensures that the number is decimal and not scientific notation even if larger number,
# because scientific notation messes up our values-units split algo below (yes, that should be improved...)
paste(formatC(10^Production, format = "fg"), "kg/yr"))) %>%
mutate(Price = ifelse(
is.na(Price),
"",
paste(formatC(10^Price, format = "fg"), "USD/kg")))
# now let's incorporate production and price into the properties dataset
properties_raw <- left_join(
properties_raw,
vesborg2012 %>% select(Symbol, Production, Price),
by = "Symbol")
Split quantities and units into separate dataframes
We now have several tasks ahead of us before the data is usable:
- what type of data each property is (character string, dimensionless quantity, or quantity with unit),
- determine and identify what part of each property value is a quantity and which is a unit,
- extract quantity and unit separately from each datapoint and save to
values and units dataframe,
respectively.
Take care to that the dataframes values and units have the same dimensions
(rows/columns). The point behind splitting each value up is to be able to treat
the quantity numerically, plot it, etc. It's therefore crucial that the dataframes
stay the same size, so that once a quantity is read from values, we can trust that
its unit is found in the corresponding cell in units.
# create empty dataframes to hold quantities and units, respectively
values <- units <-
data.frame(matrix(
nrow = dim(properties_raw)[1],
ncol = dim(properties_raw)[2],
dimnames = list(seq(1, dim(properties_raw)[1]), names(properties_raw)),
byrow = TRUE))
# The dataset on periodictable.com also has footnotes.
# These are not saved in the current implementation of `periodicdata`
Next, we list all the currently existing columns so that we can specify a type for
each one. The type will determine how we handle unit extraction etc.
# to re-create the matrix below, get started like this:
# cat(paste0(names(properties_raw), collapse = ",\n"))
property_types <- structure(data.frame(matrix(c(
"Name", "character",
"Symbol", "character",
"Atomic_Number", "unitless number",
"Atomic_Weight", "unitless number",
"Density", "number with unit",
"Melting_Point", "number with unit",
"Boiling_Point", "number with unit",
"Phase", "character",
"Absolute_Melting_Point", "number with unit",
"Absolute_Boiling_Point", "number with unit",
# contains converted values in parentheses
"Critical_Pressure", "number with unit",
"Critical_Temperature", "number with unit",
"Heat_of_Fusion", "number with unit",
"Heat_of_Vaporization", "number with unit",
"Specific_Heat", "number with unit",
"Adiabatic_Index", "character",
"Neel_Point", "number with unit",
"Thermal_Conductivity", "number with unit",
"Thermal_Expansion", "number with unit",
"Density_Liquid", "number with unit",
"Molar_Volume", "unitless number",
"Brinell_Hardness", "number with unit",
"Mohs_Hardness", "number with unit",
"Vickers_Hardness", "number with unit",
"Bulk_Modulus", "number with unit",
"Shear_Modulus", "number with unit",
"Young_Modulus", "number with unit",
"Poisson_Ratio", "unitless number",
"Refractive_Index", "unitless number",
"Speed_of_Sound", "number with unit",
"Valence", "unitless number",
"Electronegativity", "unitless number",
"ElectronAffinity", "number with unit",
"DOT_Hazard_Class", "unitless number",
"DOT_Numbers", "unitless number",
"RTECS_Number", "character",
"Alternate_Names", "character",
"Block", "character",
"Group", "character",
"Period", "character",
"Series", "character",
"Electron_Configuration", "character",
"Color", "character",
"Gas_phase", "character",
"CAS_Number", "character",
"CID_Number", "character",
"Electrical_Type", "character",
"Electrical_Conductivity", "number with unit",
"Resistivity", "number with unit",
"Superconducting_Point", "unitless number",
"Magnetic_Type", "character",
"Curie_Point", "number with unit",
"Mass_Magnetic_Susceptibility", "unitless number",
"Molar_Magnetic_Susceptibility", "unitless number",
"Volume_Magnetic_Susceptibility", "unitless number",
"Percent_in_Universe", "number with unit",
"Percent_in_Sun", "number with unit",
"Percent_in_Meteorites", "number with unit",
"Percent_in_Earths_Crust", "number with unit",
"Percent_in_Oceans", "number with unit",
"Percent_in_Humans", "number with unit",
"Atomic_Radius", "number with unit",
"Covalent_Radius", "number with unit",
"Van_der_Waals_Radius", "number with unit",
"Space_Group_Name", "character",
"Space_Group_Number", "unitless number",
# quantity includes "Stable" and num + unit. Units are a mix of time units: ms, s, m, d, y
"HalfLife", "number with unit",
# quantity includes "Stable" and num + unit. Units are a mix of time units: ms, s, m, d, y
"Lifetime", "number with unit",
"Decay_Mode", "character",
"Quantum_Numbers", "character",
"Neutron_Cross_Section", "unitless number",
"Neutron_Mass_Absorption", "unitless number",
"Autoignition_Point", "number with unit",
"Heat_of_Combustion", "number with unit",
"Gmelin_Number", "character",
"Production", "number with unit",
"Price", "number with unit"),
ncol = 2, byrow = T)), .Names = c("Data", "Type"))
# running this loop usually produces warnings "NAs introduced by coercion". We can disregard those.
for (k in 1:dim(property_types)[1]) {
# clean up or convert the quantity or unit depending on type of data in each property
# type is defined manually in the matrix in the chunk above
if (property_types$Type[k] == "character") {
values[, which(names(values) == property_types$Data[k])] <-
properties_raw %>%
pull(property_types$Data[k]) %>%
gsub("\\[note\\]", "", x = .) %>%
gsub("^N/A$", "", x = .) %>%
gsub("^None$", "", x = .)
} else if (property_types$Type[k] == "unitless number") {
values[, which(names(values) == property_types$Data[k])] <-
properties_raw %>%
pull(property_types$Data[k]) %>%
gsub("\\[note\\]", "", x = .) %>%
gsub("×10", "E", x = .) %>%
# final step, make sure the column is numeric
as.numeric()
} else if (property_types$Type[k] == "number with unit") {
values[, which(names(values) == property_types$Data[k])] <-
properties_raw %>%
pull(property_types$Data[k]) %>%
gsub("\\[note\\]", "", x = .) %>%
# replace "Stable" with Inf (only found in HalfLife, Lifetime)
gsub("^Stable$", "Inf", x = .) %>%
# replace ×10 with proper power notation
gsub("×10", "E", x = .) %>%
# extract the numeric quantity (including the string "Inf")
str_extract("Inf|[-×\\.0-9E]+") %>%
# final step, make sure this column is numeric
as.numeric()
units[, which(names(units) == property_types$Data[k])] <-
properties_raw %>%
pull(property_types$Data[k]) %>%
gsub("N/A", "", x = .) %>%
gsub("\\[note\\]", "", x = .) %>%
# replace "Stable" with Inf (only found in HalfLife, Lifetime)
gsub("^Stable$", "Inf", x = .) %>%
# remove any numbers+units within parentheses
gsub("\\([.0-9]+\\s[A-Za-z]+\\)", "", x = .) %>%
# remove the numeric quantity (including the string "Inf")
# take care not to use gsub here, since numbers could also be part of the unit (e.g., "g/cm3")
sub("Inf|[-×\\.0-9E]+", "", x = .) %>%
# remove any leading and trailing whitespace
str_trim(side="both") %>%
# replace empty string with NA
gsub("^$", NA, x = .)
} else {
message("This should never happen!")
}
}
Cleaning up and normalising the units
Now that we have separated the units and the quantities into separate dataframes, let's have a look at the units, to see what fixing is needed.
# Units by property
sapply(units, unique)
As you can see, some properties (for example HalfLife) use more than one unit.
This is problematic, since we will only be plotting against one y-axis, not several.
So we will have to convert all such occurrences to their base SI units, which means
we have to take the numerical conversion of the quantity into consideration as well.
Let's do it.
# All units in the dataset as a simple vector
cat(paste(sort(unique(unlist(sapply(units, unique)))), collapse = "\n"))
## % (percent)
## °C => K
## d => s (day to second)
## g/cm3 => kg/m^3
## g/l => kg/m^3
## GPa => Pa
## h => s (hour to second)
## J/(Kg K) => J/(kg K) (fix wrong case)
## J/(kg K)
## K (Kelvin)
## K-1 (per Kelvin)
## kJ/mol (kilojoule per mol)
## m => s (only exists in HalfLife and Lifetime property, so this is minutes not metres)
## m Ω => Ohm m (electrical resistivity, Ohm metre) # avoid Ω, causes non-ASCII char warning on package build
## m/s (metre per second)
## MPa => Pa
## ms => s (millisecond to second)
## pm => m (picometre to metre)
## S/m (Siemens per metre)
## W/(m K) (Watt per metre-Kelvin)
## y => s (year to second)
pcf <- structure(data.frame(matrix(c(
# from # to # conversion factor
"°C", "K", "+273.15",
"d", "s", "86400",
"g/cm3", "kg/m^3", "1E3",
"g/l", "kg/m^3", "1",
"GPa", "Pa", "1E9",
"h", "s", "3600",
"J/(Kg K)", "J/(kg K)", "1",
# minutes to seconds
"m", "s", "60",
"m Ω", "Ohm m", "1",
"MPa", "Pa", "1E6",
"ms", "s", "1E-3",
# be careful that this substitution doesn't precede "mins to secs"
"pm", "m", "1E-12",
"y", "s", "3.154E7"),
ncol = 3, byrow = T)),
.Names = c("pattern", "convert", "factor"))
# Replace the units and convert values according to pcf
for (k in 1:dim(units)[2]) {
# if the entire column is NA, move to the next one
if (all(is.na(units[, k]))) {next}
for (i in 1:dim(units)[1]) {
# jump to the next cell immediately if unit is empty
if (is.na(units[i, k])) {next}
# for each cell, compare the unit to pcf$pattern,
# and if they match, replace it with pcf$convert
# and apply the conversion factor on the value
# match() returns the position of the match
# in pcf$pattern, or NA if no match was found
hit <- match(units[i, k], pcf$pattern)
if (!is.na(hit)) {
# go ahead and replace unit and convert value
units[i, k] <- pcf$convert[hit]
# had to find a way to handle the odd addition operation,
# opted to do it with a string operation and this if-else
if (substr(pcf$factor[hit], 1, 1) %in% c("+", "-")) {
if (substr(pcf$factor[hit], 1, 1) == "-") {
values[i, k] <-
values[i, k] + as.numeric(pcf$factor[hit])
} else {
values[i, k] <-
values[i, k] + as.numeric(sub("^\\+", "", pcf$factor[hit]))
}
} else {
# not addition/subtraction operation
values[i, k] <-
values[i, k] * as.numeric(pcf$factor[hit])
}
}
}
}
Proof-of-concept plots and conclusion
ggplot() +
geom_point(
data = values %>% filter(!is.na(Density)),
aes(Atomic_Number, Density)) +
geom_path(
data = values %>% filter(!is.na(Density)),
aes(Atomic_Number, Density)) +
geom_point(
data = values %>% filter(!is.na(Density)) %>% filter(Group == "18"),
aes(Atomic_Number, Density),
colour = "red") +
geom_text_repel(
data = values %>% filter(!is.na(Density)) %>% filter(Group == "18"),
aes(Atomic_Number, Density, label = Symbol),
colour = "red") +
scale_y_log10() +
labs(
x = "Atomic number",
y = paste0(
"Density",
"/(",
unique(units$Density[which(!is.na(units$Density))]),
")"))
That's not a bad plot, but what if we could use the periodic table (think IUPAC's layout)
and plot the data on top of that, for example using varying degrees of fill colour,
or some such? Or even to print a typical periodic table by itself, just using the
dataset and ggplot2 (and possibly TikZ, for really nice PDF output).
To do that, we need a final preparatory step.
The IUPAC periodic table places the elements next to each other, organised in
rows (periods) and columns (groups) on a two-dimensional grid. Obviously, each
element's position on this plot is completely specified by its group and period.
values$IUPAC_Period <- values$Period %>% as.numeric()
values$IUPAC_Group <- values$Group %>% as.numeric()
values$IUPAC_Number <- values$Atomic_Number %>% as.character()
values$IUPAC_Series <- values$Series
# to maintain same dimensions as values dataframe
units$IUPAC_Period <- NA
units$IUPAC_Group <- NA
units$IUPAC_Number <- NA
# Fix coordinates for lanthanides and actinoids
# lanthanoids 57-71: Period = 8, Group = seq(4, 18)
values$IUPAC_Period[seq(57, 71)] <- 8.5
values$IUPAC_Group[seq(57, 71)] <- seq(4, 18)
values[which(values$Symbol == "Lu"), which(names(values) == "IUPAC_Series")] <- "Lanthanide"
# actinoids 89-103: Period = 9, Group = seq(4, 18)
# increase Period slightly to increase the gap up to the transition block
values$IUPAC_Period[seq(89, 103)] <- 9.5
values$IUPAC_Group[seq(89, 103)] <- seq(4, 18)
values[which(values$Symbol == "Lr"), which(names(values) == "IUPAC_Series")] <- "Actinide"
# add placeholder "element" for lanthanoids and actinoids after Ba and Ra, respectively
# (this is just to conform to the IUPAC table layout)
values[nrow(values) + 1,
which(names(values) %in% c("Name", "IUPAC_Period", "IUPAC_Group", "IUPAC_Number", "IUPAC_Series"))] <-
data.frame("lanthanoids", 6, 3, "57–71", "Lanthanide")
values[nrow(values) + 1,
which(names(values) %in% c("Name", "IUPAC_Period", "IUPAC_Group", "IUPAC_Number", "IUPAC_Series"))] <-
data.frame("actinoids", 7, 3, "89–103", "Actinide")
# add corresponding empty rows to units df to maintain the same dimensions:
units[nrow(units) + 1, ] <- NA
units[nrow(units) + 1, ] <- NA
By changing the IUPAC_Period and IUPAC_Group assignments it's possible to adjust the layout of the periodic table according to taste. Here's a reproduction of the IUPAC table of elements:
p.periodictable_ggplot2 <-
ggplot() +
# lanthanoids and actinoids background colour drawn first
geom_point(
data = values %>% filter(IUPAC_Series == "Lanthanide"),
size = 16,
shape = 15,
colour = "#cfc0c7",
aes(y = IUPAC_Period, x = IUPAC_Group)) +
geom_point(
data = values %>% filter(IUPAC_Series == "Actinide"),
size = 16,
shape = 15,
colour = "#a58394",
aes(y = IUPAC_Period, x = IUPAC_Group)) +
# boxes for all elements
geom_point(
data = values,
# size 16 leaves no gaps between the element boxes at fig.width=9,fig.height=5.25
size = 16,
shape = 0,
aes(y = IUPAC_Period, x = IUPAC_Group)) +
# element symbol
geom_text(
data = values,
size = 3,
colour = "black",
fontface = "bold",
aes(label = Symbol, y = IUPAC_Period - 0.1, x = IUPAC_Group)) +
# atomic number
geom_text(
data = values,
size = 1.8,
colour = "black",
aes(label = IUPAC_Number, y = IUPAC_Period - 0.32, x = IUPAC_Group)) +
# element name
geom_text(
data = values,
size = 1.6,
colour = "black",
aes(label = tolower(Name), y = IUPAC_Period + 0.12, x = IUPAC_Group)) +
# atomic weight
geom_text(
data = values,
size = 1.6,
colour = "black",
aes(label = Atomic_Weight, y = IUPAC_Period + 0.32, x = IUPAC_Group)) +
# period labels
geom_text(
data = values %>% filter(IUPAC_Period < 8),
size = 1.7,
colour = "black",
aes(label = IUPAC_Period, y = IUPAC_Period),
x = 0.38) +
# group labels # positions manually adjusted
geom_text(
data = data.frame(
y = c(0.38, 1.38, rep(3.38, 10), rep(1.38, 5), 0.38),
x = seq(1,18)),
size = 1.7,
colour = "black",
aes(label = x, x = x, y = y)) +
# table title # position manually adjusted
annotate(
"text",
x = 9, y = 0.6, vjust = 0,
size = 4.5,
fontface = "bold",
label = "Periodic Table of the Elements") +
# table subtitle # position manually adjusted
annotate(
"text",
x = 9, y = 0.74, vjust = 1,
size = 2.6,
label = "Created with the periodicdata package by solarchemist.se") +
scale_x_continuous(
breaks = seq(
min(values$IUPAC_Group),
max(values$IUPAC_Group)),
limits = c(
min(values$IUPAC_Group) - 1,
max(values$IUPAC_Group) + 1),
expand = c(0,0)) +
scale_y_continuous(
trans = "reverse",
breaks = seq(
min(values$IUPAC_Period),
max(values$IUPAC_Period)),
limits = c(
max(values$IUPAC_Period) + 1,
min(values$IUPAC_Period) - 1.5),
expand = c(0,0)) +
theme(
panel.grid.major = element_blank(),
panel.grid.minor = element_blank(),
plot.margin = unit(c(0, 0, -0.85, -0.85), "line"),
axis.title = element_blank(),
axis.text = element_blank(),
axis.ticks = element_blank(),
# center (roughly) over transition metal block
legend.position = c(0.42, 0.91),
legend.justification = c(0.5, 1),
legend.direction = "horizontal",
# make the legend colourbar a little longer
legend.key.width = unit(2.5, "line"),
legend.title = element_blank(),
legend.background = element_rect(fill = "transparent"))
print(p.periodictable_ggplot2)
This is just an example of the kind of plot than can be achieved with ggplot2 and this dataset.
Note that the plot is quite sensitive to changes to its final output dimension --- such
changes may necessitate changes to the size parameter in the geom_point() layers
(these effectively set the size of the box that each element occupies).
Recreation of figure 3 from @Vesborg2012:
p.data <-
values %>%
select(Symbol, Atomic_Number, Production, Percent_in_Earths_Crust)
ggplot(p.data, aes(y = 1e3*Percent_in_Earths_Crust, x = Production)) +
geom_point() +
geom_smooth(method = "lm", formula = "y~x", se = TRUE, linetype ="dashed", fill = "gray") +
scale_x_log10(
name = "Global annual production/(kg/yr)",
breaks = scales::trans_breaks("log10", function(x) 10^x),
labels = scales::trans_format("log10", math_format(10^.x))) +
scale_y_log10(
name = "Crustal abundance/ppm",
breaks = scales::trans_breaks("log10", function(x) 10^x),
labels = scales::trans_format("log10", math_format(10^.x))) +
geom_text_repel(aes(label = Symbol))
Considering that this dataset and the paper use different crustal abundance sources,
the discrepancy is not very large. The overall trend is preserved (elements below the line
are "overexploited"; here we may note gold (for obvious reasons) and nitrogen (production is
sourced from the atmosphere, thus appears overtaxed compared to crustal abundance); and
elements above the line are conversely "underexploited", if such a thing exists).
legend.mock.box <- tibble::tribble(
~x, ~y, ~text, ~id, ~label,
3.2, 2.8, "", "box", "",
3.2, 2.8-0.25, "Mo", "text", "",
3.2, 2.8+0.03, "8.4", "text", "Annual production (log)",
3.2, 2.8+0.28, "1.5", "text", "Market price (log)")
ggplot() +
# production rates above 1 Mt/yr (note, data is in kg)
geom_point(
data = values %>% filter(Production > 1e9),
size = 16, shape = 15, # alpha = 0.5,
aes(y = IUPAC_Period, x = IUPAC_Group, colour = "megaton")) +
# production rates above 33 kt/yr (note, data is in kg)
geom_point(
data = values %>% filter(Production > 33e6 & Production <= 1e9),
size = 16, shape = 15, # alpha = 0.75,
aes(y = IUPAC_Period, x = IUPAC_Group, colour = "medium")) +
# production rates above 1 kt/yr (note, data is in kg)
geom_point(
data = values %>% filter(Production > 1e6 & Production <= 33e6),
size = 16, shape = 15, # alpha = 0.5,
aes(y = IUPAC_Period, x = IUPAC_Group, colour = "low")) +
# production rates less than 1 kt/yr (note, data is in kg)
geom_point(
data = values %>% filter(Production <= 1e6),
size = 16, shape = 15, # alpha = 0.5,
aes(y = IUPAC_Period, x = IUPAC_Group, colour = "rare")) +
# boxes for all elements
geom_point(
data = values,
# size 16 leaves no gaps between the element boxes at fig.width=9,fig.height=5.25
size = 16, shape = 0,
aes(y = IUPAC_Period, x = IUPAC_Group)) +
# element symbol
geom_text(
data = values,
size = 3,
colour = "black",
fontface = "bold",
aes(label = Symbol, y = IUPAC_Period - 0.24, x = IUPAC_Group)) +
# show atomic number range for La, Ac placeholder cells
geom_text(
data = values %>% filter(is.na(Atomic_Number)),
size = 2.2,
colour = "black",
aes(label = IUPAC_Number, y = IUPAC_Period - 0.24, x = IUPAC_Group)) +
# log(production)
geom_text(
data = values,
size = 2.4, colour = "black",
aes(
label = ifelse(is.na(Production), "", formatC(log10(Production), format="f", digits=1)),
y = IUPAC_Period + 0.04, x = IUPAC_Group)) +
# log(price)
geom_text(
data = values,
size = 2.4, colour = "black",
aes(
label = ifelse(is.na(Price), "", formatC(log10(Price), format="f", digits=1)),
y = IUPAC_Period + 0.28, x = IUPAC_Group)) +
# table title # position manually adjusted
annotate(
"text",
x = 9, y = 0.6, vjust = 0,
size = 4.5,
fontface = "bold",
label = "Production and economics of the chemical elements") +
# table subtitle # position manually adjusted
annotate(
"text",
x = 9, y = 0.74, vjust = 1,
size = 2.6,
label = "Based on Vesborg & Jaramillo, RSC Advances 2, 7933–7947 (2012)") +
scale_x_continuous(
breaks = seq(
min(values$IUPAC_Group),
max(values$IUPAC_Group)),
limits = c(
min(values$IUPAC_Group) - 1,
max(values$IUPAC_Group) + 1),
expand = c(0,0)) +
scale_y_continuous(
trans = "reverse",
breaks = seq(
min(values$IUPAC_Period),
max(values$IUPAC_Period)),
limits = c(
max(values$IUPAC_Period) + 1,
min(values$IUPAC_Period) - 1.5),
expand = c(0,0)) +
# a mock element box as legend
geom_point(
data = legend.mock.box %>% filter(id == "box"),
size = 14, shape = 0, aes(x = x, y = y)) +
geom_text(
data = legend.mock.box %>% filter(id != "box"),
colour = "black", size = 2.6,
aes(label = text, x = x, y = y)) +
geom_text_repel(
data = legend.mock.box %>% filter(id != "box"),
nudge_x = 1.75, size = 2.5, point.padding = 0.75, direction = "both",
aes(label = label, x = x, y = y)) +
scale_colour_manual(
name = "Global production",
guide = guide_legend(
override.aes = list(size = 3),
title.position = "top"),
# warning, alpha is not obeyed by legend keys - I don't know why
values = c(megaton = alpha("#07B248", 0.5), medium = alpha("#C4E2BF", 0.75), low = alpha("#F4B521", 0.5), rare = alpha("#ED2025", 0.5)),
labels = c(">1 Mt/yr", ">33 kt/yr", ">1 kt/yr", "<1 kt/yr")) +
theme(
panel.grid.major = element_blank(),
panel.grid.minor = element_blank(),
plot.margin = unit(c(0, 0, -0.85, -0.85), "line"),
axis.title = element_blank(),
axis.text = element_blank(),
axis.ticks = element_blank(),
# legend is manually centered atop the transition metal block
legend.position = c(0.45, 0.82),
legend.justification = c(0.5, 1),
legend.direction = "horizontal",
legend.key = element_rect(fill = NA, colour = NA),
# key.size had no effect on the legend key size! see guides() above instead
# key.width affects the horizontal spacing of the key, not the coloured box
legend.key.width = unit(2, "mm"),
legend.text = element_text(size = 10),
#legend.title = element_blank(),
legend.background = element_rect(fill = NA))
With the dataset fully assembled, all that remains is to save the values and units dataframes.
These dataframes are part of the exported data of this package.
# use_data() saves each dataframe to `data/` as an *.rda file
use_data(values, units, overwrite = TRUE)
write_csv(values, here("inst", "extdata", "periodicdata-values.csv"))
write_csv(units, here("inst", "extdata", "periodicdata-units.csv"))
Note: the file inst/extdata/periodicdata-raw.R is created by knitting this Rmd file:
purl(input = here("vignettes", "periodicdata.Rmd"), output = here("inst", "extdata", "periodicdata-raw.R"))
The following chunk is normally not evaluated. It's only used when we need to refresh
the demo periodic table that's included in the README.
# note that the doc/ directory is only created after running devtools::build_vignettes()
ggsave(
filename = here::here("doc", "periodictable-ggplot.png"),
plot = p.periodictable_ggplot2, width = ptable.width, height = ptable.height)
ggsave(
filename = here::here("doc", "periodictable-ggplot.pdf"),
plot = p.periodictable_ggplot2, width = ptable.width, height = ptable.height)
ggsave(
filename = here("doc", "periodictable-ggplot.svg"),
plot = p.periodictable_ggplot2, width = ptable.width, height = ptable.height)
References
chepec/periodicdata documentation built on July 18, 2023, 10:53 p.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.
Introduction
In this vignette we will demonstrate how the dataset was collected, by --- drumroll --- collecting it! We will also show how you can generate the familiar IUPAC periodic table from this dataset together with some ggplot2 code.
Almost all of the properties in this dataset were scraped from the
periodictable.com website which provides "up to date, curated data provided by Mathematica's
ElementData function from Wolfram Research, Inc." as HTML/CSS tables.
The Wolfram's ElementData() function,
in turn, lists multiple sources,
both scholarly and other websites.
Despite most of the data originating in Mathematica (by Wolfram Research),
the quality of the data on periodictable.com
website is actually rather poor, appearing to be suffering from whatever
conversion was applied from the original Mathematica format.
Obvious typos and other easy-to-spot errors in the scraped data have been corrected by yours truly in this dataset.
library(dplyr) library(magrittr) # extract2() library(httr) library(rvest) library(stringr) library(readr) library(tibble) library(knitr) library(usethis) library(ggplot2) library(ggrepel) library(scales) library(here)
options( digits = 7, width = 84, continue = " ", prompt = "> ", warn = 0, stringsAsFactors = FALSE) opts_chunk$set( dev = 'svg', #out.width = "100%", fig.width = 7.10, fig.height = 4.39, fig.align = 'center', echo = FALSE, eval = TRUE, cache = TRUE, collapse = TRUE, results = 'hide', message = FALSE, warning = FALSE, tidy = FALSE) # dont reset scrape_afresh here, it is reset automatically if # inst/extdata/periodicdata-raw.rdata is not found # this just sets it to the default FALSE scrape_afresh <- FALSE # periodic table dimensions (sensitive setting, layout will likely break if changed) ptable.width <- 9 ptable.height <- 5.25
# Starting from this URL, we crawl the page and collect the URLs to all property pages, along with the name of each property. # We put them together in a dataframe, so it is clear which URL belongs to which elemental property. element_data <- read_html("http://periodictable.com/Elements/001/data.html") elemental_properties <- data.frame( property = element_data %>% html_nodes(xpath = "//a") %>% html_text()) elemental_properties$links_raw <- element_data %>% html_nodes(xpath = "//a") %>% html_attr("href") # drop any rows that do not begin "../../" elemental_properties <- elemental_properties[grep( pattern = "^\\.\\./\\.\\./", x = elemental_properties$links_raw), ] # remove the leading ../.. (relative link signifier) elemental_properties$links_trail <- gsub( pattern = "^\\.\\./\\.\\./", replacement = "", x = elemental_properties$links_raw) # keep only the rows that begin with "Properties" elemental_properties <- elemental_properties[grep( pattern = "^Properties", x = elemental_properties$links_trail), ] # make full urls elemental_properties$url <- paste0( "http://periodictable.com/", elemental_properties$links_trail) # modify the URL so it sorts the table by atomic number # (this is a feature of periodictable.com, they offer lists sortered in different ways # for each property) elemental_properties$url <- sub( pattern = "\\.html$", replacement = ".an.html", x = elemental_properties$url) # we are done, go ahead and drop the no longer required columns # links_raw, links_trail elemental_properties <- elemental_properties[, -c(2,3)] # oh, and add a sanitised version of the property names elemental_properties$sanitized <- gsub( # replace spaces with underscores "\\s+", "_", # replace "%" with "percent" gsub("%", "Percent", # remove parentheses and dashes gsub("[()-]", "", # remove apostrophes gsub("'", "", elemental_properties$property)))) # NOTE: Melting point and Boiling point are duplicated because they are displayed twice # on the data page for Hydrogen, both under "Overview" and "Thermal properties". # Other properties could be too, so we should deduplicate the dataframe. elemental_properties <- unique(elemental_properties) # reset the row numbering row.names(elemental_properties) <- seq(1, dim(elemental_properties)[1])
Collect periodic table data from periodictable.com
Each property, e.g., density, will have a value for each element of the periodic table. This value is just a string, and depending on the type of the property, it may be just a number, or a quantity with a unit, or some text with various attributes. At this stage, we don't mind the internal structure of the value, we just want to prepare for collecting them.
# get the name of all the elements (two-step process) element_names <- read_html(elemental_properties$url[1]) %>% html_nodes("table") %>% extract2(1) %>% html_nodes("table") %>% extract2(8) %>% html_nodes("td") %>% html_text() # here we initialise the properties dataframe, which we will use later # to read in all the properties (building the frame using ..._join() calls) properties_raw <- data.frame( Name = element_names[ read_html(elemental_properties$url[1]) %>% html_nodes("table") %>% extract2(1) %>% html_nodes("table") %>% extract2(8) %>% html_nodes("td") %>% html_attr("align") == "left"])
Some property pages (listed below) are difficult to parse, usually because of bad HTML. For now, we just skip those pages (no big loss).
skip_properties <- c( # has multiple values per element, plus formatting errors in HTML (cell 1 and 120) "Ionization_Energies", # replacement has 120 rows, data has 118 "NFPA_Label", # has formatting errors in HTML (cell 1 and 120) "Names_of_Allotropes", # formatting errors in HTML, specifically in cell 1 and 120 "Discovery", # formatting errors in HTML, specifically in cell 1 and 120 "Lattice_Angles", # formatting errors in HTML, specifically in cell 1 and 120 "Lattice_Constants", # replacement has 120 rows, data has 118 "Crystal_Structure", # formatting errors in HTML, specifically in cell 1 and 120 "Known_Isotopes", # formatting errors in HTML, specifically in cell 1 and 120 "Stable_Isotopes", # formatting errors in HTML, specifically in cell 1 and 120 "Isotopic_Abundances") # drop those rows from elemental_properties elemental_properties <- elemental_properties[-which(elemental_properties$sanitized %in% skip_properties), ]
Ok, let's populate our properties dataframe with some scraped data.
# This chunk does all the scraping of periodictable.com # But it tries its best not to do any scraping # If you really want to scrape the website and rebuild the entire dataset from # scratch (this takes some time!), go ahead and delete the file "inst/extdata/periodicdata-raw.rdata" # before running this chunk! if (!file.exists(here::here("inst", "extdata", "periodicdata-raw.rdata"))) { scrape_afresh <- TRUE for (k in 2:length(elemental_properties$property)) { message(paste0( "Reading property page (", k, " of ", length(elemental_properties$property), "): ", elemental_properties$property[k])) # On these property pages, the source HTML appears to be malformed (but recoverable) # The HTML source looks weird: in essence, a number of cells <td></td> are repeated (but not all) # so that the table which is expected to contain 118 *2 = 236 cells contains something like 299 instead. # To be able to ingest such a table, we need to detect and get rid of the duplicated td cells. # My approach is based on the fact that duplicated td cells always lack any CSS attributes, whereas the cells # we want to keep always specify align (either left or right). td_nodes <- read_html(elemental_properties$url[k]) %>% html_nodes("table") %>% extract2(1) %>% html_nodes("table") %>% extract2(8) %>% html_nodes("td") # The formatting of periodictable.com's HTML is unfortunately not consistent across the property pages # On most, the label and value are inside td blocks neatly aligned left or right, but on other property # pages the value part is repeated inside free-standing td blocks lacking any CSS attributes. # To distinguish between these two cases (i.e., remove the duplicate values), I check for the align attr: check_htmltable_align <- td_nodes %>% # here we are just checking for the existence of "align" attribute # will return NA for the td cells that lack align attribute, and those are the duplicates we want to remove html_attr("align") if (any(is.na(check_htmltable_align))) { # if it contains any NAs, print a message (for easier debugging) # example: https://periodictable.com/Properties/A/RTECSNumber.an.html message(paste0("Detected funky formatting on property page ", elemental_properties$property[k])) } # record the full table, element name/value pairs as a single vector property_table <- td_nodes[which(!is.na(check_htmltable_align))] %>% html_text() property <- structure(data.frame( # just the element names Name = property_table[td_nodes[which(!is.na(check_htmltable_align))] %>% html_attr("align") == "right"], # the value of the current property (this column is renamed below) value = property_table[td_nodes[which(!is.na(check_htmltable_align))] %>% html_attr("align") == "left"]), .Names = c("Name", elemental_properties$sanitized[k])) ## NOTE: You CANNOT rely on the order of the element names staying consistent across property pages, ## in particular not on "funky" property pages. Learned this the hard way. # So here we use what should be a reliable way of growing our properties df with each new property: properties_raw <- left_join(properties_raw, property, by = "Name") } properties_raw <- properties_raw %>% arrange(as.numeric(Atomic_Number)) # save this dataframe to file # (it is not large, but re-scraping the contents takes time) save(properties_raw, file = here::here("inst", "extdata", "periodicdata-raw.rdata")) write_csv(properties_raw, here::here("inst", "extdata", "periodicdata-raw.csv")) } else { load(file = here::here("inst", "extdata", "periodicdata-raw.rdata")) }
At the present time, the following property pages had formatting that deviated from the others, as detected by our logic in the previous chunk:
Reading property page (33 of 81): Ionization Energies Detected funky formatting on property page Ionization Energies Reading property page (34 of 81): DOT Hazard Class Detected funky formatting on property page DOT Hazard Class Reading property page (36 of 81): RTECS Number Detected funky formatting on property page RTECS Number Reading property page (39 of 81): Names of Allotropes Detected funky formatting on property page Names of Allotropes Reading property page (46 of 81): Discovery Detected funky formatting on property page Discovery Reading property page (49 of 81): CID Number Detected funky formatting on property page CID Number Reading property page (69 of 81): Lattice Angles Detected funky formatting on property page Lattice Angles Reading property page (70 of 81): Lattice Constants Detected funky formatting on property page Lattice Constants Reading property page (79 of 81): Known Isotopes Detected funky formatting on property page Known Isotopes Reading property page (80 of 81): Stable Isotopes Detected funky formatting on property page Stable Isotopes Reading property page (81 of 81): Isotopic Abundances Detected funky formatting on property page Isotopic Abundances
Extend the dataset with more properties
If you want to add more elemental properties to the data, that should be done at this point, i.e, before we split quantities into magnitudes and units (see below).
So far, we have extended the original dataset with the following properties:
Autoignition_Point. From periodictable.com. Has since been removed from the website.Flashpoint. From periodictable.com. Has since been removed from the website.Gmelin_Number. From periodictable.com. Has since been removed from the website.Heat_of_Combustion. From periodictable.com. Has since been removed from the website.ProductionandPrice. Global annual production and price of the elements, from @Vesborg2012.
property <- structure(data.frame(matrix(c( # Name # Autoignition_Point "Hydrogen", "535.5°C", "Lithium", "179.°C", "Carbon", "485.°C", "Sodium", "115.°C", "Magnesium", "472.°C", "Aluminum", "400.°C", "Silicon", "150.°C", "Phosphorus", "300.°C", "Potassium", "440.°C", "Titanium", "250.°C", "Chromium", "400.°C", "Iron", "100.°C", "Zinc", "460.°C", "Cadmium", "250.°C", "Tin", "630.°C", "Tellurium", "340.°C", "Samarium", "150.°C", "Thorium", "130.°C"), ncol = 2, byrow = T)), .Names = c("Name", "Autoignition_Point")) properties_raw <- left_join(properties_raw, property, by = "Name")
property <- structure(data.frame(matrix(c( # Name # Flashpoint "Hydrogen", "-18.°C", "Sodium", "4.°C", "Magnesium", "500.°C", "Aluminum", "645.°C", "Phosphorus", "30.°C", "Bromine", "-18.°C"), ncol = 2, byrow = T)), .Names = c("Name", "Flashpoint")) properties_raw <- left_join(properties_raw, property, by = "Name")
property <- structure(data.frame(matrix(c( # Name # Heat_of_Combustion "Lithium", "-298.J/(Kg K)", "Carbon", "-393.J/(Kg K)", "Magnesium","-668.J/(Kg K)", "Silicon", "-9055.J/(Kg K)", "Potassium","-182.J/(Kg K)", "Calcium", "-990.J/(Kg K)", "Germanium","-536.J/(Kg K)"), ncol = 2, byrow = T)), .Names = c("Name", "Heat_of_Combustion")) properties_raw <- left_join(properties_raw, property, by = "Name")
property <- structure(data.frame(matrix(c( # Name # Gmelin_Number "Hydrogen", "Gmelin3", "Neon", "Gmelin16211", "Sodium", "Gmelin563", "Magnesium", "Gmelin16207", "Aluminum", "Gmelin16248", "Silicon", "Gmelin61188", "Phosphorus", "Gmelin8863", "Argon", "Gmelin821", "Potassium", "Gmelin15203", "Helium", "Gmelin16294", "Calcium", "Gmelin16277", "Scandium", "Gmelin16304", "Titanium", "Gmelin16313", "Vanadium", "Gmelin49047", "Chromium", "Gmelin16274", "Manganese", "Gmelin699", "Iron", "Gmelin6845", "Cobalt", "Gmelin49017", "Nickel", "Gmelin16229", "Copper", "Gmelin16269", "Lithium", "Gmelin30", "Zinc", "Gmelin16321", "Gallium", "Gmelin16287", "Germanium", "Gmelin93634", "Arsenic", "Gmelin16247", "Selenium", "Gmelin8872", "Bromine", "Gmelin1182", "Krypton", "Gmelin16201", "Rubidium", "Gmelin16244", "Strontium", "Gmelin16302", "Yttrium", "Gmelin16319", "Beryllium", "Gmelin16256", "Zirconium", "Gmelin16322", "Niobium", "Gmelin16213", "Molybdenum", "Gmelin16205", "Ruthenium", "Gmelin16241", "Rhodium", "Gmelin16245", "Palladium", "Gmelin16239", "Silver", "Gmelin1253", "Cadmium", "Gmelin16276", "Indium", "Gmelin1081", "Boron", "Gmelin16262", "Tin", "Gmelin16300", "Tellurium", "Gmelin1153", "Iodine", "Gmelin1160", "Xenon", "Gmelin1125", "Cesium", "Gmelin15188", "Barium", "Gmelin16266", "Cerium", "Gmelin16275", "Praseodymium","Gmelin16238", "Carbon", "Gmelin8868", "Neodymium", "Gmelin16212", "Samarium", "Gmelin16301", "Europium", "Gmelin15985", "Gadolinium", "Gmelin16286", "Terbium", "Gmelin16311", "Dysprosium", "Gmelin16278", "Holmium", "Gmelin16291", "Erbium", "Gmelin16280", "Thulium", "Gmelin16307", "Nitrogen", "Gmelin150", "Ytterbium", "Gmelin16014", "Lutetium", "Gmelin16202", "Hafnium", "Gmelin16293", "Tantalum", "Gmelin16312", "Tungsten", "Gmelin16317", "Rhenium", "Gmelin16243", "Osmium", "Gmelin16234", "Iridium", "Gmelin16298", "Platinum", "Gmelin79607", "Gold", "Gmelin16246", "Oxygen", "Gmelin485", "Mercury", "Gmelin1623", "Thallium", "Gmelin16308", "Lead", "Gmelin16240", "Bismuth", "Gmelin16267", "Radon", "Gmelin16242", "Radium", "Gmelin40437", "Fluorine", "Gmelin16281", "Thorium", "Gmelin16314", "Uranium", "Gmelin16315", "Plutonium", "Gmelin40432"), ncol = 2, byrow = T)), .Names = c("Name", "Gmelin_Number")) properties_raw <- left_join(properties_raw, property, by = "Name")
# Note, I have rather lazily entered year 2010 for *all* values from Vesborg, # despite that paper noting that some values were taken from previous years. # If time permits, go back and check @Vesborg2012 and update year field in # those cases. vesborg2012 <- tibble::tribble( ~Symbol, ~Production, ~Price, ~value, ~year, ~mb, ~comment, "Fe", 12.4 , -0.24, 12.1, 2010, "M", "", "Cl", 10.7 , NA, NA, 2010, "", "NaCl", "N", 11.1 , 0.8 , 11.9, 2010, "M", "NH3 only", "Na", 11.0 , NA, NA, 2010, "M", "NaCl", "O", 10.9 , -1.4 , NA, 2010, "M", "O2 only", "H", 10.8 , 0.9 , 10.7, 2010, "M", "H2 only", "S", 10.8 , -0.7 , 10.1, 2010, "B", "Oil by-product", "Al", 10.6 , 0.3 , 10.9, 2010, "M", "", "Ca", 10.5 , NA, NA, 2010, "M", "Ca compounds", "P", 10.5 , -0.28, 10.8, 2010, "M", "Phosphate rock", "K", 10.4 , NA, NA, 2010, "M", "K compounds", "Cr", 10.3 , 0.4 , 10.7, 2010, "M", "", "Cu", 10.2 , 0.9 , 11.1, 2010, "M", "", "Mn", 10.1 , 0.6 , 10.7, 2010, "M", "", "Zn", 10.1 , 0.26, 10.3, 2010, "M", "", "Ba", 9.9 , NA, NA, 2010, "M", "Mostly BaSO4", # "Mg", 9.8 , 0.5 , 10.3, 2010, "M", "Mg compounds", # "Mg", 8.9 , NA, NA, 2010, "", "Mg metal", # I have joined Mg compounds and Mg metal from Vesborg2012 into a single row "Mg", 9.85, 0.5 , NA, 2010, "M", "Metallic and compounds", "Si", 9.8 , 0.4 , 10.2, 2010, "M", "Si only, no SiO2", "Pb", 9.6 , 0.28, 9.9, 2010, "MB", "Also a by-product of Zn", "Ar", 9.5 , -0.21, 9.3, 2010, "M", "", "F", 9.4 , NA, NA, 2010, "M", "Mostly CaF2", # "Ti", 9.4 , 1.4 , 10.4, 2010, "M", "TiO2 + FeTiO3", # "Ti", 8.1 , NA, NA, 2010, "", "Ti metal", # I have joined Ti compounds and Ti metal from Vesborg2012 into a single row "Ti", 9.42, 1.4 , NA, 2010, "", "Metallic and oxides", "B", 9.2 , 1.4 , 10.6, 2010, "M", "B compounds", "C", 9.2 , 0.18, 9.4, 2010, "M", "Graphite only", "Ni", 9.1 , 1.3 , 10.4, 2010, "M", "", "Zr", 8.9 , 0.5 , 9.4, 2010, "M", "Mostly ZrO2", "Br", 8.8 , 0.3 , NA, 2010, "M", "", "Sr", 8.6 , 0.8 , 9.4, 2010, "M", "Sr compounds", "Mo", 8.4 , 1.5 , 9.9, 2010, "BM", "By-product Cu", "Sn", 8.4 , 1.3 , 9.4, 2010, "M", "", "Sb", 8.2 , 1.2 , 9.4, 2010, "M", "", "Co", 7.9 , 1.6 , 9.5, 2010, "B", "", "Nb", 7.8 , 2.0 , 9.8, 2010, "M", "", "W", 7.8 , 2.8 , 10.6, 2010, "M", "", "V", 7.8 , 1.4 , 9.2, 2010, "M", "", "U", 7.7 , 2.0 , 9.7, 2010, "M", "Radioactive", "Ce", 7.6 , 2.0 , 9.6, 2010, "M", "With La and Nd", "As", 7.6 , 0.21, 7.8, 2010, "B", "", "La", 7.5 , 2.0 , 9.5, 2010, "M", "With Ce and Nd", "I", 7.5 , 1.4 , NA, 2010, "M", "", "Th", NA, 1.9 , NA, 2010, "M", "Radioactive", "Ag", 7.4 , 3.0 , 10.4, 2010, "B", "Mostly by-product, approx. 30% primary product", "Li", 7.4 , 1.5, 8.9, 2010, "M", "", "Nd", 7.3 , 2.5, 9.8, 2010, "M", "With La and Ce", "He", 7.3 , 0.5, 7.8, 2010, "B", "By-product of natural gas", "Cd", 7.3 , 0.4, 7.7, 2010, "B", "By-product Zn", "Y", 6.8 , 2.3, 9.1, 2010, "B", "By-product HREEs", "Bi", 6.8 , 1.5, 8.3, 2010, "B", "By-product Pb", "Pr", 6.7 , 2.5, 9.2, 2010, "B", "By-product LREEs", "Ne", 6.7 , 2.5, NA, 2010, "B", "By-product Ar", "Au", 6.4 , 4.7, 11.1, 2010, "M", "", "Sm", 6.4 , 2.2, 8.6, 2010, "B", "By-product LREEs", "Se", 6.4 , 2.1, 8.5, 2010, "B", "By-product Cu", "Gd", 6.3 , 2.3, 8.6, 2010, "B", "By-product HREEs", "Hg", 6.3 , 1.7, 8.0, 2010, "M", "", "Dy", 6.0 , 3.4, 9.4, 2010, "M", "Co-product Eu", "Ta", 6.0 , 2.7, 8.7, 2010, "M", "", "Be", 5.8 , NA, NA, 2010, "M", "", "Er", 5.8 , 2.5, NA, 2010, "B", "By-product HREE", "In", 5.7 , 2.9, 8.6, 2010, "B", "By-product Zn", "Te", 5.7 , 2.5, 8.2, 2010, "B", "By-product Cu", "Yb", 5.7 , 2.7, NA, 2010, "B", "By-product HREE", "Eu", 5.6 , 3.8, 9.4, 2010, "M", "Co-product Dy", "Kr", 5.6 , 2.6, NA, 2010, "B", "By-product Ar", "Ho", 5.4 , 3.5, NA, 2010, "B", "By-product HREE", "Tb", 5.3 , 3.7, 9.0, 2010, "B", "By-product HREE", "Pd", 5.3 , 4.3, 9.6, 2010, "B", "By-product Ni, Pt", "Pt", log10(2e5), log10(30e3), NA, 2020, "M", "@Chatenet2022", "Ge", 5.1 , 3.2, 8.3, 2010, "B", "By-product Zn", "Ga", 5.1 , 2.9, 7.9, 2010, "B", "By-product Al", "Tm", 5.1 , 4.0, NA, 2010, "B", "By-product HREE", "Lu", 5.0 , 4.0, NA, 2010, "B", "By-product HREE", "Hf", 4.9 , 2.8, 7.7, 2010, "B", "By-product Zr", "Re", 4.7 , 3.5, 8.2, 2010, "B", "By-product Mo, (Cu)", "Xe", 4.6 , 3.4, 8.0, 2010, "B", "By-product Ar", "Rh", 4.4 , 4.7, 9.1, 2010, "B", "By-product Pt", "Ru", 4.4 , 3.7, 8.1, 2010, "B", "By-product Pt", "Ir", log10(7.7e3), log10(203e3), NA, 2018, "B", "By-product of Pt. Production 2018 from @Minke2021, price 2021 from @Chatenet2022", "Tl", 4 , NA, NA, 2010, "B", "By-product Cu, Zn", "Cs", 4 , NA, NA, 2010, "M", "Co-product Li, Ta, Pb", "Os", 3.7 , 4.5, NA, 2010, "B", "By-product Pt", "Rb", 3.6 , 4.6, 8.2, 2010, "M", "Co-product Cs, Ta, Li", "Sc", 3.3 , 3.1, NA, 2010, "B", "By-product") # for purposes of quality control (easier check for duplicates, etc.) let's add # atomic number vesborg2012 %<>% left_join(properties_raw %>% select(Symbol, Atomic_Number), by = "Symbol")
As a crude check that we have not introduced any (grave) transcription errors in this data, we will now recreate some of the plots from the source paper.
p.data <- vesborg2012 %>% select(Symbol, Atomic_Number, Production) ggplot(p.data, aes(y = 10^Production, x = as.numeric(Atomic_Number))) + geom_line(colour = "gray", size = 0.75) + geom_point() + # formatting this axis title nicely required scales:: # https://stackoverflow.com/questions/10762287/how-can-i-format-axis-labels-with-exponents-with-ggplot2-and-scales scale_y_log10( name = "Global annual production/(kg/yr)", breaks = scales::trans_breaks("log10", function(x) 10^x), # note, for this trans_format() to work as intended that data needs to in absolute numbers labels = scales::trans_format("log10", math_format(10^.x))) + scale_x_continuous(name = "Atomic number, Z") + geom_text_repel( aes(label = Symbol))
Ok, that looks quite like figure 1 in @Vesborg2012. I find no obvious discrepances. Let's also recreate the only figure in the SI (log(price) vs log(production)).
p.data <- vesborg2012 %>% select(Symbol, Atomic_Number, Production, Price) ggplot(p.data, aes(y = 10^Price, x = 10^Production)) + # geom_line(colour = "gray", size = 0.75) + geom_point() + scale_x_log10( name = "Global annual production/(kg/yr)", breaks = scales::trans_breaks("log10", function(x) 10^x), labels = scales::trans_format("log10", math_format(10^.x))) + scale_y_log10( name = "Price/(USD/kg)", breaks = scales::trans_breaks("log10", function(x) 10^x), labels = scales::trans_format("log10", math_format(10^.x))) + geom_text_repel( aes(label = Symbol))
Ok, that looks about right.
# note that the production and price values above are log10(), # which is how they are presented in @Vesborg2012 # unfortunately I have not found the source values anywhere (haven't asked the authors yet, though) # this means a quite significant loss of accuracy in the following conversion, but # we do it anyway because I want to be explicit regarding the units vesborg2012 %<>% mutate(Production = ifelse( is.na(Production), "", # formatC ensures that the number is decimal and not scientific notation even if larger number, # because scientific notation messes up our values-units split algo below (yes, that should be improved...) paste(formatC(10^Production, format = "fg"), "kg/yr"))) %>% mutate(Price = ifelse( is.na(Price), "", paste(formatC(10^Price, format = "fg"), "USD/kg"))) # now let's incorporate production and price into the properties dataset properties_raw <- left_join( properties_raw, vesborg2012 %>% select(Symbol, Production, Price), by = "Symbol")
Split quantities and units into separate dataframes
We now have several tasks ahead of us before the data is usable:
- what type of data each property is (character string, dimensionless quantity, or quantity with unit),
- determine and identify what part of each property value is a quantity and which is a unit,
- extract quantity and unit separately from each datapoint and save to
valuesandunitsdataframe, respectively.
Take care to that the dataframes values and units have the same dimensions
(rows/columns). The point behind splitting each value up is to be able to treat
the quantity numerically, plot it, etc. It's therefore crucial that the dataframes
stay the same size, so that once a quantity is read from values, we can trust that
its unit is found in the corresponding cell in units.
# create empty dataframes to hold quantities and units, respectively values <- units <- data.frame(matrix( nrow = dim(properties_raw)[1], ncol = dim(properties_raw)[2], dimnames = list(seq(1, dim(properties_raw)[1]), names(properties_raw)), byrow = TRUE)) # The dataset on periodictable.com also has footnotes. # These are not saved in the current implementation of `periodicdata`
Next, we list all the currently existing columns so that we can specify a type for each one. The type will determine how we handle unit extraction etc.
# to re-create the matrix below, get started like this: # cat(paste0(names(properties_raw), collapse = ",\n")) property_types <- structure(data.frame(matrix(c( "Name", "character", "Symbol", "character", "Atomic_Number", "unitless number", "Atomic_Weight", "unitless number", "Density", "number with unit", "Melting_Point", "number with unit", "Boiling_Point", "number with unit", "Phase", "character", "Absolute_Melting_Point", "number with unit", "Absolute_Boiling_Point", "number with unit", # contains converted values in parentheses "Critical_Pressure", "number with unit", "Critical_Temperature", "number with unit", "Heat_of_Fusion", "number with unit", "Heat_of_Vaporization", "number with unit", "Specific_Heat", "number with unit", "Adiabatic_Index", "character", "Neel_Point", "number with unit", "Thermal_Conductivity", "number with unit", "Thermal_Expansion", "number with unit", "Density_Liquid", "number with unit", "Molar_Volume", "unitless number", "Brinell_Hardness", "number with unit", "Mohs_Hardness", "number with unit", "Vickers_Hardness", "number with unit", "Bulk_Modulus", "number with unit", "Shear_Modulus", "number with unit", "Young_Modulus", "number with unit", "Poisson_Ratio", "unitless number", "Refractive_Index", "unitless number", "Speed_of_Sound", "number with unit", "Valence", "unitless number", "Electronegativity", "unitless number", "ElectronAffinity", "number with unit", "DOT_Hazard_Class", "unitless number", "DOT_Numbers", "unitless number", "RTECS_Number", "character", "Alternate_Names", "character", "Block", "character", "Group", "character", "Period", "character", "Series", "character", "Electron_Configuration", "character", "Color", "character", "Gas_phase", "character", "CAS_Number", "character", "CID_Number", "character", "Electrical_Type", "character", "Electrical_Conductivity", "number with unit", "Resistivity", "number with unit", "Superconducting_Point", "unitless number", "Magnetic_Type", "character", "Curie_Point", "number with unit", "Mass_Magnetic_Susceptibility", "unitless number", "Molar_Magnetic_Susceptibility", "unitless number", "Volume_Magnetic_Susceptibility", "unitless number", "Percent_in_Universe", "number with unit", "Percent_in_Sun", "number with unit", "Percent_in_Meteorites", "number with unit", "Percent_in_Earths_Crust", "number with unit", "Percent_in_Oceans", "number with unit", "Percent_in_Humans", "number with unit", "Atomic_Radius", "number with unit", "Covalent_Radius", "number with unit", "Van_der_Waals_Radius", "number with unit", "Space_Group_Name", "character", "Space_Group_Number", "unitless number", # quantity includes "Stable" and num + unit. Units are a mix of time units: ms, s, m, d, y "HalfLife", "number with unit", # quantity includes "Stable" and num + unit. Units are a mix of time units: ms, s, m, d, y "Lifetime", "number with unit", "Decay_Mode", "character", "Quantum_Numbers", "character", "Neutron_Cross_Section", "unitless number", "Neutron_Mass_Absorption", "unitless number", "Autoignition_Point", "number with unit", "Heat_of_Combustion", "number with unit", "Gmelin_Number", "character", "Production", "number with unit", "Price", "number with unit"), ncol = 2, byrow = T)), .Names = c("Data", "Type"))
# running this loop usually produces warnings "NAs introduced by coercion". We can disregard those. for (k in 1:dim(property_types)[1]) { # clean up or convert the quantity or unit depending on type of data in each property # type is defined manually in the matrix in the chunk above if (property_types$Type[k] == "character") { values[, which(names(values) == property_types$Data[k])] <- properties_raw %>% pull(property_types$Data[k]) %>% gsub("\\[note\\]", "", x = .) %>% gsub("^N/A$", "", x = .) %>% gsub("^None$", "", x = .) } else if (property_types$Type[k] == "unitless number") { values[, which(names(values) == property_types$Data[k])] <- properties_raw %>% pull(property_types$Data[k]) %>% gsub("\\[note\\]", "", x = .) %>% gsub("×10", "E", x = .) %>% # final step, make sure the column is numeric as.numeric() } else if (property_types$Type[k] == "number with unit") { values[, which(names(values) == property_types$Data[k])] <- properties_raw %>% pull(property_types$Data[k]) %>% gsub("\\[note\\]", "", x = .) %>% # replace "Stable" with Inf (only found in HalfLife, Lifetime) gsub("^Stable$", "Inf", x = .) %>% # replace ×10 with proper power notation gsub("×10", "E", x = .) %>% # extract the numeric quantity (including the string "Inf") str_extract("Inf|[-×\\.0-9E]+") %>% # final step, make sure this column is numeric as.numeric() units[, which(names(units) == property_types$Data[k])] <- properties_raw %>% pull(property_types$Data[k]) %>% gsub("N/A", "", x = .) %>% gsub("\\[note\\]", "", x = .) %>% # replace "Stable" with Inf (only found in HalfLife, Lifetime) gsub("^Stable$", "Inf", x = .) %>% # remove any numbers+units within parentheses gsub("\\([.0-9]+\\s[A-Za-z]+\\)", "", x = .) %>% # remove the numeric quantity (including the string "Inf") # take care not to use gsub here, since numbers could also be part of the unit (e.g., "g/cm3") sub("Inf|[-×\\.0-9E]+", "", x = .) %>% # remove any leading and trailing whitespace str_trim(side="both") %>% # replace empty string with NA gsub("^$", NA, x = .) } else { message("This should never happen!") } }
Cleaning up and normalising the units
Now that we have separated the units and the quantities into separate dataframes, let's have a look at the units, to see what fixing is needed.
# Units by property sapply(units, unique)
As you can see, some properties (for example HalfLife) use more than one unit.
This is problematic, since we will only be plotting against one y-axis, not several.
So we will have to convert all such occurrences to their base SI units, which means
we have to take the numerical conversion of the quantity into consideration as well.
Let's do it.
# All units in the dataset as a simple vector cat(paste(sort(unique(unlist(sapply(units, unique)))), collapse = "\n"))
## % (percent) ## °C => K ## d => s (day to second) ## g/cm3 => kg/m^3 ## g/l => kg/m^3 ## GPa => Pa ## h => s (hour to second) ## J/(Kg K) => J/(kg K) (fix wrong case) ## J/(kg K) ## K (Kelvin) ## K-1 (per Kelvin) ## kJ/mol (kilojoule per mol) ## m => s (only exists in HalfLife and Lifetime property, so this is minutes not metres) ## m Ω => Ohm m (electrical resistivity, Ohm metre) # avoid Ω, causes non-ASCII char warning on package build ## m/s (metre per second) ## MPa => Pa ## ms => s (millisecond to second) ## pm => m (picometre to metre) ## S/m (Siemens per metre) ## W/(m K) (Watt per metre-Kelvin) ## y => s (year to second)
pcf <- structure(data.frame(matrix(c( # from # to # conversion factor "°C", "K", "+273.15", "d", "s", "86400", "g/cm3", "kg/m^3", "1E3", "g/l", "kg/m^3", "1", "GPa", "Pa", "1E9", "h", "s", "3600", "J/(Kg K)", "J/(kg K)", "1", # minutes to seconds "m", "s", "60", "m Ω", "Ohm m", "1", "MPa", "Pa", "1E6", "ms", "s", "1E-3", # be careful that this substitution doesn't precede "mins to secs" "pm", "m", "1E-12", "y", "s", "3.154E7"), ncol = 3, byrow = T)), .Names = c("pattern", "convert", "factor"))
# Replace the units and convert values according to pcf for (k in 1:dim(units)[2]) { # if the entire column is NA, move to the next one if (all(is.na(units[, k]))) {next} for (i in 1:dim(units)[1]) { # jump to the next cell immediately if unit is empty if (is.na(units[i, k])) {next} # for each cell, compare the unit to pcf$pattern, # and if they match, replace it with pcf$convert # and apply the conversion factor on the value # match() returns the position of the match # in pcf$pattern, or NA if no match was found hit <- match(units[i, k], pcf$pattern) if (!is.na(hit)) { # go ahead and replace unit and convert value units[i, k] <- pcf$convert[hit] # had to find a way to handle the odd addition operation, # opted to do it with a string operation and this if-else if (substr(pcf$factor[hit], 1, 1) %in% c("+", "-")) { if (substr(pcf$factor[hit], 1, 1) == "-") { values[i, k] <- values[i, k] + as.numeric(pcf$factor[hit]) } else { values[i, k] <- values[i, k] + as.numeric(sub("^\\+", "", pcf$factor[hit])) } } else { # not addition/subtraction operation values[i, k] <- values[i, k] * as.numeric(pcf$factor[hit]) } } } }
Proof-of-concept plots and conclusion
ggplot() + geom_point( data = values %>% filter(!is.na(Density)), aes(Atomic_Number, Density)) + geom_path( data = values %>% filter(!is.na(Density)), aes(Atomic_Number, Density)) + geom_point( data = values %>% filter(!is.na(Density)) %>% filter(Group == "18"), aes(Atomic_Number, Density), colour = "red") + geom_text_repel( data = values %>% filter(!is.na(Density)) %>% filter(Group == "18"), aes(Atomic_Number, Density, label = Symbol), colour = "red") + scale_y_log10() + labs( x = "Atomic number", y = paste0( "Density", "/(", unique(units$Density[which(!is.na(units$Density))]), ")"))
That's not a bad plot, but what if we could use the periodic table (think IUPAC's layout) and plot the data on top of that, for example using varying degrees of fill colour, or some such? Or even to print a typical periodic table by itself, just using the dataset and ggplot2 (and possibly TikZ, for really nice PDF output). To do that, we need a final preparatory step.
The IUPAC periodic table places the elements next to each other, organised in rows (periods) and columns (groups) on a two-dimensional grid. Obviously, each element's position on this plot is completely specified by its group and period.
values$IUPAC_Period <- values$Period %>% as.numeric() values$IUPAC_Group <- values$Group %>% as.numeric() values$IUPAC_Number <- values$Atomic_Number %>% as.character() values$IUPAC_Series <- values$Series # to maintain same dimensions as values dataframe units$IUPAC_Period <- NA units$IUPAC_Group <- NA units$IUPAC_Number <- NA # Fix coordinates for lanthanides and actinoids # lanthanoids 57-71: Period = 8, Group = seq(4, 18) values$IUPAC_Period[seq(57, 71)] <- 8.5 values$IUPAC_Group[seq(57, 71)] <- seq(4, 18) values[which(values$Symbol == "Lu"), which(names(values) == "IUPAC_Series")] <- "Lanthanide" # actinoids 89-103: Period = 9, Group = seq(4, 18) # increase Period slightly to increase the gap up to the transition block values$IUPAC_Period[seq(89, 103)] <- 9.5 values$IUPAC_Group[seq(89, 103)] <- seq(4, 18) values[which(values$Symbol == "Lr"), which(names(values) == "IUPAC_Series")] <- "Actinide" # add placeholder "element" for lanthanoids and actinoids after Ba and Ra, respectively # (this is just to conform to the IUPAC table layout) values[nrow(values) + 1, which(names(values) %in% c("Name", "IUPAC_Period", "IUPAC_Group", "IUPAC_Number", "IUPAC_Series"))] <- data.frame("lanthanoids", 6, 3, "57–71", "Lanthanide") values[nrow(values) + 1, which(names(values) %in% c("Name", "IUPAC_Period", "IUPAC_Group", "IUPAC_Number", "IUPAC_Series"))] <- data.frame("actinoids", 7, 3, "89–103", "Actinide") # add corresponding empty rows to units df to maintain the same dimensions: units[nrow(units) + 1, ] <- NA units[nrow(units) + 1, ] <- NA
By changing the IUPAC_Period and IUPAC_Group assignments it's possible to adjust the layout of the periodic table according to taste. Here's a reproduction of the IUPAC table of elements:
p.periodictable_ggplot2 <- ggplot() + # lanthanoids and actinoids background colour drawn first geom_point( data = values %>% filter(IUPAC_Series == "Lanthanide"), size = 16, shape = 15, colour = "#cfc0c7", aes(y = IUPAC_Period, x = IUPAC_Group)) + geom_point( data = values %>% filter(IUPAC_Series == "Actinide"), size = 16, shape = 15, colour = "#a58394", aes(y = IUPAC_Period, x = IUPAC_Group)) + # boxes for all elements geom_point( data = values, # size 16 leaves no gaps between the element boxes at fig.width=9,fig.height=5.25 size = 16, shape = 0, aes(y = IUPAC_Period, x = IUPAC_Group)) + # element symbol geom_text( data = values, size = 3, colour = "black", fontface = "bold", aes(label = Symbol, y = IUPAC_Period - 0.1, x = IUPAC_Group)) + # atomic number geom_text( data = values, size = 1.8, colour = "black", aes(label = IUPAC_Number, y = IUPAC_Period - 0.32, x = IUPAC_Group)) + # element name geom_text( data = values, size = 1.6, colour = "black", aes(label = tolower(Name), y = IUPAC_Period + 0.12, x = IUPAC_Group)) + # atomic weight geom_text( data = values, size = 1.6, colour = "black", aes(label = Atomic_Weight, y = IUPAC_Period + 0.32, x = IUPAC_Group)) + # period labels geom_text( data = values %>% filter(IUPAC_Period < 8), size = 1.7, colour = "black", aes(label = IUPAC_Period, y = IUPAC_Period), x = 0.38) + # group labels # positions manually adjusted geom_text( data = data.frame( y = c(0.38, 1.38, rep(3.38, 10), rep(1.38, 5), 0.38), x = seq(1,18)), size = 1.7, colour = "black", aes(label = x, x = x, y = y)) + # table title # position manually adjusted annotate( "text", x = 9, y = 0.6, vjust = 0, size = 4.5, fontface = "bold", label = "Periodic Table of the Elements") + # table subtitle # position manually adjusted annotate( "text", x = 9, y = 0.74, vjust = 1, size = 2.6, label = "Created with the periodicdata package by solarchemist.se") + scale_x_continuous( breaks = seq( min(values$IUPAC_Group), max(values$IUPAC_Group)), limits = c( min(values$IUPAC_Group) - 1, max(values$IUPAC_Group) + 1), expand = c(0,0)) + scale_y_continuous( trans = "reverse", breaks = seq( min(values$IUPAC_Period), max(values$IUPAC_Period)), limits = c( max(values$IUPAC_Period) + 1, min(values$IUPAC_Period) - 1.5), expand = c(0,0)) + theme( panel.grid.major = element_blank(), panel.grid.minor = element_blank(), plot.margin = unit(c(0, 0, -0.85, -0.85), "line"), axis.title = element_blank(), axis.text = element_blank(), axis.ticks = element_blank(), # center (roughly) over transition metal block legend.position = c(0.42, 0.91), legend.justification = c(0.5, 1), legend.direction = "horizontal", # make the legend colourbar a little longer legend.key.width = unit(2.5, "line"), legend.title = element_blank(), legend.background = element_rect(fill = "transparent")) print(p.periodictable_ggplot2)
This is just an example of the kind of plot than can be achieved with ggplot2 and this dataset.
Note that the plot is quite sensitive to changes to its final output dimension --- such
changes may necessitate changes to the size parameter in the geom_point() layers
(these effectively set the size of the box that each element occupies).
Recreation of figure 3 from @Vesborg2012:
p.data <- values %>% select(Symbol, Atomic_Number, Production, Percent_in_Earths_Crust) ggplot(p.data, aes(y = 1e3*Percent_in_Earths_Crust, x = Production)) + geom_point() + geom_smooth(method = "lm", formula = "y~x", se = TRUE, linetype ="dashed", fill = "gray") + scale_x_log10( name = "Global annual production/(kg/yr)", breaks = scales::trans_breaks("log10", function(x) 10^x), labels = scales::trans_format("log10", math_format(10^.x))) + scale_y_log10( name = "Crustal abundance/ppm", breaks = scales::trans_breaks("log10", function(x) 10^x), labels = scales::trans_format("log10", math_format(10^.x))) + geom_text_repel(aes(label = Symbol))
Considering that this dataset and the paper use different crustal abundance sources, the discrepancy is not very large. The overall trend is preserved (elements below the line are "overexploited"; here we may note gold (for obvious reasons) and nitrogen (production is sourced from the atmosphere, thus appears overtaxed compared to crustal abundance); and elements above the line are conversely "underexploited", if such a thing exists).
legend.mock.box <- tibble::tribble( ~x, ~y, ~text, ~id, ~label, 3.2, 2.8, "", "box", "", 3.2, 2.8-0.25, "Mo", "text", "", 3.2, 2.8+0.03, "8.4", "text", "Annual production (log)", 3.2, 2.8+0.28, "1.5", "text", "Market price (log)") ggplot() + # production rates above 1 Mt/yr (note, data is in kg) geom_point( data = values %>% filter(Production > 1e9), size = 16, shape = 15, # alpha = 0.5, aes(y = IUPAC_Period, x = IUPAC_Group, colour = "megaton")) + # production rates above 33 kt/yr (note, data is in kg) geom_point( data = values %>% filter(Production > 33e6 & Production <= 1e9), size = 16, shape = 15, # alpha = 0.75, aes(y = IUPAC_Period, x = IUPAC_Group, colour = "medium")) + # production rates above 1 kt/yr (note, data is in kg) geom_point( data = values %>% filter(Production > 1e6 & Production <= 33e6), size = 16, shape = 15, # alpha = 0.5, aes(y = IUPAC_Period, x = IUPAC_Group, colour = "low")) + # production rates less than 1 kt/yr (note, data is in kg) geom_point( data = values %>% filter(Production <= 1e6), size = 16, shape = 15, # alpha = 0.5, aes(y = IUPAC_Period, x = IUPAC_Group, colour = "rare")) + # boxes for all elements geom_point( data = values, # size 16 leaves no gaps between the element boxes at fig.width=9,fig.height=5.25 size = 16, shape = 0, aes(y = IUPAC_Period, x = IUPAC_Group)) + # element symbol geom_text( data = values, size = 3, colour = "black", fontface = "bold", aes(label = Symbol, y = IUPAC_Period - 0.24, x = IUPAC_Group)) + # show atomic number range for La, Ac placeholder cells geom_text( data = values %>% filter(is.na(Atomic_Number)), size = 2.2, colour = "black", aes(label = IUPAC_Number, y = IUPAC_Period - 0.24, x = IUPAC_Group)) + # log(production) geom_text( data = values, size = 2.4, colour = "black", aes( label = ifelse(is.na(Production), "", formatC(log10(Production), format="f", digits=1)), y = IUPAC_Period + 0.04, x = IUPAC_Group)) + # log(price) geom_text( data = values, size = 2.4, colour = "black", aes( label = ifelse(is.na(Price), "", formatC(log10(Price), format="f", digits=1)), y = IUPAC_Period + 0.28, x = IUPAC_Group)) + # table title # position manually adjusted annotate( "text", x = 9, y = 0.6, vjust = 0, size = 4.5, fontface = "bold", label = "Production and economics of the chemical elements") + # table subtitle # position manually adjusted annotate( "text", x = 9, y = 0.74, vjust = 1, size = 2.6, label = "Based on Vesborg & Jaramillo, RSC Advances 2, 7933–7947 (2012)") + scale_x_continuous( breaks = seq( min(values$IUPAC_Group), max(values$IUPAC_Group)), limits = c( min(values$IUPAC_Group) - 1, max(values$IUPAC_Group) + 1), expand = c(0,0)) + scale_y_continuous( trans = "reverse", breaks = seq( min(values$IUPAC_Period), max(values$IUPAC_Period)), limits = c( max(values$IUPAC_Period) + 1, min(values$IUPAC_Period) - 1.5), expand = c(0,0)) + # a mock element box as legend geom_point( data = legend.mock.box %>% filter(id == "box"), size = 14, shape = 0, aes(x = x, y = y)) + geom_text( data = legend.mock.box %>% filter(id != "box"), colour = "black", size = 2.6, aes(label = text, x = x, y = y)) + geom_text_repel( data = legend.mock.box %>% filter(id != "box"), nudge_x = 1.75, size = 2.5, point.padding = 0.75, direction = "both", aes(label = label, x = x, y = y)) + scale_colour_manual( name = "Global production", guide = guide_legend( override.aes = list(size = 3), title.position = "top"), # warning, alpha is not obeyed by legend keys - I don't know why values = c(megaton = alpha("#07B248", 0.5), medium = alpha("#C4E2BF", 0.75), low = alpha("#F4B521", 0.5), rare = alpha("#ED2025", 0.5)), labels = c(">1 Mt/yr", ">33 kt/yr", ">1 kt/yr", "<1 kt/yr")) + theme( panel.grid.major = element_blank(), panel.grid.minor = element_blank(), plot.margin = unit(c(0, 0, -0.85, -0.85), "line"), axis.title = element_blank(), axis.text = element_blank(), axis.ticks = element_blank(), # legend is manually centered atop the transition metal block legend.position = c(0.45, 0.82), legend.justification = c(0.5, 1), legend.direction = "horizontal", legend.key = element_rect(fill = NA, colour = NA), # key.size had no effect on the legend key size! see guides() above instead # key.width affects the horizontal spacing of the key, not the coloured box legend.key.width = unit(2, "mm"), legend.text = element_text(size = 10), #legend.title = element_blank(), legend.background = element_rect(fill = NA))
With the dataset fully assembled, all that remains is to save the values and units dataframes.
These dataframes are part of the exported data of this package.
# use_data() saves each dataframe to `data/` as an *.rda file use_data(values, units, overwrite = TRUE) write_csv(values, here("inst", "extdata", "periodicdata-values.csv")) write_csv(units, here("inst", "extdata", "periodicdata-units.csv"))
Note: the file inst/extdata/periodicdata-raw.R is created by knitting this Rmd file:
purl(input = here("vignettes", "periodicdata.Rmd"), output = here("inst", "extdata", "periodicdata-raw.R"))
The following chunk is normally not evaluated. It's only used when we need to refresh the demo periodic table that's included in the README.
# note that the doc/ directory is only created after running devtools::build_vignettes() ggsave( filename = here::here("doc", "periodictable-ggplot.png"), plot = p.periodictable_ggplot2, width = ptable.width, height = ptable.height) ggsave( filename = here::here("doc", "periodictable-ggplot.pdf"), plot = p.periodictable_ggplot2, width = ptable.width, height = ptable.height) ggsave( filename = here("doc", "periodictable-ggplot.svg"), plot = p.periodictable_ggplot2, width = ptable.width, height = ptable.height)
References
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.