MRL @Illinois

The materials research laboratory (MRL) is an interdisciplinary research laboratory focused on fundamental issues in materials science. It hosts a large collection of equipments for researchers to use. I, as a graduate student focusing on nanoscale heat transfer in materials, have been using the micro fabrication tools since I started. However, recently many of the tools I use have been down for a prolonged time. This is frustrating especially when I am finishing up my last set of experiments for my PhD defense. So, I decided to scrape the data from MRL instrument schedule and see how much usage for each tool in the past year.

Data type

Let’s first take a look at the webpage.

Schedule

The data stored in this page is in an HTML table. However, the table only gives the text info on the schedule.

library(XML)
library(dplyr)
library(ggplot2)
library(knitr)
library(lubridate)

usage.url <- 'http://cmmserv.mrl.illinois.edu/microfabschedule/guest/npTimeSlots.asp?BW=2016-01-17&GI=0'
usage <- htmlTreeParse(usage.url, error=function(...){}, useInternalNodes = TRUE)
readHTMLTable(usage, as.data.frame=TRUE, stringsAsFactors=FALSE)[[1]]
##   12:00AM Sun Mon Tue Wed -- SHANG Fri Sat
## 1 12:30AM Sun Mon Tue Wed      Thu Fri Sat
## 2 01:00AM Sun Mon Tue Wed      Thu Fri Sat
## 3 01:30AM Sun Mon Tue Wed      Thu Fri Sat
## 4 02:00AM Sun Mon Tue Wed      Thu Fri Sat
## 5 02:30AM Sun Mon Tue Wed      Thu Fri Sat
## 6 03:00AM Sun Mon Tue Wed      Thu Fri Sat
## 7 03:30AM Sun Mon Tue Wed      Thu Fri Sat

To get the time slots that have been booked, I need to go into the file and look for bgcolor attributes using the following code inUse <- length(xpathSApply(usage, "//td[@bgcolor !='#FFFFFF']",xmlValue)). Once I have the time slots with bgcolor != '#FFFFFF', the total booked time of that week can be calculated by \(Number_{nonwhite} \times 0.5 hr\). Now let’s find all the equipments and loop through past year.

Usage in past year

To get the equipment list:

eq.url <- "http://cmmserv.mrl.illinois.edu/microfabschedule/guest/npMonthView.asp?QM=1&QY=2016&GI=0"
eq <- htmlTreeParse(eq.url, error=function(...){}, useInternalNodes = TRUE)
eq <- xpathSApply(eq,"//select[@name ='GI']/option",xmlValue)

eq.df <- data.frame(GI = seq(0, 38, by = 1), equipment = eq)
head(eq.df)
##   GI            equipment
## 1  0         Raith e-Line
## 2  1   AJA Sputter Coater
## 3  2 AJA Sputter Coater 2
## 4  3  E-beam Evaporator 1
## 5  4  E-beam Evaporator 2
## 6  5  E-beam Evaporator 3

To find all the booked time slots as well as down time for each equipment:

dates <- seq(from=as.Date("2015-01-01"), to=as.Date("2016-01-23"), by="day")

dates <- strptime(dates, "%Y-%m-%d")
sunday <- as.character(format(dates[dates$wday==0], '%D'))
usage.df <- data.frame(matrix(ncol = 4, nrow = 0))


for (i in eq.df[,1]){
  
  for (day in sunday){
  usage.url <- paste('http://cmmserv.mrl.illinois.edu/microfabschedule/guest/npTimeSlots.asp?BW=', day, '&GI=', i, sep = '')
  usage <- htmlTreeParse(usage.url, error=function(...){}, useInternalNodes = TRUE)
  inUse <- length(xpathSApply(usage, "//td[@bgcolor !='#FFFFFF']",xmlValue))
  down <- length(c(xpathSApply(usage, "//td[@bgcolor ='#999999']",xmlValue), xpathSApply(usage, "//td[@bgcolor ='#777777']",xmlValue)))
  use.time <- inUse * 0.5
  down.time <- down*0.5
  usage.df<- rbind(usage.df, data.frame(day, as.character(eq.df[i+1, 2]), use.time, down.time, use.time-down.time))
}
}

colnames(usage.df) <- c('startingSunday', 'equipment', 'hours', 'down', 'realHours')

Let’s plot the equipments that have been used over 100 hours in the past year.

usage.by.eq <- usage.df %>%
  group_by(equipment) %>%
  summarise(hours = sum(hours), down = sum(down), realHours = sum(realHours)) %>%
  arrange(realHours)

filtered <- filter(usage.by.eq, realHours>100)

ggplot(data = filtered, aes(x=equipment, y=realHours, fill = realHours)) + geom_bar(stat = 'identity') +
  scale_x_discrete(limits=unique(filtered$equipment)) + coord_flip() +
  ylim(0, 3500) + theme_bw(15) + theme(axis.title.y = element_blank(), legend.position = c(0.8, 0.5))

And the equipments suffered a significant amount of down times.

filtered <- filtered %>%
  filter(down>0) %>%
  arrange(down)

ggplot(data = filtered, aes(x=equipment, y=down, fill = down)) + geom_bar(stat = 'identity') +
  scale_x_discrete(limits=unique(filtered$equipment)) + coord_flip() +
  ylim(0, 3000) + theme_bw(15) + theme(axis.title.y = element_blank(), legend.position = c(0.8, 0.5))

usage.by.month <- usage.df %>%
  mutate(month = month(as.Date(startingSunday, "%m/%d/%y"))) %>%
  group_by(equipment, month) %>%
  summarise(hours = sum(hours), down = sum(down), realHours = sum(realHours)) %>%
  filter(equipment %in% c('Atomic Layer Deposition', 'Raith e-Line', 'E-beam Evaporator 2'))

ggplot(data = usage.by.month, aes(x=month, y=realHours, fill = realHours)) + geom_bar(stat = 'identity') + 
  scale_x_discrete(limits=unique(usage.by.month$month)) +
  facet_wrap(~equipment, ncol = 1) +
  theme_bw(15) + theme(axis.title.y = element_blank())